Handbook of Thermal Analysis and Calorimetry: Recent Advances, Techniques and Applications

RECENT ADVANCES, TECHNIQUES AND APPLICATIONS HANDBOOK OF THERMAL ANALYSIS AND CALORIMETRY SERIES EDITOR PATRICK K. ...

Author: Michael E. Brown | Patrick K. Gallagher

88 downloads 1869 Views 92MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

RECENT ADVANCES, TECHNIQUES AND APPLICATIONS

HANDBOOK OF THERMAL ANALYSIS AND CALORIMETRY

SERIES EDITOR

PATRICK K. GALLAGHER DEPARTMENTS OF CHEMISTRY AND MATERIALS SCIENCE & ENGINEERING OHIO STATE UNIVERSITY USA

ELSEVIER AMSTERDAM – BOSTON – HEIDELBERG – LONDON – NEW YORK – OXFORD PARIS – SAN DIEGO – SAN FRANCISCO – SINGAPORE – SYDNEY – TOKYO

HANDBOOK OF THERMAL ANALYSIS AND CALORIMETRY VOLUME 5

RECENT ADVANCES, TECHNIQUES AND APPLICATIONS EDITED BY

MICHAEL E. BROWN DEPARTMENT OF CHEMISTRY RHODES UNIVERSITY GRAHAMSTOWN 6140 SOUTH AFRICA

PATRICK K. GALLAGHER DEPARTMENTS OF CHEMISTRY AND MATERIALS SCIENCE & ENGINEERING OHIO STATE UNIVERSITY COLUMBUS, OH, 43210 USA

ELSEVIER AMSTERDAM – BOSTON – HEIDELBERG – LONDON – NEW YORK – OXFORD PARIS – SAN DIEGO – SAN FRANCISCO – SINGAPORE – SYDNEY – TOKYO

Elsevier Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands Linacre House, Jordan Hill, Oxford OX2 8DP, UK

First edition 2008 Copyright © 2008 Elsevier B.V. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 978-0-444-53123-0

For information on all Elsevier publications visit our website at books.elsevier.com

Printed and bound in Hungary 07 08 09 10 11

10 9 8 7 6 5 4 3 2 1

FOREWORD The applications and interest in thermal analysis and calorimetry have continued to grow with the arrival of the 21st century. The use of computers and automation of equipment have become commonplace, but the need to characterize the thermal behaviour of materials of all kinds has continued to increase and the increased complexity of some current materials has required that the techniques for their characterization become more imaginative and sophisticated. Successes in the areas of thermal analysis and calorimetry have added to the stature of these subjects and to the numbers of scientists and engineers who have come to rely upon the information provided. The first four volumes of this series provided an unrivalled source of information, prepared by recognized experts, on the basic principles and practice of thermal analysis and calorimetry over the widest interdisciplinary fields. Volume 1 described the basic background information common to the broad subject in general. Thermodynamic and kinetic principles are discussed along with the instrumentation and methodology associated with thermoanalytical and calorimetric techniques. The three subsequent volumes described applications based on general categories of materials. Volume 2 concerns the wide range of inorganic materials, e.g., chemicals, ceramics, metals, etc. Volume 3 describes applications to polymer materials, and biological applications are described in Volume 4. The Editors of each of the four initial volumes chose authors with great care in an effort to produce a readable informative handbook. This practice has continued with the present volume. Again, the chapters are not intended to be comprehensive reviews of the specific subject, but they should enable the reader to glean the essence of the subject and form the basis for further critical reading or actual involvement in the topic. Our goal is to continue to spur your imaginations to recognize the potential application of these methods to your specific goals and efforts. PATRICK K. GALLAGHER Series Editor

PREFACE TO VOLUME 5 This volume contains 19 chapters on recent advances in the techniques and applications of thermal analysis and calorimetry. The intention was to extend and update the original four volumes in this series. Unfortunately, the editors of Volumes 3 and 4 did not find it possible, in the current time frame, to survey, in general terms, the recent advances in the study of polymers and in calorimetry, since publication of their respective Volumes, although many of the present chapters cover some important aspects of these topics. Thus, lurking in the future, is a possible Volume 6 to concentrate more fully on these two areas? Contributions were invited on the basis of the expertise of the authors in each particular area. Some overlap between chapters has been unavoidable, but the differences in points-of-view do add to the interest. Arrangement of the chapters is rather arbitrary, but does emphasize the still vigorous growth of the subject. The contents of each chapter have been designed to make the topic selfcontained, although reference is made, wherever possible, to useful sections of the previous Volumes. As always, in the preparation of such a compilation, there have been model authors who worked entirely to schedule and then had to wait for rather a long time for their "not-so-model" colleagues to produce their chapters. To the former, our sincere apologies, and to all, our thanks for contributing. MICHAEL E. BROWN PATRICK K. GALLAGHER Volume Editors

vii CONTENTS Foreword - P.K. Gallagher Preface - M.E. Brown and P.K. Gallagher Contributors

v vi XX

CHAPTER 1. INTRODUCTION TO RECENT ADVANCES, TECHNIQUES AND APPLICATIONS (Michael E. Brown and Patrick K. Gallagher)

1. THE HANDBOOK OF THERMAL ANALYSIS AND CALOPJMETRY 2. THE LITERATURE OF THERMAL ANALYSIS AND CALORIMETRY 2.1. 2.2. 2.3. 3. 4. 4.1. 4.2. 4.3. 5. 5.1. 5.2. 5.3. 5.4. 5.5. 5.6. 6. 7. 7.1. 7.2. 7.3. 7.4. 7.5. 7.6. 8.

Books Major conferences and their proceedings Websites NOMENCLATURE RECENT ADVANCES IN TECHNIQUES Micro-Thermal Analysis Pulsed thermal analysis Fast scanning calorimetry ADVANCES IN APPLICATIONS Quartz-crystal microbalances Electrical techniques Heating-stage spectroscopy Rheology Catalysis Nanoparticles KINETICS ADDITIONAL TOPICS Thermochemistry Coordination compounds and inorganics Thermophysical properties Polymorphism Medical applications Dental materials QUALITY CONTROL

1 2 2 3 5 6 6 6 7 7 7 7 8 8 8 9 9 9 10 10 10 11 11 11 12 12

viii

CHAPTER 2. DEVELOPMENTS IN NOMENCLATURE (Jean Rouquerol, I. Wadso, T. Lever and P. Haines) 1. INTRODUCTION 13 2. 2006 ICTAC NOMENCLATURE OF THERMAL ANALYSIS 14 2.1. Scope 14 15 2.2. Intent 15 2.3. Definition of the field of Thermal Analysis (TA) 15 2.4. Techniques 16 2.5. Terminology and Glossary 22 2.6. Experimental conditions 22 2.7. Symbols used specifically in Thermal Analysis 23 2.8. Overview and historical matters 24 2.9. Recent Members of the ICTAC Nomenclature Committee 3. COMMENTS ON THE 2006 ICTAC NOMENCLATURE OF THERMAL ANALYSIS 24 4. A CONVENIENT NOMENCLATURE FOR CALORIMETERS 28 28 4.1. Basic representation, criteria and categories 30 4.2. "Passive" adiabatic calorimeters 32 4.3. "Active" adiabatic calorimeters 34 4.4. "Passive" diathermal calorimeters 35 4.5. "Active" diathermal calorimeters 37 5. OTHER POSSIBLE NOMENCLATURES FOR CALORIMETERS 37 5.1. Nomenclature proposed by Swietoslawski in 1933 37 5.2. Nomenclature proposed by Calvet and Prat in 1956 39 5.3. Nomenclature proposed by Evans in 1969 39 5.4. Nomenclature proposed by Skinner in 1969 40 5.5. Nomenclature proposed by Rouquerol and Laffitte in 1972 41 5.6. Nomenclature proposed by Hemminger and Hohne in 1984 44 5.7. Nomenclature proposed by Rouquerol and Zielenkiewicz in 1986 44 5.8. Nomenclature proposed by Tachoire and Medard in 1994 45 5.9. Nomenclature proposed by Wadso in 1997 46 5.10. Nomenclature proposed by Hemminger and Sarge in 1999 47 5.11. Nomenclature proposed by Hansen in 2001 48 5.12. Nomenclature proposed by Matsuo in 2004 50 5.13. Nomenclature proposed by Zielenkiewicz in 2004 51 6. CONCLUSIONS 7. REFERENCES 52-54

IX

CHAPTER 3. MICRO-THERMAL ANALYSIS AND RELATED TECHNIQUES (Duncan M. Price) 1. 2. 2.1. 2.2. 2.3. 2.4. 2.5. 3. 3.1. 3.2. 3.3. 3.4. 3.5. 4. 4.1. 4.2. 4.3. 4.4. 5. 6.

INTRODUCTION SCANNING THERMAL MICROSCOPY (STHM) Introduction Instrumentation for SThM Probe design Quantitative SThM Other SThM techniques LOCALISED THERMAL ANALYSIS Principles Calibration Features Terminology Applications LOCALISED CHEMICAL ANALYSIS Introduction Localised evolved gas analysis Near-field photothermal spectroscopy Thermally-assisted micro-sampling CONCLUSIONS REFERENCES

55 57 57 58 59 61 66 67 67 68 69 71 71 78 78 78 82 83 84 84-92

CHAPTER 4. PULSE THERMAL ANALYSIS (M. Maciejewski and A. Baiker) 1. INTRODUCTION 2. EXPERIMENTAL 3. CALIBRATION OF SPECTROMETRY SIGNALS IN HYPHENATED THERMOANALYTICAL TECHNIQUES 3.1. Calibration of gases 3.2. Verification of the calibration 3.3. Calibration of liquids 4. QUANTIFICATION OF THE SPECTROMETRY SIGNALS IN A TA-MS-FTIR SYSTEM 4.1. Determination of the intrinsic fragmentation in a TA-MS system 4.2. Application of PulseTA® for quantification of gas-solid reactions 5. INJECTION OF A GAS WHICH REACTS WITH THE SOLID 5.1. Investigations of the reduction and oxidation of solids

93 94 95 95 98 99 101 101 104 112 112

5.2. Investigation of the redox behaviour of solids: reduction and re-oxidation of CeCO2 116 5.3. Investigation of gas-solid reactions 118 5.4. Miscellaneous applications 123 6. INJECTION OF A GAS WHICH ADSORBS ON THE SOLID 124 6.1. Adsorption of ammonia on HZMS-5 zeolite 124 6.2. Investigation of the adsorption and desorption of NH3 on a titania-silica aerogel 125 6.3. Investigation of adsorption combined with gas-solid reaction 126 6.4. Miscellaneous applications 129 7. CONCLUSIONS 129 8. REFERENCES 130-132 CHAPTER 5. THE QUARTZ CRYSTAL MICROBALANCE (Allan L. Smith) 1.

HIGH SENSITIVITY BALANCES: THEIR ROLE IN THERMAL ANALYSIS AND CALORIMETRY 2. EARLY HISTORY OF THE QUARTZ CRYSTAL MICROBALANCE 3. THE LITERATURE OF THERMAL ANALYSIS AND OF THE QUARTZ CRYSTAL MICROBALANCE 4. PRINCIPLES OF OPERATION OF THE QUARTZ CRYSTAL MICROBALANCE (QCM) 5. DETECTION ELECTRONICS 5.1. Simple QCM driving circuits 5.2. Frequency and damping measurements 5.3. Impedance analysis 6. IS THE TRANSVERSE SHEAR MODE RESONATOR A TRUE MICROBALANCE? 7.

PRACTICAL DETAILS

7.1. 7.2. 7.3. 8.

Calibration Comparison of gravimetric and Sauerbrey masses Sample Preparation CHEMICAL AND BIOLOGICAL APPLICATIONS OF THE QCM Film-thickness monitors in vacuum deposition The metal/solution interface in electrochemical cells Faraday Society Discussion No. 107, 1997 Determination of shear and loss modulus at QCM frequencies

8.1. 8.2. 8.3. 8.4.

133 134 135 142 147 147 148 148 148 150 150 151 152 152 152 153 154 155

XI

8.5. Chemical sensors and biosensors 8.6. Biological surface science 9. SENSORS 9.1. Acoustic microsensors - the challenge behind microgravimetry 9.2. Piezoelectric sensors 10. THE QUARTZ CRYSTAL MICROBALANCE/HEAT CONDUCTION CALORIMETER 10.1. Introduction 10.2. Beginnings of QCM/HCC 10.3. Development of QCM/HCC 10.4. Biological applications 10.5. The Masscal Scientific Instruments Gl Microbalance/Calorimeter 10.6. Recent applications 10.7. Conclusion 11. REFERENCES

156 158 159 159 159 161 161 161 163 164 164 165 165 166-170

CHAPTER 6. HEATING STAGE SPECTROSCOPY: INFRARED, RAMAN, ENERGY DISPERSIVE X-RAY AND X-RAY PHOTOELECTRON SPECTROSCOPY (Ray L. Frost and J. Theo Kloprogge) 1. INFRARED EMISSION SPECTROSCOPY 1.1. Introduction 1.2. The theory behind infrared emission spectroscopy (IES) 1.3. Infrared emission spectroscopy of alunite 2. HEATING STAGE RAMAN SPECTROSCOPY 2.1 Heating stage Raman spectroscopy of weddellite 3. THERMAL STUDIES OF MATERIALS USING HEATING AND COOLING STAGE SCANNING ELECTRON MICROSCOPY AND ENERGY DISPERSIVE X-RAY ANALYSIS 3.1. Apparatus 3.2. Thermal decomposition of weddelite by heating stage SEM and infrared emission spectroscopy (IES) 3.3. Sublimation of urea CH4N2O 3.4. Wetting/drying of montmorillonite 4. HEATING STAGE PHOTOELECTRON SPECTROSCOPY (XPS) 4.1. Dehydration of calcium oxalate monohydrate CaC 2 0 4 .H 2 0 4.2. Calcination of titania/PVA expanded hectorite 5. CONCLUSIONS

171 171 173 179 182 186

188 188 191 196 198 200 201 202 206

6. ACKNOWLEDGEMENTS 7. REFERENCES

206 206 - 208

CHAPTER 7. ELECTRICAL TECHNIQUES (Madalena Dionisio and Joao F. Mano) 1. INTRODUCTION 209 209 1.1. Dielectric materials in the presence of static electric fields 211 1.2. Application of alternating electric fields 216 2. MEASUREMENT TECHNIQUES 216 2.1. Introduction 217 2.2. Equivalent circuits 219 2.3. Time-domain measurements 220 2.4. Cells 2.5. Temperature calibration in dielectric and electrical measurements 222 3. DIELECTRIC SPECTROSCOPY IN MODEL SYSTEMS AND ASSIGNMENT OF MOLECULAR MOTIONS 224 3.1 Sub-glass mobility 225 231 3.2. a - Relaxation 235 3.3. Crossover region 240 3.4. Low-frequency processes 247 3.5. Dielectric response in semi-crystalline polymers 4. THERMALLY STIMULATED DEPOLARIZATION CURRENTS 253 5. CONCLUSIONS 259 6. REFERENCES 260 - 268 CHAPTER 8. BENEFITS AND POTENTIALS OF HIGH PERFORMS DIFFERENTIAL SCANNING CALORIMETRY (HPer DSC) (Vincent B.F. Mathot, Geert Vanden Poel and Thijs F.J. Pijpers) 1. INTRODUCTION 2. MAJOR CHALLENGES 2.1. Introduction 2.2. Measuring under realistic conditions 2.3. The study of metastability and reorganization 3. HIGH-SPEED CALORIMETRY 3.1. Instrumental aspects 3.2. Temperature calibration 3.3. Constancy of the scan rate

269 270 270 271 271 276 276 277 282

3.4. Linking experiment with practice and processing 3.5. Quantitative measurements 3.6. Higher sensitivity; working on minute amounts of material 4. CONCLUSIONS 5. REFERENCES

284 291 293 295 295 - 298

CHAPTER 9. DYNAMIC PULSE CALORIMETRY - THERMOPHYSICAL PROPERTIES OF SOLID AND LIQUID METALS AND ALLOYS (C. Cagran and G. Pottlacher) 1. INTRODUCTION - THERMOPHYSICAL PROPERTIES 299 2. DYNAMIC PULSE CALORIMETRY (PULSE-HEATING) 301 2.1. Historical development and brief description of pulse-heating 301 2.2. Classification of pulse-heating systems and existing systems 302 3. EXPERIMENTAL DESCRIPTION 304 304 3.1. General information about pulse-heating 308 3.2. Experiment - Basic electrical quantities 310 3.3. Experiment - Derived thermophysical properties 324 3.4. Experiment - Levitation 325 4. EXPERIMENTAL DATA - IRIDIUM 5. RECENTLY DEVELOPED (SPECIAL) APPLICATIONS OF PULSE CALORIMETRY 329 5.1. Extended temperature range by a pulse-calorimeter/DSC combination 329 5.2. Mechanical properties with a Kolsky bar apparatus 330 331 5.3. Pulse-heating/ laser flash combination 332 5.4. Pulse-heating microcalorimetry 333 6. UNCERTAINTIES 333 6. FURTHER READING 334 7. CONCLUSIONS 334 8. ACKNOWLEDGEMENTS 9. REFERENCES 335 - 342 CHAPTER 10. SURFACE PROPERTIES OF NANOPARTICLES (Piotr Staszczuk) 1. 1.1. 1.2. 1.3. 2.

INTRODUCTION Nanotechnology and nanostructures Total (energetic and structural) heterogeneity of surfaces Fractal dimensions of nanoparticles PHYSICOCHEMICAL PROPERTIES OF SELECTED NANOMATERIALS 2.1. Carbon nanotubes

343 343 345 348 349 349

XIV

2.2. 2.3. 2.4. 3. 3.1. 3.2. 3.3. 3.4. 3.5. 4. 4.1. 4.2. 4.3. 4.4. 5. 6.

Montmorillonites 349 Zeolites 350 Superconductor materials 350 TECHNIQUES USED 351 Q-TG thermogravimetry 351 Surface adsorption 356 Porosimetry 356 Calculation of fractal dimensions from sorptometry and porosimetry data 357 Atomic force microscopy, (AFM), Scanning electron microscopy (SEM) and Energy dispersive X-ray spectroscopy (EDX) 358 EXAMPLES OF STUDIES ON SELECTED MATERIALS 359 Carbon nanotubes 359 Montmorillonites 370 Aluminas 371 Fractal dimensions 381 SUMMARY 382 REFERENCES 384 - 386

CHAPTER 11. HETEROGENEOUS CATALYSIS ON SOLIDS (Ljiljana Damjanovic and Aline Auroux) 1. INTRODUCTION 2. EXPERIMENTAL 2.1. Some limitations of the technique for characterizing catalytic sites 2.2. Probe molecules most commonly used to characterize catalytic surfaces 2.3. The role and the influence of the probe molecule in determining adsorption heats 3. ACID-BASE PROPERTIES OF CATALYST SURFACES 3.1. Zeolites and related materials 3.2. Bulk, doped, supported and mixed oxides 4. REDOX PROPERTIES OF CATALYST SURFACES 4.1. Metals and supported metals 4.2. Oxides and supported oxides 5. CORRELATION WITH CATALYTIC ACTIVITY 6. CONCLUSIONS 7. REFERENCES 431-

387 388 394 396 398 401 401 408 421 421 424 426 430 438

XV

CHAPTER 12. COORDINATION COMPOUNDS AND INORGANICS (Stefano Materazzi) 1. 2. 3.

INTRODUCTION REVIEWS USE OF COORDINATION COMPOUNDS AND INORGANICS TO DEVELOP NEW METHODS 4. INORGANICS 4.1. Alloys 4.2. Arsenates 4.3. Borates 4.4. Carbonates 4.5. Chromates 4.6. Iodides 4.7. Nitrates and Nitrites 4.8. Oxalates 4.9. Oxides 4.10. Perchlorates 4.11. Phosphates 4.12. Stannates 4.13. Sulfides, Sulfites and Sulfates 5. METAL-ORGANIC FRAMEWORKS: COORDINATION POLYMERS 5.1. Introduction 5.2. Bismuth 5.3. Cadmium 5.4. Cobalt 5.5. Copper 5.6. Iron 5.7. Lanthanides 5.8. Lead 5.9. Lithium 5.10. Magnesium 5.11. Manganese 5.12. Nickel 5.13. Palladium 5.14. Silver 5.15. Sodium 5.16. Strontium 5.17. Zinc 6. REFERENCES 493

439 440 441 445 445 449 450 451 453 453 454 456 460 463 464 465 466 469 469 469 469 470 472 477 478 482 483 484 485 486 488 488 490 490 491 - 502

XVI

CHAPTER 13. ISOCONVERSIONAL KINETICS (Sergey Vyazovkin) 1. INTRODUCTION 2. ISOCONVERSIONAL METHODS 3. CONCEPT OF VARIABLE ACTIVATION ENERGY 4. KINETICS OF PHYSICAL PROCESSES 4.1. Crystallization 4.2. Melt and glass crystallization of polymers 4.3. Second-order transitions 4.4. Glass transition 5. KINETICS OF CHEMICAL PROCESSES 5.1. Reversible decompositions 5.2. Thermal and thermo-oxidative degradation of polymers 5.3. Crosslinking 6. ISOCONVERSIONAL METHODS AND THE KINETIC TRIPLET 6.1. Is it really needed? 6.2. Isoconversional kinetic predictions 6.3. Evaluating the pre-exponential factor and the reaction model 7. CONCLUSIONS 8. REFERENCES 535 -

503 504 508 512 512 516 518 519 522 522 525 526 529 529 529 532 534 538

CHAPTER 14. THERMOCHEMISTRY (M.V. Roux and M. Temprado) 1. 1.1. 1.2. 2. 2.1. 2.2. 2.3. 2.4. 2.5. 3. 4. 5.

INTRODUCTION The objectives of thermochemistry Short historical introduction EXPERIMENTAL DETERMINATION OF THE ENTHALPIES OF FORMATION OF ORGANIC COMPOUNDS Introduction Combustion calorimetry Reaction calorimetry Thermochemistry of phase changes Additional techniques REFERENCE MATERIALS THERMOCHEMICAL DATA BASES FOR ORGANIC COMPOUNDS RECENT DEVELOPMENTS IN EXPERIMENTAL TECHNIQUES

539 539 541 542 542 542 550 551 554 557 558 559

XV11

5.1. 5.2. 6. 7. 7.1. 7.2. 7.3. 7.4. 7.5. 7.6. 7.7. 7.8. 8. 9.

Combustion calorimetry Enthalpies of sublimation and vaporization COMPUTATIONAL THERMOCHEMISTRY THERMOCHEMISTRY AS A POWERFUL TOOL TO SOLVE ACTUAL CHEMICAL PROBLEMS Thermochemistry of cyclobutadiene: Enthalpy of formation, ring strain, and anti-aromaticity Thermochemistry of cubane and cuneane Enthalpy of formation of Buckminsterfiillerene, Ceo Steric, estereolectronic and electrostatic interactios in oxanes, thianes and sulfone and sulfoxide derivatives Keto-enol tautomerism and enthalpy of mixing between tautomers of acetylacetone Radical generation by using organometallic complexes of Group 6 metals Application to biochemical systems Thermochemistry of reactions in gas phase for compounds with important implications as catalysts. CONCLUSIONS REFERENCES 567 -

559 560 561 562 562 563 563 564 565 566 566 566 567 578

CHAPTER 15. THERMAL ANALYSIS AND RHEOLOGY (Mustafa Versan Kok) 1. INTRODUCTION 2. PARAFFIN WAXES 3. EXPERIMENTAL TECHNIQUES 3.1. Introduction 3.2. Differential scanning calorimetry (DSC) 3.3. Thermomicroscopy and rheology 4. APPLICATIONS 5. CONCLUSIONS 6. REFERENCES

579 580 581 581 582 584 584 595 595 - 596

CHAPTER 16. POLYMORPHISM (Mino R. Caira) 1. INTRODUCTION 2. RECENT DEVELOPMENTS IN POLYMORPHIC RESEARCH 2.1. Introduction

597 599 599

XV111

3. THERMAL ANALYSIS IN STUDIES OF CRYSTAL POLYMORPHISM 3.1. Introduction 4. RECENT STUDIES 4.1. Characterization of polymorphs and polymorphic transformations 4.2. Characterization of solvates and desolvation processes 5. CONCLUSIONS 6. ACKNOWLEDGEMENTS 7. REFERENCES 626 -

603 603 611 611 621 626 626 630

CHAPTER 17. DENTAL MATERIALS (W.A. Brantley) 1. 2. 2.1. 2.2. 2.3. 3. 3.1. 3.2. 3.3. 3.4. 4. 5.

INTRODUCTION NICKEL-TITANIUM ALLOYS IN DENTISTRY Metallurgy background Nickel-titanium endodontic instruments Nickel-titanium orthodontic wires DENTAL POLYMER MATERIALS Silicone maxillofacial materials Elastomeric impression materials Orthodontic elastomeric modules Resin composites and other dental polymers ACKNOWLEDGMENTS REFERENCES

631 631 631 632 641 647 647 650 654 656 658 658 - 662

CHAPTER 18. MEDICAL APPLICATIONS OF THERMAL METHODS (Beverley D. Glass) 1. INTRODUCTION 2. APPLICATION TO PENETRATION OF DRUGS INTO THE SKIN 2.1. Introduction 2.2. Thermoanalytical techniques and the skin 2.3. Thermoanalytical techniques and drug penetration (penetration enhancers) into the skin 3. APPLICATION TO DRUG DELIVERY 3.1. Introduction 3.2. Thermoanalytical techniques used in drug delivery 4. APPLICATION TO IMPLANTS 4.1. Introduction 4.2. Thermoanalytical techniques used in implants

663 664 664 665 668 675 675 675 677 677 677

XIX

5. APPLICATIONS TO PROSTHETICS 5.1. Introduction 5.2. Bioprostheses used in heart valves 5.3. Bioprostheses used in aortic valves 6. MISCELLANEOUS APPLICATIONS 6.1. DSC studies on albumins 6.2. DSC studies on the human intervertebral disc 6.3. DSC studies of human skin from patients with diabetes mellitus (DM) 6.4. DSC studies on cartilage destruction by septic arthritis 6.5. DSC studies on the effect of tetracaine on erythrocyte membranes 6.6. DSC studies on modified poly(urethaneurea) blood sacs 7. CONCLUSIONS 8. REFERENCES 691 -

685 685 685 686 687 687 688 689 689 689 690 690 694

CHAPTER 19. QUALITY CONTROL (Donald J. Burlett) 1. INTRODUCTION 2. GENERAL CONSIDERATIONS 3. POLYMERS 4. ORGANIC CHEMICALS 5. PHARMACEUTICALS 6. FOODS 7. INORGANIC CHEMICALS 8. METALS 9. OTHER REFERENCES 10. FUTURE OPPORTUNITIES 11. REFERENCES

695 696 698 704 709 715 722 724 728 729 729- 732

INDEX

733 - 756

CONTRIBUTORS Auroux, Aline [I11

Baiker, A. [41

Institut de Recherches sur la Catalyse et 1'Environnement de Lyon, UMR 5256 CNRS - UniversitC Lyon 1 , 2 avenue Einstein, 69626 Villeurbanne Cedex, France. e-mail: aline.auroux@,ircel~on.univ-lyonl.fr Institute for Chemical and Bioengineering, Swiss Federal Institute of Technology, ETH Honggerberg, CH-8093 Zurich, Switzerland. e-mail: baiker@,chem.ethz.ch

Brantley, William A.

College of Dentistry, The Ohio State University, 305 West 12th Avenue, Mailbox 191, P.O. Box 182357, Columbus, OH 43218-2357, USA. e-mail: wbrantle@,columbus.rr.com

Brown, Michael E.

Chemistry Dept., Rhodes University, Grahamstown, 6 140 South Africa. e-mail: m.brown@,ru.ac.za

[ll Burlett, Donald [I91 Cagran, Claus [91 Caira, Mino [I61 Damjanovic, Ljiljana [Ill

Gates Corporation, 2975 Waterview Drive, Rochester Hills, Michigan 48309, USA. e-mail: baikalteal13@,netzero.com Institut fir Experimentalphysik, Technische Universitat Graz, Petersgasse 16, A - 80 10 Graz, Austria. e-mail: [email protected] Dept of Chemistry, University of Cape Town, Private Bag, Rondebosch, 770 1 South Africa. e-mail: Mino.Caira@,uct.ac.za Faculty of Physical Chemistry, University of Belgrade, Studentski trg 12-16, 11000 Belgrade, Serbia. e-mail: liiliana@,f,ffh.bg.ac.yu -

xxi

Dionisio, Madalena [71

Frost, Ray L. [61

Gallagher, Patrick K. [ll Glass, Beverley D. [181

REQUIMTEICQFB, Departamento de Quimica, FCT, Universidade Nova de Lisboa, 2829-5 16 Caparica, Portugal. Inorganic Materials Research Program, School of Physical and Chemical Sciences, Queensland University of Technology, G.P.O. Box 2434, Brisbane, Qld 400 1, Australia. e-mail: r.frost@,qut.edu.au 409 South Way Court, Salem, South Carolina 29676-4625, USA. e-mail: p gallagh@,bellsouth.net School of Pharmacy and Molecular Sciences, James Cook University, Townsville, 481 1, Australia. e-mail: beverlev.glass@,~icu.edu.au

Haines, Peter [21

3 8 Oakland Avenue, Farnharn, GU9 9DX, UK.

Kloprogge, Theo

Inorganic Materials Research Program, School of Physical and Chemical Sciences, Queensland University of Technology, G.P.O. Box 2434, Brisbane, Qld 4001, Australia.

Kok, Mustafa V.

Petroleum & Natural Gas Engineering, Department, Middle East Technical University, 0653 1 Ankara, Turkey. e-mail: kok@,metu.edu.tr

Lever, Trevor

Trevor Lever Consulting, 1 Hope Close, Wells, Somerset, BA5 2FH, UK.

[21

xxii

Maciejewski, M.

[41

Mano, J.F.

[71

Materazzi, Stefano

1121 Mathot, Vincent B.F.

[81 Pijpers, Thijs F.J.

181 Pottlacher, Gernot

[91 Price, Duncan M.

[31 Rouquerol, Jean

121

Institute for Chemical and Bioengineering, Swiss Federal Institute of Technology, ETH Honggerberg, CH-8093 Zurich, Switzerland. e-mail: macieiewski@,chem.ethz.ch 3B's Research Group - Biomaterials, Biodegradables and Biomimetics, Department of Polymer Engineering, University of Minho, Campus de Gualtar, 4710-053 Braga, Portugal. e-mail: jmano@,de~.uminho.pt Dept of Chemistry, University "La Sapienza", p.le A.Moro, 5-00 185 Rome, Italy. .it e-mail: stefano.materazzi@,uniromal SciTe B.V., Ridder Vosstraat 6, 6 162 AX Geleen, The Netherlands. e-mail: vincent.mathot@,scite.eu Burg. Savelberglaan 54,6461 GR Kerkrade, The Netherlands. e-mail: thiis.piipers@,tiscali.nl Institut fir Experimentalphysik, Technische Universitat Graz, Petersgasse 16, A - 80 10 Graz, Austria. e-mail: pottlacher@,tugraz.at Exhaust Management Systems, BOC Edwards, Kenn Business Park, Kenn Road, Clevedon, North Somerset, BS2 1 6TH, UK. e-mail: Duncan.Price~,bocedwards.com Madirel, UniversitC de Provence - CNRS, Centre de St JCrBme, 13397 Marseille Cedex 20, France. e-mail : iean.rousuerol@,up.univ-mrs.fi

xxiii

Roux, M. Victoria

[I41 Smith, Allan L.

[51 Staszczuk, P.

[lo1

Temprado, Manuel

[I41 Vanden Poel, Geert

[81 Vyazovkin, Sergey

[I31

Institute of Physical Chemistry "Rocasolano", CSIC, Serrano 119,28006 Madrid, Spain. e-mail: victoriaroux~,iqfr.csic.es Masscal Corporation, 96 A. Leonard Way, Chatham, MA, 02633, USA. e-mail: [email protected] Department of Physicochemistry of Solid Surfaces, Chemistry Faculty, Maria CurieSklodowska University, Maria CurieSklodowska Sq. 3, 20-03 1 Lublin, Poland. e-mail: [email protected] Institute of Physical Chemistry "Rocasolano", CSIC, Serrano 119,28006 Madrid, Spain. e-mail: m.tempradoOiqfr.csic.es DSM Research, P.O. Box 18,6160 MD, Geleen, The Netherlands. e-mail: [email protected] Department of Chemistry, University of Alabama at Birmingham, 901 S. 14th Street, Birmingham, AL 35294, USA. e-mail: vyazovkin(ii,uab.edu Chemical Center, Physical Chemistry 1, Lund University ,Sweden.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter I

INTRODUCTION TO RECENT ADVANCES, TECHNIQUES AND APPLICATIONS OF THERMAL ANALYSIS AND CALORIMETRY Michael E. Browna and Patrick K. ~ a l l a ~ h e r ~ Chemistry Department, Rhodes University, Grahamstown, 6140 South Africa

a

b409 South Way Court, Salem, SC 29676, U.S.A. 1. THE HANDBOOK OF THERMAL ANALYSIS AND CALORIMETRY

Details of this well-established series are given at the following website: h~p:llwww.elsevier.codwpslfind/bookseriesdescription.cws home/BS HATACIdescription The Series Editor is Emeritus Professor Patrick Gallagher. VOLUME 1 - PRINCIPLES AND PRACTICE (Edited by Michael Brown) was published in 1998; VOLUME 2 - APPLICATIONS TO INORGANIC AND MISCELLANEOUS MATERIALS (Edited by Patrick Gallagher and Michael Brown) was published in 2003; VOLUME 3 - APPLICATIONS TO POLYMERS AND PLASTICS (Edited by Stephen Cheng) was published in 2002; VOLUME 4 - FROM MACROMOLECULES TO MAN (Edited by Richard Kemp) was published in 1999. Detailed Contents of these four volumes are given in the preliminary pages of this volume.

In the nine years since the appearance of Volume 1 in 1998, the techniques of thermal analysis and calorimetry have naturally advanced and the applications have broadened even further. Excellent surveys of these developments are the bi-annual reviews in the journal Analytical Chemistry, originally compiled by the late Connie Murphy and extended by the late David Dollimore and colleagues [D. Dollimore and S. Lerdkanchanaporn, Anal. Chem., 70 (12) (1998)27-36; D. Dollimore and P. Phang, Anal. Chem., 72 (12) (2000) 27-36.] and now compiled by Sergey Vyazovkin [S. Vyazovkin, Anal. Chem., 74 (2002) 2749-2762; 76 (2004) 3299-3312; 78 (2006) 3875-3886.1.

2. THE LITERATURE OF THERMAL ANALYSIS AND CALORIMETRY In Volume 1 [p 49-72] of this series, Hemminger and Sarge surveyed the literature available at that time. Since 1998 the major additions to the literature include the following. 2.1. Books

E.A. Turi (Ed.), Thermal Characteristics of Polymeric Materials, 2nd Edn, Academic, San Diego, 1997, Vols. 1 and 2. T. Hatakeyama and F.X. Quinn, Thermal Analysis: Fundamentals and Applications to Polymer Science, 2nd Edn, John Wiley & Sons, Chichester, 1999. A.K. Galwey and M.E. Brown, Thermal Decomposition of Ionic Solids, Elsevier, Amsterdam, 1999. P. Haines (Ed.), Principles of Thermal Analysis and Calorimetry, Royal Society of Chemistry: London, 2002. O.T. Sorensen and J. Rouquerol (Eds), Sample Controlled Thermal Analysis: Origin, Goals, Multiple Forms, Applications and Future, Kluwer, Dordrecht, 2003. M.E. Brown, Introduction to Thermal Analysis, 2nd Edn, Kluwer, Dordrecht, 200 1.

B. Wunderlich, Thermal Analysis of Polymeric Materials, Springer, Berlin, 2005. G.W. Ehrenstein, G. Riedel and P. Trawiel, Thermal Analysis of Plastics: Theory and Practice, Hanser Gardner, Cincinnatti, 2004.

T. Hatakeyama and H. Hatakeyama, Thermal Properties of Green Polymers and Biocomposites, Kluwer, Dordrecht, 2004. D. Lorinczy (Ed.), The Nature of Biological Systems as Revealed by Thermal Methods, Kluwer, Dordrecht, 2004. M. Sorai (Ed.), Comprehensive Handbook of Calorimetry and Thermal Analysis, John Wiley, Hoboken, NJ, 2004. B.V. L'vov, Thermal Decomposition of Solid and Liquid Substances,St. Petersburg Polytechnic University, Publisher, 2006, in Russian. English edition in preparation to be published by Springer. 2.2. Major conferences and their proceedings

2.2.1. International 1998 7th European Symposium on Thermal Analysis and Calorimetry (ESTAC), Balaton-Fiired, Hungary. [J. Krist6f and C. Novhk (Eds), J. Therm. Anal. Cal., 56 (1999) 1-1479.1 2000 12th International Congress on Thermal Analysis and Calorimetry (ICTAC), Copenhagen, Denmark. [O. Toft Scirensen and P. Juul Mciller (Eds), J. Therm. Anal. Cal., 64 (2001) 1-1339.1 2000 2nd International Symposium on Calorimetry and Thermal Effects in Catalysis, Lyon, France [Thermochim.Acta, 379 (2001) 1-2781. 2002 8th European Symposium on Thermal Analysis and Calorimetry (ESTAC), Barcelona, Spain [R. Nomen and J. Sempere (Eds), J. Thermal Anal., 72 (2003) 1-11811.

2004 13th ICTAC, Sardinia. [A. Schiraldi and B. Marongiu, (Eds), J. Them. Anal. Cal., 80 (2005) 1801.1 2004 13th Conference of the International Society for Biological Calorimetry, Wiirzburg-Veitshijchheim, Germany [Thermochim. Acta, 422 (2004) 1-130.1 2004 3rd International Symposium on Calorimetry and Thermal Effects in Catalysis, Lyon, France [Thermochim. Acta, 434 (2001) 1-1821. 2006 9th ESTAC in Krakow, Poland. Forthcoming conferences in this series are: 2008 14th ICTAC in Brazil. 2.2.2. The annual conferences of the North American Thermal Analysis Society (NATAS)

1998 26th NATAS conference, Cleveland, Ohio [Thermochim. Acta, 3571358 (2000) 1-334.1 1999 27th NATAS conference, Savannah, Georgia (This conference was cancelled due to a hurricane, but the proceedings, dedicated to David Dollimore, were published in Thermochim. Acta, 367 (200 1) 1-455.) 2000 28th NATAS conference, Orlando, Florida 2001 29th NATAS conference, St. Louis, Missouri [Thermochim. Acta, 396 (2003) 1-2341. 2002 30th NATAS conference, Pittsburgh, Pennsylvania 2003 3 1st NATAS conference, Albuquerque, New Mexico 2004 32nd NATAS conference, Williamsburg, Virginia 2005 33rd NATAS conference, Universal City, California 2006 34th NATAS conference ,Bowling Green, Kentucky Other regional and specialist group conferences have also been held. Medicta 2003, Porto, Portugal [Thermochim. Acta, 420 (2004) 1-1801.

2.3. Websites The offical website of the International Confederation for Thermal Analysis and Calorimetry (ICTAC) is: www.ictac.org

Check out t h e I C T A C Web site @

...

Designed t o meet the ever-increasing needs of ICTAC members, this site will hopefully become your first port of call for all the latest information from t h e world of thermal anolysis and calorimetry! ICTAC News Contact information for ICTAC Courcil, Committees & Affiliated 6twps Membership Information Upcoming Conferences and Meetings The ICTAC Statutes Thermal Analysis Links Advertising Information But it doesn't end there! We want this site t o become t h e if 1resource f o r thermal analysts around the world ... and that's where we need your help!

I f you have any ideas or suggestions f o r what you would like t o see on t h e ICTAC web site then please contact Don Burlett,Stuart du Kamp -1'. And when you visit the site, don't forget t o sign up f o r the ListBot! Join our mailing list and automatically receive all the latest information from ICTAC.

Go check us out! The ICTAC m b site was developed and is tnnintaincd by hddpwebdesign.com. I f you have any problems with accessing the site please contact Pam du Kamp a t [email protected].

The 40th Anniversary Issue (1965-2005) of ICTAC News contains a wealth of information on the history of ICTAC and its Affiliated Societies. The THERMAL list-server: http://~~~.egr.msu.edu/mailman/listinfo/thermal initiated by Michael Rich at Michigan State University, provides a forum of experts willing to advise on thermoanalytical problems.

3. NOMENCLATURE The nomenclature of thermal analysis continues to be a controversial and incompletely resolved topic. The system proposed by the ICTAC Nomenclature Committee chaired by W. Hemminger and described in detail by Hemminger and Sarge in Volume 1 [Vol 1, p 1-31] has since been reconsidered by a further Committee and the latest recommendations, which have been approved by the Council of ICTAC and by ASTM, are presented by Rouquerol et al. in Chapter 2 of this Volume. The current Committee has followed the advice of the late Robert Mackenzie in that: - terminology should be simple; - abbreviations kept to a minimum; - names based on particular instruments should be discouraged.

Definition of the field of Thermal Analysis (TA) "Thermal Analysis (TA) is the study of the relationship between a sample property and its temperature as the sample is heated or cooled in a controlled manner". The nomenclature of Calorimetry is also dealt with in Chapter 2. The basic criterion that the authors use to distinguish different types of calorimeters is on the basis of the heat exchanges between the system and the thermostat. Two broad families result, namely: "Adiabatic" calorimeters, where the aim is to avoid any exchange of heat between the system S and the surrounding thermostat T "Diathermal" calorimeters, where the aim is, on the contrary, to favour the above heat exchange 4. RECENT ADVANCES IN TECHNIQUES Amongst the important advances in thermal analysis and calorimetric techniques have been the following. 4.1. Micro-Thermal Analysis The introduction and development of Micro-Thermal Analysis are described and discussed by Duncan Price in Chapter 3. The atomic force microscope (AFM) forms the basis of both scanning thermal microscopy (SThM) and instruments for performing localised thermal analysis. The principles and operation of these techniques, which exploit the abilities of a thermal probe to act both as a very small heater and as a thermometer, in the surface characterisation of materials are described in detail. The

coupling of thermal probes with methods of localised evolved gas analysis by pyrolysis and/or near-field photothermal infrared microscopy is also described. 4.2. Pulsed thermal analysis The technique, named rather ambiguously as pulsed thermal analysis, is described by Maciejewski and Baiker in Chapter 4. Pulse thermal analysis ( ~ u l s e T ~is@based ) on the injection of specific amounts of gases or liquids into the inert carrier gas stream of a thermobalance or differential scanning calorimeter and subsequent monitoring of the changes in the mass andlor enthalpy of the sample and composition of the evolved gases resulting from the incremental amount of reaction between the sample and the injected gas. The method is also particularly suitable, using a gas which does not react with the sample, for quantification of evolved gas analysis by mass spectrometry (MS) or Fourier transform infrared spectroscopy (FTIR). All types of gas-solid interactions can be studied by injecting gases that adsorb on, or react with, the sample. PUIS~TA@ applications are illustrated using examples from various fields such as catalysis and materials science. 4.3. Fast scanning calorimetry Most calorimeters have problems in achieving high controlled, constant rates on heating and, even more so, during cooling. In Chapter 8. Mathot, Vanden Poel and Pijpers describe and discuss means for producing controlled, fast cooling and heating by way of high-speed calorimetry. The use of a commercially available version of high-speed calorimetry: High Performance DSC (HPer DSC) is described to provide insight into the commonly occurring metastable states of substances, such as polymers and pharmaceuticals, which result fiom their thermal treatment and which can drastically change their end properties. Phenomena related to metastability include supercooling, amorphization, 'hot' crystallization (from the melt), cold crystallization (from the glass state), recrystallization (after melting), annealing, etc., are discussed with examples and the many advantages of HPer DSC are described.

5. ADVANCES IN APPLICATIONS

5.1. Quartz-crystal microbalances Although quartz-crystal microbalances are not a new invention, they have not been used extensively in thermoanalytical investigations. In Chapter 5, Alan Smith provides a historical background to quartz-crystal microbalances and describes the principles of their operation, before

surveying their chemical and biological applications in thermal analysis and calorimetry. He then discusses quartz crystal microbalance/heat conduction calorimetry, (QCM/HCC), which is a new measurement technology that permits high-sensitivity measurements, in real- time, of the mass changes, the heat generated and the change in loss compliance of nanoscale coatings or films undergoing chemical reaction. The high sensitivities enable the energetics of the formation of a self-assembled monolayer, as well as the thermodynamics of the chemical processes in nanoscale polymer coatings, to be studied. 5.2. Electrical techniques In Chapter 7, Mano and Dionisio describe how electrical methods, and particularly dielectric relaxation spectroscopy (DRS) and thermally stimulated depolarisation current (TSDS) techniques, play a major role as tools for exploring molecular mobility. DRS enables molecular relaxational processes (both slow and fast) to be studied. For example, the localized motions of glass formers in the glassy state give rise to local fluctuations of the dipole vector that are the origin of the secondary relaxation processes detected by dielectric relaxation spectroscopy, while above, but near, the glass transition, cooperative motions result in a distinguishably different relaxation process (the a-relaxation). DRS is also valuable for studying the translational motion of charge carriers. These effects are important in inhomogeneous materials such as biological systems, emulsions and colloids, porous media, composite polymers, blends, crystalline and liquid crystalline polymers and electrets. The results of DRS may be complemented by TSDC studies, which provide a way of probing the mobility of dipoles and electric charges over a wide temperature range. 5.3. Heating-stage spectroscopy In Chapter 6, Frost and Kloprogge, describe the use of various heatingstages coupled with infrared, Raman, energy dispersive X-ray, or X-ray photon spectroscopy. Such techniques strongly complement the results obtained from traditional thermoanalytical techniques and enable the changes in the composition and molecular structure of minerals and materials to be obtained in situ at elevated temperatures.

5.4. Rheology In Chapter 15, Kok describes how thermal analysis techniques can play a key role in examining the rheological behaviour of crude oils. For crude oil samples with different contents of wax, differential scanning calorimetry, thermomicroscopy and rheometry provide excellent

methods of measuring the wax appearance temperatures. These techniques, together with viscometry, can be used to study the flow properties of crude oils below the wax appearance temperature and the effects of the addition of flow improvers.

5.5. Catalysis Techniques such as TG, DTA or DSC enable the thermal behaviour of a catalyst to be examined as it undergoes heating at a constant rate (or by steps of temperature) and are particularly adapted for studying the decomposition of catalyst precursors or for desorption studies involving poisoned catalysts. Damjanovic and Auroux in Chapter 11 describe how, by the use of adsorption microcalorimetry, where the sample is kept at a constant temperature while a probe molecule adsorbs onto its surface, values of the enthalpy of adsorption can be determined and valuable insights into the mechanism of adsorption can be obtained. Very sensitive flow microcalorimeters provide a means of measuring the energy of adsorption as a function of coverage, thus revealing details of the surface heterogeneity and information on the strength and distribution of the different sites on the surface of a catalyst. The chapter gives some illustrations of applications in different fields of catalysis, focusing mainly on redox and acid-base catalysis. 5.6. Nanoparticles Nanotechnology covers the search for and synthesis of new materials of advanced technology which possess the sizes of nanometres; the determination of their characteristics, and their practical application. This broad and fast-developing topic is tackled by Staszczuk in Chapter 10. Typical nanomaterials include soil mineral components, adsorbents, silica gels, latexes, synthetic zeolites modified by ions, molecular sieves, fullerenes, carbon nanotubes, active carbons, semiconductors, hightemperature superconductors, modified zeolites and adsorbents with deposited proteins, etc. Studies of the surface properties of some of these materials, including determination of their fractal dimensions, using mainly quasi-isothermal (Q-TG) thermogravimetry, are described. 6. KINETICS

The techniques of thermal analysis stimulated an interest in the estimation of reaction kinetic parameters from programmed temperature experiments. Some of this background was covered by Brown and Galwey in Chapter 3 of Volume 1 [Vol.l, Ch.31. A major advance was

the ICTAC Kinetics Project, reported on in Thermochimica Acta [ TCA ref]. Participants in the project were supplied with sets of numerical data on which to test their favourite methods of kinetic analysis. In Chapter 13 of this Volume, Sergey Vyazovkin focuses on the development and application of isoconversional methods of kinetic analysis. This approach has made a major impact on the field of kinetics and allows the kinetic model (or conversion function) to be temporarily eliminated from the analysis of sets of data obtained at different heating rates and hence provides apparent values of the activation energy and the pre-exponential factor. The way in which the activation energy appears to vary with extent of reaction is then a good guide to the complexity of the reaction model. This sort of approach has been adopted by several major manufacturers of thermal equipment. These isoconversional methods (often unwisely referred to as "model-free") have been applied with great success to a wide variety of processes including polymerization, copolymerization (curing), crystallization, and relaxation. 7. ADDITIONAL TOPICS A preliminary survey of the Contents of Volumes 1 to 4 (see the Elsevier website) revealed that some topics had, perhaps, not received the attention which they deserved. In an attempt to rectify this, the following contributions have been provided. 7.1. Thermochemistry In Chapter 14, Roux and Temprado have provided a detailed survey of the many facets of thermochemistry, including the history of the subject, methods of measurement of thermodynamic quantities, calibration of instruments, estimation of accuracy and the application of correction factors. Reference materials are discussed and data bases of thermodynamic properties are described, including an introduction to computational thermochemistry. The chapter concludes with some examples of the solving of thermochemical problems. 7.2. Coordination compounds and inorganics Chapter 12 by Materazzi is a survey of the thermal analysis literature on coordination compounds and inorganics published in the years 2000-2006 and is limited to the most recent representative publications. Coordination compounds and inorganics are extensively studied for a variety of reasons, including their use as simplified models for understanding the behaviour of the more complex molecules that are involved in biological reactions or that are of biomedical interest.

The conventional thermoanalytical techniques (thermogravimetry TG, differential scanning calorimetry DSC, differential thermal analysis DTA, etc.) can provide fundamental data concerning the thermal behaviour of these substances, but the addition of mass spectrometry (TG-MS) or Fourier transform infrared spectroscopy (TG-FTIR) has permitted the identification of gaseous species evolved during thermal processes. 7.3. Thermophysical properties "Thermophysical properties' are defined, in Chapter 9 by Cagran and Pottlacher, as a selection of mechanical, electrical, optical, and thermal material properties of metals and alloys (and their temperature dependencies) that are relevant to industrial, scientific, and metallurgical applications, and this covers a wide range of different material properties obtained by numerous different measurement techniques. The focus in Chapter 9 is, however, on thermophysical properties that are accessible through dynamic pulse calorimetry Other non-calorimetric techniques have been developed but, with the exception of levitation (needed to measure technologically important properties like viscosity and surface tension) have been excluded from consideration.

Three Chapters complement the outstanding review of the thermal analysis and calorimetry of pharmaceuticals by Ford and Willson in Chapter 17 of Volume 4.

7.4. Polymorphism The identification, structural and thermal characterization of new polymorphs is an important topic in solid-state chemistry and requires a battery of techniques that includes X-ray diffraction and spectroscopic methods, in addition to thermal analysis methods and dissolution techniques to determine solubility trends. Such studies are described by Caira in Chapter 16, as well as more recent theoretical techniques aimed at the prediction of the crystal structures of new polymorphs. Crystal polymorphism is particularly important in pharmaceutical products, so there is an emphasis on this area. Systems displaying solvatomorphism (the ability of a substance to exist in two or more crystalline phases arising from differences in their solvation states) molecular inclusion and isostructurality (the inverse of polymorphism) are also given due attention in this chapter. 7.5. Medical applications In Chapter 18, Glass attributes the recent increased use of thermal methods in tackling medical problems to the improved sensitivity and usability of the instrumentation, especially differential scanning

calorimetry (DSC). Investigations of the thermal properties of human and animal skin have aided understanding of drug penetration of the skin and methods of drug delivery. Isothermal titration calorimetry (ITC) has proved useful in the evaluation of thermal stability and to measure the heat generated in physical and chemical reactions. Knowledge of the effects of alteration of biological systems is not only important in medical research but also in advanced patient-care. The applications of thermoanalytical techniques to determining the thermal behaviour of prosthetics and implants are also discussed. 7.6. Dental materials Thermal analysis techniques have had only limited use in the study of the various dental materials used for restorative, prosthetic and implant applications. The innovative research by Brantley's group, using conventional and temperature-modulated DSC (TMDSC) to examine the thermal behaviour of several metallic and polymeric dental materials, is described in Chapter 17 and numerous matters requiring additional research are identified.

8. QUALITY CONTROL A very important contribution to the Volume, of interest to all practitioners, is the review, by Burlett in Chapter 19, of the uses of thermal analysis and calorimetry in quality control. During the production of materials for use by consumers, the nature of the starting materials and their responses to processing and storage conditions must be fully understood. The products must then meet certain criteria to satisfy the user. Quality control (QC), or quality assurance (QA), is the name of the operation conducted to assure that the nature of a material is within certain specifications that are critical for subsequent processing or use. Legal and financial considerations are an important part of the quality control operation. Accreditation systems provide standards of operation covering the quality that must be considered. All these aspects are discussed fully, with examples, and this chapter thus forms a very fitting conclusion to the Volume.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 2

DEVELOPMENTS IN NOMENCLATURE Jean ~ o u ~ u e r o lI.' ,wadso', T.J.

ever^ and P.J. ~ a i n e s ~

1

Madirel, UniversitC de Provence-CNRS, Centre de St JCr6me 13397 Marseille Cedex 20, France 2 Chemical Center, Physical Chemistry 1, Lund University , Sweden Trevor Lever Consulting, 1 Hope Close, Wells, Somerset, BA5 2FH, UK 4 38 Oakland Avenue, Farnham, GU9 9DX, UK 1. INTRODUCTION The development of a clear, extensive and well-accepted nomenclature can be considered as a sign of good health for any field of science. Scientists need to understand each other precisely and to draw the best from the available literature. Unfortunately, nomenclature can also be considered as a "final product" which only becomes fully available when the science is mature and when the scientists, after a long period of discussion, finally agree. This is what has happened with calorimetry, in spite of its early start at the end of the 18th century (when Lavoisier coined the term "calorimeter" [I]). For this reason, it is not possible to provide a single system of nomenclature for calorimetry. In this chapter, after a general introduction to the field of calorimetry, a relatively simple and pedagogical system of nomenclature is provided, and is followed by descriptions of a number of other nomenclature systems used for calorimeters during the past decades. The interest and objective of each nomenclature system is stressed, so as to favour its use in the most suitable context. By contrast, thermal analysis, which started to be recognized only one century later, developed a broadly accepted system of nomenclature more quickly. This was the result of collective work, conducted in the scope of the International Confederation of Thermal Analysis (ICTA), under the guidance of Robert Mackenzie from 1968 onwards. Once adopted by the Council of ICTA, this nomenclature was published in various papers [2-81 and also in booklets edited by ICTA [9- 101. Most importantly, it was endorsed by IUPAC [l l-121 and, also, used as a reference by ASTM. This thermal analysis nomenclature, which was progressively built up from the existing techniques, has needed to be updated to include more recent techniques

and to offer a more definitive and broader frame, able to embrace techniques which still do not exist to-day. This work, which started 15 years ago in the scope of ICTAC (with the second "C" indicating that it then included calorimetry in its fields of interest), took more time than expected. This was due to the need to introduce changes, to adjust them at an "acceptable" level for the scientific community and to leave time for a broad, international, exchange of views, including scientists from non English-speaking countries. Such a duration is not abnormal and, in some respect, is even desirable: the role of an international nomenclature committee is to represent, or at least to attentively examine and take into account, if possible, the many viewpoints of scientists. Time spent on collecting these viewpoints, in evaluating them, and in building a consistent ensemble is, in reality, time saved at the end of the process, because this will make the acceptance by the scientific community easier and broader. Under these conditions, the proposed nomenclature becomes a tool really useful for a majority of people, which is the real objective of such an endeavour. Such an updated system of nomenclature for thermal analysis was finally adopted by the ICTAC Council in August 2006. The corresponding document was prepared in its final form by Trevor Lever and Peter Haines (the Chairman and Secretary, respectively, of the ICTAC Nomenclature Committee). Those who were in charge of the Committee during the years during which the maturation and adaptation of the project took place, include the former Chairman, Wolfgang Hemminger, and the former Secretaries Fred Wilburn, Stefan Sarge and Duncan Price. The h l l list of members of the Committee for the period 1992-2006 can be found at the end of Section 2. The present official nomenclature for Thermal Analysis is thus contained in the Report of the ICTAC Nomenclature Committee, which is reproduced (with the permission of ICTAC) in Section 2. In Section 3 a number of extra comments are provided in order to justify, explain or complement the material in Section 2. After that, the nomenclature in Calorimetry is examined and a detailed nomenclature system, devised for pedagogical purposes, is presented in Section 4, which serves as an introduction to other systems described in Section 5. Section 6 concludes the chapter with a few general comments about these two systems of nomenclature.

2. 2006 ICTAC NOMENCLATURE OF THERMAL ANALYSIS 2.1. Scope The scope of this document is to provide scientists working in the field of thermal analysis with a consistent "definitions of terms" that are commonly used within the field to allow precise communication and understanding. Since the scope of ICTAC also covers calorimetry, a further document dealing with this

latter part of nomenclature is planned, once a satisfactory international consensus is reached on this matter. In considering all the matters of nomenclature, the current Committee has followed the advice of the late Robert Mackenzie in that: - terminology should be simple; - abbreviations kept to a minimum; - names based on particular instruments should be discouraged. 2.2. Intent This document acknowledges that nomenclature develops - without regulated definition - as the field of thermal analysis develops. Some terms used by authors and scientists rapidly become accepted by the scientific community, even if the term is not consistent with past definitions, science or grammatically correct. However, if such a term is widely used and understood, it is reported here. 2.3. Definition of the field of Thermal Analysis (TA) "Thermal Analysis (TA) is the study of the relationship between a sample property and its temperature as the sample is heated or cooled in a controlled manner".

2.4. Techniques A technique exists for each property or physical quantity that is measured versus temperature - a summary of some of these are presented below.

temperature difference between a

Deformations are measured. Thermodilatometry

TD Dimensions are measured

Electrical Properties

Dielectric Thermal Analysis

DEA

Thermally Stimulated Current

TSC

Dielectric Constantt Dielectric Loss measured Current Often combined with TGA

Magnetic Properties

Thermomagnetometry

Gas flow

Evolved Gas Analysis

EGA

The nature andlor amount of gas I vapour is determined.

Emanation Thermal Analysis

ETA

Trapped radioactive gas within the sample is released and measured. Evolution of gas is detected by pressure change.

Pressure

Thermomanometry Thermobaromehy

Optical Properties

Thermoptometry

Acoustic Properties

Thermoluminescence Thermosonimetry or Thermoacoustimetry

Structure

Thermodiffractomehy Thermospectrometry

Pressure exerted by a dense sample on the walls of a constant volume cell is studied. A family oftechniques in which an optical characteristic or property of a sample is studied TL

Emitted light measured Techniques where the sound emitted (sonimetry) or absorbed (acoustimetry) by the sample is studied Techniques where the compositional or chemical nature of the sample are studied

2.5. Terminology and Glossary NOTE: For all the techniques listed here, the terminology defines the property that is measured, and each definition may be completed by adding... "as a function of temperature". For example: dynamic mechanical analysis (DMA), n- a technique where moduli are determined as a function of temperature. adiabatic, adj- indicating that the experiment is carried out so that no heat enters or leaves the system. atmosphere, n, - the gaseous environment of the sample, which may be controlled by the instrumentation or generated by the sample. calorimetry, n- techniques where temperature.

heat is measured

as a

function of

combined, adj - the application of two or more techniques to different samples at the same time. This can include thermal and non-thermal analytical techniques. controlled-rate thermal analysis (CRTA), n- a sample-controlled method where the heating is exclusively controlled by the rate of transformation controlled temperature program, n - the temperature history imposed on the sample during the course of the thermal analysis experiment. cooling curve, n - the experimental result of measuring the temperature of the sample as a function of time during cooling. The technique is thermometry, and heating curves are obtained for temperature-time experiments during heating. derivative, adj - pertaining to the lStderivative (mathematical) of any curve with respect to temperature or time. dielectric thermal analysis (DEA), n- a technique where dielectric properties are measured. differential, adj - pertaining to a difference in measured or measurable quantities usually between a sample and a reference or standard material. differential scanning calorimetery (heat-flow DSC), n - technique where the heat flow rate difference into a sample and a reference material is measured. differential scanning calorimetery (power compensation DSC), n - technique where the electrical power difference into a sample and a reference material is measured. differential thermal analysis (DTA), n - a technique where the temperature difference between a sample and a reference material is measured. dynamic, adj- a prefix indicating that a parameter changes continuously during the experiment. The opposite of static. dynamic mechanical analysis (DMA), ndetermined.

a technique where moduli are

emanation thermal analysis (ETA), n- a special type of EGA where the emanation of previously trapped radioactive gas is measured. evolved gas analysis (EGA), n- a family of techniques where the nature and/or amount of gas or vapour evolved is determined. The term evolved gas detection (EGD) has also been used where the nature of gas is not determined. gas flow, n - the passage of gas from one part of the system to another, either by sorption by the sample, evolution from it, or chemical reaction. high pressure, (HP.. .), adj - a prefix applied to the technique name to indicate that the pressure of the experiment is above ambient. Note: As an example a TGA experiment carried out under elevated pressures would be High Pressure Thermogravimetric Analysis (HPTGA) isobaric, adj- a prefix indicating the experiment is carried out at constant pressure. isothermal, adj, - a prefix applied to a technique to indicated that the temperature is maintained constant throughout the experiment material, n-, the substance which is studied and from which the sample is taken.. micro-, adj,- a prefix used to denote that the technique measures small quantities, either with respect to the amount of sample studied, or with respect to the change in the properties measured. Note: This prefix has been applied to many thermal methods, and the equipment associated with them, for example micro-balance, microreactor, micro-calorimeter and also to the technique itself : microthermal, microscopic and the property studied: micro-structural. Note: The opposite prefix, macro- is also occasionally used. modulated, adj- a prefix indicating that a parameter changes in a periodic manner during the experiment.

modulated temperature, (MT.. .), adj, - a prefix applied to the technique name to indicate that a temperature modulation has been applied to the temperature program. Note 1: As an example a DSC experiment carried out with a modulated temperature program would be Modulated Temperature Differential Scanning Calorimetry (MT-DSC) Note 2: Other modulated techniques are possible, such as modulated force TMA ,modulated rate SCTA etc. Note 3: The use of the prefix MT is preferred to TM. photo-, adj, - prefix to indicate that the experiment involves the illumination of the sample or measures the amount of light emitted from a sample. Where possible the wavelength range of the light should be specified. sample, n- the material under study during the entire experiment (starting material, intermediates and final products) and its close atmosphere. This is equivalent to the thermodynamic system. sample-controlled, adj - prefix applied to the technique to indicate that a property of the sample is used either continuously or discontinuously to control the sample heating. With no prefix, it is assumed that the experiment is following a controlled-temperature program. Note: the generic term for all TA techniques making use of such a feedback is Sample-Controlled TA (SCTA), whereas specific names will be of the form Sample-Controlled TGA (SC-TGA) etc. scan, n- a term used to describe the data produced from a thermal analysis experiment. More correct usage is a thermal analysis curve, or, for a specific technique thermogravimetric curve, etc. scanning, adj, - a prefix indicating a specified experimental parameter, usually temperature, is changed in a controlled manner.

simultaneous, adj - the measurement of two or more properties of a single sample at the same time. Note: A hyphen is used to separate the abbreviations of the techniques; for example, simultaneous measurement of mass and heat flow rate (thermogravimetric analysis and differential scanning calorimetry) would be TGA-DSC. static, adj- indicating a constant parameter during the experiment. The opposite of dynamic stepwise, adj - prefix indicating discrete, discontinuous changes in an experimental parameter, e.g. force, temperature etc. tan 6, n - is the dimensionless ratio of energy lost to energy returned during one cycle of a periodic process. For example tan 6 = E" I E', in DMA. temperature-programmed desorption (TPD), n - EGA using an inert atmosphere or vacuum, in the absence of sample decomposition. temperature-programmed oxidation (TPO), n - Experiment using an oxidising atmosphere, usually oxygen. Oxidation is monitored by any appropriate technique (EGA, TGA, gas sorption, etc.). temperature-programmed reduction (TPR), n- Experiment using a reducing atmosphere, usually hydrogen. Reduction is monitored by any appropriate technique (EGA, TGA, gas sorption, etc.). thermal curve, n - any graph of any combination of property, time, temperature derived from a thermal analysis technique. Note: thermal curve is a loose abbreviation of the more correct term thermoanalytical curve thermally stimulated current (TSC), n- a technique where the current from the relaxation of sample polarisation is measured. thermo-, adj- a prefix indicating the use of changing temperature during the experiment.

thermoacoustimetry, n- a technique where the characteristics of sound waves passing through the sample are measured. thermoanalytical, adj - of, or pertaining to, thermal analysis. thermodiffractometry, nsample is measured.

a technique where the X-ray diffraction of the

thermodilatometry,(TD), n- a technique where one or more dimensions of the sample is measured under negligible load thermogravimetric analysis, (TGA), n- a technique where the mass of the sample is measured. thermogravimetry, (TG), n- see thermogravimetric analysis, which is to be preferred thermoluminescence, n- a technique where light emission from the sample is measured. thermomagnetometry, n- a technique where a magnetic property of the sample is measured. thermomanometry, n- a technique where the pressure is measured. thermomechanical analysis, (TMA), n- a technique where the deformation of the sample is measured under constant load. thermometry, n- a technique where the temperature of the sample is measured. thermomicroscopy, n- a technique where the optical properties of the sample are observed and measured through a microscope. thermoptometry, n- a technique where the optical properties of a sample are measured. thermosonimetry, n- a technique where the sound emitted by the sample is measured. thermospectrometry, n- a group of techniques where a spectrum of the sample is measured.

torsional braid analysis (TBA), n- a dynamic mechanical analysis technique where the sample is supported on a braid.

2.6. Experimental conditions The specifics of how the technique is used, additional experimental parameters and constraints should, of course, be reported alongside the data in all published work. It is important to separate the technique (instrumentation) from the way in which it is used (experiment). The make and model number should be included in all reports, papers and studies as well as an experimental section that describes in full all experimental parameters. Note: For example in a Thermomechanical Analysis (TMA) experiment the sample may be subjected to no force, a constant force, an increasing force or a modulated force - or any combination of the above - during a single experiment. The technique (TMA) has not changed, only the experimental variables for that technique. The reader is referred to the ICTA publication [9] specific guidelines. It must also be stressed that it should be normal practice to use the standard IUPAC quantities, units and symbols when reporting any work in thermal analysis. These are listed, more especially, in the IUPAC "Green Book" [13].

2.7. Symbols used specifically in Thermal Analysis 2.7.1.Symbols for quantities and units

2.7.2. Symbols describing speciJic events or materials In general, symbols for physical quantities should be in italic type, or, if vectors, in bold italic type. The symbols for units do not take plural. Subscripts should generally be restricted to single letters. If the subscript relates to an object or property, it should be a CAPITAL letter: ms = mass of sample S. TR = temperature of reference R. Tc = Curie temperature If the subscript refers to a phenomenon, it should be lower case: Tfus = melting temperature T, = glass transition temperature. If the subscript refers to a specific event, time or point, it should be lower case or figures: Ti = initial temperature rnf = final mass T, = peak temperature t1,2 = time of half reaction Changes in extensive thermodynamic quantities X due to an event y should be represented by A& : AVa,$l = enthalpy of vaporization A,G = Gibbs energy of reaction. Symbols for the physical state of the material should be put in brackets after the formula symbol: AvapH = H(g) - H(1) 2.8. Overview and historical matters This document is concerned with providing definitions of common terms that are used by thermal analysts to report, present and explain their work. The ICTAC Nomenclature Committee was initiated in 1965 under the guidance of Robert Mackenzie and with the secretarial expertise of Cyril Keattch. This document acknowledges the debt to previous members of the Committee under their succeeding Chairmen, including John Sharp (1984-8), Ed Gimzewski (1988- 1992) and Wolfgang Hemminger (1992-200 1) who continued the discussions and published their findings as listed in the references. The task of the current committee has been to rationalise the work of all proceeding committees and to deliver a document that covers current practice in thermal analysis that can be accepted internationally.

Thanks are due to recent members of the Nomenclature Committee listed below for their contributions to the deliberations and to others for the advice received: 2.9. Recent Members of the ICTAC Nomenclature Committee Roger Blaine (200 1-6) Don Burlett (200 1-6) Edward Charsley (200 1-6) Valter Fernandez (200 1-6) B.O.Haglund (1992-200 1) P.C. Gravelle (1992-2001) Gerrit Hakvoort (1992-200 1) Peter Haines (1997-2006, Secretary 2003-6) Wolfgang Hemminger(Chairman, 1992-2001) Trevor Lever (Chairman, 2001-6) Marianne Odlyha (199 1-2001) Takeo Osawa (2001-6) Michael Reading (199 1-7) Duncan Price (2001-6, Secretary, 200 1-3) Stefan Sarge(1992-2001,Secretary,2000-1) Judit Simon (1992-200 1) Fred Wilburn (1992-2006, Secretary, 1991-2000)

3. COMMENTS ON THE 2006 ICTAC NOMENCLATURE OF THERMAL ANALYSIS Our comments will be focussed on the general definition of thermal analysis, which has had to be modified over the last 40 years and which has attracted much interest, discussion and even controversy. The following definitions of Thermal Analysis have been successively proposed: In 1969 [2] the official ICTA definition, later endorsed by IUPAC [ l 11, was: " Thermal Analysis. A term covering a group of techniques in which a physical property of a substance andor its reaction product(s) is measured as a function of temperature whilst the substance is subjected to a controlled temperature programme"

This definition covered what is called to-day "temperature-controlled thermal analysis" [14 ] and whose principle is represented in Figure l(1eft). Here, the heating control loop uses the sample (or furnace) temperature T to impose a " ~ ~ n t r ~ ltemperature led programme", whereas the ' ~ h y s i c apropertyyy l X of the "substance andor its reaction product(s)" is simply "measured as a function of temperature".

Heating Heating Algorithm

time

'-.-....."

time

Figure 1. Principles of temperature-Controlled (or conventional) thermal analysis (left) and of sample-controlled thermal analysis (right), from [14]. The above definition did not embrace sample-controlled thermal analysis (SCTA), which was still in its infancy and on which very few papers had been published at that time. What makes sample-controlled thermal analysis special is that the physical property X (related to the behaviour of the sample) is directly involved in the heating control loop, as represented in Figure 1 (right). The heating-control algorithm can also make use of the sample (or furnace) temperature T, but this is not compulsory, hence the dotted control loop on the right hand of the figure. In such an experiment, there is no pre-determined temperature programme, because the temperature-time profile will depend directly on the - a-priori unknown - behaviour of the sample. The definition proposed in 1991 in the ICTAC booklet "For Better Thermal Analysis and Calorimetry" [lo] was as follows:

"Thermal Analysis. A group of techniques in which a property of the sample is monitored against time or temperature while the temperature of the sample, in a speciJied atmosphere, is programmed. The programme may involve heating or cooling at a fuced rate of temperature change, or holding the temperature constant, or any sequence of these.

This definition is more complicated than the first one and, in reality, although proposed to the ICTAC membership, was never officially endorsed by the ICTAC Council. It still does not embrace the case of sample-controlled thermal analysis, although SCTA is referred to in that 1991 booklet, in the form of controlled rate thermal analysis (CRTA). The above proposal also mixes the role of a definition and that of a recommendation, by stating "in a speciJied atmosphere" . It is certainly desirable that the experimenter reports the type of atmosphere surrounding the sample during the experiment, but the absence of such an information does not, of course, put the experiment outside of the scope of thermal analysis. The definition also describes the most common temperature programmes, which "involve heating or cooling at ajixed rate of temperature change, or holding the temperature constant, or any sequence of these". This sentence is not really needed in a definition. Its intended purpose was probably to provide examples of common ways to operate, but it unnecessarily restricted the scope of thermal analysis to experiments carried out with certain types of heating. Then came the definition endorsed in 2006 by the ICTAC Council: "Thermal Analysis (TA) is the study of the relationship between a sample property and its temperature as the sample is heated or cooled in a controlled mannery'

This definition is shorter than the two previous ones and it corresponds to the general representation of thermal analysis given in Figure 2. The above definition and representation leave the choice of the heating control for the thermal analysis experiment totally open. Such a definition therefore embraces any type of control, present or future, and includes both temperature-controlled and sample-controlled thermal analysis. A major interest of the full official definition given above is that it was built up and accepted by a broad, international, community of scientists who had to make compromises, because at least ten other, slightly different, versions of the definition were sound enough to be accepted. It should be stressed that this definition of thermal analysis clearly excludes from the scope of thermal analysis any totally isothermal experiment. This makes sense, because, otherwise, thermal analysis would embrace any isothermal recording of any physical property of the sample as a function of time, i.e. most experiments in physics and biology! Along these lines, one should restrict the term "thermogravimetry" to the recording of mass vs temperature. For instance, studying the gaseous corrosion of a metal at 400 OC by following its mass uptake, even if carried out with a "thermobalance", is not, according to the definition, a thermal analysis experiment. Therefore, it

should not be referred to as "thermogravimetry" but, more simply, as "gravimetry" or, to be more precise, as "isothermal gravimetry". Also, one should restrict the term "differential scanning calorimetry" (DSC) to an experiment where there is effectively a temperature scan (to which the term "scanning" explicitely refers). For instance, using a standard DSC instrument to study the isothermal behaviour of a living material, or an adsorption or desorption process, should not be considered as carrying out a DSC or thermal analysis experiment. The correct name of this isothermal technique would simply be "Differential Calorimetry" or "Isothermal Differential Calorimetry".

(3(3 Property "X'

Temperature

Figure 2. Principle of Thermal Analysis in general, from [14]: recording of physical property X as a function of temperature T, whatever the type of heating control.

4. A CONVENIENT NOMENCLATURE FOR CALORIMETERS The classification of calorimeters presented hereafter aims at being simple and clear enough to be relatively easy to teach, learn and remember. This classification was arrived at after a time of maturation of about 35 years. There were several reasons for such a slow maturation. One was the great variety of calorimetric devices developed over the last two centuries, probably due to the "labile" or "evanescent" character of heat, so difficult to store, which excited the imagination of many scientists. More than 100 different types of calorimeters can easily be listed. This made it difficult to establish any comprehensive and ordered presentation of the calorimeters, so that the famous Polish thermochemist Swietoslavski could write, in 1933, in his handbook on thermochemistry: "There is such a variety of different forms of calorimeters that we must renounce to list them, even hastily" [15]. This great variety of calorimeters led to a large number of possible classifications (a number of them are described in section 5), depending on the criteria chosen. 4.1.

Basic representation, criteria and categories

A calorimetric device in a simplified form is presented in Figure 3.

Figure 3. Schematic representation of a calorimeter

Here, the system "S" includes the sample and its container, with which it is in good thermal contact (both of them being at temperature Ts). The system "S" is thermally connected to the surrounding thermostat "T" (at temperature TT ) through the thermal resistance "R". Let us point out that the word "thermostat"

designates here a surrounding system at reasonably uniform temperature, which is an experimental prerequisite for any measurement of heat; that temperature is neither necessarily constant (like in DSC experiments) nor even controlled (like in the Thomsen or Berthelot calorimeters, cf Section 4.2 hereafter), though at least monitored. The basic criterion we shall use to distinguish different types of calorimeters will be that of the heat exchanges between system S and thermostat T. This immediately divides the calorimeters into two broad families, namely: 1. "Adiabatic" calorimeters, where the aim is to avoid any exchange of heat between the system S and the surrounding thermostat T. 2. "Diathermal" calorimeters, where the aim is, on the contrary, to favour the above heat exchange. The word adiabatic comes from the greek "adiabatos", meaning "which cannot be crossed". The word diathermal comes from the greek "dia" and "thermos", meaning "through" and "hot", respectively. In adiabatic calorimeters, one thus tries to store the heat in the system S, so as to measure it in situ from the temperature change of the system. This temperature change is therefore imposed by the heat produced or absorbed in the system S and is not known a priori. In diathermal calorimeters, where one tries to exchange most of the heat with the surroundings, the surroundings determine the temperature which can therefore be controlled by the experimenter. The above adiabatic or diathermal hnctioning of a calorimeter can be achieved (irrespective of the final efficiency) either by simple design and construction (by means of appropriate thermal conductors or resistors R) or (see below) with the help of electronic control. Hence each of the two families above, can be further divided into two groups, which can be named "passive" and "active", respectively. We therefore end up with four groups of calorimeters: Adiabatic, passive Adiabatic, active Diathermal, passive Diathermal, active The above classification is now applied to the "real world" of calorimeters.

4.2. "Passive" adiabatic calorimeters Heat exchange between the system S and the surrounding thermostat T (cf Figure 3) can be lowered at will by using two routes, i.e. either by increasing the thermal resistance R or by decreasing the temperature difference Ts - TT. "Passive" adiabatic calorimeters make use of only the first route, by means of an appropriate thermal insulation. Because an infinite thermal resistance does not exist, there will always be a residual heat leak which will have to be taken into account, usually as a correction term. These calorimeters can therefore also be called, quite appropriately, "quasi-adiabatic" or "semi-adiabatic". During half a century they were often referred to as "isoperibol", after the term coined by Kubachevski and Hultgren [16] to stress that, although the system temperature Ts varies (due to the thermal phenomenon under study), the surrounding thermostat usually remains at a constant temperature (as suggested from the greek "iso" and "peribolos", meaning "equal" and "surrounding space"). Now, this condition of constant surrounding temperature, although frequent, makes the term "isoperibol" of less general application than the term "quasi-adiabatic". Furthermore, the term "quasi-adiabatic" is more easily understood and selfexplanatory, so that it is preferred . The largest part of this group is made up by calorimeters, which were also called "water calorimeters" because the sample is in good thermal contact with an amount of water which efficiently collects any heat evolved and limits the resulting temperature change of the system S to a small, but measurable value,. These calorimeters are known under various names, such as the "Thomsen reaction calorimeter" or the "Berthelot calorimeter" (Figure 4), or, still, the "Ordinary calorimeter" (after Swietoslavski ) or the "Dewar calorimeter" (which makes use of the good thermal insulation of a Dewar vessel, like Eucken and Nemst's calorimeter shown in Figure 5,left). This category embraces simultaneously very simple and even crude calorimeters, as well as the most accurate tools of thermochemistry, like the "Argonne rotating bomb calorimeter", whose inaccuracy can be as low as a few ten-thousandths of the total heat measured, or modem reaction calorimeters whose roots can be found in the one devised by Sunner et al. [17]. It also embraces the ultra-fast calorimeters from Rodriguez-Viejo et a1. [18] and from Adamowsky and Schick [19] (Figure 5, right). In the latter, the thermopile is used to assess the instantaneous temperature of the micro-sample, rather than to measure the heat-flow (because the latter quantity is determined with a time-lag inconsistent with the objective of rapidity of the experiment). The heating rate can be as high as 1o4K min-'.

Figure 4. Two historical calorimeters of the "passive adiabatic" group: the Thomsen (left) and Berthelot (right) calorimeters

*unyp"n

mcmhmw

hcatcr hrrllunctmn

Figure 5. Dewar vessel calorimeter from Eucken, 1909 [20] (left) and ultra-fast calorimeter from Adamowsky and Schick, 2004 [19] (right).

mld ~llllcllhfl

4.3. "Active" adiabatic calorimeters The specific feature of these calorimeters is that they make use of the second route to lower the heat exchanges, which is to decrease the temperature dzflerence Ts - TT. This is achieved by servo-controlling the thermostat temperature TT to follow that of the sample Ts . Because it is more feasible to cancel a temperature difference than to build an infinite thermal resistance, these calorimeters, especially because they usually also make use of good thermal insulation, are able to provide excellent adiabatic conditions. For instance, they enable heat to be stored, and hence phenomena to be followed, over periods of weeks. They can therefore be named true >) adiabatic calorimeters. The first historical calorimeters of this type were manually-operated (see Figure 6). Person, in 1849 decided to cancel the temperature difference Ts - TT which he could measure by means of a double-bulb differential thermometer. For this purpose, he made use of a second, external, water bath, at a higher or lower temperature, in which he could partly immerse the water thermostat surrounding the system. He adjusted the depth of immersion by means of the crank visible in the drawing. He called his set-up a "cancelled heat-loss calorimeter" [21]. This is the first known "active adiabatic" calorimeter.

Figure 6 . The first "active adiabatic calorimeters": Person's, 1849 (left, figure from [2 11) and Richards', 1905 (right, figure from [22]).

In 1905 Richards proposed a chemical way to achieve the desired temperature increase of the thermostat jacket. The latter was filled with a soda solution into which sulphuric acid was dropped fiom burettes at the desired rate [22]. Appropriate mixing of warm and cold water was used by Swietoslawski to let the water jacket follow the sample temperature [15] but, later, an automatic cancellation of Ts - TT by electrical control and heating was progressively preferred. This is used in modern calorimeters such as the low-temperature calorimeters used to determine heat capacities, like those shown in Figure 7 (calorimeters designed by Westrum [23], Suga and Seki [24], Stnrlen et al. [25], Gmelin and Rodhammer [26] ) or the Accelerating Rate Calorimeters (ARC) used to study spontaneous self-heating and explosion during the storage or processing of new chemicals [27]. A few calorimeters have a double active control: in addition to the automatic cancellation of Ts - TTby control of the thermostat temperature TT, there is also an automatic cancellation of the temperature difference between the sample and a reference (which also forms part of the system S) by supply of heat to the cooler side. The first control is to provide the adiabatic conditions, whereas the second control is to provide a heat compensation of the phenomenon studied. The measurement of heat is not derived fiom a temperature increase of the system but from the electrical energy provided for the compensation. This principle was followed by Clarebrough [28], Bonjour [29] and Privalov [30].

-

Figure 7. Active adiabatic calorimeters now in use: multi-shielded lowtemperature calorimeter (left, after [31] ) and ARC calorimeter (right, after ~71).

"Passive" diathermal calorimeters Passive diathermal calorimeters are those in which good heat-exchange between the system S and the surrounding thermostat T is achieved by good thermal conduction. The sample temperature passively follows, here, the thermostat temperature and ,except in transient situations, there is no heat stored in the system S. These calorimeters can also be called, quite correctly, "thermal conduction calorimeters". The most common way, to-day, of measuring the heat exchanged in a passive diathermal calorimeter is to make use of a heatflowmeter (Tian-Calvet, [32-331, Wadso, [34], Du Pont heat-flux DSC, cf Figure 8). 4.4.

Figure 8. Three heat-flowmeter calorimeters: Tian-Calvet, 1950, leR, after [33] Wadso, middle, after [34]and Du Pont DSC, right, after [35]. A heat-flowmeter is used in the combination of a diathermal, passive, microcalorimeter with a quartz crystal microbalance, proposed by Smith and Shirazi [36], to study the adsorptive behaviour of samples weighing a fraction of a milligram. Another way to measure the heat exchanged in a passive diathermal calorimeter is to make use of a phase transformation taking place in the surrounding thermostat. This is the principle of the ice calorimeter devised and used by Lavoisier and Laplace (1783) [37] and of its modification by Bunsen (1870) [38], where the amount of molten ice is no longer determined from the mass of water produced, but from the constriction of the system, followed by the displacement of a mercury index in a glass capillary. Other phase changes proposed for operation at other temperatures are melting of di-phenyl ether to operate around room temperature, and vaporization of liquid air or liquid nitrogen.

Figure 9. Two ice calorimeters: Lavoisier and Laplace, 1783, left, after [37] and Bunsen, 1870, right, after [381.

A third way of measuring the heat exchanged in a passive diathermal calorimeter is to make use, in the surrounding thermostat, of aperipheral liquid flow (usually water) whose temperature change is determined. This is the principle followed by Junkers' flame calorimeter [39], the Picker liquid mixing calorimeter [40], and modern reaction calorimeters developed for safety studies of chemical reactions. 4.5. "Active" diathermal calorimeters These are still "diathermal" calorimeters, where everything happens as if there was excellent heat exchange between the system S and the surrounding thermostat T. This means that the system temperature Ts closely follows that of the thermostat, TT. The good heat exchange is, in reality, (( simulated )). For that purpose, the thermal transfer between sample and thermostat is complemented or replaced by a physical phenomenon taking place in close contact with the sample and able to generate or remove heat in situ.

ezlE

didd

Figure 10. Pneumatic compensation calorimeter, Ter-Minassian, 1983, after [441-

This (( power compensation )) is carried out so as to servo-control the sample temperature to follow the thermostat temperature. This is, therefore, exactly the opposite situation to the (( active )) adiabatic calorimeter, where it is the thermostat temperature which is servo-controlled to follow that of the sample. Here, one measures the heat provided to (or removed from) system S (sample + container) by the power compensation device. Heat compensation can be achieved by the Joule effect, if the studied phenomenon is endothermic and by the Peltier effect if it is exothermic. This was used by Tian, in 1924, to compensate the major part of the heat (by means of a constant power, which is easy to measure), whereas the remainder was measured with a heat-flowmeter [41]. A simpler and more elegant way to compensate either exothermic or endothermic effects was used by Dzhigit et al. in 1962 [42]. A Joule effect is produced in the system and this is controlled to ensure a constant temperature difference (say, 5 K ) between the system S and the thermostat T and, hence, a constant heat-flow. If the phenomenon studied is endothermic, the Joule effect is automatically increased to ensure the constant heat flow required. If it is exothermic, the Joule effect is simply decreased. These changes in the Joule effect are recorded and provide a direct measurement of the heat involved in the phenomenon studied. This was also used by Hansen et al. in 1982 in their high temperature "battery calorimeter" [43]. An original route is that proposed by Ter-Minassian and Milliou in 1983 [44] with their pneumatic compensation calorimeter, represented in Fig 10. The tubular sample cell 4 is in good thermal contact with four metallic bulbs. Two of them operate like bulb 1 in the figure, i.e. as pneumatic thermal detectors. They are filled with gas, say around 1 bar, and their pressure is compared, by means of a differential manometer, with the constant pressure of a reference reservoir 3 immersed in the surrounding thermostat block 5. Therefore, they detect any temperature change of the sample. The two other bulbs operate like bulb 2, i.e. as pneumatic energy-compensating devices. They are also filled with gas, say around 1 bar, but they are connected to the piston-cylinder 7 which enables the heat of compression (or decompression) necessary to cancel the temperature difference between the sample and thermostat (as detected with the first set of bulbs) to be produced in the bulb. More recently, Zimmermann and Keller built a comparable pneumatic compensation calorimeter whose calorimetric performances were carefully examined [45].

5. OTHER POSSIBLE NOMENCLATURES FOR CALORIMETERS A number of other nomenclatures have been proposed in calorimetry, sometimes with specific goals or viewpoints which can explain the differences. A number of them are described below, with short comments given in italics.

5.1. Nomenclature proposed by Swietoslawski in 1933 [15] Given the complexity of the topic, this author decided to limit himself to the description of calorimeters of the most general type. He did not aim to be comprehensive and proposed three main groups:

1) Ordinary, with a central vessel and double-walled surroundings. Typical calorimeters mentioned in this group are the Thomsen mixing calorimeter and the Dewar vessel calorimeter. 2) Adiabatic, with cancellation of the temperature difference between the sample and the surrounding thermostat by appropriate temperature control of the thermostat. Swietoslawki mentions, among others, Richards' calorimeter (1905) and his own calorimeters where the temperature control of the thermostat was achieved either electrically (19 14) or by addition of cold or warm water (19 19). 3) Isothermal which meant, for Swietoslawski, that the surrounding thermostat was kept at constant temperature. He included in this category the adiabatic systems in which the temperature of the thermostat, instead of being brought identical to that of the sample (to keep adiabatic conditions) was simply kept constant. He also included here phase-change calorimeters (the Bunsen ice calorimeter or the water vapour calorimeter successively described, in the same year 1887, by Joly and Bunsen).

Comments: I ) The "Ordinary" group totally enters the group of 'passive adiabatic "calorimeters described in section 4.2. 2) The "Adiabatic" group is identical to the group of "active adiabatic "calorimeters described in section 4.2. 3) The "isothermal" group is more heterogeneous,because, apart from the phase-change calorimeters, it curiously includes those 'passive adiabatic" calorimeters whose surrounding shield is kept isothermal "within 0.002 OC ".. . 5.2. Nomenclature proposed by Calvet and Prat in 1956 [46] The classification proposed by Calvet and Prat is based on the heat-exchange between the calorimetric vessel and its surroundings :

1) Adiabatic calorimeters: the thermal conductance between the calorimetric vessel and the surrounding thermostat equals 0. These calorimeters integrate all heat effects. The heat measurement is based on the measurement of the sample temperature. The "most perfect" adiabatic calorimeters are those for which the temperature of the thermostat is brought to follow that of the internal vessel, like those proposed by Person (1849) and Richards (1905). In this group of "adiabatic" calorimeters, the authors also mention the Dewar vessel calorimeter (which they call quasi-adiabatic") and the Berthelot calorimeter. 2) Isothermal calorimeters: the thermal conductance is here very high. They consider that the most perfect example is the Bunsen ice calorimeter. They also quote the Junkers flame) calorimeter with heat exchanger and water countercurrent but consider that it is less perfect than the ice calorimeter and should rather be considered as "semi-isothermal". 3) Conduction calorimeters: typically, the Tian-Calvet calorimeter. They comment that, in spite of its good external thermal insulation, this calorimeter is not adiabatic, because the "calories" produced are continuously eliminated from the calorimetric vessel. They also consider that, in spite of the very small temperature variations of the sample cell, this calorimeter is not "strictly isothermal", which justifies a separate group, except when a Peltier compensation totally cancels the temperature variations in the sample.

Comments: I) It would probably be more correct and general to define an adiabatic calorimeter as having a very low heat transfer (which takes into account simultaneously the thermal conductance and the temperature difference) rather than as having a very low heat conductance. Also, the definition given above (thermal conductance equal to zero) only applies to an ideal, non-realistic, system and is not amenable, in principle, to existing calorimeters. Iin spite of these inconsistencies, it is noteworthy that the general content of this group coincide swell with that of the "adiabatic familyJ'presented in sections 4.1 and 4.2 and includes both the "active adiabatic" (or "true adiabatic'y and the '@assiveadiabatic" (or quasi-adiabatic) calorimeters. 2) The distinction between "Isothermal" and "Conduction" calorimeters, which they base on heat-conductance differences, does not withstand a close analysis. In a Bunsen ice calorimeter the temperature differences between the sample and the melting ice are of the same order of magnitude as in a conduction calorimeter, i.e. rangingfrom less than 10'~K up to I K, depending on the rate of the reaction studied,the thermal conductivity of the sample, and the

construction of the calorimeter (Bunsen glass-made or brass-made, for instance).

5.3. Nomenclature proposed by Evans in 1969 [ 471 .Evans introduced his classification in the following way: "We need only briefly consider classification. Calorimeters are either isothermal or adiabatic, though some calorimeters do not belong strictly to either category". He finally distinguishes: 1) Isothermal calorimeters, which make use of a phase change, the best example being the Bunsen ice calorimeter. 2) Adiabatic calorimeters, where the heat exchange with the environment is eliminated. 3) Heat conduction calorimeters of the Calvet type Comment: this is similar to the Calvet and Prat classij?cation (1956), still with an unjustiJied separation between isothermal and heat conduction calorimeters. The adiabatic calorimeters are more correctly defined (i.e. after the heat exchange, not after the thermal conductance) but, still, in an idealized way, because the author restricts the term to calorimeters where the heat exchange is "eliminated". 5.4. Nomenclature proposed by Skinner in 1969 1481 Skinner considered "Two extreme types: either the adiabatic or the conduction calorimeter.... Between the two extremes lies the well-known and commonly used isothermally jacketed calorimeter". He ended with the following classification: 1) Adiabatic calorimeters. Isothermally-jacketed calorimeters are also examined in this section 2) Conduction calorimeters. a) Phase-change calorimeters b) Mechanical-conduction calorimeters making use of a thermoelectric pile to transfer heat.

Comment: after saying that isothermally jacketed calorimeters are an intermediate category, Skinner nevertheless ends with only two main groups, associating the isothermally jacketed calorimeters with the adiabatic ones (as is proposed in Sections 4.1 to 4.3 ). Also, he puts phase-change and conduction calorimeters in the same category. This idea was kept, under the name of '>passive diathermal", in section 4.4. The calorimeters forming, in section 4.5,

the category of "active diathermal" have no place in the above classification, which therefore looks like a "truncated" classiJication) Nomenclature proposed by Rouquerol and Laffitte in 1972 [49] The aim of this classification was "to unquestionably classify any calorimeter within only one main class. The relationship existing between the sample temperature and that of the surrounding thermostat seems to be an appropriate criterion". For that purpose, a clear distinction is made between two parts of the calorimeter, namely (i) the sample together with the container with which it is in good thermal contact and (ii) the surrounding thermostat. To make this distinction clear, these two distinct parts are shown on the schematic representation of 11 types of calorimeters. This finally leads to the following classification:

5.5.

1) Adiabatic calorimeters: the temperature of the surrounding thermostat is controlled by that of the sample. 2) Isothermal calorimeters: the sample temperature is controlled by that of the surrounding thermostat. 3) Isoperibol calorimeters: no such temperature link exists between the sample and the surrounding thermostat. The temperature of the thermostat is kept constant.

Comments: 1) The "separating power" of the above classzjication is high, with very few overlaps. 2) The distinguishing of a central system consisting of only the sample and its container which are in good thermal contact is a simple one,and is easy to apply. It was therefore kept in the basic representation of a calorimeter given in section 4.1. 3) The definition of Adiabatic calorimeters, in spite of being simplified (it does not mention the value of the thermal resistance, always referred to in the past) effectively covers all calorimeters considered by thermochemists as being really adiabatic. This definition was kept for the "active adiabatic calorimeters" described in section 4.3. 4) The dejinition of the Isoperibol calorimeter is the one given by Kubaschewski and Hultgren, who coined the term in 1962 [I 61. 5) The definition of the Isothermal calorimeter raises a problem because the term "isothermal" is not used with its usual meaning (temperature constant over time and space) but with the restricted meaning of temperature constant over space. The sample follows the surrounding thermostat temperature, but the

thermostat can itself be programmed to vary with time. The word "isothermal" is therefore not very appropriate. Nomenclature proposed by Hemminger and Hohne in 1984 [50] These authors were aware of the difficulty of establishing a comprehensive classification of calorimeters: " In every classification there are certain calorimeters which do not clearly belong to a particular category. ...The Calvet calorimeter, for instance, can be used either isothermally with electric compensation.. .or in an isoperibol manner involving the measurement of a local temperature difference. Moreover, a number of existing calorimeters remain outside our classification. One example is a calorimeter involving a compensation of the thermal effect other than by thermoelectric means or by phase transition. But such devices can be easily included in our classification by analogy." They proposed various types of classifications, depending on the criterion used. 5.6.

A) Classification based on the measuring principle (pp 5-19). It lead them to distinguish two major "methods of calorimetry" : 1) Calorimetry with compensation of the thermal effects a) By a phase transition: ice calorimeter (Lavoisier, Bunsen) b) By electric effects: Joule effect (Stenwehr, 1901; Bronsted, 1906)' thermoelectric (Peltier) effect (Tian's isoperibol calorimeter with Peltier effect) c) By chemical heat of reaction (Regnault,1870) 2) Calorimetry by measurement of temperature differences a) Time-dependent temperature differences. This involves the heat capacity of the instrument. Heat leaks must be taken into account. b) Local temperature differences. Simultaneous determination in two different positions: - Flow calorimeter for the reaction of two liquids - Heat flux calorimeter making use of a heat conduction path Comments: 1) The "compensation"family includes phase-change calorimeters. This results from the place where the authors locate the "boundary surface' between "measuring system" and surroundings. They consider that '?arts which constitute or traverse the boundary surface belong to the measuring system insofar as they are involved in one way or another in the changes caused by the sample reaction". Such a "measuring system" can embrace much more than the central system S described in section 4.1: for instance, in an adiabatic calorimeter, the surrounding thermostat is still part of the measuring system,

because its temperature is dictated by the sample reaction; also, in the Jungers ' flame calorimeter with a liquid flow heat exchanger, one should then consider that the surrounding liquid flow is part of the measuring system, because its temperature changes depend upon the sample reaction. Along the same lines, in a phase-change calorimeter, the phase change is considered to take place within the "measuring system". In the three examples above, the whole calorimeter is included in the "measuring system". Here, provided that power compensation takes place within that broad "measuring system", the calorimeter is said to be of the compensation type: this applies to the phase-change calorimeters. This differs from the common usage, which restricts the term 'Ipower compensation calorimeter" to calorimeters where power compensation takes place within, or in the very close vicinity oJ; the sample, i.e., only within system S as shown in Figure 3. 2) The "temperature difference" family includes most adiabatic and quasiadiabatic calorimeters (in the "time dependent temperature" group) together with most heat-flowmeter calorimeters. The total probably represents between 80 and 90 % of the calorimeters used today, so that,for practical use, the above class$cation looks somewhat unbalanced. Moreover, the calorimeters just mentioned shift to the first family as soon as they also make use of heat compensation, hence a real overlap exists between the two mainfamilies.

B ) Classification based on the mode of operation (pp 80-88) : 1. Isothermal : the "measuring system" and the surroundings are at the same temperature; the thermal resistance, R, between the "measuring system" and the surrounding thermostat is supposed to be "infinitesimally small.. ..which is not feasible in calorimetry". Consequently, isothermal operation is said to necessitate a compensation of the heat released, either by a phase transition (passive measuring system) or by thermoelectric effects (active measuring system). 2. Isoperibol : with constant temperature of the surroundings. The thermal resistance, R, is said to be here "of finite magnitude". This category includes the "classic liquid calorimeter" and the "heat-flux calorimeter". 3. Adiabatic: the thermal resistance R is said to be here "very large" or "infinitely large", whereas the temperature of the surroundings is so controlled as to be always equal to that of the "measuring system". 4. Adiabatic scanning : with a "linear heating rate" of the "measuring system" 5. Isoperibol scanning : same heating as for adiabatic 6. Scanning of the surroundings : with a "linear heating rate" of the surroundings

Comments: 1) The scanning: because the criterion used is the mode of operation, it makes sense, atJirst sight, to make a special family with the calorimeters used in the temperature scanning mode. However this only considers a single aspect of the mode of operation (namely, the mode of heating), whereas many other aspects could be considered (mode of introduction of the sample, mode of initiation of the phenomenon studied, open or closed system, static or dynamic experiment ...). One may therefore wonder whether calorimeters 4 to 6 should not be simply considered as sub-groups of theJirst three? 2) The thermal resistance R is supposed to increase from the isothermal to the isoperibol and then to the adiabatic type of calorimeter. It would probably be more correct and general to base the distinction between the adiabatic and the isoperibol calorimeters on the heat transfer (involving simultaneously the thermal conductance and the temperature difference) rather than on the value of the thermal resistance. For instance, a simple Dewar vessel calorimeter provides a very high thermal resistance between the central system and the surroundings, though it is simply an isoperibol calorimeter (called "quasiadiabatic" in section 4.2.), whereas Swietoslawski's adiabatic calorimeters, which do not use any vacuum insulation, certainly provide a much lower thermal resistance [15]. C) Classification based on the construction principle (p 131) : 1) Twin calorimeters 2) Single calorimeters

Comments: 1) Twin calorimeters are often referred to as "differential" calorimeters. 2) For many readers, the term "constructionprinciple" may mean much more than a twinning: the calorimeter can also be built liquid or aneroid, closed or open, in a Dewar, with a liquid thermostat or a furnace, with thermopiles or thermistors, with a batch or liquidflow mixing system, with a steady flame or a bomb combustion device, with a sample dropfacility and with a multi-sampler). Hemminger and Hohne finally proposed (p 132) to simultaneously use the three classifications above to classify any calorimeter.

Nomenclature proposed by Rouquerol and Zielenkiewicz in 1986 [51] The aim of this classification was to be "simple, easy to apply (both for actual and hypothetical calorimeters)" and to give rise to the minimum overlaps. It starts fiom the general equation:

5.7.

andr from the three ideal categories of calorimeters which can be derived fiom it, namely (i) the "perfect adiabatic", where all the heat is accumulated, (ii) the "perfect heat-flow", where all the heat is exchanged, and (iii) the "ordinary" calorimeters, where the heat is partly accumulated and partly exchanged. To be applicable, however, a nomenclature system should not request "perfection" for two categories out of three. This is why this classification comes back to the criterion proposed by Rouquerol and Laffitte in 1972, namely the relationship between temperature Ts of calorimetric vessel and temperature TT of the first surrounding thermal shield. This criterion is (i) readily applicable to either real and imperfect calorimeters and (ii) does not require any subjective judgement about the "quality" of the calorimeter. It leads to the three following groups: A) Adiabatic calorimeters: Ts is controlled to follow Tc, From the viewpoint of heat-exchange and dynamic behaviour these are integrating systems. B) Ordinary calorimeters: no fixed relationship between Tc and Ts . These are inertial systems, characterized by a time-constant, and mainly include "isoperibol" calorimeters. C) True Isothermal (i.e. both in space and in time) or extended isothermal (i.e. only isothermal in space) calorimeters: Tc follows Ts ; these are proportional systems and include phase-change, power-compensation and heat-flowmeter calorimeters.

Comment: This classification still separates fiom each other the Adiabatic and the Ordinary (or Isoperibo)l calorimeters, but, under the heading of "Isothermal or Extended Isothermal", introduces, as a whole, the family of calorimeters which are called, in sections 4.4. and 4.5 above, "Diathermal", a term certainly more appropriate than "Isothermal". 5.8. Nomenclature proposed by Tachoire and MCdard in 1994 [52] These authors do not classify calorimeters, but rather "calorimetric methods in thermochemistry", which is a more general term able to include things other than the principles of construction or detection. Also, their classification is

explicitly aimed at classifying the main calorimetric methods in use prior to 1912. A) The mixing method Because the heat to be measured is usually brought to a vessel containing water, most calorimeters used in this method can be called "water calorimeters". Examples: James Watt (1786), Regnault (1840), Favre and Silbermann (1845), Berthelot (1875). Other liquids can be used: Favre and Silberrnann's mercury calorimeter. B) The isothermal method 1. Phase transition calorimeters (water, phenyl oxide) 2. Compensation isothermal calorimeters: Mathias (1888) C) The adiabatic method They also suggest to distinguish, among the calorimetric methods: A) Those operating under constant pressure (usually atmospheric pressure) B) Those operating under constant volume (like in Berthelot's calorimetric bomb) Comments: 1) The group making use of the "mixing method" mainly involves Ordinary or Isoperibol calorimeters, but, curiously, not all of them. This is because the term "mixing method" eliminates the isoperibol experiments which do not make use of a liquid medium to store the heat evolving from the sample, like cement hydration carried out in a Dewar vessel ... 2) The aim of this classification is mainly historical. It introduces the reader to the way of thinking of the thermochemists prior to 1912 and does not pretend to be comprehensive and modern. Quite logically, it does not include, for instance the Tian-Calvet microcalorimeter. 5.9. Nomenclature proposed by Wadso in 1997 [53] From the point of view of heat measurement principles, the calorimeters are divided into three main groups:

A) Adiabatic calorimeters. These can be: - Either ideal (or close to ideal), with temperature control of the surrounding shield so as to keep at zero the temperature difference between the calorimetric vessel and the shield. - Or "semi-adiabatic", often called "isoperibol" B) Heat conduction calorimeters, normally making use of a thermopile as a sensor of the heat flow.

C) Power compensation calorimeters, where the thermal process is balanced by a cooling or heating power. Comments: 1) Adiabatic calorimeters include here, all at once, the "Ideal adiabatic" and the "Semi-adiabatic". The same general meaning for the term adiabatic was kept to defne the adiabaticfamily in sections 4.2. and 4.3. 2) Being in the scope of a review of modern trends, the above classzjication did not need to explicitly list the phase-change calorimeters. 3) This classzjication is also used in the IUPAC Technical Document on " Standards in Isothermal Microcalorimetry" prepared by Wadso and Goldberg in 2001 [54].

5.10. Nomenclature proposed by Hemminger and Sarge in 1999 [55] This nomenclature is close to that proposed by Hemminger and Hohne in 1984. It makes use of the same three primary criteria: the principle of measurement, the mode of operation and the construction principle. Each criterion leads to its own classification, as shown hereafter. The main difference from the 1984 classification is that, instead of only proposing two major "methods of calorimetry" (compensation of the thermal effects and measurement of the temperature differences, respectively) there are now three. This is obtained by splitting the second one into calorimeters that measure a heat-accumulation (including the adiabatic and the isoperibol calorimeters) and calorimeters that measure a heatflow. A) Theprinciple of measurement 1) heat-compensatingprinciple: determination of the energy (power) required for compensating the heat (heat flow rate) to be measured; a) passive compensation: phase transition Example: Ice calorimeter p) active control system. Examples: Adiabatic scanning calorimeter, explained to be a heat compensation, single measuring system - Power comvensating DSC, explained to be isoperibol scanning with a twin measuring system. 2) heat-accumulatingprinciple: measurement of the temperature change caused by the heat to be measured. Examples: Drop calorimeter ,explained to be heat accumulation, isoperibol, single measuring system Adiabatic bomb calorimeter, Flow calorimeter , explained to be heat accumulation, isoperibol and either single or twin measuring systems. 3 ) heat-exchanging principle: measurement of the temperature difference between the sample and the surroundings caused by the heat (heat flow rate) to

be measured. Examples: Heat-flux DSC, explained to be heat exchange, twin measuring systems, with scanning of the surroundings. B ) The mode of operation (temperature conditions) 1) static: isothermal isoperibol adiabatic 2) dynamic: scanning of surroundings; isoperibol scanning adiabatic scanning.

C) The construction principle 1 ) single measuring system; 2) twin or differential measuring system. Comment: The criteria above are aimed at describing the calorimeters rather than at helping to classzfL them. Most comments made about the previous nomenclature by Hemminger and Hohne, 1984, apply here. 5.11. Nomenclature proposed by Hansen in 2001 [56] A systematic nomenclature for describing calorimeters is proposed. The instrument name and description should include four parts: the method of heat measurement, a description of the temperature control of the surroundings, a description of the means of initiating the heat effect, and a description of the operation of the calorimeter including data analysis.

I. Method of heat measurement Only two principles upon which heat measurement can be based are known, namely: A) the law of conservation of energy 1) compensation calorimeters a) "phase change compensation" b) "power compensation" 2) temperature change calorimeters : a near-adiabatic system. Suggested name "temperature change calorimetry"

B) Newton's law for the rate of heat transfer: heat-conduction calorimeters

I1 Temperature control of the surroundings (ie everything outside the reaction vessel) A) Isothermal control B) Adiabatic control (to minimize the temperature difference between the system and surroundings) C) Other control

I11 Means of initiating the heat effect A) Changing the temperature "Temperature scanning calorimetry" B) Changing the pressure or volume "Pressure/volume scanning calorimetryn C) Changing the composition of the system "Titration calorimetry", "Flow calorimetry", "photochemical calorimetry", "sample insertion calorimetry" IV Operation of the calorimeter A) "Twin" (or "differential") B) Constant pressure or constant volume C) Mode of operation : temperature scanning, isothermal Comments: I ) Here, the author stresses the existence of "Only two principles upon which heat measurement can be based", namely A) the law of conservation of energy and B) Newton's law for the rate of heat transfer. The same two principles are used, in sectionl.l., to distinguish two main families of calorimeters, namely the adiabatic and the diathermal. 2) The above nomenclature aims at describing the calorimeters rather than at classzfiing them. It is comprehensive and certainly useful as a check-list for the author of a paper, which was probably a main objective of the author. 5.12. Nomenclature proposed by Matsuo in 2004 [57] The author first distinguishes between static and dynamic methods: A) Static methods 1. Isothermal calorimeters: only phase change 2. Conduction calorimeters: "the heat to be measured is conducted down to a thermostated component of the calorimeter through a thermomodule or thermopile. 3. Isoperibol calorimeters : the surroundings are held at constant temperature. Heat leakage is corrected for.

4. Adiabatic calorimeters, with temperature control of the surroundings to follow the sample temperature. B) Dynamic methods 1. DSC 2. Adiabatic Twin Differential Scanning Calorimetry: Privalov's calorimeter 3. AC Calorimetry: periodic heating (Joule effect, light) on one side of the sample, measurement of the amplitude of the temperature response on the other side (which is related to a heat sink through a thermal resistance). Provides values of the heat capacity. 4. Relaxation calorimetry: heating of sample, left then to cool down. The time constant of the exponential decay provides values of the heat capacity. He also distinguishes between : A) Non-reaction calorimetry: measurement of physical modifications, change of state.. . B) Reaction calorimetry: heats of mixing, dissolution, chemical reaction (combustion, biological reactions...). Adsorption and immersion can be included under either reaction or non-reaction calorimetry. And also: A) Absolute and B) Relative measurements. Absolute measurements are related directly to the measurement of basic SI quantities. Only possible, at present, with the "static" methods B) Relative measurements A) Batch calorimeters and B) Flow calorimeters. Both are used in calorimetry on fluids. A) Single and Twin calorimeters

Comment: The above classz~cationinvolves many ideas, much experience and many criteria. It cannot avoid, unfortunately, many overlaps, and shifts of a "calorimetric method" from one category to another one by simply changing

the experimental conditions. It is more a classiJicationof calorimetric methods than of calorimeters.

5.13. Nomenclature proposed by Zielenkiewicz in 2004 [58] The classification given here is based on two criteria: 1) heat exchange conditions between the calorimeter proper and the shield (surroundings) 2) temperature conditions under which a studied process occurs (i.e. the way in which the calorimeter temperature T,, the shield temperature To and their difference AT are held constant or vary with time). The author stresses that the above criteria were already used by Lange and Mishchenko in 1930 [59], and then led to four groups. Here, these criteria are applied in a more refined way which leads to two main groups and seven categories:

I Adiabatic calorimeters (AT= 0) 1) Both Tcand To vary. This is a standard adiabatic calorimeter, either scanning or not. It is said to be "characterized only by heat accumulation and dynamic properties of integral objects". 2) Tc and To are constant (and equal, because AT = 0). Here we have a compensation calorimeter, which can be: a) Either aphase change calorimeter (Lavoisier and Laplace, Bunsen.. .) b) Or an "adiabatic-isothermap calorimeter making use of a Peltier or Joule effect, like the ITC Microcal calorimeter. This is often known as a "power-compensation calorimeter"

I1 Non-adiabatic calorimeters (AT# 0) 3) Tc and To are constant and AT is also constant ( this constant AT creates a constant, intended, heat leak).This is called an isothermal-non adiabatic calorimeter. Examples: Dzhigit et a1.[42] , Zielenkiewicz and Chajn [60], Hansen et a1.[43], Christiansen and Izatt [61]. Like the preceding one, this is also often known as a "power-compensation calorimeter" 4) Both Tc and To vary but AT is again constant. (This can be considered as the "temperature scanning version" of the preceding calorimeter with constant heat leak) 5) Tc varies, To is constant and, as a consequence, AT varies. These "nonadiabatic non-isothermal calorimeters" include:

a) Dewar vessel calorimeters b) Water calorimeters for combustion calorimetry c) Conduction calorimeters "in which the heat exchange between the calorimetric vessel and shield is very good" 6 ) T, and Tovary, and so does AT. Examples: DTA and DSC 7) Finally, T, is constant, whereas To varies and, as a consequence, so does A T . This is a theoretical possibility, with no representative known.. .. Comments: I ) The above comprehensive classiJication covers the whole ensemble of existing or possible calorimeters. 2) It is systematic and leads to seven different groups, from which we could expect a useful classification. 3) Most unfortunately, when applied to existing calorimeters, this classification is unbalanced, since one group (NO 7) is void, whereas another one (NO5) embraces more than 60% of the real calorimeters, i.e. all isoperibol and most conduction calorimeters.

6. CONCLUSIONS Our conclusions are somewhat different for Thermal Analysis and for Calorimetry. As already pointed out, the community of Thermal Analysis organized itself, 40 years ago, under the scope of the International Confederation of Thermal Analysis (ICTA), to produce a nomenclature acceptable by all scientists and equipment manufacturers. This nomenclature was - and is still- updated under the scope of the ICTAC Nomenclature Committee (the "C", standing for "Calorimetry", was added in the acronym in 1992). This does not mean that agreement is easy to reach when one starts from the terms used or coined by various countries and various manufacturers. Nevertheless, a permanent, acknowledged, body exists (the Committee above) which, taking time, eventually arrives at a proposal acceptable by a great majority. The ingredients needed for success are time (10 years being an order of magnitude of the time needed to get general agreement), logic (but this is not always the major factor), pragmatism (which means one should not change a term already widely used and understood, if not absolutely needed) and authority (like that given by an international body of the type of ICTAC). Thermal analysts are therefore advised to follow the latest nomenclature proposed (and reported in sections 2. and 3. above) and also to convey to the standing Nomenclature Committee of ICTAC any problems of application, or any suggestions for improvements,

because nomenclature is, for any scientific community, an essential tool which needs to be kept efficient and up-to-date. The field of Calorimetry is much older than that of Thermal Analysis and the terminology used is oRen confusing. This is partly due to the fact that a large number of calorimeters, many of which are only slightly different, have been reported. Further, commercial manufacturers of calorimeters have sometimes promoted their own terminology. Also, the community of Calorimetry does not have the same unity as that of Thermal Analysis; it is distributed indeed between several overlapping International Groups (one for Biocalorimetry, another one for North-America, a third one which took over from the former IUPAC Commission on Chemical Thermodynamics...) and it also forms, with the Thermal Analysis community, more than 20 national groups which accommodate the two communities and which are federated in ICTAC. Hence there are a series of technical and social difficulties in achieving agreement on nomenclature. Calorimetrists should, therefore, choose their nomenclature from among those presented above, depending on their own objectives and fields of interest: - For a complete presentation and description of a new calorimeter: Hansen (2001), Hemminger and Hohne (1984), Matsuo (2004), Hemminger and Sarge (1999) - For special attention to the heat transfer and response of the calorimeter: Zielenkiewicz (2005) - For a simple, and pedagogical presentation of calorimetry in general: the classification presented in section 4 above. 7. REFERENCES

1. A.L. Lavoisier, Trait6 61Cmentaire de Chimie, Cuchet, Paris,1789. 2. R.C. Mackenzie, Talanta, 16 (1969) 1227. 3. R.C. Mackenzie, C.J. Keattch, D. Dollimore, J.A. Forrester, A.A. Hodgson and J.P. Redfern, Talanta, 19 (1972) 1079. 4. R.C. Mackenzie, C.J. Keattch, T. Daniels, D. Dollimore, J.A. Forrester, A.A. Hodgson, J.P. Redfern and J.H. Sharp, J. Thermal Anal., 8 (1975) 197. 5. R.C. Mackenzie,Thermochim. Acta, 28 (1979) 1. Acta, 46 (1981) 333. 6. R.C. Ma~kenzie~Thermochim 7. R.C. Mackenzie, in P.J. Elving (Eds), Treatise on Analytical Chemistry, Part 1, Vol. 12, Wiley, New York, 1983, p.1. 8. J.H. Sharp, Thermochim. Acta, 104 (1986) 395. 9. G. Lombardi, For Better Thermal Analysis, ICTA, Rome, 1977. 10. J.O. Hill, For Better Thermal Analysis and Calorimetry, Edition 111, ICTAC, 1991. 11. Pure Appl.Chem., 37 (1974) 439.

12. G.C. Guilbault, Pure Appl.Chem., 52 (1980) 2385. 13. I. Mills, T Cvitas, K Homann, N. Kallay and K. Kuchitsu (Eds), Quantities, Units and Symbols in Physical Chemistry 2nd Edn.,, IUPAC, Blackwell, Oxford, 1993. 14. J. Rouquerol and O.Toft Ssrensen, in 0.Toft Serrensen and J. Rouquerol (Eds), Sample Controlled Thermal Analysis: origin, goals, multiple forms, applications and future, Kluwer, Dordrecht/Boston/London, 2003, pp. 8-15. 15. W. Swietoslawski, Thermochimie, Librairie Alcan, Paris, 1933, pp 3- 10. 16.0. Kubachevski and R. Hultgren, in H.A. Skinner (Ed.), Experimental Thermochemistry, Vol. 2, Wiley-Interscience,New York, 1962,35 1. 17. S. Sunner and I. Wadso, Acta Chem. Scand., 13 (1959) 97. 18. J. Rodriguez-Viejo, M. Chacbn, A.F. Lopeandia, M.T. Clavaguera-Mora, Leone1 R. Arana, and K.F. Jensen, J. Materials Res., (2002) 192. 19. S. Adamovsky 1, C. Schick. Thermochim. Acta, 41 5 (2004) 1. 20. A. Eucken, Physikal.Zeitschrift, 10 (1909) 586. 2 1. C.C. Person, C.R. Acad. Sci., 29 (1849) 302. 22. T.W. Richard and G.S. Forbes, Proc. Am. Acad. Arts Sci., 41 (1905) 8. 23. E.F. Westrum Jr., J. Thermal Anal., 14 (1978) 5. 24. H. Suga and S. Seki, Bull. Chem. Soc. Japan, 38 (1965) 1000. 25. S. Stslen, F.Grsnvold and E.F. Westrum, Jr., J. Chem. Thermodyn., 22 (1990) 1035. 26. E. Gmelin and P. Rodhammer, J. Phys. E: Sci.Instrum., 14 (1981) 223. 27. D.I. Townsend, Chem. Eng. Prog., 73 (1977) 80 and D.I. Townsend and J.C.Tou, Thermochim. Acta, 37 (1980) 1. 28. L.M. Clarebrough, M.E. Hargreaves, D. Michell and G.W. West, Proc. Roy. Soc. London, Ser. A, 215 (1952) 507. 29. E. Bonjour in J.P. Redfern (Ed.), Thermal Analysis, Macmillan, London, 1965, pp 170-173. 30. P.L. Privalov, V.V. Plotnikov and V.V. Filimonov, J. Chem. Thermodyn., 7 (1975) 41. 3 1. T. Atake,Adiabatic Heat Capacity Calorimetry, in M Sorai (Ed.), Calorimetry and Thermal Analysis, Wiley, 2004, p.70. 32. A. Tian, Bull.Soc.Chim.Fr., 33 (1923) 427. 33. L. MCdard and H. Tachoire, Histoire de la Thermochimie, Pub.UniversitC de Provence, Aix-en-Provence, 1994, p.475. 34. J. Suurkuusk and I. Wadso, Chemica Scripta, 20 (1982) 155. 35. W. Hemminger and G. Hohne, Calorimetry, Fundamentals and Practice, Verlag Chemie,Basel, 1984, p.226. 36. A.L. Smith and H.M. Shirazi, Thermochim. Acta, 432 (2005) 202. 37. A.L. Lavoisier and P.S. de Laplace, MCmoire sur la chaleur, C.R. Acad. Royale des Sciences, 2gthJune1783.

38. R.W. Bunsen, Ann.Phys., 141 (1970) 1. 39. H. Junkers, Z. Instrumentenkd, 15 (1895) 408. 40. P. Picker, C. Jolicoeur and J.E. Desnoyers, J. Chem. Thermodyn., 1 (1969) 469 41. A. Tian, J. Chim. Phys., 30 (1933) 665. 42. O.M. Dzhigit, A.V. Kiselev and G.G. Muttik, J. Phys. Chem., 66 (1962) 2 127 43. L.D. Hansen, R.H. Hart, D.M. Chen and H.F. Gibbard, Rev. Sci. Instrum., 53 (1982) 503. 44. L. Ter-Minassian and F. Milliou, J. Phys. E: Sci. Instrum., 16 (1983) 450. 45. W. Zimmermann, J. U. Keller, Thermochim. Acta, 405 (2003) 3 1. 46. E. Calvet and H. Prat, MicrocalorimCtrie ; applications physico-chimiques et biologiques, Masson, Paris, 1956, pp 7-10. 47. W.J. Evans in H.D. Brown (Ed.), Biochemical Microcalorimetry, Academic Press, 1969, pp 257-273. 48. H.A. Skinner in H.D. Brown (Ed.), Biochemical Microcalorimetry, Academic Press, 1969, pp 1-32. 49. J. Rouquerol and M. Laffitte in Thermochimie , CNRS, Paris, 1972, pp. 181-188 50. W. Hemminger and G. Hohne, Calorimetry, Fundamentals and Practice, Verlag Chemie,Basel, 1984, pp 5-19 , 130-133 and 275. 5 1. J. Rouquerol and W. Zielenkiewicz, Thermochim. Acta, 109 (1986) 121. 52. H. Tachoire and L. MCdard, Histoire de la Thermochimie, UniversitC de Provence, Aix-en-Provence, 1994, pp 26-28. 53. I. Wadso, Chem. Soc. Rev., (1997) 79. 54. I. Wadso and R. Goldberg, Pure Appl. Chem., 73 (200 1) 1625. 55. W. Hemminger and S. Sarge in P.K. Gallagher and M.E. Brown (Eds.), Handbook of Thermal Analysis and Calorimetry, Vol. 1, Elsevier, Amsterdam, 1999, Ch. 1. 56. L.D. Hansen, Thermochim. Acta, 371, (2001) 19. 57. T. Matsuo in M. Sorai (Ed.), Comprehensive Handbook of Calorimetry and Thermal Analysis, Wiley, New York, 2004, pp 63-85. 58. W. Zielenkiewicz, Calorimetry, Inst. Phys. Chem. of the Polish Academy of Sciences, Warsaw, 2005, pp 121-150. 59. E. Lange and K.P. Mishchenko, Zeits. Phys. Chem. A, 148 (1930) 161. 60. W. Zielenkiewicz and J. Chajn, Proceedings lStCalorimetry Conference, Zakopane, Inst. Phys. Chem. of the Polish Academy of Sciences, Warsaw, (1973) 1. 61 J.J. Christiansen and R.M. Izatt, Thermochim. Acta, 71 (1984) 117.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors 2008 Elsevier B.V.

Chapter 3

MICRO-THERMAL ANALYSIS AND RELATED TECHNIQUES Duncan M. Price Exhaust Management Systems, BOC Edwards, Kenn Business Park, Kenn Road, Clevedon, North Somerset, BS21 6TH, UK e-mail: [email protected] 1.

INTRODUCTION

The major difficulty with conventional thermal analysis techniques is that they measure the response of the whole sample. If one observes a broad change in behaviour on heating a specimen, this could be the result of a genuine effect in a homogeneous system or be due to a series of overlapping responses from a heterogeneous system, where there may be a gradation in properties throughout the sample. Alternatively, a weak effect seen in the entire sample could arise from a strong response from a minority component (e.g. an impurity) within in the bulk of the material. The same statements are true of other classical analytical techniques which require moderately sized amounts of material for sampling. It is not surprising, therefore, that many investigators have combined studies by thermal analysis with such techniques as optical [I] or electron microscopy. For example, Price and Bashir [2] studied the thermal behaviour of poly(acrylonitri1e) which had been compression moulded in the presence of water. Under conditions of high pressure, water acts as a solvent for this otherwise intractable polymer (which normally degrades on melting) allowing the material to flow. On cooling, the specimen has an open structure with the water trapped in pores within the polymer mass. Differential scanning calorimetry (DSC) was used to study the formation of ice within the pores on cooling. Analysis of the shape of the DSC curves show that the pore size followed an approximately bimodal distribution, whereas scanning electron microscopy indicated that a third category of larger pores existed which could not be detected due to the fact that they were incompletely filled with water [2]. Thermomicroscopy is the direct observation of a sample (usually) by optical microscopy (often with polarised light) as a function of temperature [I]. Frequently called "hot-stage (optical) microscopy", this is a relatively common

technique which is used widely for the study of historical artefacts [3], polymorphic transitions in drugs [4], pyrotechnics [5], and also for following the crystallisation of polymers [6,7]. Other forms of high-resolution imaging such as electron and scanning probe microscopies have been combined with the use of a variable temperature sample stage. Of particular relevance to this discussion is the use of atomic force microscopy at temperatures other than ambient [8,9]. Studies by Grandy et al. [lo] illustrate the potential of mechanical property imaging above and below transition temperatures of two phase polymer systems. Particularly novel is the application of a method of image analysis which seeks to assign a statistical probability of each pixel comprising the image to be one phase or the other (or an interphase). This is based upon the assumption that the contrast can be described using overlapping distributions of signal intensity arising fi-om each component. Fasolka et al. [ l 11 have published similar work using resonant frequency (rather than forced oscillation) non-contact atomic force microscopy where they also obtain images at a temperature between the glass-rubber transitions of a diblock polymer system. The major disadvantage of any method of thermomicroscopy (whether optical, electron or scanning probe technique) is that the whole sample is heated at the same time. Changes in the sample which occur outside the field of view are not observed and if these changes are irreversible (e.g. the sample crystallises, degrades, two phases mix or separate) then a new sample must be prepared. Ideally, one would desire a means to obtain a high-resolution image of a specimen under ambient conditions and then carry out thermal analysis of a specific region, phase or contaminant identified using the image as a guide. Reading [12], in a patent application filed in May 1992, proposed an instrument based upon a scanning thermal microscope modified to perform spatially resolved modulated(-temperature) differential thermal analysis. Although such an apparatus was never constructed, subsequent development in a collaboration with Pollock and Hammiche resulted in a working instrument and three more patents [13-151. Development and prototyping of this concept took place through collaboration between Loughborough and Lancaster Universities in the UK in conjunction with commercial and UK Research Council support. TA Instruments Inc. (Newcastle, Delaware, USA) launched this equipment as a commercial product in 1998. The instrument received the Gold Award at the 1998 PittCon analytical instrumentation conference and the R&D 100 award for innovation. In addition, the UK government awarded the instrument "Millenium Product" status. Before considering the development of what has become known as "microthermal analysis", it is appropriate to review the history, technology and

applications of scanning thermal microscopy so as to place this family of techniques in context.

2.

SCANNING THERMAL MICROSCOPY (STHM)

2.1. Introduction The inventions of the scanning tunnelling microscope (STM) [16] and the atomic force microscope (AFM) [17] have allowed sub-micrometre and, at times, atomic-scale spatially-resolved imaging of surfaces. Spatially-resolved temperature measurements using optical systems are diffraction limited by the wavelength of the radiation involved, which is about 5-10 ym for infrared thermography and about 0.5 ym for visible light [IS]. The spatial resolution of near-field techniques (such as AFM) is only limited by the active area of the sensor (which in the case of STM may be only a few atoms at the end of a metal wire). The first experiments in scanning thermal microscopy (SThM) were carried out by Williams and Wickramasinghe [20] who employed a heated thin-film thermocouple fabricated from a conventional STM tip. As the tip approached a surface, it was cooled due to tip-substrate heat transfer. By using the temperature sensed by the thermocouple as a feedback to maintain a constant tip-substrate gap, this scanning thermal profiler could overcome the limitations of STM and be used to image electrically insulating surfaces with a lateral resolution of 100 nrn. Since the feedback signal was based on maintaining a constant probe temperature, this device could not be used to obtain true thermal images of surfaces. Instead, the system measured a convolution of topography of the specimen and its thermal conductivity in a non-contact mode by means of heat leak from the thermocouple to the sample through a small air gap. In an attempt to overcome the limitations of this method of SThM, Majumdar [21] described the use of an AFM cantilever fashioned from a pair of dissimilar metal wires (chrome1 and alumel) which met to form a thermocouple junction at the tip. In this way, the conventional AFM force feedback mechanism could be used to measure surface topography whilst at the same time mapping the temperature distribution of energized electronic devices with sub-micrometre resolution. Since this demonstration, a number of different probe designs have been developed and progress has been made towards the measurement of absolute thermal conductivities and 3-dimensional tomographic imaging. A general discussion of SThM can be found in the reviews by Gmelin [22] and Majumdar [ 18,191.

2.2. Instrumentation for SThM The atomic force microscope forms the basis of both scanning thermal microscopy and instruments for performing localised thermal analysis. A schematic diagram of an AFM is shown in Figure 1 [23]. The instrument consists of a sharp tip mounted on the end of a cantilever which is scanned across the specimen by a pair of piezoelectric elements aligned in the x-y plane. As the height of the sample changes, the deflection of the tip in contact with the surface is monitored by an optical lever arrangement formed by reflecting a laser beam from the back of the cantilever into a photodetector [24]. The tip is then moved up and down by a feedback loop connected to a z-axis piezo which provides the height of the sample at each x,y position. Besides the topographic information provided by rastering the tip across the sample, other properties can be obtained by measuring the twisting of the cantilever as it is moved across the sample (lateral force microscopy) [25]. This provides image contrast based on the frictional forces generated from the sample-tip interaction. Other imaging modes, such as force modulation and pulsed force modes can indicate the stiffness of the sample [26]. An advantage of the AFM over the scanning electron microscope is that little sample preparation is required because the sample is not exposed to a high vacuum and electrically insulating materials can be examined.

sample

Figure 1. Elsevier.)

Atomic force microscope (schematic) [23]. (With permission from

2.3. Probe design Three methods have been used to combine the conventional AFM cantilever with a means of localised thermometry.

2.3.1. Thermocouple cantilevers The use of AFM with a thermoelectric element at the tip has been described above. In an effort to improve the performance of a bare thermocouple tip, Majurndar et al. [27] cemented a diamond shard to the junction so as to give a harder tip with improved spatial resolution and reduced thermal resistance. The same group also describe depositing successive layers of different metals so as to make thermocouple pair on top of a standard "A-fkame" AFM cantilever 1281. Fish et al. [29] borrowed technology from near-field scanning optical microscopy to make a thermocouple derived from gold-coated glass micropipettes containing a platinum core. Workers at Glasgow University [30,311 have fabricated thermocouple probes using electron beam lithography and silicon micromachining in order to deposit one or more thermocouple junctions at the AFM tip. Such work leads to the possibility of building thermopile sensors (with the incorporation of a heater) analogous to a miniature heat flux calorimeter 132,331. 2.3.2. Resistance thermometry

Figure 2. Schematic diagrams of resistive SThM probes: a) Wollaston wire type [34,35], b) micro-machined coated Si cantilever [37] and c) "data storage" doped Si probe [55]. In 1994, Dinwiddie and Pylkki [34,35] described the first combined SThMIAFM probes that employed resistance thermometry to measure thermal properties. These were fashioned from Wollaston process wire. This consists of a thin platinum/5% rhodium core (about 5 pm in diameter) surrounded by a thick (about 35 pm) silver sheath. The total diameter of the wire is thus about 75 pm. A length of wire is formed into a "V" and the silver is etched away at the apex to reveal a small loop of Pt/Rh which acts as a miniature resistance thermometer (Figure 2(a)). A bead of epoxy resin is added near the tip to act

serve as mechanical reinforcement and sliver of silicon wafer is glued to the top of the probe to act as a target mirror for the laser employed in the cantilever deflection detection mechanism [36]. This can be operated in two modes: a) as a passive thermo-sensing element (by measuring its temperature using a small excitation current) or b) as an active heat flux meter. In the latter case, a larger current (sufficient to raise the temperature of the probe above that of the surface) is passed through the wire. The power required to maintain a constant temperature gradient between the tip and sample is monitored by means of an electrical bridge circuit. In essence, this is equivalent to a power compensation calorimeter. Mills et al. [37] describe similar probes in which the resistance element is deposited across the apex of a silicon nitride pyramid similar to a conventional AFM cantilever (Figure 2(b)). In a passive mode, such devices function like thermocouple probes described above. These can be used (for example) to map the temperature distribution in energised electronic devices simultaneously with their topography [38,39]. If the surface is illuminated with infrared radiation, the photo-thermal effect arising from the absorption of energy specific to the infra-red (IR) active modes of the specimen may be used to obtain the sample's IR spectrum [40-471. In the active mode, the heat flow from the tip can be used to detect surface and subsurface defects of different thermal conductivity than the matrix [48,49]. Other resistive probe designs have been reported. For example, Li and Gianchandani [50] have fabricated SThM probes using polyimide rather than silicon as a substrate in a process similar to that of Mills et al. [37] except that it exploits the mechanical flexibility of the polyimide to implement an assembly technique that eliminates the need for probe removal or wafer dissolution. Use of polyimide rather than silicon as a substrate offers a greater degree of thermal isolation since its thermal conductivity is three orders of magnitude lower than silicon [5 I]. These probes have been manufactured in a differential arrangement with one probe being used as a reference. Edinger et al. [52,53] describe a sensor consisting of a nanometre-sized filament formed at the end of a piezoresistive atomic force microscope type cantilever. The freestanding filament is deposited by focussed electron beam deposition using methylcyclopentadienyl trimethyl platinum as a precursor gas. The authors claim a spatial resolution of < 20 nm and, due to its small thermal mass, a high sensitivity and fast response time. Leinhos et al. [54] have fabricated thermal elements from silicon with a Schottky diode integrated into the probe tip. One design of such a probe may also be used for simultaneous scanning near-field optical microscopy as well as SThM and AFM. Chui et al. [55] also describe high resolution silicon (piez0)resistive cantilevers designed not for SThM, but for high-density data storage. An array of probes are used to "write" 10-50 nm diameter pits in a spinning polymer coated silicon

disc [56]. Read-back is achieved by making use of the increased cooling of the tip when it encounters the indentations when the probes are operated in SThM mode. Such an approach is essentially the reverse of micro-thermal analysis in that the marks produced as a consequence of localised thermomechanical experiments are detected by the effect of their topography on the heat flux rate between the probe and the sample [57]. Schematic diagrams of the Wollaston, micromachined silicon and "data storage" probes are shown in Figure 2(a-c). The spatial resolution of these probes is of the order of 2, 0.2 and 0.02 pm respectively. All of these designs are now commercially available. 2.3.3. Bimetallic sensors Nakabeppu et al. [58] describe the use of composite cantilevers made from tin or gold deposited on conventional silicon nitride AFM probes to detect spatial variations in temperature across an indiumltin oxide heater. Differential thermal expansion of the bimetallic elements causes the beam to bend. This movement is monitored using the AFM optical lever deflection detection system. In order to separate thermal deflection of the beam from displacement of the cantilever caused by the sample topography, an intermittent contact mode of operation is employed. Measurements were made under vacuum so as to minimize heat loss. A more practical use of this technology is in the form of miniature chemical and thermal sensors [59]. This approach has been used to perform thermal analysis on picolitre volumes of material deposited on the end of a bimetallic cantilever [60]. Arrays of such devices have applications as highly sensitive electronic "noses". 2.4. Quantitative SThM Small-scale measurements of thermal conductivity and thermal difhsivity would benefit the semiconductor and other industries where thermal transport properties are significantly different to, and cannot be inferred from, measurements at higher scales. Examples of key areas of modem technology and science which might be expected to benefit include microelectronics, cellular biology, forensic, pharmaceutical and polymer science etc. In theory, heated thermal probes are capable of measuring the absolute thermal conductivity of materials by the heat flux between the tip and the surface. In practice, heat losses also occur both within the probe and to the atmosphere. Furthermore, the contact area between the tip and the specimen is usually unknown. Ruiz et al. [61] developed a simple method for converting heat flux to thermal conductivity by using hard materials of known thermal conductivity to calibrate the system. This procedure was used to determine the thermal conductivity of diamond-like nanocomposites to a precision of *15%.

Gorbunov et al. [62-641 measured the change in heat flux as the probe approached the sample surface, or was ramped in temperature in contact with the specimen, so as to derive its thermal conductivity - again by calibration with samples of known response. This approach has been used to study IR receptors in snakes with a view towards designing artificial sensors which mimic those found in nature [65]. Fiege et al. [66] used AC heating of the tip to measure the thermal conductivity of silver and diamond, using gold as a single-point reference material in order to estimate the contact area of the tip. Blanco et al. [67] have used SThM to investigate qualitative differences in thermal conductivity of carbon-carbon composites arising from different processing histories.

Figure 3. Illustration of the convolution of surface roughness on the apparent thermal conductivity (k) measured by the tip during a line scan. Darker coloured material (left) has lower bulk thermal conductivity than lighter coloured material. See text for additional explanation. The thermal conductivity contrast image obtained by scanning thermal microscopy represents a convolution of the true thermal transport properties of the specimen with artefacts arising fkom changing tip-sample thermal contact area caused by any surface roughness of the specimen [48]. When the probe encounters a depression on the surface, the area of contact between the tip and sample increases, resulting in increased heat flux from the tip to the sample. More power is required to maintain the tip temperature at the set-point value and

this is recorded on the image as an apparent increase in the local thermal conductivity. The opposite is true when the probe meets an asperity. Visual comparison of the thermal image with the topographic image (recorded simultaneously) often shows that "features" in the former are correlated with changes in height of the specimen - particularly at edges of such features where the surface relief changes rapidly. Even with careful sample preparation (e.g. cutting or polishing) it is almost impossible to avoid artefacts in the thermal image not attributable to topography. In cases where there is little inherent thermal contrast between parts of a heterogeneous specimen, interpretation of the thermal image can be problematic. An example of this is illustrated in Figure 3 for three cases: 1 - a homogenous material flat surface; 2 - the same material with a depression and an asperity; 3 - a smooth interface between low thermal conductivity and high thermal conductivity phases & 4 as 3, but with added surface roughness. In theory, it would be possible to construct an analytical model of the probesurface interaction based upon the local geometry of the probe tip and surface [62-641. Rather than develop a general model for heat transfer, an alternative approach can be developed using a neural net algorithm [68]. Each digitised image consists of a grid of points (pixels) which define the height of the sample and the heat flux from the probe to the sample. For each pixel the local topography is characterised by subtracting its height from the heights of the surrounding points. This defines whether the probe rests in a depression, asperity, flat surface, etc. The training set for the neural network comprises images acquired on materials of homogeneous thermal properties with a range of surfaces roughnesses. From these samples a table of topographic parameters correlated with the thermal measurement is obtained for a wide range of surface roughness. The 'true' thermal measurement is the value obtained on a perfectly smooth surface. In practice, an average value from the smoothest available surface can be taken as the 'true' value. In each case, the required value is given as the measurement on a smooth surface. The usual procedures for training, testing and validation of a neural network can be applied to result in a method to take a pair of images (thermal and topographic) and effectively "subtract" artefacts arising from surface roughness. As an alternative to lengthy training with a variety of specimens, it is feasible to use the images of a test specimen itself to train the network. This method relies upon the sample having discrete areas of homogenous thermal conductivity to provide internal calibration values for network training. Figure 4 shows the results of carrying out this type of image processing on a sample of polyolefin packaging film. Despite careful sample preparation, differences in mechanical properties result in a non-flat sample after cutting with a microtome. This generates false contrast in the raw thermal conductivity image which is largely

eradicated by processing with a neural net which was trained using a series of polymers of differing surface roughnesses.

Figure 4. Left - topography of microtomed cross-section through a multilayer polyolefin packaging film. Centre - raw thermal conductivity contrast image. Right - thermal conductivity image post-processing with trained neural net program. Other methods of image processing have been devised which apply a statistical analysis of pixel intensity distribution to enhance image contrast. Royal1 et al. [69] have used such a method to discriminate between the substrate and coating of a pharmaceutical compact whereby each pixel is defined to be one or the other component according to the heat flux from the tip to the sample. This simple "on/ofI" data treatment can be extended to assign a probability (displayed as a grey-scale level) of being a certain component based upon the position of the pixel's intensity within the total distribution of values. This procedure is illustrated using a non-prescription analgesic tablet containing paracetarnol (acetaminophen) described below.

Figure 5. (a) Raw thermal conductivity contrast image of an analgesic tablet containing paracetarnol [70], (b) histogram showing distribution of pixel intensity fitted to two overlapping Gaussian peaks assigned to the drug and filler components.

Figure 5(a) shows the raw thermal conductivity image of an analgesic tablet containing the drug paracetamol (acetaminophen). Localised thermal analysis of the sample indicates that the bright area in the bottom right of the image is some form of thermally inert filler whereas the dark regions comprise the drug [70]. Inspection of a histogram of pixel distribution in Figure 5(b) shows that the shape of this distribution can be modelled using two overlapping Gaussian peaks corresponding to the drug and filler responses.

Figure 6. (a) Simple black and white separation of image shown in Figure 5(a) using a threshold of 1.66 mW, (b) Grey-scale separation of image shown in Figure 5(a) using a more complete statistical analysis of the original histogram distribution in Figure 5(b).

A simple means of highlighting the two phases might be to assign a zero greyscale intensity to pixels with an original value below the cross-over between the two peaks at 1.66 mW and a grey-scale intensity of unity to all the pixels above this threshold. Thus, points which are statistically more likely to be comprised of the drug appear black and those which are more likely to be the filler appear white (Figure 6(a)). This is the same, simple method of image processing that has also been found to be particularly applicable to mechanical property imaging modes (pulsed force mode atomic force microscopy) in combination with a variable temperature sample stage [lo]. A more refined approach is to generate a grey-scale value for each pixel based upon the probability of it belonging to the drug or filler component derived from the ratio of the value of the fitted peak height assigned to the filler to the total fitted histogram intensity. The resulting transformation of the thermal image in Figure 5(a) via this process is shown in Figure 6(b) and exhibits a more satisfactory discrimination of phases -

particularly at the interface between the two domains where there exists some uncertainty over their assignment. De Cupere and Rouxhet [71] have recently published a similar method of image contrast enhancement based upon splitting a surface friction image into two components using the pixel intensity histogram. The resulting image was cleaned by selectively erasing any foreground pixel in contact with a background pixel ("erosion") and the final image was obtained by reconstruction of the objects left after erosion. This procedure, which is similar to that described above, was used to measure the surface fraction of spherulites grown on amorphous poly(ethy1ene terephthalate) film after different annealing times.

2.5. Other SThM techniques 2.5.1. 3-D tomographic imaging The decay length of thermal waves produced by AC heating of a tip varies as a function of the reciprocal of its frequency [72]. Thus, it is possible to detect variations in thermal response at shallower depths by using a high frequency temperature modulation superimposed on the conventional DC heating of the tip. Several groups have employed this technique to study the thermal diffusivity variations in materials [49,73,74]. Gombs et al. have examined this process theoretically [75] and there is potential for the use of multiple frequency modulated-temperature SThM as a means to provide non-destructive threedimensional imaging of optically opaque samples using similar principles to those employed for medical imaging by electrical impedance tomography [76791. 2.5.2. Thermal expansivity imaging An additional imaging mode has been demonstrated whereby the thermal expansion of a specimen is detected whilst AC heating is applied to the probe as it is scanned over the surface [23,70,80,81]. The resulting z-axis modulation of the probe arising from thermal expansion and contraction of the surface is detected and used to construct an image based on the thermal expansivity of the sample. Although Majumdar [82-841 has also described similar measurements, this approach does not require electrically conductive specimens and employs the same configuration used for thermal conductivity imaging. Again, there is the potential for using this approach for tomographic imaging.

3. LOCALISED THERMAL ANALYSIS 3.1. Principles Active, resistively-heated, thermal probes readily lend themselves to localised thermal analysis. Although it is feasible to control the temperature of a thermocouple-based probe by passing a current through it [86], the simultaneous measurement of its temperature is non-trivial and requires filtering out of the current providing the heating. Probes based on resistance thermometers are more readily controlled and this technology forms the basis of the well-known power-compensation differential scanning calorimeter [87]. Using a previously acquired topographic andlor thermal image obtained by the SThM facility, it is possible to place the probe at one or more selected locations in sequence on the sample and program the tip's temperature in order to make localised measurements of transition temperatures. By monitoring the power required to follow the temperature programme, a form of spatially resolved calorimetry may be carried out [49,89]. In addition, since the vertical deflection of the tip can be determined using the AFM stage, localized thermomechanical analysis (TMA) may be performed concurrent with the calorimetry [72,90]. A linear temperature ramp of the order of tens of degrees Celsius per second is the most commonly employed programme, often with a superimposed sinusoidal modulation of the probe temperature at kilohertz frequencies about the mean set-point. The use of AC heating allows two extra calorimetric signals to be obtained - the AC power and phase difference between the applied modulation and probe response - akin to AC calorimetry [88]. Such high heating (and cooling rates) are possible as a consequence of the small size of the probe and the region (typically a few pm square) of material that it contacts. The apparatus developed by Fryer [91] employs lower heating rates (of the order of a few degrees Celsius per minute) because of manual control of the probe set-point temperature and lack of automatic data logging. At present, only the analogues of DSC and TMA are commercially available. Localised dynamic mechanical analysis (DMA) has also been demonstrated [23,70,72,80,92]. Here, the force between the thermal probe is modulated during the temperature ramp. Such a procedure may also be used as an imaging mode in order to obtain a map of variations in mechanical properties across the specimen [80,92]. Localised modulated-temperature TMA has been performed whereby the amplitude and phase difference of the modulation in z-axis displacement of the probe is detected whilst a temperature ramp with overlaid AC modulation is applied to the tip [81]. An indirect form of thermogravimetry has been reported whereby the mass of material remaining adhered to the tip was monitored indirectly using the AC calorimetric signal [70]. By employing stiffer cantilevers with integrated heaters, the mechanical resonance frequency

of the beam has also been used to measure the mass of material on the tip during heating [93]. 3.2. Calibration [94] Localised thermal analysis has generally been used as a qualitative tool; that is, it is primarily used to identify the material being examined or for examining compositional gradients within materials. The temperature at which a transition takes place is often of primary interest, although it may be possible to add further semi-quantitative interpretation of the data from the magnitude of the change (e.g. amount of probe penetration) observed. Like all thermal analysis techniques, interpretation of results requires detailed temperature calibration procedures and an understanding of the precision of the temperature measurement. A good reference material has a number of desirable properties including a well-documented value, availability in a suitable form for analysis, homogeneity, stability, low toxicity, and traceability to a national reference laboratory (NRL). In traditional DSC and TMA, metals like indium, tin, and zinc meet these criteria. These metals are not suitable for temperature calibration for localised thermal analysis, however, as they may contaminate the probe tip thus changing its resistance and defeating the object of calibration.. Organic calibration materials are more suitable for calibration as they do not react with the probe and the tip may be easily cleaned at the end of an experiment by heating to above 500 "C in air. This is sufficient to decompose most organic substances. The British Laboratory of the Government Chemist (LGC), a national reference laboratory, conveniently offers eleven organic reference materials with melting temperatures ranging from 41 to 285 "C. However, in these materials are generally unsuitable to be used directly and large flat polycrystalline surfaces must be prepared by melting a small amount of substance in a suitable holder (such as small cup aluminium foil) and then cooled. Many investigators, particularly those working in the field of polymer science, prefer to use films of semi-crystalline thermoplastics such as poly(capro1actone or polyethylene terephthalate) as melting point standards. Whilst less desirable from a theoretical standpoint, temperature calibration with polymeric films offers a number of ease-of-use advantages. Moreover, many polymers have high melting temperatures that provide a calibration range of nearly 300 "C, a range difficult to achieve with organic chemicals. Generally, a two point temperature calibration is carried out spanning the experimental range of interest. Whatever protocol the investigator adopts, it is essential that the method is well-documented so that it can be replicated.

3.3. Features There are several important differences between the localised calorimetric and thermomechanical measurements and their more conventional "bulk" analogues. Firstly, for the calorimetric measurements, the sample size is poorly defined. This is because the contact area between the tip and specimen is ill-defined. Furthermore, should the sample soften during the measurement, the probe will sink into the specimen thus aggravating this effect. Therefore, there is often a strong correspondence between the shape of the calorimetric response and the displacement of the probe [95]. A second-order effect, not often considered, is that the temperature gradient extending away from the heated tip will be affected by the properties of the specimen and the heating rate employed during the measurement. This is analogous to the modulated temperature SThM mode of imaging which has been considered by Pollock and Hammiche [85] whereby the depth of penetration of the thermal wave emanating from the tip decays more rapidly for higher modulation frequencies. Slough [96] has considered the case of a heat pulse lasting the duration of the temperature programme of the probe for a typical polymeric sample. His calculations suggest that for an instantaneous heat pulse of 200 "C delivered for one second to the surface of a sample of polystyrene at 25 "C, the specimen's temperature is not significantly affected 10 ym fiom the probe. Inoue and Uehara [97] have modelled this process at a more sophisticated level in the context of data writing and erasing in a phase-change optical disk. Melting of a crystalline material at a part of the surface causes surface rippling around the molten area and subsequent rapid cooling generates an amorphous spot. This amorphous material is maintained at low temperature and subsequent localised thermal analysis can be used to determine the glass-rubber transition temperature of this region. The way in which localised thermomechanical measurements are performed also differs from conventional measurements. In the usual implementation of these measurements, the feedback loop between the cantilever force and the zaxis piezo control is disabled at the start of the experiment. Thus, the probe is initially applied to the surface with a user-defined cantilever deflection which will change (thus giving the displacement of the tip by means the optical lever formed by reflecting a laser spot from the back of the probe into a photodetector) during the course of the measurement. Should the specimen soften during the experiment, the probe will indent into the sample and the force on the tip will decrease (possibly to the extent of losing contact with the specimen). A final effect often observed during localised thermal analysis arises from the high heating-rates that can be employed. Many thermal transitions are governed by kinetic laws which define the time dependence of such processes as devitrification and thermal degradation. As a consequence of this, some

transitions can appear at elevated temperatures when compared to bulk measurements at more modest heating rates - even after careful calibration of the instrument. This effect is not generally seen for melting phenomena, and the rapid heating affords a means of measuring the melting temperature of metastable materials without them undergoing rearrangement to more stable forms [98].

50

100

150

200

250

C0.0 300

probe temperature I "C

Figure 7. Localised thermal analysis of semi-crystalline poly(ethy1ene terephthalate) showing two consecutive measurements at the same location (solid line is the probe displacement, broken line is the probe power, filled symbols denote first upwards temperature scan, open symbols denote second upwards temperature scan, heating rate 10 "C s-'). Some of the above aspects of the technique are illustrated in Figure 7. This shows two measurements performed consecutively on the same region of a sample of poly(ethy1ene terephthalate) of approximately 40% bulk crystallinity at a heating rate of 10 "C s-'. During the first upwards temperature scan, very little effect is seen in the probe displacement until melting of the polymer occurs around 250 "C. This is mirrored by an increase in probe power which is largely arises from the increased contact area between the tip and the sample as it indents the surface. A small decrease in probe power occurs around 125 "C, well above the "normal" glass-rubber transition temperature of this polymer of 70-80 "C. This feature might be ascribed to devitrification of amorphous material present. Immediately the first measurement was completed, the tip was lifted clear of the (now molten) material. The polymer in this area is rapidly quenched to room temperature by the surrounding specimen. On carrying out a second scan of the same region, the glass-rubber transition is observed readily in both signals, without any evidence for further rearrangement to crystalline material above this temperature or subsequent melting of any crystals so formed.

3.4. Terminology The nomenclature of this new method of analysis is still under development. The originators (Hammiche, Pollock and Reading) devised the terms "Calorimetric Analysis by Scanning Microscopy" (CASM) as a name for the measurement of probe power, and "Mechano-thermal Analysis by Scanning Microscopy" (MASM) as part of an effort to coin a systematic hierarchy of terms for this family of techniques [72]. The commercial term for this field is "pTA" which is a registered trade mark of TA Instruments Inc. (Newcastle, Delaware, USA) who also holds trademarks on the terms "pDTA" and "pTMA" corresponding to the calorimetric and thermomechanical methods [99]. Many authors have used the terms: "micro-thermal analysis" (with or without the hyphen, abbreviated to "micro-TA") and corresponding terms "micro-DTA" and "micro-TMA". Such usage goes against the recommendations of the International Confederation of Thermal Analysis and Calorimetry nomenclature committee who claim that there is ambiguity over the term "micro" because it is not clear whether the term is related to sample size (mass, volume), the size of the instrument, or the value of the measured signal (which may be amplified) or quantity detected [loo]. (At least one company markets a "micro-DSC" which is a high sensitivity instrument for measuring heat flow in the pW range.) The general term which seems to be gaining popularity is "local(ised) thermal analysis" and the derivations "local(ised) calorimetry" and "local(ised) thermomechanical analysis" in order to differentiate this technique from traditional forms of thermal analysis which measure the global response of the sample [loll. It this discussion, the term "micro-thermal analysis" (or "microTA") is used to denote SThM and localised measurements, whereas the individual measurements are described in full. Until there has been time for the terminology to develop and become accepted, searching the growing literature on this subject will be difficult. 3.5. Applications Since its development and subsequent commercialisation, localised thermal analysis has been employed in a number of different disciplines. A review by Pollock and Hammiche [85], summarises the wide range of materials which have been studied by these techniques. An earlier paper by Price et al. [I021 describes measurements on polymers, electronic components and biological specimens in the same publication as an illustration of the wide applicability of such measurements. Articles in popular technical publications have also served to illustrate this point [72,90, 103-1051.

3.5.1. Polymers

Localised thermal analysis has been used for the characterisation of multicomponent polymer thin films [101,109,110] or polymer blends [102,111-1131, for the investigation of surfaces and interfaces between materials [92,98,1141261, and compositional gradients brought about by the specimen's processing history [90,126-1311. In particular, localised thermal analysis is useful for probing the behaviour of polymers used for the fabrication of microelectronic components, where the bulk response of the polymer may not be representative of the same material when it is present in thin layers [126,132-1361. However, care must be taken to ensure that an apparent elevation in transition temperature is not due to the substrate acting as a heat sink, thus leading to errors in temperature measurement [137]. One interesting application of localised thermal analysis is to use the thermal probe for in-situ processing of materials, whereby heat is used to cross-link or decompose the substrate [126,138,139]. Again, this technology lends itself to data storage and micro-machining [140].

probe temperature I "C

Figure 8. Localised thermal analysis of multi-layer polyolefin film shown in Figure 4 [105]. Two examples of the use of localised thermal analysis are provided in order to illustrate the generic applications of this approach. Figure 8 shows localised thermomechanical analysis of the surface of the multi-layer film in Figure 4. Measurements were made at points within this image describing the bulk polymer, the central gas-barrier layer and the thin tie-layer between this and the bulk film. The melting transition temperatures are consistent with high density polyethylene, poly(ethy1ene-co-vinyl alcohol) and medium density polyethylene for the bulk, gas barrier and tie layers, respectively [loll.

Figure 9. Optical micrograph of a cross-section of a "gel" particle embedded in a 75 pm low density polyethylene film. Also visible are craters remaining following localized thermal analysis of the specimen. Another example is illustrated in Figures 9 and 10 for a blemish or "gel" particle in a blown polyethylene film. Figure 9 shows an optical micrograph of the film which has been carehlly cross-sectioned across the feature. Also visible are small craters in the film which are a consequence of a series of localised thermal analyses along the specimen. These also serve to illustrate the potential spatial resolution of the technique. Whilst the Wollaston wire probe has a diameter of 5 pm at the tip and is capable of detecting thermal transitions in the order of a few square micrometres, heat from the measurement process spreads out and disrupts a large region (about 20 pm square) around the tip. This ultimately restricts the proximity of a sequence of measurements on a specimen. For the analysis of the sample shown in Figure 9 the specimen was translated under the instrument using a micrometer stage. Over an order-ofmagnitude improvement (both in terms of sampling area and proximal placement of multiple analyses) can be achieved using semi-conductor probes such as those shown in Figure 2(c). These have recently become commercially available and promise to move localised thermal analysis into the nano-scale with sub-micrometre resolution.

probe temperature I "C

Figure 10. Results from localised thermal analysis of the specimen shown in Figure 9. The solid symbols denote measurements on the normal film whereas the open symbols denote measurements made in the "gel" particle [103]. Figure 10 shows results measurements made on the bulk film and in the region of the defect. Whilst it can be observed that the temperatures of the onset of probe penetration into the low density polyethylene film are broadly similar for all measurements. The degree of penetration into the sample into the "gel" particle is much lower. This implies that the material here has a much higher melt viscosity than the normal perhaps as a consequence of some problem during polymer synthesis [104]. 3.5.2. Pharmaceuticals The applications of micro-thermal analysis within the pharmaceutical industry have been reviewed by Craig et al. [141]. Most tablets are not composed of the pure active ingredient, but contain the drug dispersed within an excipient package (such as micro-crystalline cellulose, starch, glucose etc.) with added processing aids (such as magnesium stearate) which act as fillers and lubricants for the formation of a mechanically stable compact. Furthermore, the tablet may be coated with a polymer or sugar film to prevent the drug being released into the body before it enters the stomach [142]. Imaging by SThM can be used to identify discrete phases containing these ingredients and their identification can be confirmed by localised thermal analysis [23,69,70, 104,143-1471. Figure 11 illustrates this by means of measurements on the specimen shown in Figure 5(a). Localised thermal analysis detects the melting of the drug paracetamol

(acetaminophen) around 180 "C for the low thermal-conductivity region of the image. No change in response is observed when an area from the high thermalconductivity region of the same image is examined. This suggests that the material here is comprised of a filler or excipient - most probably microcrystalline cellulose. The distribution of drug within a pharmaceutical compact can have very important implications for the dissolution of the tablet within the digestive system and micro-thermal analysis is a useful characterisation tool.

- drug (9.8,27.2) ---- excipient (41.2,39.4)

-6 -I 100

120

140

160

180

200

220

probe temperature I "C

Figure 11. Localised thermal analysis of the analgesic tablet shown in Figure 5(a) with measurements made in the low thermal conductivity (dark) region (solid line) and the high thermal conductivity (bright) region (broken line) of the image [70]. Many drugs can be produced in more than one crystalline modification. Microthermal analysis has been shown to be a viable means of differentiating between such polymorphs [148,149], or between crystalline and amorphous regions in drugs [150,1511. Royal1 et al. [I521 have employed localised thermal analysis to detect surface segregation of progesterone encapsulated in poly(1actic acid) microspheres (thus confirming earlier work by DSC), while Zhang et al. have studied the distribution of poly(ethy1ene glycol) in the same polymer [153].

3.5.3. Biology Micro-thermal analysis has been used to examine a number of specimens of biological origin. Gorbunov et al. have reported the study of the infrared receptors of snakes using SThM and localised thermal analysis, whereby the

thermal transport properties of the receptors were found to be lower than the surrounding skin [65]. The surfaces of plant leaves have also been examined by localised thermal analysis, to study the waxy coating which protects the leaf from water loss. Melting of the cuticular wax has been detected [102] and the effect of surfactant packages on this behaviour investigated as a means of improving the absorbtion of agrochemicals [154]. Studies of historical parchment derived fiom the dermis of animal skin have been employed to measure the softening temperature of the material. It is a particular advantage of the technique to be able to examine small quantities of material [155]. SThM has been used in food science to image the surface of caramel [156]. In this instance pulsed-force microscopy and infrared spectroscopy were employed to characterise surface structures, although it would be a natural extension to use the SThM to study local thermal transitions 3.5.4. Inorganic materials This category includes metals, ceramics and electronic materials which are typically of high thermal conductivity whereby the discrimination between phases afforded by SThM can be compromised [157]. Furthermore, it is difficult to envisage that point-source heating afforded by active thermal probes would be sufficient to heat highly-conductive bulk specimens such as metals. However, thin-film NiTi shape-memory alloys have been studied by localised thermal analysis, whereby measurements of the martensitic to austenitic transformation were made and the spatial variation in transition temperature corresponding to compositional variations within the specimen identified [15 81. Micro-thermal analysis has also been employed to resolve differences in thermal conductivity and softening temperatures that arise during the processing of carbon fibres [67,129,159]. These were related to the local oxygen content of the fibre measured by electron probe micro-analysis.

Figure 12. Shaded topography (left) and thermal conductivity contrast (right) images of a light emitting diode [102]. (With the permission of AkadCmiai Kiad6.)

probe temperature l "C

Figure 13. Localised thermal analysis of the LED shown in Figure 12. The upper set of curve show measurements made within the high thermal conductivity central region and the lower set of curves show measurements made outside this area [102]. (With the permission of AkadCmiai Kiad6.) As described earlier, passive SThh4 techniques have proven popular in the microelectronics industry for the identification of hot-spots within components. For example, Boroumand et al. have made measurements of the temperature distribution across a polymer light-emitting diode (LED) [160]. Figure 12 shows the topography and thermal conductivity contrast image of a siliconbased LED from a batch of components which had failed under testing. Localised thermomechanical measurements of the centre and outside of specimen are shown in Figure 13. Although no thermal transitions are observed, the thermal expansivity is different between the two areas. This would lead to thermal stresses building up when to the device was energised. This could ultimately lead to failure of the LED. Measurements on a LED which passed testing showed no variation in properties [I 02,1041.

4. LOCALISED CHEMICAL ANALYSIS

4.1. Introduction The measurements physical properties afforded by existing forms of microthermal analysis can be insufficient to discriminate between different materials. Incorporation of some means of chemical analysis of the specimen is therefore highly desirable. This has been achieved by two processes; localised pyrolysisevolved gas analysis, and near-field photothermal spectroscopy. 4.2. Localised evolved gas analysis 4.2.1. Introduction It has been demonstrated that the Wollaston wire probe tip used for localised thermal analysis measurements can be heated rapidly and repeatedly to temperatures in excess of 600 "C. This is sufficient to bring about localised pyrolysis of most organic materials [72,157]. This process generates a small plume of gaseous decomposition products characteristic of the substrate. Chemical analysis of these products can be performed via two alternative routes: by absorbing them on a suitable substrate and subsequent investigation using thermal desorption gas chromatography-mass spectrometry (td-GC-MS), or by direct sampling by mass spectroscopy (MS) [23,70,106,162-1651. Furthermore, the thermal probe may be used to soften and remove material from the surface of the specimen for subsequent characterisation [70]. These approaches are described in more detail in the following sub-sections.

4.2.2. Offline localisedpyrolysis-td-GC-MS Realisation of the first approach - trapping and offline analysis - employs a miniature gas-sampling tube packed with a mixture of TenaxB (molecular sieve) and CarbopakB (activated charcoal) absorbent material. Such tubes are routinely used for environmental monitoring of hazardous industrial atmospheres whereby operators during the normal course of their duties carry a small tube (about the size of a pen) clipped to their clothes. A pump may be used to draw gas through the tube at a controlled rate and, at the end of the work period, the tube is sealed and sent for analysis. Heating the sorbent tube drives off the trapped material into a gas chromatograph for separation and quantification. For micro-pyrolysis-td-GC-MS, the sorbent tube is modified to end in a short section of stainless steel hypodermic tubing the open end of which can be placed immediately adjacent to the heated thermal probe using a micro-manipulator. As the tip is heated, a pump is used to draw gas through the tube. After sampling, the tube is placed in a suitable carrier that fits into a standard thermal desorption unit interfaced to a GC-MS system. Blank desorption runs of the

sorbent tube are carried out before and after each pyrolysis experiment to confirm the cleanliness of the detection system. Such tubes are re-usable since the thermal desorbtion process cleans them of trapped material. The lifetime of such tubes is at least 1000 cycles. A dedicated design of micro-manipulator is used for positioning the sampling tube which can be interfaced with a variable-temperature sample stage upon which the microscope is placed. Thus, the sample can be cooled or heated independently of the thermal probe, using a small heated sample holder coupled to a Dewar vessel containing liquid nitrogen for cooling the specimen. Generally, the specimen is only required to be under ambient conditions. Therefore, an x-y-z translator can be used in place of the heated stage. This configuration is very versatile and affords independent positioning of the sampling tube, microscope and sample. One of the obvious benefits of off-line trapping and analysis of pyrolysis products is the ability to take samples of evolved gases from more than one location on the sample. For example, if the region of interest covers a sufficiently wide area, then multiple points may be selected for pyrolysis and the evolved gases gas can be trapped in the same tube, thus increasing the yield of material for subsequent analysis. Alternatively, line or area scans of surfaces may be made with a heated probe to drive off any volatiles into the tube [165]. This approach has particular benefits for filled systems, such as paints and coatings, which may contain only a small fraction of organic binders. Another benefit of analysis by GC-MS is to use the ability of gas chromatography to separate the mixture of decomposition products yielded by all but the simplest of substrates. This allows complex systems (such as paints and coatings described above) to be identified or at least "fingerprinted" by the characteristic mixture of volatile materials formed during thermal degradation [166]. Alternatively, one may only be interested in the presence (or absence) of a particular component at a specific location. For example, this technique has been used to locate the source of camphor extracted from plant leaves [164]. It is possible to exploit this technique as a means of surface-specific pyrolysis. For example, Figure 14 shows total-ion chromatograms for the decomposition products from samples of the same household paint. The top curve ("macroEGA") shows material collected from a bulk sample (about 1 mg) of paint using a thermobalance to decompose the material and gas sampling tube in the purge gas outlet to collect the evolved gases. The lower curve ("micro-EGA") shows material collected by scanning a hot thermal probe over at 25 x 50 pm area of the surface of the paint. Both curves show common features arising from the binder in the paint (polyamide and acrylic polymers), but also differences between the surface and the bulk indicating a reduction in antioxidants and

volatile plasticisers (particularly the peak at 21.3 min due to dibutyl phthalate) at the exposed surface 11651.

1041 0

5

10

15

20

25

retention time / rnin

Figure 14. Comparison of evolved gases from bulk (macro-EGA) and surface (micro-EGA) of paint film [165].

4.2.3. Online localisedpyrolysis-MS As an alternative to analysis of evolved gases by GC-MS, mass spectroscopy by itself can be used. This has the disadvantage of lacking the specificity given by the chromatographic separation, but the advantage of considerably reducing the time for analysis. One of the drawbacks of the td-GC-MS approach is the time taken for the analysis of collected gases from the micro-pyrolysis experiment. The thermal probe itself may be heated at over 100 OC/s to the required pyrolysis temperature and several locations may be examined within a few minutes. Desorption and separation of the trapped material is limited by the cycle time of the GC-MS system. Even with an optimised oven programme for the GC, a typical analysis can take 30 minutes. Furthermore, it is advisable to conduct a "blank" experiment with the tube prior to sampling in order to ensure that no residues are present in the system. One way around this problem is to dispense with the trapping and separation stage by continuous sampling of the atmosphere around the thermal probe. This can be achieved by using a small-bore silica-glass capillary transfer line which also serves to reduce the pressure from atmospheric to that which could be accommodated by the ion source of the mass spectrometer. The capillary tube

must be surrounded by a heated jacket for most of its length in order to prevent condensation of volatiles on the walls of the tube. Only a few centimetres of the end of the capillary are left unheated - these were passed through an empty micro-sorbent tube (described above) so as to enable the same micromanipulator assembly to be used to position the end of the transfer line close to the thermal probe. For online micro-pyrolysis-MS, three modes of pyrolysis have been developed. Firstly, the temperature of the probe may be rapidly pulsed to the required temperature - the amount of material liberated depending upon the duration and temperature of the heat pulse. Secondly, a conventional temperature ramp can be applied to the probe whilst monitoring for evolved species. This approach has the benefit that the operator can select several locations within the field of view of the microscope in order to carry out compositional mapping via evolved gas detection. Finally, a heated thermal probe may be brought into contact with a specimen whilst monitoring gas evolution. As the heat source nears the surface, material is progressively decomposed, affording a means of depthprofiling though the sample so as to reveal buried layers beneath the surface [106,164]. Alternatively, successive pyrolysis measurements may be made using either of the first or second methods of heating the probe with either direct sampling to the mass spectrometer, or the more time-consuming offline tdGCMS sampling [106]. As stated earlier, online MS sampling of evolved gases lacks the separation stage afforded by td-GC-MS and is unsuitable for complex systems. Furthermore, the sensitivity of the technique is improved by monitoring for specific species rather than collecting mass spectra across a wide mass range during pyrolysis. For example, materials containing aromatic species often evolve benzene amongst their decomposition products, whereas aliphatic materials often give alkene fragments. These may be detected by monitoring single-ion masses rather than acquiring a full mass spectrum. The ability to examine a succession of points within a few minutes can enable a compositional map of the specimen to be obtained, which shows the spatial distribution of phases in a specimen via its pyrolysis products [165]. An example of this is shown in Figure 15 for a poly(methy1 methacrylate)/polystyrene laminate. Data from a 6 x 6 array of pyrolysis measurements were used to reconstruct images based upon the ion yield of the respective monomers. Unlike similar methods of chemical imaging (e.g. secondary ion mass spectrometry and laser ionisation mass spectrometry), the sample is examined under ambient conditions rather than high vacuum [167]. Three-dimensional tomographic imaging may also be considered by using the probe to ablate the surface.

Figure 15. Left: topographic image of poly(methylmethacry1ate) /polystyrene laminate - polystyrene layer is to the right of the image. Centre: evolved gas "image" reconstructed from multiple pyrolysis measurements monitoring for methyl methacrylate monomer (molecular ion m/z 100). Right: similar image reconstructed for styrene (m/z 104) [165]. 4.3. Near-field photothermal spectroscopy When a material is illuminated with infrared radiation it will tend to absorb energy corresponding to specific infia-red (IR) active modes and increase in temperature. This is known as the "photo-thermal effect" and can be exploited to obtain the IR spectrum of a specimen. Passive temperature-sensing SThM probes can be used to measure the specimen response to IR radiation over a very small area below that realisable using conventional optics because the spatial resolution is limited largely by the contact area of the tip and the thermal difisivity of the specimen being of the order of 1 pm which is at least one order-of-magnitude improvement over conventional techniques. Near-field photothermal spectroscopy using resistive probes of the types shown in Figure 2(a-b) has been exploited by Hammiche and co-workers [40-471, who used a mirror system to bring the IR beam to focus on the specimen surface while it was in the SThM. Two approaches have been used to generate the spectrum. In the first method a high-intensity tunabIe IR source has been used to scan the spectrum in a manner analogous to dispersive IR spectroscopy [42]. The second approach uses a broadband source in an interferometer and subsequent Fourier transform analysis to obtain the spectrum [40]. The results of this method, in addition to the work described above, have also been combined with measurements by SThM and localised thermal analysis. Polymers [106] and pharmaceuticals [23] represent the largest classifications of materials that have been investigated, although this technique has been used in cellular biology to monitor the life-cycles of cells [168].

4.4. Thermally-assisted micro-sampling

time 1s Figure 16. Online MS single-ion monitoring for methyl methacyrlate ion fragment m/z = 69) removed from the monomer (using the CH~C(CH~)CO'' surface as a consequence of heating the probe in contact with the specimen.

The ability of the heated thermal probe to soften the surface of a specimen and remove a small amount of material has also been demonstrated. By placing the probe on the sample and raising its temperature, the substrate is softened [70]. As the probe is withdrawn from the sample, some material adheres to tip. This residue can then be characterised by pyrolysis-evolved gas analysis (td-CG-MS or MS). Figure 16 illustrates this process using online-MS monitoring for the evolution of monomer from a specimen of poly(methy1 methacrylate). The tip is first heated rapidly to 1000 "C to ensure that it is free of any contaminants. Then, the probe is brought into contact with the specimen and the tip is heated to 400 "C at 10 "C s-', whereupon a small amount of monomer is evolved as the hot tip is removed from the surface. A second probe-cleaning cycle results in a larger amount of monomer being detected as a result of the decomposition of material adhering to the tip. Subsequent cleaning of the tip by heating to 1000 "C does not result in the detection of any residue. Experiments indicate that it is only necessary to heat the specimen above its glass-rubber transition or melting temperature in order to transfer material from the surface to the tip. Analogous measurements have been demonstrated using photothermal IR spectrometry to detect material removed in this way [169]. Thermal dip-pen nanolithography

using a heated AFM tip has also been developed [170]. In this case, the tip is coated with an organic material which is transferred to the surface by heating the tip so as to melt the "ink". This is essentially the inverse of the micro-sampling process described above.

5. CONCLUSIONS This chapter presents an overview of micro-thermal analysis. Scanning thermal microscopy has gained acceptance in many areas of physical science. The commercial availability of instrumentation will continue to broaden its scope of usage into more areas of characterisation of materials. A better understanding of the mechanisms of heat transport from the tip to the surface can be expected to make routine measurements of absolute thermal properties via scanning thermal microscopy possible. The thermodynamic limit of measurement (kgis about lo-'' J at room temperature [27]. The current spatial resolution of scanning probe microscopy is around lo-'' m. The maximum temperature resolution of the most sensitive thermal probes (bimetallic cantilevers) is K with an estimated sensitivity limit of =lo-l2 J and a spatial resolution of m. There is, therefore, plenty of room for improvement. Advances in thermal probe design are also expected to lead to applications for localised thermal analysis, thus enabling the behaviour of materials to be investigated over even smaller dimensions. The integration of chemical analysis (by pyrolysis andlor near-field photothermal infrared microscopy) raises the possibility of constructing a versatile instrument capable of performing a wide range of analyses that exploits abilities of the thermal probe to act as a very small heater and thermometer. Such an instrument might well be termed "the laboratory on a tip". 6. REFERENCES 1. H. G. Wiedemann and S. Felder-Casagrande, in M. E. Brown (Ed.), Handbook of Thermal Analysis and Calorimetry. Vol. 1: Principles and Practice, Elsevier Science B. V., Amsterdam, 1998, Ch.l0,473. 2. D. M. Price and Z. Bashir, Thermochim. Acta, 249 (1995) 351. 3. J. H. Townsend, Thermochim. Acta, 365 (2000) 79. 4. J. 0. Henck, J. Bernstein, A. Elern and R. Bose, J. Am. Chem. Soc., 123 (2001) 1834. 5. B. Berger, A. J. Brammer and E. L. Charsley, Thermochim. Acta, 269 (1995) 639.

6. S. Andjelic, D. Jailolkwski, J. McDvitt, J Fischer and J Zhou, J. Polym. Sci. Polym. Phys., 39 (2001) 3073. 7. J. Wagner and P. J. Phillips, Polymer, 42 (2001) 8999. 8. C. Basire and D. A. Ivanov, Phys. Rev. Lett., 85 (2000) 5587. 9. Y. K. Godovsky and S. N. Magonov, Polym. Sci., A 43 (2001) 647. 10. D. B. Grandy, D. J. Hourston, D. M. Price, M. Reading, G. Goulart Silva, M. Song and P. A. Sykes, Macromolecules, 33 (2000) 9348. 11. M. J. Fasolka, A. M. Mayes and S. N. Magonov, Ultramicroscopy, 90 (2001) 21-31. 12. M. Reading, US Pat., 5,248,199 (1993). 13. A. Hammiche, H. M. Pollock and M. Reading, US Pat., 6,095,679 (2000). 14. A. Hammiche, H. M. Pollock and M. Reading, US Pat., 6,200,022 (2001). 15. A. Hammiche, H. M. Pollock, M. Reading and M. Song, US Pat., 6,491,425 (2002). 16. G. Binnig and H. Rohrer, Helv. Phys. Acta, 55 (1982) 726. 17. G. Binnig, C.F. Quate and Ch. Gerber, Phys. Rev. Lett., 56 (1986) 930. 18. A. Majumdar, Ann. Rev. Mat. Sci., 29 (1999) 505. 19. L. Shi and A. Majumdar, Microscale Therm. Eng., 5 (2001) 25 1. 20. C.C. Williams and H. K. Wickramasinghe, Appl. Phys. Lett., 49 (1986) 1587. 21. A. Majumdar, J.P. Carrejo and J. Lai, Appl. Phys. Lett., 62 (1993) 2501. 22. E. Gmelin, R. Fisher and R. Stitzinger, Thermochim. Acta, 3 10 (1998) 1. 23. D. M. Price, M. Reading, A. Hammiche and H. M. Pollock, Int. J. Pharm., 192 (1999) 85. 24. G. Meyer and A. M. Amer, Appl. Phys. Lett., 53 (1988) 1045. 25. J. S. G. Ling and G. J. Leggett, Polymer, 38 (1997) 2617. 26. A. Rosa-Zeiser, E. Wieland, S. Hild and 0. Marti, Meas. Sci. Technol., 8 (1997) 1333. 27. A. Majumdar, J. Lai, M. Chandrachood, 0. Nakabeppu, Y. Wu and Z. Shi, Rev. Sci. Instrum., 66 (1995) 3584. 28. K. Lou, Z. Shi, J. Varesi and A. Majumdar, J. Vac. Sci. Technol. B, 15 (1997) 349. 29. G. Fish, 0. Bouevitch, S. Kokotov, K. Lieberman, D. Palanker, I. Turovets and A. Lewis, Rev. Sci. Instrum., 66 (1995) 3300. 30. G. Mills, H. Zhou, A. Midha, L. Donaldson and J.M.R. Weaver, Appl. Phys. Lett., 72 (1998) 2900. 3 1. H. Zhou, G. Mills, B.K. Chong, A. Midha, L. Donaldson and J.M.R. Weaver, J. Vac. Sci.Techno1. A, 17 (1999) 2233. 32. B. Wunderlich, Thermochim. Acta, 355 (2000) 43. 3 3 . 0 . Nakabeppu and T. Suzuki, J. Therm. Anal. Cal., 69 (2002) 727. 34. R.B. Dinwiddie, R.J. Pylkki and P.E. West, in T.W. Tong (ed.), Thermal Conductivity 22, Technomics, Lancaster PA, 1994, pp. 668.

35. R. J. Pylkki, P. J. Moyer and P. E. West, Jap. J. Appl. Phys., 33 (1994) 784. 36. D. Sarid, Scanning Force Microscopy, Oxford University Press, Oxford, 1994. 37. G. Mills, J. M. R. Weaver, G. Harris, W. Chen, J. Carrejo, L. Johnson and B. Rogers; Ultramicroscopy, 80 (1999) 7. 38. A. Buck, B. K. Jones and H. M. Pollock; Microelectron. Reliab., 37 (1997) 1495. 39. Y. Ji, Z. G. Li, D. Wang, Y. H. Cheng, D. Luo and B. Zong, Microelectron. Reliab., 41 (2001) 1255. 40. A. Hammiche, H.M. Pollock, M. Reading, M. Claybourn, P.H. Turner and K. Jewkes, Appl. Spectrosc., 53 (1999) 810. 41. A. Harnmiche, L. Bozec, M. Conroy, H, M. Pollock, G. Mills, J. M. R. Weaver, D. M. Price, M. Reading, D. J. Hourston and M. Song, J. Vac. Sci. Technol. B, 18 (2000) 1322. 42. L. Bozec, A. Hammiche, H. M. Pollock, M. Conroy, J. M. Chalmers, N. J. Everall and L. Turin, J. Appl. Phys., 90 (2001) 5 159. 43. L. Bozec, A. Hammiche, H. M. Pollock and M. Conroy, Anal. Sci., 17 (2001) S494. 44. M. Claybourn, A. Hammiche, H. M. Pollock and M. Reading, US Pat., 6,260,997 (2001). 45. L. Bozec, A. Hammiche, M. J. Tobin, J. M. Chalmers, N. J. Everall and H. M. Pollock, Meas. Sci.Technol., 13 (2002) 1217. 46. A. Hammiche, L. Bozec, H. M. Pollock, M. German and M. Reading, J. Microsc.-Oxford, 2 13 (2004) 129. 47. A. Harnmiche, L. Bozec, M. J. German, J. M. Chalmers, N. J. Everall, G. Poulter, M. Reading, D. B. Grandy, F. L. Martin and H. M. Pollock, Spectroscopy, 19 (2004) 20. 48. A. Hammiche, H.M. Pollock, M. Song and D.J. Hourston, Meas. Sci.Technol., 7 (1996) 142. 49. A. Hammiche, D. J. Hourston, H. M. Pollock, M. Reading and M. Song, J. Vac. Sci. Technol. B, 14 (1996) 1486. 50. M-H. Li and Y. B. Giachandani, J. Vac. Sci. Technol. B, 18 (2000) 3600. 5 1. J. H. Lee and Y. B. Giachandani, Rev. Sci. Instrum., 75 (2004) 1222. 52. K. Edinger, T. Gotszalk and I. W. Rangelow, J. Vac. Sci. Technol. B, 19 (2001) 2856. 53. I. W. Rangelow, T. Gotszalk, N. Abedinov, P. Grabiec and K. Edinger, Microelectron. Eng., 57-8 (2001) 737. 54. T. Leinhos, M. Stopka and E. Oesterschulze, Appl. Phys. A - Mater., 66 (1998) S65. 55. B. W. Chui, T. D. Stowe, T. W. Kenny, H. J. Mamin, B. D. Terris and D. Rugar, Appl. Phys. Lett., 69 (1996) 2767.

56. W. P. King, T. W. Kenny, K. E. Goodson, G. L. W. Cross, M. Despont, U. T. During, H. Rothuizen, G. Binnig and P. Vettiger, J. Microelectromech. Sys., 11 (2002) 765. 57. G. Binnig, M. Despont, U. Drechsler, W. Haberle, M. Lutwyche, P. Vettiger, H. J. Mamin, B.W. Chui and T.W. Kenny, Appl. Phys. Lett., 74 (1999) 1329. 5 8 . 0 . Nakabeppu, M. Chandrachood, Y. Wu, J. Lai, and A. Majumdar, Appl. Phys. Lett., 66 (1995) 694. 59. J. K. Gimzewski, Ch. Gerber, E. Meyer and R. R. Schlittler, Chem. Phys. Lett., 217 (1994) 589. 60. R. Berger, Ch. Gerber, J. K. Gimzewski, E. Meyer and H. J. Giintherodt, Appl. Phys. Lett., 69 (1996) 40. 61. F. Riuz, W. D. Sun, F. H. Pollak and C. Venkatraman, Appl. Phys. Lett., 73 (1998) 1802. 62. V. V. Gorbunov, N. Fuchigami, J. L. Hazel and V. V. Tsukruk, Langmuir, 15 (1999) 8340. 63. V. V. Gorbunov, N. Fuchigami and V. V. Tsukruk, Probe Microsc., 2 (2000) 53. 64. V. V. Gorbunov, N. Fuchingarni and V. V. Tsukruk, Probe Microsc., 2 (2000) 65. 65. V. Gorbunov, N. Fuchigami, M. Stone, M. Grace and V. V. Tsukruk, Biomacromol., 3 (2002) 106. 66. G. B. M. Fiege, A. Altes, R. Heiderhoff and L.J. Balk, J. Phys. D, Appl. Phys., 32 (1999) L13. 67. C. Blanco, S. P. Appleyard and B. Rand, J. Microsc.- Oxford, 205 (2002) 21. 68. M. Reading and D. M. Price, US Pat. Appl. 20030004905 (2003). 69. P. G. Royall, D. Q. M. Craig and D. B. Grandy, Thermochim. Acta, 380 (2001) 165. 70. D. M. Price, M. Reading, A. Hammiche and H. M. Pollock, J. Them. Anal. Cal., 60 (2000) 723. 71. V. M. De Cupere and P. G. Rouxhet, Polymer, 43 (2002) 5571. 72. M. Reading, D. J. Hourston, M. Song, H. M. Pollock and A. Hammiche, Am. Lab., 30(1) (1998) 13. 73. E. Oesterschultze, M. Stopka, L. Ackerman, W. Sholtz and S. Werner, J. Vac. Sci. Technol. B, 14 (1996) 832. 74. L. J. Balk, M. Maywald and R. J. Pylkki, 9th Conf. on Microscopy of Semiconducting Materials (Inst. Phys. Conf. Ser. 146), Oxford, 1995, pp. 655. 75. S. Gom&s,F. Depasse and P. Grossel, J. Phys. D, Appl. Phys., 31 (1998) 2377. 76. P. Metherall, D. C. Barber, R. H. Smallwood and B. H. Brown, Nature, 380 (1996) 509.

77. R. Smallwood, P. Metherall, D. Hose, M. Delves, H. Pollock, A. Hammiche, C. Hodges, V. Mathot and P. Willcocks, Thermochim. Acta, 385 (2002) 19. 78. A. Hammiche, M. Reading, H. M. Pollock, M. Song, and D. J. Hourston, Rev. Sci. Instrum., 67 (1996) 4268. 79. D. P. Almond and W. Peng, J. Microsc.-Oxford, 201 (2001) 163. 80. A. Hammiche, D. M. Price, E. Dupas, G. Mills, A. Kulik, M. Reading, J. M. R. Weaver and H. M. Pollock, J. Microsc.-Oxford, 199 (2000) 180. 81. D. M. Price, J. Therm. Anal. Cal., 64 (2001) 323. 82. A. Majumdar, Microelectron. Reliab., 28 (1998) 559. 83. A. Majumdar and J. Varesi, J. Heat Transf., 120 (1998) 297. 84. J. Varesi, and A. Majumdar, Appl. Phys. Lett., 72 (1998) 37. 85. H. M. Pollock and A. Hammiche, J. Phys. D Appl. Phys., 34 (2001) R23. 86. R. P. Miller and G. Somner, J. Sci. Instrum., 43 (1966) 293. 87. M. J. O'Neill, Anal. Chem., 36 (1964) 1238. 88. Y. Kraftmakher, Phys. Rep., 356 (2002) 3-1 17. 89. D. S. Fryer, J. J. dePablo and P. F. Nealey, Proc. SPIE, 3333 (1998) 1031. 90. D. M. Price, M. Reading, A. Caswell, H. M. Pollock and A. Hammiche, Eur. Microsc. Anal., 65 (1998) 17. 9 1. R. S. Tate, pers. comm. 92. F. Oulevey, N. A. Burnham, G. Gremaud, A. J. Kulik, H. M. Pollock, A. Hammiche, M. Reading, M. Song and D. J. Hourston, Polymer, 41 (2000) 3087. 93. E. 0 . Sunden, T. L. Wright, J. Lee, W. P. King and S. Graham, Appl. Phys. Lett., 88 (2006) 033 107. 94. R. L. Blaine, C. G. Slough and D. M. Price, Proc. 27th NATAS (1999) pp. 691. 95. V. V. Tsukruk, V. G. Gorunov and N. Fuchigami, Thermochim. Acta, 395 (2003) 151. 96. C. G. Slough, pers. comm. 97. T. Inuoue and T. Uehara, J. Phys. Paris, 9 (1999) 341. 98. A. T. Kwan, M. Yu. Efremov, E. A. Olson, F. Schiettekatte, M. Zhang, P. H. Geil amd L. H. Allen, J. Polym. Sci. B, Polym. Phys., 39 (2001) 1237. 99. R. H a l e r and E. zur Muhlen, Thermochim. Acta, 36 1 (2000) 113. 100. W. Hemminger and S. M. Sarge, in M. E. Brown (ed.), Handbook of Thermal Analysis and Calorimetry. Vol. 1: Principles and Practice, Elsevier Science B. V., Amsterdam, 1998, Ch.1, p.12. 101. D. M. Price, M. Reading, A. Hammiche, H. M. Pollock and M. G. Branch, Thermochim. Acta, 332 (1999) 143. 102. D. M. Price, M. Reading and T. J. Lever, J. Therm. Anal. Cal., 56 (1999) 673. 103. T. J. Lever and D. M. Price, Am. Lab., 30(16) (1998) 15.

104. M. Reading, D. M. Price, H. M. Pollock, A. Hammiche and A. Murray, Am. Lab., 31(1) (1999) 13. 105. D. M. Price and M. Reading, Adv. Polym. Technol., 18 (1999) 67. 106. M. Reading, D. M. Price, D. B. Grandy, R. M. Smith, L. Bozec, M. Conroy, A. Hammiche and H. M. Pollock, Macromol. Symp., 164 (2001) 45. 107. M. J. Abad, A Ares, L. Barral, J. Cano, F. J. Dieze, J Lopez and C. Ramirez, J. Appl. Polym. Sci., 85 (2002) 1553. 108. M. El Fray and V. Altstadt, Des. Monomers Polym., 5 (2002) 353. 109. I. Woodward, S. Ebbens, J. Zhang, S. Luk, N. Pate1 and C. J. Roberts, Packaging Technol. Sci. 17 (2004) 129. 110. G. Van Assche, A. Ghanem, 0. Lhost and B. Van Mele, Polymer, 46 (2005) 7132. 111. M. Song, D. J. Hourston, D. B. Grandy and M. Reading, J. Appl. Polym. Sci., 81 (2001) 2136. 112. W. Xie, J. Liu, C. W. M. Lee, and W-P Pan, Themochim. Acta, 367-378 (2001) 135. 113. S. Okamoto and H. Ishida, Macromolecules, 34(21) (2001) 7392. 114. H. M. Pollock, A. Hammiche, M. Song, D. J. Hourston and M. Reading, J. Adhesion, 67 (1998) 193. 115. M. S. Tillman, T. Takatoya, B. S. Hayes and J. C. Seferis, J. Therm. Anal. Cal., 62 (2000) 599. 116. H. Matsumura and W. G. Glasser, J. Appl. Polym. Sci., 78 (2000) 2254. 117. M. S. Tillman, B. S. Hayes and J. C. Seferis, J. Appl. Polym. Sci., 80 (2001) 1643. 118. J. C. Seferis, M. S. Tillman and B. S. Hayes, J. Macromol. Sci. Phys., B40 (2001) 923. 119. M. S. Tillman, B. S. Hayes and J. C. Seferis, Thermochim. Acta, 392-393 (2002) 299. 120. L. Barral, J Cano, F. J. Diez, J. Lopez, C. Ramirez, M. J. Abad and A. Ares, J. Polym. Sci. B Polym. Phys., 40 (2002) 284. 121. G. Van Assche and B. Van Mele, Polymer, 43 (2002) 4605. 122. J. Ye, N. Kojima, K. Furuya, F. Munakatu and A. Okadu, J. Therm. Anal. Cal., 69 (2002) 1031. 123. S. A. Edwards, M. Provitas, M. Ginic-Markovic and N. R. Chowdhury, Polymer, 44 (2003) 366 1. 124. S. A. Edwards and N. R. Chowdhury, Polym. Eng. Sci., 44 (2004) 96. 125. S. Mallarino, J. F. Chailan and J. L. Vernet, Composites A, 36 (2005) 1300. 126. S. Mallarino, J. F. Chailan and J. L. Vernet, J. Polym. Sci. B, Polym. Phys., 44 (2006) 205. 127. J. P. Ye, T. Hasegawa and A. Suzuki, J. Appl. Polym. Sci., 100 (2006) 1306.

128. G. F. Meyers. M. T. Dineen, E. 0. Shaffer, T. Stokich and J. H. Im, Macromol. Symp., 167 (2001) 213. 129. C. Blanco, S Lu, S. P. Appleyard and B. Rand, Carbon, 40 (2002) 132. 130. M. Provitas, S. A. Edwards and N. Roy Chowdhury, Thermochim. Acta, 392393 (2002) 339. 131. T. Grossetete, L. Gonon and V. Verney, Polym. Degrad. Stab., 78 (2002) 203. 132. D. S. Fryer, P. F. Nealey and J. J. de Pablo, Macromolecules, 33 (2000) 6439. 133. D. S. Fryer, P. F. Nealey and J. J. de Pablo, J. Vac. Sci. Technol. B, 18 (2000) 3376. 134. D. S. Fryer, R. D. Peters, E. J. Kim, J. E. Tomaszewski, J. J. de Pablo, P. F. Nealey, C. C. White and W. L. Wu, Macromolecules, 34 (2001) 5627. 135. R. S. Tate, D. S. Fryer, S. Pasqualini, M. F. Montague J. J. de Pablo and P. F. Nealey, J. Chem. Phys., 115 (2001) 9982. 136. V. V. Gorbunov N. Fuchigami and V. V. Tsukmk, High Perf. Polym., 12 (2000) 603. 137. D. B. Grandy, pers. comm. 138. G. Yang, H. Tokuhisa, E. Koyama, S-I Matsuzono and K. Hiratani, J. Appl. Polym. Sci., 84 (2002) 2287. 139. M. J. Krupers, H. R. Fischer, A. G. A. Schuurman and F. F. Vercauteren, Polym. Adv. Technol., 12 (2001) 561. 140. T. H. Fang and W. J. Chang, Appl. Surf. Sci., 240 (2005) 312. 141. D. Q. M. Craig, V. L. Kett, C. S. Andrews and P. G. Royall, J. Pharm. Sci., 91 (2002) 1201. 142. V. Pillay and R. Fassihi, J. Control. Release, 59 (1999) 243. 143. P. G. Royall, D. Q. M. Craig, D. M. Price, M. Reading and T. J. Lever, Int. J. Pharm., 192 (1999) 97. 144. L. Bond, S. Allen, M. C. Davies, C. J. Roberts, A. P. Shivji, S. J. B. Tendler, P. M. Williams and J. Zhang, Int. J. Pharm., 243 (2002) 71. 145. K. Six, J. Murphy, I. Weuts, D. Q. M. Craig, G. Verreck, J. Peters, M. Brewster and G. Van den Mooter, Pharm. Res., 20 (2003) 135. 146. J. R. Murphy, C. S. Andrews and D. Q. M. Craig, Pharm. Res., 20 (2003) 500. 147. M. Galop, Pharm. Res., 22 (2005) 293. 148. G. H. W. Sanders, C. J. Roberts, A. Danesh, A. J. Murray, D. M. Price, M. C. Davies, S. J. B. Tendler and M. Wilkins, J. Microsc.- Oxford, 198 (2000) 77. 149. E. Yonemochi, T. Hoshino, Y. Yoshihashi and K. Teada, Thermochim. Acta, 432 (2005) 70. 150. P. G. Royall, V. L. Kett, C. S. Andrews and D. Q.M. Craig, J. Phys. Chem. B, 105 (2001) 7021.

151. S. Hasegawa, T. Hamaura, N. Furuyama, S. Horikawa, A. Kusai, E. Yonemochi and K. Terada, Int. J. Pharm., 280 (2004) 39. 152. P. G. Royall, V. L. Hill, D. Q. M. Craig, D. M. Price and M. Reading, Pharm. Res., 18 (200 1) 294. 153. J. Zhang, C. J. Roberts, K. M. Shakesheff, M. C. Davies and S. J. B. Tendler, Macromolecules, 36 (2003) 1215. 154. M. C. Perkins, C. J. Roberts, D. Briggs, M. C. Davies, A. Friedman, C. Hart and G. Bell, Appl. Surf. Sci., 243 (2005) 158. 155. M. Odlyha, N. S. Cohen, G. M. Foster, A. Aliev, E. Verdonck and D. Grandy, J. Therm. Anal. Cal., 71 (2003) 939. 156. D. N. Morton, C. J. Roberts, M. J. Hey, J. R. Mitchell, J. Hipkiss and J. Vercautheren, J. Food Sci., 68 (2003) 1411. 157. H. Fischer, Thermochim. Acta, 425 (2005) 69. 158. J. Zhang, N. W. Botterill, C. J. Robert and D. M. Grant, Thermochim. Acta, 401 (2003) 111. 159. C. Blanco, S. Lu, S. P. Appleyard and B. Rand, Carbon, 41 (2003) 165-171. 160. F. A. Boroumand, M. Voigt, D. G. Lidzey, A. Hammiche and G. Hill, Appl. Phys. Lett., 84 (2004) 4890. 161. M. Reading, US Pat., 6,405,137 (2002). 162. R. M. Smith, D. J. Hourston, D. M. Price, M. Reading, A. Hammiche and H. M. Pollock, Proc. 6th Int. Symp. Hyphenated Techniques in Chromatography and Hyphenated Chromatographic Analyzers, Brugge, Belgium, February 911,2000, p69. 163. D. M. Price, M. Reading, T. J. Lever, A. Hammiche and H. M. Pollock, Thermochim. Acta, 3671368 (2001) 195, Thermochim. Acta, 376 (2001) 95. 164. D. M. Price, M. Reading, R. M. Smith, A. Hammiche and H. M. Pollock, J. Therm. Anal. Cal., 64 (2001) 309. 165. D. M. Price, D. B. Grandy, M. Reading, L. Bozec, A. Hammiche and H. M. Pollock, Proc. 29th NATAS, (2001) St Louis, Missouri, pp. 574. 166. W. H. McClennan, R. M. Buchannan, N. S. Arnold, J. P. Dworzanski, H. L. C. Meuzelaar, Anal. Chem., 65 (1993) 2819. 167. D. Briggs, Surface Analysis of Polymers by XPS and Static SIMS, Cambridge University Press, Cambridge, 1998. 168. A. Hammiche, M. J. German, R. Hewitt, H. M. Pollock, F. L. Martin, Biophys. J., 88 (2005) 3699. 169. M. Reading, D. Grandy, A. Hammiche, L. Bozec and H. M. Pollock, Vib. Spectrosc., 29 (2002) 257. 170. P. E. Sheehan, L. J. Whitman, W. P. King and B. A. Nelson, Appl. Phys. Lett., 85 (2004) 1589.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 4 PULSE THERMAL ANALYSIS M. Maciejewski and A. Baiker Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology, ETH Honggerberg, 8093 Zurich, Switzerland 1. INTRODUCTION Thermal analysis (TA) is generally used for characterizing gas-solid reactions or decompositions of solids in the range from zero to full conversion, i.e. the entire process is monitored. In conventional TA, the experiments are carried out isothermally or with a linear temperature ramp. In isothermal experiments, some time is required for the system to reach the desired temperature. During this heating period the reaction may start, rendering proper investigation of the initial transformations almost impossible. For the same reasons, it is impossible to quench the reaction at a desired time (or at a desired extent of reaction). Also, the most frequently used non-isothermal mode of thermal analysis does not allow controlled decomposition of the sample to a certain degree or quenching of reaction at a specified time. Pulse thermal analysis pulse^^@) [I] eliminates, or at least reduces, the difficulties mentioned above. P U ~ S ~ T Ais@based on the injection of a specific amount of the gases or liquids into the inert carrier gas stream and subsequent monitoring of changes in the mass, enthalpy and gas composition, resulting from the incremental reaction extent. Because a known amount of the selected gas, which can be used for calibration, is injected into the system, the method is also suitable for quantification of the evolved gas by mass spectrometry (MS) or Fourier transform infrared spectroscopy (FTIR). In contrast to conventional TA and all its modifications, the reaction is controlled not only by the temperature, but also by a distinct change in the composition of the reactive atmosphere. pulse^^@ offers three principal opportunities for thennoanalytical studies, depending on the kind of injected gas: injection of a gas which does not react with the sample. This (i) procedure facilitates quantitative calibration of the mass spectrometric signals and increases the sensitivity of TA

(ii)

(iii)

measurements so that species in amounts lower than 0.01 mass% can be detected. injection of a gas which reacts with the solid sample. This mode provides the opportunity of investigating all types of gas-solid reactions. injection of a gas which adsorbs onto the sample surface. This mode offers the study of adsorption phenomena under atmospheric pressure and at specified temperature.

The most distinct feature of P U ~ S ~ T Ais@that changes of the gaseous atmosphere and the solid sample occur during a well-defined period of time after each injected pulse of the gas or liquid. The transient character of the pulse technique offers interesting opportunities for P U ~ S ~ T A applications @ which will be illustrated using examples from various fields such as catalysis and materials science. 2. EXPERIMENTAL

TA-MS experiments reported here were carried out isothermally or nonisothermally (generally heating rates were in the range of 5-10 Wmin) on a Netzsch STA 409 simultaneous thermal analyzer equipped with a gas pulse device (commercially available as a P U ~ S ~ T Abox @ by Netzsch GmbH, http:llwww.netzsch.com), enabling injection of a known amount of two different pure gases or gaseous mixtures into the system. The amount of injected gas could be varied from 0.01 to 10 cm3. Primarily, volumes of 0.25, 0.50 and 1.OO cm3 were used. Gases evolved during reaction and/or injected into the system were monitored on-line using a Balzers QMG 420 quadrupole mass spectrometer, connected to the thermal analyzer by a heated (about 200°C) capillary. TA-FTIR experiments were carried out on a Netzsch STA 449 simultaneous thermal analyzer connected by a heated capillary to the Vector 22 FTIR spectrometer (Bruker, Germany) equipped with MCT detector and an especially developed low-volume gas cell (8.7 cm3) with a 123 mm path length and ZnSe window. The resolution of the collected spectra was set between 1 and 32 cm-' and co-addition of 4 scans per spectrum was applied. As a consequence spectra were recorded with a time resolution of about 2-20 s, depending on the integration methods and the spectral resolution of the FTIR instrument used.

3. CALIBRATION OF SPECTROMETRIC SIGNALS IN HYPHENATED THERMOANALYTICAL TECHNIQUES 3.1. Calibration of gases Injection of a known amount of gas or liquid that does not react with the investigated sample provides a quantitative calibration by relating the spectrometric signals to the injected quantity of the probe gas. Previously published results [2,3] indicate that the kind of carrier gas and its temperature affect only the shape of the MS and FTIR signals but not their integral intensities. The main feature of the coupled techniques TA-MS and TA-FTIR is the identification of the gaseous products, which together with the thermal effects (DTA) and mass changes (TG), aids in interpreting the course of the investigated reactions. The qualitative analysis is routinely done by comparing recorded spectra with key fragment ions and their relative intensities for known elements and compounds (MS) and with reference spectroscopic signals (FTIR). Spectroscopic data are available in the large libraries, see e.g. Gas Phase IR Spectra and Mass Spectra in htt~://webbook.nist.nov/chemistry/. Examples of FTIR and MS spectra of H20, CO and C 0 2 are presented in Figures 1 A and B, respectively.

4000

3500

3000

2500 wavenumber

2000

I

1500

1000

500

cm-1

Figure 1. (A) FTIR and (B) MS spectra of water, carbon monoxide and carbon dioxide.

Due to the fragmentation of the investigated molecules during the ionization process in the mass spectrometer (for example, the mfz peak at 28 for CO and C02 in Figure 1B) and possible overlapping of characteristic IR bands, the identification of the gaseous species, especially in multicomponent systems, is sometimes difficult. However, an even more complicated problem is the quantitative interpretation of spectrometric data, which needs the calibration of the system, i.e. the determination of the relationship between the observed intensities of the ion current (MS) or absorbance (IR-spectra) and the amount of the analyzed species. The common method of calibration of spectrometric signals is time consuming and requires the application of gaseous mixtures with well-defined composition. An additional difficulty is that such calibration is usually performed at room temperature. During thermal decomposition the temperature of the gas phase often changes considerably and can render the results obtained by calibration at low temperature doubtful. The P U ~ S ~ T A method @ of calibration is based on introducing a known amount of the calibration gas into the TA-MS or TA-FTIR systems. Specific amounts of the probe molecules can be introduced into the system by two methods: (i) by decomposing solids via well-known, stoichiometric reactions [2,4-81, or (ii) by injection of a calibration gas into the carrier gas stream flowing with constant rate through the system. Both methods can be applied for quantification of the spectroscopic signals in both systems: TA-MS and TA- FTIR. Quantification using MS combined with the pulse calibration technique is facilitated by the fact that a linear relationship between the observed intensities of the ion current and the amount of the analyzed species holds over a wide concentration range. Furthermore the MS data acquisition time is generally short so that recording of evolved gases can be achieved with a high time resolution. In contrast, quantification of evolved gases in TG-FTIR systems is complicated by the fact that Lambert-Beer's law is often valid only in a small concentration range, and consequently the amount of injected gas in a single-point calibration must closely match that evolved during decomposition. Additionally the data acquisition time in FTIR can only be made comparable with that used in MS if a low spectral resolution (e.g. 32 cm") is chosen. Proper adjustment of other parameters, such as residence time in the IR-cell, is also important for achieving high accuracy in data collection [9,10] Figure 2A presents results from the investigation of the decomposition of sodium bicarbonate (NaHC03,gaseous products C 0 2 and HzO) with a rate of 10 Klmin using TA-FTIR. To calibrate evolved C02 traces, two 1.00 cm3 pulses of C 0 2 were injected before and after the decomposition process. The relationship between the mass of the reactant and the intensity of the C 0 2 signals is shown in Figure 2B.

time 1 s

sample mass I mg

Figure 2. A) Relationship between the mass of decomposed NaHC03 and the intensity of C 0 2 traces recorded using TA-FTIR; B) amount of C 0 2 found by FTIR as a function of the reactant mass. A linear relationship was found between the amount of the analysed species and the corresponding spectrometric signals even when the sample mass was increased by a factor of ten. This conclusion is illustrated by the results shown in Figure 2B. Similar behaviour was observed using TA-MS. Results presented in Figure 3A illustrate the linear relationship also observed between the intensities of the MS signals m/z=44 and the amount of injected COz. For capillary coupling the sensitivity of the spectrometric signals does not depend on the temperature in the range of 20-1000°C. This enables quantitative evaluations of MS- and FTIR spectra obtained at any temperature using a singlepoint calibration. The dependence between the integral intensity of the mass spectrometric signals (m/z = 44 resulting from the injection of 1.OO cm3pulses of COz into the carrier gas helium) and temperature is presented in Figure 3B.

C' 1

; 'pulse volume 7m1)

I

A

0

.

I 10

.

I 20 time

.

I 30

I min

.

I 40

.

I 50

B

d

260

'

460

'

660

'

8A0

1600 I

temperature I "C

Figure 3. A) Intensities of MS signals as a function of the amount of injected C 0 2 B) Temperature dependence of the intensities of MS signals 3.2. Verification of the calibration Verification of the calibration procedure carried out by P U ~ S ~ T A can @be done by applying a decomposition process with well-known stoichiometry. Figure 4 depicts the calibration of the m/z=44 MS signal using the decomposition of NaHC03 and injection of C02. The reaction proceeds according to the equation:

The theoretical total mass loss during decomposition of NaHC03 amounts to 36.90 mass% with the contributions of C02 and H 2 0 being 26.19 mass% and 10.7 1 mass%, respectively. The mass of decomposed reactant was 35.66 mg; the mass loss (after correction of the TG-base-line) was 13.01 mg corresponding to 36.48 mass% (0.99 of the stoichiometric value). The integral signal of a 1 cm3 pulse of CO2 was 3.30 a.u. (A s). The temperature of the injecting loop was 30 "C, therefore the mass of injected C02 amounted to 1.781 mg. The integral signal m/z = 44 resulting from the evolution of C02 had an intensity of 17.33

a.u., which, after comparing to the integral intensity of the injected pulse, is equivalent to 9.35 mg of CO2, The stoichiometric amount of CO2 in the sample was 35.66 x 0.2619 = 9.34 mg. The error in the determination of the C02content by MS was thus comparable to the error of the thermogravimetric analysis.

0

-

TG

-10-

.

s

V) m/z = 18 07 q -20-V)

0

-30- m/z = 44 -40

-36.5% I

I

I

I

I

I

50

100

150

200

250

300

temperature I "C

Figure 4. Decomposition of NaHC03 during heating with a rate of 10 Wmin, P = pulse of 1.00 cm3 of C 0 2 injected for the calibration of the mlz=44 MS signal. In routine experiments the applied method provides an accuracy of 2-3 % and can be very easily verified, as described above, by comparing results obtained by conventional thermoanalytical techniques (TG) with results obtained by quantification of spectroscopic signals (MS or FTIR) calibrated by injected pulses of the probe gas. 3.2. Calibration of liquids In all experiments described above, injected calibrating species were gases. Quantification of spectroscopic signals for liquids is described in detail in ref. [lo]. This study presents the results of pulse calibration and two methods based on the evaporation of liquids: pinhole and differential methods. The latter method is based on the isothermal or non-isothermal vaporization of a liquid compound at different temperatures, with low heating rates and simultaneous monitoring in differential time periods of corresponding mass losses and intensities of FTIR signals. The correlation of the integral intensity of the FTIR signal with the exact amount of evaporated liquid measured by thermobalance in arbitrarily chosen periods of time allowed the extension of the method for the simultaneous calibration and measurement of evolved species in one experimental set-up. The target sample and the calibration sample are placed in two separate pans which are weighed continuously as during conventional thermogravimetric runs. The first calibration period is performed in the isothermal mode at relatively low

temperature. During this period the relationship between the intensity of the FTIR signal and the amount of evaporated liquid can be determined in a few differential time periods which increases the accuracy of the calibration. After total evaporation of the calibration liquid, which is indicated by the end of the mass loss on the TG curve, the second experimental stage is started, during which the temperature in the system is raised according to the chosen temperature ramp. This in situ calibration is shown in Figure 5 which depicts the quantification by FTIR of the evolved water formed during decomposition of sodium bicarbonate.

time / min

Figure 5. Differential calibration: water traces recorded during decomposition of NaHC03 related to the integral intensity of recorded during evaporation of water at 50 "C prior to the decomposition. Details of experiment see [lo]. Figure 5 shows the water trace recorded during calibration and decomposition of 200 mg NaHC03 according to reaction (1). The integral of the FTIR absorbance over the interval of 10 min was 2039 a.u. while the integral intensity of the signal of the evolved H 2 0 was 5045 a.u. The amount of evolved water during the calibrating stage was 8.91 mg. The amount of H20 formed during the decomposition calculated from these data corresponds to 10.98 wt.%, which agrees well with the stoichiometric value of 10.69 wt.%, corroborating the very good accuracy of the differential quantification method.

4. QUANTIFICATION OF THE SPECTROMETRIC SIGNALS IN A TAMS-FTIR SYSTEM

4.1. Determination of the intrinsic fragmentation in a TA-MS system ~ u l s e provides ~ ~ @ easy access to creation of a user's own library of injected species for both pure components and multicomponent mixtures. This is very important when applying FTIR and reference data are not accessible, or if it is necessary to check the influences of some components on the spectra of mixture. However, setting up a user's own library is even more important for the interpretation of mass spectrometric data. The experimental settings in a TA-MS system can distinctly influence the intensities of the recorded ion fragments. Even for such a simple case as, for example, the MS spectrum of C 0 2 one can find in the literature a large variety of data concerning the intensities of the fragmentation patterns which may significantly differ from those recorded by the users.

Table 1. Intensities of main peaks and fragmentation patterns for the pure gases and their mixtures. Integral intensities of ion currents are given in A . ~ o - s. '

Application of reference intensities of ion fragments taken from the literature for the quantification of experimental signals can cause significant errors. The properties of the applied TA-MS system have to be carefully checked before beginning with quantification of recorded spectra. This can be easily done by injecting the required gases and comparing the intensities of fragments with corresponding literature data. Results of such an experiment are presented in Figure 6 and Table 1 for illustration.

?

saJ (

6U

'

-c! E

.=

* 4.U)

C

a , ' *

.-C

2-

time I min

Figure 6. Intensities of mass spectrometric signals resulting from the injection of the respective gases and their mixtures into the carrier gas stream (He). Gas flow 50.0 cm3 min-', temperature 30 "C, injected volumes: 1.00 cm3 for pure gases, and 2.00 or 3.00 cm3 for gas mixtures.

The common mistake in quantitative or semi-quantitative evaluation (comparison of the intensities only without applying the inner standard for the normalization of the intensities) is due to the assumption that the MS intensities of two gases with the same concentration are the same. The fact that gases possess different relative probabilities of ionization is very often forgotten. The gas ionization depends mainly on the electron energy and the reference data applied in analytical mass spectrometry are usually collected at 70 eV. However, the properties of the mass analyzer can also change the degree of the ionization in the applied system. This phenomenon, very important for the quantitative evaluation of the MS spectra has also to be taken into account in order to avoid misleading interpretation of the recorded signals. The determination of the intensities of the main peaks, being proportional to the probability of ionization in the applied TA-MS set-up, can be easily carried out by applying P U ~ S ~ TThe A ~ results . of such an experiment are shown in Figure 7. All the above remarks and results of the experiments presented in Figures 6 and 7 and included in Table 1 can help in correct quantitative interpretation of the MS spectra and in understanding of the common errors.

J

I

I

I

I

20

40

60

80

I

time I min

Figure 7. The integral intensities, expressed on curves in A.IO-' sec, for different gases injected into the carrier gas stream in the TA-MS system. Injected volume 1.00 cm3, gas flow 50 cm3 min-', temperature 30 OC.

Let us assume that for a mixture of CO and C 0 2 two signals: m/z=28 and m/z=44 are recorded under experimental conditions identical to those used in the present study, and both spectra have the same integral intensity (peak area) of 100 a.u. This result does not mean that the amount of both gases in the analyzed mixture is the same: one has to take into account that part of the m/z=28 signal that is due to fragmentation of C02. From the NIST library, the intensity of m/z=28 in the cracking pattern of COz is 10.0% of the main signal. One can thus calculate the contribution of C 0 2 to the intensity of the MS signal of CO. After this correction the calculated ratio between C 0 2 and CO amounts to 100 / (10010) =I. 11. But this ratio is incorrect because it does not account for the real characteristics of the TA-MS system used, i.e, the intensities of fragmentation patterns and ionization probabilities. Using the data presented above, the following calculations can be done: - amount of C02: 100 / 13.8=7.25 cm3 (13.8 is the integral intensity of 1.OO cm3 of C 0 2 in the experimental set-up used). - the contribution of C 0 2 to the integral intensity of the m/z=28 signal amounts to 2.1 / 13.8=0.153 (see Table 1 for the intensities of the m/z=28 and m/z=44 peaks in the fragmentation patterns for COz). - the integral intensity of CO is equal to 100-15.3=84.7 a.u. which results in the amount of CO being equal to 84.7 /25.3=3.34 cm3 (25.3 is taken as the intensity of 1.00 cm3 of CO).

These calculations show that the ratio of C02/ CO amounts to 7.25/3.35=2.16, which is two times larger than that calculated without taking into account the properties of the mass spectra (fragmentation, probability of ionization) obtained in the TA-MS system used. 4.2. Application of ~ u l s e ~ ~quantification ~ f o r of gas-solid reactions 4.2.1 Decomposition of ZnC2O4.2H20 After calibration by means of PUIS~TA@ it is possible to properly quantify the MS spectra not only for one evolved gas but also for multicomponent mixtures. This is illustrated by comparing the results obtained with both spectroscopic methods for the decomposition of zinc oxalate dihydrate (gaseous products: water, CO and C02). These results clearly show the advantages and drawbacks of both MS and FTIR for qualitative and quantitative analysis of the gas phase. Due to fragmentation of the evolved C 0 2the results obtained by TA-MS (Figure 8) can not be used for unambiguous confirmation of CO formation without calibration of the mass spectrometric signals.

0

100

200

300

400

temperature

500

600

700

I "C

Figure 8. Thermal decomposition of ZnC204-2H20investi ated by TAMS. Pulses of injected CO, C 0 2 (4.00 cm3)and H2 (1.00 cm ) are marked as " P . Intensities of MS signals are expressed in arbitrary units.

f

The main mass-to-charge ratio m/z=28 of CO is overlapped by the same signal occurring due to ionization of C02. For CO the abundance of mIz=28 signal amounts to 100. According to the reference data, the relative abundance of

m/z=44 and m/z=28 signals in the mass spectrum of C 0 2 are 100 and 10-16 (depending on the intrinsic properties of the mass spectrometer used), respectively. The interpretation of the m/z=28 signal for mixtures of CO and C 0 2 can thus be misleading if the correct contributions of the C 0 2 fragments are not taken into account. Quantitative evaluation of MS signals is possible after injection of CO and C 0 2 pulses before (or after) the decomposition of the oxalate. Firstly, quantifying fragmentation of injected C 0 2 allows the determination of the ratio between m/z=28 and 44 signals which can be used for the correction of the intensity of m/z=28 traces resulting from the evolution of CO. Secondly, the signal intensities of both probe gases, compared to intensities of MS signals of evolved gases, allow the determination of the composition of the gas phase. Results of the evaluation of the mass spectrometric signals are illustrated in the following. During the decomposition proceeding according to the reaction: ZnC204.2H20--, 2H20 + CO + C 0 2+ ZnO the mass losses due to the evolution of water, CO and C 0 2 amount to 19.00%, 14.78 and 23.23 mass%, respectively. The mass of decomposed reactant was 36.95 mg. The intensities of 4.00 cm3 pulses of CO and C 0 2 were 24.95 and 14.97 a.u. respectively. The ratio of the intensities of the MS signals, m/z=28 and m/z=44, during injection of C 0 2 was 0.153. The intensities of the m/z=28 and 44 signals during oxalate decomposition were 31.30 and 18.90 a.u., respectively. After correction of the m/z=28 signal due to C02 fragmentation, i.e. after subtracting 0.153 x 18.90 from 31.30 and taking as a reference the intensities of injected pulses, the amount of CO and COz evolved during decomposition was found to be 5.13 and 8.99 mg, which correspond to 0.94 and 1.03 of the stoichiometric values. The results obtained using TA-FTIR agree well with those described above. After quantification of the CO and COz traces, the ratio between the evolved CO and C 0 2 was found to be 1.07 (1.09 determined by MS). The qualitative interpretation of the spectroscopic signals was in this case easier than by MS due to the presence of characteristic bands for both gases. However, the inability y of FTIR to detect homonuclear diatomic molecules did not allow the occurrence of the secondary reaction between CO and traces of water to be confirmed. The formation of hydrogen in this reaction was confirmed using TA-MS. Detection of H2 explained the lower than expected amount of evolved CO and the higher amount of C 0 2 due to the reaction: CO + H 2 0 S H2 + C02.

4.2.2 Calcination of a mixture of y-alumina and calcium carbonate Quantitative analysis by means of combined TA-MS and TA-FTIR is especially useful in the case of multicomponent systems, where two or more processes can overlap due to simultaneous reactions. A typical example is shown in Figure 9 which depicts the calcination of mixtures of calcium carbonate and yalumina with known compositions.

temperature I "C

Figure 9. Determination of the CaC03 content in a mixture with y-A1203by TG and MS. The mass spectrometric signals of evolved CO2 (m/z=44) were calibrated by pulses of C02 injected before andlor after carbonate decomposition. The real CaC03 content in the samples is given in italics, the results of TG and MS analysis are quoted in rectangles. Because of the continuous mass change originating from the evolution of water from alumina, the exact determination of the mass loss due to CaC03 decomposition is uncertain. This leads to an imprecise determination of the amount of CaC03 in the mixture. The onsets of the TG curves taken for the determination of the evolution of COz have to be chosen arbitrarily, leading to substantial errors in the calculation of the real composition of the analyzed samples. The smaller the content of CaC03, the greater is the error of thermogravimetric analysis. For the sample containing only 0.21 mass% carbonate the difficulties in assessing the beginning and the end of the CO2

evolution cause a gross overestimation of the calcium carbonate content. The results obtained by quantification of the MS signals are much more accurate. 4.2.3. Determination of the C 0 2content in Au/Zr02 and Au/Ti02 catalysts The quantitative interpretation of the TG curve is even more uncertain when the temperature ranges of the various decomposition stages overlap so that the specific stages of the decomposition are not discernible. Also, when the total mass change is below 1-2 mass%, quantitative interpretation is difficult due to the buoyancy effect.

= . * m

A

-

9 03

I 1 cm3 pulse of CO,

309

coz

. g,

1 .o

Hz0

.

0.0-

CO*

.

$

0.5H,O

a,

-0.1 %

-0.5: . rn-1.00

-1.4%

g-1.5-

-2.0

E

4

.

.

-1.0-

I

I

I

200

400

600

temperature 1 "C

-1.2%

'

800

0

160

260

I

300

.

,

400

'

,

500

'

600

temperature "C

Figure 10. Determination of carbon-containing species in ZrO2 (A) and Ti02-(B) supported gold catalyst.

Figure 10 depicts the quantitative determination of the carbon-containing species in supported gold catalysts Au/Zr02 and Au/Ti02 [ll]. During calcination in 20 vol% O2 the water and carbon dioxide evolve simultaneously and the very small mass losses observed amounted to 1.2 and 1.4 mass%, respectively The quantification of the mass spectrometric signal of evolved C02 by injection of 1.00 cm3 calibration pulses allowed the organic residue in the catalysts to be determined. The carbon contents in the Au/Zr02 and Au/Ti02 samples were 0.04 and 0.05 mass%, respectively.

4.2.4 Determination of the diamond content in grinding tools The application of TG curves for interpretation of mass changes can be difficult when processes resulting in mass loss and mass gain occur simultaneously. The advantage of the pulse technique in this case is illustrated by the determination of the diamond content in nickel-cobalt alloys. Some special grinding tools contain synthetic diamonds embedded in the bonding matrix of a Ni-Co alloy. During heating in air [12] the metals are oxidised (mass gain) and diamond reacts with oxygen forming CO2 (mass loss). The formation of metal oxides is overlapped at higher temperatures by the decomposition of Co3O4into COO (mass loss). So the course of the exothermic reactions, namely oxidation of the diamonds and metals, is overlapped by the endothermic reaction of Co3O4 decomposition as presented in Figure 11.

temperature 1 "C

Figure 11. Determination of diamond content by its combustion in air and quantification of the MS signal m/z=44 (C02). All these simultaneously occurring processes result in quite complicated shapes of both the TG and the DTA curves. The determination of the mass loss from the TG curve is uncertain, therefore the only method for analysis of the diamond content is the quantification of the evolved C02. The comparison of the integral intensities (the areas of the MS signals) of the injected pulses (1.00 cm3) with the C 0 2 signals due to diamond oxidation allowed quantification of the evolved C02, which in turn gave the amount of the diamond in the investigated sample. The diamond content was 12.3 mass%.

4.2.5. Determination of the amount of organic residue in a ZrOz aerogel Figure 12 illustrates the application of P U ~ S ~ T A for@the determination of the amount of organic residues in a Zr02 aerogel prepared by the acid-catalyzed alkoxide sol-gel route [13]. The observed mass loss is caused by the evolutionof physisorbed water and the oxidation of the organic residues present in the aerogel. The formation of C02, indicated by the exothermic effect on the DTA curve and the accompanying signal of m/z=44 centred at 357OC, indicates the combustion of organic residues present in the aerogel matrix. The calibration of the mass spectrometric signal by injection of 1.00cm3C 0 2 at 550 "C allowed the carbon content (1.9 mass%) to be determined, which was confirmed by elemental analysis (2.0 mass%). =I .6 -

s2

-

7

CO, pulse

1

.

4

i

K

-8 6

5 c d

m/z=18 - 4

$

E

" 1'2 2

*I .o .E

0 -

:2 m/z=44-0

- 20

.

cn-6In 2-87 -10-12 0

r

2-

.g

.

-0 9.6% I

200

I

400 temperature I "C

I

600

. -20

Figure 12. Determination of the carbon content in a Zr02 aerogel. Inset A depicts the magnification of a very small C02peak centred at 492 OC. The sample investigated was amorphous to X-rays. Its crystallisation is represented by the sharp exothermic peak centred at 492 OC. During crystallisation, the residual species trapped in the amorphous network (water (m/z=18) and a small peak of CO2 - its magnification marked as "A" in the inset in the figure) were evolved. The amount of the carbon released during the crystallisation of the zirconia aerogel amounts to 0.007 mass%.

4.2.6. QuantiJication of COz in a cuprate superconductor of the Y ~ B u ~ C U ~ + , , O , ~ + ~ family The possibility of exact calibration of the MS signal by means of ~ u l s e ~ ~ @ significantly increases the potential of the coupled TA-MS and TA-FTIR techniques. Using this method to determine the content of certain species in the investigated system has considerable advantages over conventional elemental analysis. With conventional analysis, only the total amount of an analyzed species can be measured, and it is impossible to resolve the species evolved in multistage reactions.

=! . m

I Icm3CO, pulse

w-

921

706

(V

c-3 II

m/z = 32

-C! E

'

4 C .-

.-

mlz = 44 DTA

-3-

1-20 - -40 - -60 - -80

t

4

4 -6

-0

705

TG

-5

1A I

I/

$ I

0

861

... ........$ 0.36 % 0.005%

200

I

400

I

I

600 800 temperature PC

. 2 5

I

1000

Figure 13. Calcination of a Y2Ba4C~6+n014+n superconductor in argon. Inset A presents the magnification of the m/z=44 signal. The pulse of C 0 2 was used for calibration. This advantage of ~ u l s e is~ illustrated ~ @ by the quantification of the evolved gases during the TA-MS measurement of a cuprate superconductor of the Y2Ba4Cu6+,,0L4+,, family [14]. These superconductors often contain a remarkable ~ can - strongly influence their properties. In the quantity of ~ 0ions,~which studies of the carbon dioxide incorporation process, a cardinal question is whether it is possible to remove C 0 2 from the superconductor without its decomposition. Also the knowledge of the substitution of certain species in the crystallographic lattice of the superconductor by ~ 0can~help~ in -the interpretation of YBaCuO properties in relation to the preparation method. The MS signals clearly indicate that evolution of carbon dioxide occurs at temperatures higher than that of superconductor decomposition, this is reflected

by the temperature of the beginning of oxygen evolution. The 1.OO cm3 pulse of C02, used later for the calibration of the m/z=44 signal, was injected before the carbon dioxide release, starting at about 600 OC. As emerges fiom the above A@ signals of C 0 2 with intensity ratios results, the application of P U ~ S ~ T allowed as small as about 1:70 (0.36 vs. 0.005 mass%) to be distinguished and helped to quantify the two steps of C02 evolution.

4.2.7. Miscellaneous applications The method of quantification of spectroscopic data based on P U ~ S ~ T Ahas @ been applied for the investigation and/or the determination of: - content of COOin the solid products of cobalt oxalate decomposition [15]. - amount of sulfur and organic matter in rock samples collected during prospecting [12]. - quantification of the organic residue in spent Pd-PtITS-1 catalysts used for continuous epoxidation of propylene [I 61. - C02 content in metastable flame-made monoclinic barium carbonate [17]. - amount of C 0 2 and traces of NO in flame-made calcium carbonate [IS]. - quantification of the evolved nitrogen formed during the thermal decomposition of InN and nitrided Ti02/In203[19]. - quantitative description of the thermal behaviour of fluorides and/or fluoride hydrates [20]. - distribution of barium-containing phases in NO, storage-reduction (NSR) catalysts [21,22]. - formation of barium aluminates and cerates during thermal deterioration of NO, storage-reduction catalysts 1231. - secondary reactions occurring during the thermal decomposition of a cobalt amino complex [24]. - carbon content in Ir samples resulting fiom the reduction of Ir02 with CO, propane and propene 1251 and in samples used for radiocarbon dating procedures [26]. - silver carbonate decomposition [27].

5. INJECTION OF A GAS WHICH REACTS WITH THE SOLID 5.1. Investigations of the reduction and oxidation of solids 5.1.I. Reduction of CuO supported on y-A1203

Injection of a gas which reacts with the solid enables the differential changes of the solid phase and the gas composition resulting from the injected pulse to be investigated. The interesting feature of p u l s e T ~ @is its suitability for investigating complex reaction systems at any desired temperature. Artifacts caused by temperature and gas phase composition settings are avoided and this allows better distinction of simultaneous processes. A representative example is the reduction of oxides in multicomponent systems which is often overlapped by evolution of physisorbed or chemisorbed species, or occurs already at room temperature, during first contact of the solid sample with the reducing agent. Because it is possible to carry out the reaction at a well-defined temperature, PulseT~@ allows the separation of the reactions occurring simultaneously. This feature is illustrated in Figure 14 by the determination of the copper content of an alumina-supported catalyst by the reduction of CuO with hydrogen.

50

100

150

200

250 300 temperature IOC

0'0~

0.5 -

20

!15 ~ I O -5

. 2

1.0:

lo

Am,, = 1.21%

1.5-

--lo +

V) V)

0.00

0.8-'

5

1-5

m/~18

0

3

5 (6

. '3

e

1.2 -

.'2 E

h

-1.09%

1.6-

.B

.a,

I

I

1

25

50

75

'

1

100 time I min

'

1

125

I

I

150

175

.-C

Figure 14. Reduction of alumina-supported CuO with hydrogen, carried out by A@ 350 "C (A) conventional TA, and (B) P U ~ S ~ T at

The reduction of CuO, investigated by conventional TA, is shown in Figure 14A.The TG, DTA, and m/z=18 curves indicate that two processes occur simultaneously. Dehydration of alumina begins at room temperature as seen from the continuous mass loss starting at 30 OC and the accompanying weak endothermic effect observed by DTA, and the broad peak of the m/z=18 curve. Starting at about 150 OC, the dehydration of the support is overlapped by the exothermic reduction of CuO (note the exothermic effect in DTA together with the mass loss and evolution of water). The mass loss due to CuO reduction calculated from the conventional TG curve (1.21 mass%) is about 10% greater than that obtained by pulse^^@ because calculation of the exact mass loss resulting from the reduction is uncertain, resulting in inaccurate determinationof the amount of CuO. Using ~ u l s e (Figure ~ ~ @ 14 B), the sample was dehydrated at 350 "C, until the mass loss caused by the water evolution was negligible, and then reduced progressively by 1.OO cm3 pulses of hydrogen. The observed mass loss due to the pulses of hydrogen results from the reduction only and the determination of the Cu content of the sample is much more accurate. 5.1.2. Reduction of zeolite-supported iridium oxide When gas-solid reactions (reduction, oxidation) occur already at room temperature, conventional TA can only be properly applied by starting the experiment below ambient temperature. Using pulse^^@ one can carry out the reaction at any temperature because gas-solid reactions, when investigated by ~ u l s e ~generally ~ @ , occur only during the pulse. This feature is illustrated by the determination of the Ir02 content in a zeolite-supported iridium catalyst. The results obtained by conventional TA and pulse^^@ are compared in Figures 15A and 15B, respectively. The determination of the mass change due to the reduction of iridium oxide with hydrogen is impossible in a conventional TA measurement. The reduction of Ir02 under the applied conditions (20 vol% Hz, balance He, heating rate of 10 K min-') starts already at room temperature and occurs over a wide temperature range. The very small mass loss due to the reduction (0.15 mass%, marked in Figure 15A) is overlapped by the mass loss due to desorption of water from the support. The exact determination of the Ir02 content was possible after removal of all water under inert atmosphere at 250 "C followed by hydrogen pulses (Figure 15B). The observed mass loss due to the hydrogen pulse originated from the reduction only. The reduction was complete after the first pulse, the following H2 injections did not change the mass of the sample further.

50 0.0

100 I

,

150 I

.

200 I

.

250 I

20 vol% H,

I

300 I

.

350 l

.

400

temperature I "C

time I min

Figure 15. Reduction of a zeolite-supported iridium catalyst by (A) conventional TA (10 K min-', H2 20 ~ 0 1 %/He) and (B) PulseTA (250 "C, 1.00 cm3HZpulses). 5.1.3 Preparation of solids with well-defined degrees of reduction 5.1.3.1 Reduction of CuO with hydrogen ~ u l s e ~allows ~ @ for dosing of the reacting gas in small quantities. This facilitates the study of gas-solid reactions in differential mode. This feature provides interesting opportunities for investigating gas-solid reactions, as e.g. the preparation of solids with well-defined extents of reaction. Depending on the temperature and pulse volume, any required amount of reaction progress can be achieved. Figure 16 presents the results of the reduction of CuO with hydrogen at two, significantly different temperatures. There is controversy in the literature concerning the course of the reduction of copper (+2) oxide with hydrogen. Generally it is assumed that CuO is directly reduced to metallic copper without intermediates [28], but there are suggestions that Cu20 is formed in the first stage of the reduction 1291. To compare the phase composition of the reduction products, two samples of CuO were reduced to exactly the same extent (ca 0.33) by pulses of hydrogen at 250 and 450 "C, respectively. Because of the much lower rate of the reduction, each pulse at 250 "C resulted in about 1.5% change of the reaction extent only. To reach the same extent of reduction, 22 injections were required at 250 "C compared with nine pulses at 450 "C (see Figure 16A).

100

50

34

36

38

time 1 min 150

40 two theta 1

200

42

250

44

46

O

Figure 16. (A) Reduction of CuO with H2 pulses at 250 OC and 450 "C to similar reduction extents (marked on TG curves), (B) XRD patterns of the reduced samples. The results of the XRD analysis of both reduction products presented in Figure 16B confirmed that, despite the same overall degree of reduction (one third of the amount of oxygen present in the copper oxide was removed) the phase composition of the products differed significantly. The sample prepared at the lower temperature contained much less metallic copper and more, poorly crystalline (broad XRD reflections) of Cu20. 5.1.3.2Reduction of Mn203 with carbon monoxide The unique feature of P U I ~ ~ T that A @allows reaction to be stopped at any point between the pulses, and facilitates the elucidation of the relationship between the composition of the solid and the reaction extent, is also illustrated by the reduction of Mn-oxides by CO pulses shown in Figure 17. The XRD patterns presented in the lower part of Figure 17 indicate that sample 3, corresponding to the stoichiometric mass loss of the reduction of M%03 to Mn304, contains three phases: Mn203, Mn304 and MnO. Even at very low degree of reduction (sample 1, mass loss 0.67 mass%), the final reduction product, MnO, was already formed. The preparation of samples corresponding to such a low and well-defined reaction progress is extremely demanding by any conventional, iso- or non-isothermal TA technique.

100

110

120

time 1 rnin. 130 140

150 #

do

32

34

36 038 two-theta I

160 ,

I

40

170 .

42

Figure 17. Reduction of Mn203by 1.00 cm3 pulses of CO at 500 OC and XRD patterns of the products taken at the points marked by 1,2 and 3, corresponding to different extents of the reduction. 5.2. Investigation of the redox behaviour of solids: reduction and reoxidation of CeOz A comparison between conventional and differential characterisation of gassolid processes is illustrated by the investigation of the redox properties of CeOz. The redox behaviour of this oxide is used in automotive exhaust catalysts. The ease of its reduction and re-oxidation makes it useful as an oxygen buffer during catalytic reactions [30]. Because automotive exhaust gas has a cyclic fluctuation of lean-rich composition, a component that can store oxygen and that readily undergoes redox cycles, can provide oxygen for oxidation of CO and hydrocarbons in the fuelrich region. In the reduced state the component can remove oxygen from the gas phase when the exhaust gas cycles into the lean region. Thus, cerium oxide not only promotes the oxidation activity of the catalyst, but also widens the air-fuel ratio composition range where all three major pollutants, CO, hydrocarbons, and NO can be removed. The difficulty in applying conventional TA techniques for investigating redox phenomena in CeOz is caused by the fact that only very small changes of the composition of ceria occur during the catalytic process. The composition of ceria during catalytic reactions changes only slightly during subsequent reduction and reoxidation cycles and, consequently, the characterization of the redox processes

of ceria by investigating the total reduction and oxidation processes are of little value. 400

temperature I "C 800 1000

600 I

I

200

.

I

400

.

I

'

1200

I

I

I

600 800 temperature / "C

'

1

1000

I

'

I

.

1200

Figure 18. (A) mass changes of Ce0.6Zr0.402 (50.0 mg) resulting from 4.00 cm3 pulses of hydrogen followed by 2.00 cm3 pulses of oxygen at different temperatures, (B) progress of the oxygen removal as a function of sample composition. The inset in (A) explains the determination of the Am-T relationship applied for the construction of the plots in Figure 18 (B). Much better characterisation of the redox behaviour of the catalyst is achieved ~ @ , out in incremental steps at temperatures at which the by ~ u l s e ~ carried catalytic reactions occur. The results presented in Figure 18 indicate that, after each hydrogen pulse, there is a distinct mass loss due to ceria reduction marked as AH2. The determination of the important property called "dynamic oxygen exchange capacity" is illustrated in Figure 18B showing the investigation of the redox properties of flame-made Pttceria-zirconia samples with different compositions (Ce:Zr ratio) [31]. The samples were subjected to a series of reducing (hydrogen) or oxidizing (oxygen) pulses while the temperature increased. The mass loss after the hydrogen pulse, due to oxygen removed from the CeO2 lattice, was fully reversible: re-oxidation by an oxygen pulse brought the sample

back to the original mass before the reduction. The inset in Figure 18B explains the determination of the mass loss for a given temperature. The method used for the differential characterization of the ceria reduction (allowing determination of a reduction degree lower than 0.01) indicated that the lowtemperature activity of flame-made materials was better than those derived by wet chemical methods [32]. 5.3. Investigation of gas-solid reactions A very valuable feature of P U ~ S ~ T Ais@the possibility of in-situ investigation of the course of reactions. This offers an important tool for elucidating mechanisms of complicated decomposition processes such as the decompositions of oxalates and for investigating, in differential mode, reactions between gases and solids such as occur in heterogeneous catalysis.

5.3.1 Decomposition of CoC204.2H20 The complexity of the processes occurring during oxalate decomposition indicates that an interpretation of the mechanism based only on TG results is of little value. Interpretation of mass changes alone does not allow deeper insight into all of the potential primary and secondary reactions that could occur. The secondary reactions, only avoidable when working under high vacuum conditions, distinctly influence the course of the reaction as illustrated by the results of the decomposition of cobalt oxalate dehydrate (COD) in inert, oxidizing and reducing atmospheres [15]. The solid products of COD decomposition in an inert atmosphere contain a substantial amount of COO. This contradicts the commonly held opinion that only metallic cobalt is formed under these conditions. Moreover, the COO content varies according to the secondary redox reactions caused by (i) the oxidation of very active, freshly formed metallic cobalt by the unavoidable traces of oxygen present in any thermoanalytical system, and (ii) the reduction of COO by hydrogen, carbon monoxide or methane being the gaseous products of the COD decomposition. The determination of the amount of COO in solid products, which can be used as a proof of the quantitative contributionof the particular mechanisms of COD decomposition, has therefore to be done in situ, just after completion of the decomposition process. The thermal decomposition of COD in helium is presented in Figure 19A showing TG, DTA and MS signals recorded during the decomposition with a heating rate of 10 K min-'. Note the presence of the unexpected gaseous products, hydrogen (m/z=2) and methane (m/z=15), formed from the primary gaseous products (CO, C 0 2 and H20) via the water-gas shift and FischerTropsch reactions, respectively.

temperature 1 "C

400 450 500 550 600 650 700 750 0.5 - ' . ' ~ ' ~ ' . ' . '

-

end of decomposition

0.0 -66.6 %

V)

I

3

mlz=18

.-

T = 391 O C

rn=66.09 mg

.

s -20P -40.!=.

:-60E -80

I

0

A

.

100

I

200

.

I

300

temperature I "C

.

I

400

.

I

500

10

B

20

.

~

30

.

I

40

.

50

I

.

60

I

.

70

time I min

Figure 19. (A) TA and MS signals recorded during decomposition of COD in He; (B) In situ determination of the composition of the solid products by COO reduction due to pulses of hydrogen (1) and CO (2) immediately after the end of COD decomposition. The determination of the amount of COO in the solid products, immediately after the decomposition ceased, is presented in Figure 19B. The reduction of COO was achieved by injection of 1.00 cm3 pulses of H2 (top) or CO (bottom) into the carrier gas stream. The recorded mass loss due to COO reduction amounted to 367 pg for the 32.58 mg COD samples which indicates that the solid products beside the main phase of metallic Co contains 15.8 mass% of COO,corresponding to 12.9 mol%. The unexpected detection of Hz, methane and water during the main course of the COD decomposition is explained by the fact that separation of the dehydration and oxalate decomposition is impossible. The removal of two water molecules is not complete before the beginning of the oxalate decomposition, which results in the formation of CO and C 0 2 and then hydrogen is formed by the water-gas shift reaction: CO + H20 +T C 0 2 + H2. Note that the equilibrium is shifted to the right at the temperatures of the COD decomposition.

. !

Figure 20. Formation of methane and water due to the water-gas shift and Fischer-Tropsch reactions (A) heating of metallic Co in C02:H2atmosphere and (B) pulses of C 0 2 over Cometheated in 20 vol% HZ. The results shown in Figure 20 depict the investigation of the reactions between CO, C02, H 2 0 and H2 in the presence of the active metallic cobalt under conditions of COD decomposition. The metallic cobalt obtained by the decomposition of COD in hydrogen was heated at a rate of 10 K min-' in an atmosphere of C 0 2 and H2 mixed in the ratio 1:2 (Figure 20A). In transient experiments (Figure 20B), the pulses of C02 were injected into a system in which Co,,, was heated in an atmosphere of 20 vol% HZ,balance He. The first experiment confirmed that, under the applied conditions, the formation of methane begins at about 190 "C. The maximum yield of CH4, calculated from ~ @ shown in Figure 20B, was observed at the results of ~ u l s e ~experiments about 400 "C. 5.3.2 Reaction of NO, with Pt/Ba/A1203storage reduction catalysts

~ u l s e can ~ ~ be @ applied for investigating gas-solid reactions in differential steps, which is particularly powerful for elucidating the mechanism of heterogeneous catalytic reactions. This is illustrated by the investigations of NO,

storage-reduction (NSR) catalysts used for the cleaning of automotive exhaust gases. (NO, is a generic term for the various nitrogen oxides produced during combustion, namely: N 2 0 nitrous oxide, NO nitric oxide and NO2 nitrogen dioxide). Generally, these catalysts contain the alkaline-earth components (mainly BaO or BaC03) supported on A1203. The removal of NO, from exhaust gases occurs in cycling conditions: the storage of NO, takes place during the oxidizing cycle (excess of air in the air-fuel mixture, so called lean conditions) when the barium-containing species are transformed into barium nitrate. During a short period of rich conditions (excess of fuel) the catalyst is regenerated to BaC03 and evolved NO is reduced by HZ,CO or propene to nitrogen. two-theta I"

temperature I "C

Figure 21. Reaction of NO, with Pt/Ba/A1203catalyst studied by P U ~ S ~ T A @ . Pulses of NO in 5% 02/He were passed over the catalyst at 300 OC resulting in the formation of Ba(N03)2 from BaC03; (B) Thermal stabilities of BaC03 phases present in the catalyst as a function of Ba loading. Knowledge of the nature, distribution and activity of the barium-containing species, in the process of NO, storage, is one of the key issues for optimizing the catalyst performance. Systematic studies [33] of the build-up and thermal stability of Ba-containing phases in differently loaded Pt-Ba/A1203catalysts has shown that the catalyst contains three active phases for NO, storage, each with different thermal stability: BaO, low-temperature (LT-BaC03) and hightemperature barium carbonate (HT-BaC03). P u l s e ~ ~ was ' applied to assess the activity of respective Ba-phases in the NO, storage process.

The interaction of Pt-Ba/A1203with NO, is presented in Figure 21A, while Figure 21B depicts the difference in the thermal stability of LT and HT-BaC03 as a function of Ba loading. In order to check which phase, LT or HT-BaC03, is more active in NO, storage, NO, pulses were injected at 300 OC over the catalysts in an inert atmosphere. The composition of the catalysts after 1 , 2 and 4 pulses and after the total saturation by NO, was investigated by thermal decomposition of the samples after each series of pulses.

0

200

400

600

temperature PC

800

1000

400

600

800

temperature I "C

Figure 22. (A) MS signals of m/z=44 recorded during the decomposition of a raw PtIBa~aloxcatalyst (bold) and after 1, 2 and 4 NO pulses at 300 OC. (B) Decrease of the amount of LT-BaC03 in the catalyst after four pulses of NO (bold), note the evolution of NO during decomposition of Ba(N03)2 formed upon exposure to previous NO pulses. Progress of nitrate formation (see Figure 22A) was calculated based on quantification by P U ~ S ~ T A of@ the MS signals of evolved C 0 2 (resulting from the decomposition of unreacted BaC03) and of NO evolved from the increasing amount of Ba(N03)2formed. Less thermally stable LT-BaC03 (decomposed in the range of about 420-800 OC) is more active in NO, storage: note the disappearance of the first C 0 2 peak characteristic for the decomposition of less stable LT-carbonate (Figure 22A and 22B). This important finding concerning the different activity of Ba-containing phases present in the catalyst helped in

elucidating the influence of the loading and kind of the support on the properties of this type of NSR catalyst.

5.4. Miscellaneous applications ~ u l s e ~has ~ @ been applied in investigations of the following gas-solid reactions: - interaction of NO, in the presence of water with barium aluminate and barium cerate [23]. - reduction by propene of the NO, stored on differently supported (A1203, Ce02,Zr02) PtBa storage-reduction catalysts [34]. - determination of the oxidation state of Pd during methane combustion over Pd/Zr02 catalysts [35]. - uncovering of the role of Pd-particle size and PdO lattice oxygen in methane oxidation over Pd/Zr02 [36,37]. - CO disproportionation over metallic Co during thermal decomposition of the cobalt oxalate dehydrate [15]. - low-temperature CO oxidation on Au/Ti02 and Au/Zr02 catalysts [38]. - selective reduction of NO by NH3over manganese-cerium mixed oxides [39]. - enhanced oxygen exchange capacity in flame-made nanocrystalline cerialzirconia doped with alumina or silica [40] and conventionally prepared ceria doped with silica [30]. - redox behaviour of promoted Ru-hydroxyapatite [41]. - oxidation of butadiene by lattice oxygen during catalytic oxidation of butadiene to furan by sol-gel bismuth-molybdenum-titania mixed oxides [42] - redox behaviour of zeolite-supported Ir and the influence of the particle size of iridium on the yield of this reactions [43]. - effect of sodium on the catalytic properties of Ir black in the selective reduction of NO, by propene [44]. - structural properties and catalytic behaviour of iridium in the selective reduction of NO by hydrocarbons [45]. - influence of the kind of reducing agent on the reduction of nitrogen oxides over unsupported iridium [25]. - structure sensitivity of NO reduction over iridium catalysts [46]. - effect of hydrogen on the selective catalytic reduction of NO in excess oxygen over alumina supported Ag [47].

6. INJECTION OF A GAS WHICH ADSORBS ON THE SOLID

6.1. Adsorption of ammonia on HZMS-5 zeolite Due to simultaneous monitoring of mass and thermal effects, ~ u l s e can ~ ~be@ applied for investigating adsorption phenomena occurring at atmospheric pressure. An example showing the adsorption of ammonia at 200 OC on HZSM-5 zeolite is presented in Figure 23.

40

60

80

100 120 time I min

140

160

180

Figure 23. Adsorption of ammonia on zeolite investigated by PUIS~TA@ Both reversible (physisorption) and irreversible adsorptions occur on the zeolite surface. The irreversible adsorption (chemisorption) occurs during the first pulse. The next pulses of ammonia are only weakly adsorbed. After about 20 minutes the sample mass returns to its initial value before pulsing, indicating that no further ammonia is adsorbed. The observed exothermal effect in the DTA curve, in conjunction with the mass gain resulting from the NH3 chemisorption, allows the determination of the heat of adsorption per mole of adsorbed ammonia. During weak, reversible adsorption the exo- effect on the DTA curve is followed by the endothermic ammonia desorption. Decreasing the pulse volume gives access to the determination of the differential heat of adsorption [481.

6.2 Investigation of the adsorption and desorption of NH3 on a titania-silica aerogel. The application of P U ~ S ~ T A allows @ not only the amount of strongly adsorbed adsorbate to be determined, but also quantification of the desorption process and characterization of the adsorption strength and kind of active sites. Such an example is presented in Figure 24, depicting the adsorption of ammonia on titania-silica aerogels. time I min

, $ 3-

0 I

50 .

I

100 .

I

150 ,

I

200 .

250

I

,

A

300

I

.

I

2.32% ......

1.23%

T = 50°C

" B . =?

CMA-13

.-s 50

I

I

I

I

I

100

150

200

250

300

350

temperature I "C

Figure 24. A) Adsorption of NH3 on modified titania-silica aerogels at 50 OC followed by B) desorption of ammonia during heating with a rate of 10 Wmin.

Mesoporous titania-silica mixed oxides containing different amount of covalently bound methyl groups were prepared in a sol-gel process, followed by low-temperature supercritical extraction with C02. The catalysts were used for the epoxidation of various olefins and allylic alcohols [49]. The olefin epoxidation activity was generally lowered by the methyl modification, whereas in the epoxidation of allylic alcohols a maximum in activity was observed with increasing methyl content. To gain more insight into the role of methyl groups in the catalytic behaviour, the acid properties of the surface were characterised by the adsorption of ammonia at 50 OC. Methyl groups located at the surface replace part of the silanol groups, resulting in a decreased ability to adsorb ammonia. The unmodified aerogel (CMA-13) adsorbed twice the amount of NH3 than a

corresponding catalyst containing 2.4 mg of methyl groups per gram of catalyst (CMA- 18). The shape of the desorption signals (Figure 14B) confirmed that in both samples Brernsted and Lewis acid sites are present: the desorption of ammonia occurred in the same temperature range confirming similar distribution and strength of active sites. The amount of desorbed ammonia was additionally quantified by injection of a NH3 pulse (not shown in Figure 14). 6.3. Investigation of adsorption combined with gas-solid reaction 6.3.1.Adsorption of CO on a zirconia-supported Au catalyst ~ u l s e can ~ ~ be @ applied not only for determining the strongly adsorbed (chemisorbed) species, but also for investigating weak, reversible adsorption. Results of a study of weakly adsorbing gases are presented in Figure 25 depicting the interaction between CO, O2 and COz and an Au/Zr02 catalyst [ll]. The catalytic properties of the supported gold catalysts were found to depend on calcination temperature.

10

20

time I min 100 110 40

30

m/z=44

I

50

120

130

140

CO, pulse

150

1

1

100 150 temperature I "C

200

Figure 25. Admission of 1.00 cm3 pulses of CO at 30 OC to an Au/Zr02 catalyst calcined at 200 OC (A) and 475 OC (B). Plot C depicts the formation of C02 due to injection of 1.00 cm3 pulses of CO during heating the catalyst pre-calcined at 200 OC ( 0 )and 475 OC (O) with a heating rate of 2 "C/min. 1.OO cm3 pulses of C 0 2were used for determining the extent of CO oxidation.

During calcinations, organic residues present in the catalyst after synthesis (see. Figure 10) are removed. The adsorption of CO was also strongly dependent on this pretreatment. The catalyst precalcined at 200 OC adsorbed much less CO than that calcined at 475 OC. The adsorption of CO is weak and fully reversible at 50 OC, CO desorbs after 10-15 minutes. All experiments shown in Figure 25 were carried out in an atmosphere of 20 vol% Oz/He, therefore, simultaneous with adsorption, catalytic oxidation of the injected CO occurred and was monitored. The mass spectrometric signals of the COz produced (rn/z=44) show that the catalytic activity of Au/Zr02 was strongly influenced by the catalyst pretreatment. At room temperature (Figure 15B), as well as during slow temperature change (Figure 15C), the catalyst calcined at 475 OC (traces 0)was much more active than that calcined at 200 OC (traces O in Figure 15C). 6.3.2 Adsorption of NH3 and NO on manganese-cerium mixed oxides P U ~ S ~ T Acan @ be used not only for determination of the amount of adsorbed species, as described above, but also for gaining some insight into the mechanisms of gas-solid reactions in which adsorption of reactive species plays an important role. This is illustrated by studies of the selective catalytic reduction (SCR) of NO by ammonia over manganese-cerium mixed oxide that elucidated the relation between adsorption, redox and catalytic behaviour of the investigated samples [39]. All these reactions were investigated by PU~S~TA'. To investigate the mechanism of low-temperature SCR over Mn-Ce mixed oxides catalysts a series of experiments were performed in which the sequence of reactant (NO and NH3) pulses was changed. Representative examples are shown for catalyst (MnOx)o.s(CeOz)o.~ in Figure 26A. First the catalyst was exposed to NO pulses till saturation by NO, was achieved, then NH3 pulses were injected and, because they resulted only in adsorption, no nitrogen production was observed. However, upon injection of further NO, pulses the adsorbed ammonia reacted with the NO, in the gas phase until complete consumption of the adsorbed ammonia by the SCR reaction occurred. After four NO, pulses no further significant nitrogen formation could be detected and the catalyst mass (TG) reached the same value as before the introduction of ammonia, corroborating that all the adsorbed ammonia had reacted with NO,. To further elucidate the role of adsorbed ammonia in the SCR reaction, several NH3pulses were injected in the presence of an excess of NO, under an oxidative atmosphere (5% O2 in helium) after complete adsorption of both NO, and ammonia as presented in Figure 26B. First, one pulse of ammonia and four pulses of NO, were simultaneously injected resulting in the best nitrogen yield (72 %). The TG curve shows first a distinct mass gain due to mainly physisorption of both gases, followed by mass loss due to the consumption of

adsorbed ammonia, which reacts with the gaseous NO, species. Then one pulse of ammonia was introduced (pulse No. 2) in order to readsorb the ammonia removed in the preceding SCR reaction. 4NOx

B

4

=! .

0 4 ~ 0 , m

~

i

II. NH' N2 \

D

Imin

\

72%

Uj

At

At Imin

1-k

a: 7 0 0

37% 0%

h

55% 0%

sII

-P? E

.-

time I min

Figure 26. Pulses of NO and NH3 injected at 100 OC over a (MnOX)o.5(Ce02)o.5 catalyst. For detailed explanation of the procedure applied see text. As a consequence, the mass of the catalyst reached the same value as before the SCR reaction. During the ammonia pulse no nitrogen production was observed, indicating that NH3 was adsorbed and did not react in the gas phase with adsorbed NO, species. In the third pulse (No. 3 in Figure 26B) 4.00 cm3 NO, was injected and this led to significant nitrogen production, but the yield was half of that observed when the gaseous NO, reacted with chemi- and physisorbed ammonia (pulse No. 1). Then the ammonia pulse was repeated (pulse No. 4) in order to reach maximal irreversible ammonia adsorption. During the fifth pulse (No. 5 in Figure 26B) injection of an ammonia pulse was followed by injection of 4.00 cm3 NO, delayed by one minute. This delay decreased the amount of physisorbed ammonia by about 20% (see Figure 26 B) due to desorption and lead to about 24% lower nitrogen yield than obtained during simultaneous introduction of both gases. However, the yield was still higher compared to that originating from pulse No. 3 where only irreversibly adsorbed ammonia reacted with NO,(g). These and a series of analogous investigations indicated that in the absence of NO, in the gas phase, no significant nitrogen formation occurred. Based on this observation we could conclude that the dominant mechanism of SCR on the manganese-cerium mixed oxides is an Eley-

Rideal type mechanism where adsorbed ammonia reacts with NO, from the gas phase and that a Langmuir-Hinshelwood type mechanism seems to be of minor importance. When NO, gas was injected into the system containing manganese-cerium oxides on which ammonia had been preadsorbed, the nitrogen formation was significantly lower (about 50%) than obtained during simultaneous introduction of NO, and ammonia into the carrier gas. This finding confirms that weakly, i.e. reversibly adsorbed ammonia species also contributed significantly to the higher yield of the SCR reaction. This is further corroborated by the fact that after removal of weakly adsorbed ammonia (about 10 minutes after ammonia injection), a NO, pulse resulted in much lower nitrogen production (see pulse 3, Figure 26B).

6.4. Miscellaneous applications Pulse thermal analysis has also been applied in investigating the adsorption : - of NO over Pt/A1203and A1203[50 ] - of NO over Ir black [46 ] - combined with decomposition of hydrocarbons on unsupported Ir [25] - of ammonia on unsupported Co-based catalyts [5 11 7. CONCLUSIONS Pulse thermal analysis extends the versatility of conventional thermoanalytical methods by providing a means for studying differential reaction progresses. This advantage is combined with all the opportunities of thermogravimetry, differential thermal analysis or differential scanning calorimetry and evolved gas analysis. The primary benefits of the new method are: (i) Injection of a known amount of gas into the system during measurement allows in situ quantitative calibration of the MS or FTIR signals which greatly increases the opportunities especially for the analysis of complex and multicomponent systems. pulse^^@ enables the introduction of a well-defined amount of probe gas into TA-MS or TA-FTIR systems at a desired temperature (nonisothermal-) or time (isothermal mode). Because the calibration can be done during the course of the reaction, the accuracy is significantly improved compared to off-line calibration methods. Influences of changing experimental conditions can be accounted for. The possibility of in situ calibration of the spectroscopic signals with pulse^^' increases greatly the potential of the coupled methods. By conventional analysis, only the total amount of an analyzed species can be measured, and it is impossible to resolve the species evolved during multistep reactions. The range

in which the target species can be determined is very broad, between tens and 0.1 mass%, or, in certain cases, even 0.01 mass%. The method is fast (only one or two pulses are required), simple (no extensive calculations) and possesses good reproducibility. Measurements can be easy confirmed by in situ calibration by using decompositions of solids with wellknown stoichiometry. User libraries of spectroscopic data can be set up, which aid in qualitative and quantitative interpretation of the experiments. (ii) Monitoring gas-solid processes, corresponding to a specific extent of reaction at a desired temperature, is possible. The reaction can be stopped at any point between pulses, enabling the relationship between the composition of the solid and the reaction progress to be determined. This differential feature of investigating gas-solid reactions allows very small extents of the processes to be studied. (iii) The potential for monitoring simultaneous changes in mass, thermal effects, composition and amount of gaseous reactants and products under pulse conditions enables information concerning both gas and solid phases to be obtained simultaneously. In studies of adsorption phenomena, the total characterization of the interaction between adsorbate and adsorbent requires the pretreatment of the solid at an optimal temperature. Proper selection of the temperature of the adsorption and the determination of the strength of the adsorbed species is done by temperature programmed desorption. P U ~ S ~ T A @ makes it possible to carry out all these procedures in the same experimental setup, a commercial thermoanalyser. 8. REFERENCES

M. Maciejewski, C. A. Muller, R. Tschan, W.-D. Emmerich and A. Baiker,Thermochim. Acta, 295 (1997) 167. M. Maciejewski and A. Baiker, Thermochim. Acta, 295 (1997) 95. B. Roduit, J. Baldyga, M. Maciejewski and A. Baiker, Thermochim. Acta, 295 (1977) 59. J. Wang and B. McEnaney, Thermochim. Acta, 190 (1991) 143. D. Price, D. Dollomore, N.S. Fatemi and R. Whitehead, Thermochim. Acta, 42 (1980,323. M. Yoshimura and E.Tajma, Thermal Analysis, Vol. 1 (Ed. H. Chihara) Heyden, London, 1978, p.71. M. Muller-Vonmoos, G. Kahr and A. Rub, Thermochim. Acta, 20 (1977) 387. W. Diinner and H. Eppler, Thermal Analysis, Vol. 3 (Ed. I. Buzas), Akademiai Kiado, Budapest, 1975, p.1049.

F. Eigenrnann, M. Maciejewski and A. Baiker, J. Therm. Anal. Cal., 83 (2006) 321.

F. Eigenmann, M. Maciejewski and A. Baiker, Thermochim. Acta, 440 (2006) 8 1.

M. Maciejewski, P. Fabrizioli, J.-D. Grunwaldt, O.S. Becker and A. Baiker, Phys. Chem. Chem. Phys., 3 (2001) 3846. M. Maciejewski, A. Baiker, E. Kapsch, M. Hollering and W.-D. Emmerich, GIT Laboratory Journal, 45 ( 200 1) 6 16. C. Stocker, M. Schneider, and A. Baiker, J. Porous Mater., 2 (1966) 323. M. Maciejewski, A. Baiker, K. Conder, Ch. Kriiger, J. Karpinski and E. Kaldis, Physica, C227 (1994) 343. M. Maciejewski, E. Ingier-Stocka, W.-D. Emmerich and A. Baiker, J. Therm. Anal. Cal., 60 (2000) 735. G. Jenzer, T. Mallat, M.Maciejewski, F. Eigenmann and A. Baiker, Appl. Catal. A, 208 (2001) 125. R. Strobel, M. Maciejewski, S.E. Pratsinis and A. Baiker, Thermochim. Acta, 445 (2006) 23. M. Huber, W.J. Stark, S. Loher, M. Maciejewski, F. Krumeich and A. Baiker, Chem. Commun., 2005,648 H.M.Yu,Q.H.ZhangQH, L . J . Q i , C . W . L u , T . G . X i a n d L . L u o , Thermochim. Acta, 440 (2006) 195. M. Feist and E. Kemnitz, Thermochim. Acta, 446 (2006) 84. M. Piacentini, M. Maciejewski and A. Baiker, Appl. Catal., B 60 (2005) 265.

M. Piacentini, M. Maciejewski and A. Baiker, Appl. Catal., B 66(2006) 126.

M. Casapu, J.-D. Grunwaldt, M. Maciejewski, M. Wittrock, U. Gobel and A. Baiker, Appl. Cat., B 63 (2005) 232 E. Ingier-Stocka and M. Maciejewski, Thermochim. Acta, 432 (2005) 56. C. Wogerbauer, M. Maciejewski and A. Baiker, Appl. Cat., B 34 (2001) 11. D. Brandova, M. Maciejewski and W. Keller, J. Therm. Anal. Cal., 56 (1999) 893.

M. Maciejewski, W.-D. Emmerich and A. Baiker, J. Therm. Anal. Cal., 56 (1999) 627.

H.H. Voge and L.T. Atkins, J. Catal., 1 (1962) 171. M.J. Murray, J. Phys.Chem., 35 81931) 1011. E. Rocchini, A. Trovarelli, J. Llorca, G. W. Graham, W.H. Weber, M. Maciejewski and A. Baiker, J. Catal., 194 (2000) 46 1.

W.J. Stark, M. Maciejewski, L. Madler, S.E. Pratsinis and A.Baiker, J. Catal., 220 (2003) 35. W.J. Stark, J.-D. Grunwaldt. M. Maciejewski, S.E. Pratsinis and A. Baiker, Chem. Mater., 17 (2005) 3352. M. Piacentini, M. Maciejewski and A. Baiker, Appl. Catal., B 59 (2005) 187. M. Piacentini, M. Maciejewski and A. Baiker, Appl. Catal., B, submitted. J.-D. Grunwaldt, M. Maciejewski and A. Baiker, Phys. Chem. Chem. Phys., 5 (2003) 1481 C.A. Miiller, M. Maciejewski and A. Baiker, Catal. Today, 47 (1999) 245. C.A. Muller, M. Maciejewski, R. Koeppel, R. Tschan and A. Baiker, J. Phys.Chem., 100 (1966) 20006. J.-D. Grunwaldt, M. Maciejewski, O.S. Becker, P. Fabrizioli and A. Baiker, J. Catal., 186 (1999) 458 F. Eigenrnann, M. Maciejewski and A. Baiker, Appl. Catal., B 62 (2005) 311. H. Schulz, W.J. Stark, M. Maciejewski, S.E. Pratsinis and A. Baiker, J. Mater. Chem., 13 82003) 2979. Z. Opre, J.-D. Grunwaldt, M. Maciejewski, D. Ferri, T. Mallat and A. Baiker, J. Catal., 230 (2005) 406. M.D. Wildberger, M. Maciejewski, J.-D. Grunwaldt, T.Mallat and A. Baiker, Appl. Catal., A 179(1999) 189. C. Wogerbauer, M. Maciejewski, A. Baiker and U. Gobel, Top. Catal., 16/17 (2001) 181. C. Wogerbauer, M. Maciejewski, M.M. Schubert and A. Baiker, Cat. Letters, 74 (200 1) 1. C. Wogerbauer, M. Maciejewski, A. Baiker and U. Gobel, J. Catal., 201 (2001) 113. C. Wogerbauer, M. Maciejewski and A. Baiker, J. Catal., 205 (2002) 157. M. Richter, U. Bentrup, R. Eckelt, M. Schneider, M.M. Pohl and R. Fricke, Appl. Catal., B 5 1 (2004) 261. F. Eigenmann, M. Maciejewski and A. Baiker, Thermochim. Acta, 359 (2000) 131. C.A. Muller, M. Maciejewski, T. Mallat and A.Baiker, J. Catal, 184 (1999) 280. M. Piacentini, M. Maciejewski and A. Baiker, Top. Catal., 30-3 l(2004) 71. A. Fischer, M. Maciejewski, T. Burgi, T. Mallat and A.Baiker, J. Catal., 183 (1999) 373.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 5 THE QUARTZ CRYSTAL MICROBALANCE Allan L. Smith Masscal Corporation, 96 A. Leonard Way, Chatham, MA, 02633 USA 1. HIGH SENSITIVITY BALANCES: THEIR ROLE IN THERMAL ANALYSIS AND CALORIMETRY

In a brief history of classical analytical chemistry, Beck [I] defines gravimetry and emphasizes it central role in chemistry: "Gravimetry is the determination of an element or species through the measurement of the mass of a well-characterized insoluble product of a definite chemical reaction involving that element or species. ... Tracing the history of gravimetry amounts to tracing the early history of chemistry." After all, without a good balance John Dalton would not have been able to measure the ratio of combining masses which led him to the atomic theory of matter. Virtually every laboratory today doing chemistry or materials science has access to balances of varying precision, and many companies exist to provide balances and to develop newer, more sensitive ones. The term microbalance is often used to describe hi h precision analytical balances of the type offered by several manufacturers a The mechanical technology used to achieve these sensitivities has been described as a "force-to-current converter"". A practical readability and reproducibility limit of mechanical microbalances balances commercially available is g or 1 pg, and these balances are delicate and sensitive to vibrations. Balances and thermogravimetry are described in Chapter 4 of Volume 1 of this Handbook.

'.

a

Thenno Scientific, Cahn http:Nwww.thermo.com/com/cda~product/detail/l,1055,1000001000408,00.html Mettler-Toledo, XP-26 Microbalance series, http://us.mt.com/mt~products/

Most chemists and materials scientists are not aware that an alternative gravimetric technology has been available for decades that provides sensitivities at least 3 orders of magnitude lower than mechanical balances, using a piezoelectric sensor that is rugged, inexpensive, and operates at high frequencies relatively immune to vibrations: the quartz crystal microbalance. The purpose of this chapter is to describe the capabilities and limitations of the quartz crystal microbalance and to discuss its uses in thermal analysis and calorimetry. 2. EARLY HISTORY OF THE QUARTZ CRYSTAL MICROBALANCE

Piezoelectricity is defined as electric polarization produced by mechanical strain in certain crystals, the polarization being proportional to the strain [2]. The Curies first observed piezoelectricity in 1880 as a potential difference generated across two surfaces of a quartz crystal under strain [3]. The converse piezoelectric effect, the deformation of a piezoelectric material by an applied electric field, was predicted by Lippman [4]. Thus, when a thin wafer from a piezoelectric crystal such as quartz is placed in an alternating electric field of the right frequency it will oscillate in a mechanically resonant mode of the wafer. The resonance frequency depends upon the angles with respect to the optical axis at which the wafer was cut from a single crystal and inversely on the crystal thickness. The angle most commonly chosen is referred to as the AT cut, 35' 15' from the Z or optic axis of the crystal. The AT-cut angles are chosen so that the temperature dependence of the resonant frequency is essentially zero at 25°C. These thin quartz plates with attached electrodes are called transverse shear mode (TSM) resonators. The development and applications of quartz plate TSM resonators is a venerable field in electrical engineering. In the early 1920s the National Bureau of Standards (U.S.) began studies of quartz-crystal oscillators as frequency standards [5]. To meet the growing demand for better accuracy, NBS sought outside partners, and began collaboration on oscillators with the Naval Research Laboratory and Bell Telephone Laboratories. In 1929, Bell Labs delivered four complete temperaturecontrolled 100 kHz oscillators to NBS, and these oscillators quickly became the national primary standard of radio frequency. By 1952 the NBS laboratory had a large number of oscillators, and the measurement uncertainty had been reduced to about 2 parts in lo8. An entertaining account of the history of the quartz crystal industry in the U.S [6] indicates the critical role that quartz resonators played in the development of radio

communications during World War 11. Quartz resonators are presently found in many commercial products from quartz timepieces to ultra-stable frequency counters. Quartz plate resonators have been used as sensitive microbalances with subnanogram sensitivity for thin adherent films since the late 1950s, following the pioneering work of Sauerbrey (7), who coined the term quartz crystal microbalance (QCM) (see Figure 1).

-

I=

'W'

0-

-

Cmmonly called the "QCM" or quartz crystal microbalance 2-20 MHz fundamental resonance frequency

rt,

Applying an RF voltage across the piezoelectric quartz crystal induces oscillation in the crystal at its acoustical resonant frequency

AT-cut quartz

!.:thickness-shear mode v

Figure 1. Thickness shear-mode oscillators. Based on "Resonant Piezoelectric Devices as Physical and Biochemical Sensors", F. Josse, and R. W. Cemosek, presented at the 2002 IEEE International Frequency Control Meeting as a Symposium Tutorial, New Orleans, LA. http://www.ieee-uffc.org/fkacontrol/tutse-Cnosekfiles/frame.htm

3. THE LITERATURE OF THERMAL ANALYSIS AND OF THE QUARTZ CRYSTAL MICROBALANCE

It is useful to compare the history of the development of thermal analysis with the history of the development of the quartz crystal microbalance. Scopus (www.Scopus.com) is the largest online abstract and citation database of scientific and technical research literature and quality web sources. A search using the term thermal analysis and the terms QCM or quartz crystal microbalance in Scopus, found all references dating to 1966. A Scopus literature search on the latter terms

only showed about 4500 citations since 1966. Figure 2 shows the number of publications per year for thermal analysis and for the quartz crystal microbalance. The QCM publications remain at 1-4% of the thermal analysis publications until about 1990, when they gradually increase to 1518% by 2005 (reasons for this expansion are discussed below). Yet the number of publications involving BOTH thermal analysis and QCM is miniscule (5, all in the last five years). The number involving both QCM and calorimetry is not much larger (28, again in the last few years). These data show that the techniques of thermal analysis and the quartz crystal microbalance have developed independently of each other until very recently.

TA pubs -- n OCM pubs %

I

Figure 2. The number of publications per year for thermal analysis and for the quartz crystal microbalance. The following figures present a breakdown of the published references in thermal analysis and in QCM by professional field and how they have changed since 1971. Figures 3-6 show all thermal analysis publications in the time intervals 1966-1980, 1981-1990, 1991-2000, and 2001-2003. For comparison, Figures 7-10 show the fields assigned for QCM in the same time intervals.

I3 Material Science 47%

El Chemistry 4% 3 b Chemical Engineering 3% Engineering (other than Chemical) 37%

N Physics & Astronomy 3%

13 Biochemistry 2% E4 Energy 2% R Medicine 2%

Figure 3. Fields of thermal analysis publications 1966-1980.

1 Material Science 32% Chemistry 7%

Chemical Engineering 4%

Engineering (other than Chemical)

43% N Physics & Astronomy 5% I3 Energy 3%

MBiochemi~try2%

EEarth 8 Planetary Science 2%

Figure 4. Fields of thermal analysis publications 198 1-1990.

138

Bl Chemlcal Engineering 9% Bl PhyslcS8 Astronomy 9%

sl Biochemlstly 2%

Figure 5. Fields of thermal analysis publications 1991-2000.

0 I3 Chemistry 22% mchemlcal Englnrnng 12% slEngineenng(other than Chemical) 11% WPhyrlcr a Arfmnomy 10%

mBiochemlslry 3%

Figure 6. Fields of thermal analysis publications 2001-2003.

I3 Chemlslry 22%

m Chemical Englneerlng 125h Fd Engineering (other than Chemical) 11%

El Physiw & Astronomy 10% Q Energy 3%

El B1ochem#stw3% B Earn 8 Planetary Science 2%

El Pharmaceubcais 1%

I

I

Figure 7. Fields of QCM publications 1966-1980.

13 Chemistry 14%

Ea

Chemical Engineering 11%

m

Engineering (other than Chemical) 32%

Physics & Astronomy 23% Medicine 1%

Figure 8. Fields of QCM publications 1981- 1990.

El Chemlstw 36%

u Chemical Englneerlng15% rn ~ngineering(omer than chemicd) 1 1 % 5 Physa a listammy ,I%

u Blochem. Genetics a MolscuiarBi0 W

Figure 9. Fields of Q C M publications 1991-2000.

1

Matelial Science 18%

I

Chemistty 37%

Chemical Engineering16%

Engineering(other than Chemical) 9%

H Physics &Astronomy 13% Q Biochem, Genetics & Molecular Bio 7%

I

I

Figure 10. Fields of QCM publications 2001-2003. For thermal analysis, materials science has the largest number of publications in each time interval. The next largest category is engineering in 1966-1980, but this shifts to chemistry and chemical engineering in 2001-2003.

For the quartz crystal microbalance, however, the pattern of dominant fields is quite different. 78% of the early literature (1966-80) is in engineering (predominantly electrical) and physics and astronomy, with only 18% in materials science and chemistry. The 61 references before 1980 are in engineering and physics journals, and deal with applications of the QCM in vacuum science and technology to determine the mass and thickness of deposited metallic films. The 1980s show a significant increase in the fraction of QCM publications in materials science and chemistry (to 33%) and in chemical engineering (to 11%). In the 1990s the field with the largest number of publications on QCM is chemistry, with materials science second. This trend is continued in 2001-2003. During the mid-tolate 1980s, as discussed below, the QCM was shown to be able to function as a sensor when immersed in water or other liquids or solutions. This discovery led to rapid increases both in the total number of publications (Figure 2) and the number in chemistry and chemical engineering in such sub-disciplines as electrochemistry, analytical chemistry, and surface chemistry. Table 1 shows the journals with the most thermal analysis publications, 19652005, and Table 2 shows the Journals with the most QCM publications, 1965-2005. The complete lack of overlap between these two lists is another indication that the quartz crystal microbalance is relatively unknown to the thermal analysis and calorimetry community. A later section in this chapter describes research that bridges the two fields. Table 1. Journals with the most thermal analysis publications, 1965-2005, in decreasing order Journal title Journal of Thermal Analysis and Calorimetry Journal of Thermal Analysis Thermochimica Acta Journal of Applied Polymer Science Journal of Alloys and Compounds Journal of Materials Science Polymer Journal of Noncrystalline Solids Journal of the American Ceramic Society Journal of Materials Science Letters Materials Letters Materials Research Bulletin

Journal of Solid State Chemistry Solid State Ionics Journal of Crystal Growth

Table 2. Journals with the most QCM publications, 1965-2005, in decreasing order. Journal name Langmuir Journal of the Electrochemical Society Electrochimica Acta Sensors and Actuators, B: Chemical Analytical Chemistry Biosensors and Bioelectronics Proc. SPIE - intl. Soc. For Optical Engineering Journal of Physical Chemistry Synthetic Metals Analytical Chimica Acta Thin Solid Films Journal of Colloid and Interface Science Journal of the American Chemical Society Macromolecules

4. PRINCIPLES OF OPERATION OF THE QUARTZ CRYSTAL MICROBALANCE (QCM) The resonant frequency of the fundamental acoustic mode of vibration of a quartz TSM resonator of thickness hq is

where ~q and pqare the shear modulus and the density of quartz, respectively [S]. The shift in frequency due to deposition of a film of the same acoustic impedance

as quartz is proportional to the deposited mass per unit area of the film, AmlA, a relationship first given by Sauerbrey [7]:

In equation (2), pf and hf are the density and the thickness, respectively, of the deposited film. For an AT-cut, 5 MHz crystal at room temperature, C = 56.6 ~ z / ( ~ g / c r nBecause ~). it is easy to measure frequencies to %.01Hz, changes in mass per unit area of < 1 ng/cm2 are measurable with the QCM. The electrical characteristics of a QCM are well represented by a simple RLC damped resonator equivalent circuit [S], termed the Butterworth-Van Dyke equivalent circuit (Figure 11).

Figure 11. Butterworth - Van Dyke equivalent circuit for the QCM. The series resonant frequency& of an RLC resonant circuit is given by:

f=LF "

2n L C ,

Here, L is the dynamic inductance, a measure of the oscillating mass of the quartz, and C, is the dynamic capacitance or series capacitance, a measure of the elasticity of the oscillating body [9]. The resistance of the RLC circuit is related to the quality factor Q (the width of the resonance), the dissipation D [lo], and the full width at halfmaximum Sf of the resonance by the relationship:

where R, is the motional or dynamic resistance of the quartz resonator. For an uncoated resonator, R,, is a measure of the internal frictional damping of the quartz. Coatings provide additional damping. The width of the resonance for an uncoated 5 MHz resonator is about 50 Hz, (i.e. Q = lo5), and the damping within the quartz that gives rise to this broadening can be determined by measuring the motional resistance R of the uncoated resonator, typically 10 ohms. When thin, stiff films are deposited on the QCM surface the increase in R is small, but softer, thicker films (i.e., rubbery polymers 5-20 pm thick) can increase R by hundreds or even thousands of ohms. Since the mid 1980s [ l l ] , it has been recognized that TSM resonators can also operate in fluid media if electronic oscillator drivers of suitable gain are employed to excite the resonator and to offset the losses due to damping of the resonator by the fluid. When immersed in water, the motional resistance of a 5 MHz TSM resonator increases to -360 ohms. For an infinite viscoelastic liquid in contact with the TSM, the frequency shift is:

where pl and 111 are the density and the viscosity of the liquid, respectively. For a 5 MHz QCM immersed in water at 25"C, Ahq = -710 Hz. The frequency responses of a QCM to different media are illustrated in Figure 12. More complete theories of the operation of transverse shear mode resonators have been given by Kanazawa [11,12], Martin et al. [13,14], Lucklum and Hauptmann [IS- 171, Voinova, Jonson and Kasemo [IS], Johannsmann [19,20], Tsionsky [2 11 and Arnau [22]. In these theories, the electrical impedance of the QCM, Z, , is complex. The impedance of a TSM resonator damped by a finite viscoelastic film can be described as the sum of two complex impedances:

where the acoustic load impedance due to the film, ZL,contains both an inductive and a resistive part. Equations are given relating the complex impedance of a TSM resonator, damped by a finite viscoelastic film, to four parameters characterizing

Rigid Layer

Air

Viscoelastic Layer

- Air ----- Rigid Layer

.

I

Viscous Liquid

I I I

-

i

C

.E

2

I 11

...'.. .................... --I

I

I

I

I

I

1 I

I

~

I

~

I

n

I

I

Frequency

Figure 12. The frequency responses of a QCM to different media. the film: the thickness hfi the density pf, the shear storage modulus G; and the shear loss modulus Gf". Shear moduli are functions of the frequency at which they are measured, so for TSM resonators, G j and Gf"are determined at the QCM series resonant fi-equencyx or one of its overtones. The acoustic load impedance, ZL, of a film measured at frequency w = 27cJ is:

where qis the (complex) acoustical phase shift:

Here the modulus, Gf, is also complex: Gf= Gf' + iGf".

Two convenient experimental measures of the film properties are: (a) the difference in resonant frequency Af =Acrystal + film) -Acrystal) (b)the difference in motional resistance AR = R(crysta1 + film) - R(fi1m). For thin films, the frequency shift Af is proportional to pfhf,the mass per unit area of the film (the Sauerbrey relation, equation (2)). It is possible to define an "ideal rigid mass layery'with small acoustic phase shifts [23], for which the acoustic load impedance is purely imaginary and the Sauerbrey limit is reached. The relationship between ZLand the measured quantities is [24]:

Herefo, Z,,, and L, are the resonant frequency, acoustic impedance, and motional inductance of the bare quartz crystal. Both Af and AR are zero in the limit of zero film thickness. Voinova et al. [I 81 and Johannsmann [19,20] present equations for both quantities in a power series expansion in the thickness hf. The result for Af and AR, to third order in hf, is:

The compliance can be used instead of the modulus to quantify storage and loss behaviour in viscoelastic solids. The shear storage compliance is defined as:

and the shear loss compliance is defined as:

Thus, the thin-film limit equations can be rewritten as:

and

Equation (14) is useful in estimating the thickness of compliant films at which deviations from the Sauerbrey equation are noticeable. Equation (15) is useful in interpreting motional resistance measurements of thin films. In the thin-film limit, the motional resistance change is proportional to the square of the film density, the cube of the film thickness, and the loss compliance of the film. For a 5 MHz QCM, typical values for L, and 2, are 0.0402 Henry and 8.84 x lo6 pa slm, respectively.

5. DETECTION ELECTRONICS 5.1. Simple QCM driving circuits Simple circuits to drive quartz resonators were developed in the 1960s and 70s. The most common and least expensive detection electronics involves such a driving circuit and a frequency counter. The series resonance frequency is recorded. For Figure 11 the series resonant frequency is given by equation (3), whereas the parallel resonance frequency is given by:

where

The static parallel capacitance, C,, is the capacitance between the quartz electrodes, the crystal holder, and the leads to the driving circuit. Typical values are between 4 pF and 30 pF. For a typical 5 MHz quartz oscillator with C, = 20 pF, L1

= 0.033 H, C , = 0.0307 pF, and R,= 10 S1, the series resonant frequency& = 5.0000

MHz, the quality factor is 96500, the difference between parallel and series resonant frequencies is 3.30 kHz, and the resonance line width Gf is 52 Hz. When QCMs began to be used with liquids, there was a need to develop more specialized circuits. The damping produced by liquids causes a decrease in Q by 23 orders of magnitude and a corresponding broadening of the resonance. Eichelbaum et al. [9] give references to handbooks that present standard crystal oscillator circuits. They discuss the developments needed for interface circuits to operate with a fluid in contact with one of the crystal faces. 5.2. Frequency and damping measurements Beginning in the 1990s, driving circuits that produce both resonance frequency outputs and analogue outputs that measure the motional resistance began to be commercially available. A list of companies now providing such electronics is given in Table 3. The advent of motional resistance data leads, through equation (15) above, to a direct determination of the shear-loss compliance, J",of the sample film at 5 MHz.

5.3. Impedance analysis Using an impedance analyzer, it is possible to measure both the real and imaginary parts of 2, for TSM resonators [25] at the resonant frequency and many of its overtones. With suitable analysis software these measurements yield the shear modulus, G', and the loss modulus, G", of polymer films deposited on the QCM [19,20,26,27].

6. IS THE TRANSVERSE SHEAR MODE RESONATOR A TRUE MICROBALANCE? This is still a controversial question, after many years of discussion. After all, analytical balances do not operate linearly when immersed in water because mechanical motions are strongly damped, and the QCM is no exception. Here are verbatim quotes from some of the leaders in the field: V. Tsionsky [28]: " Is the quartz crystal rnicrobalance really a microbalance? For one thing, it should rightly be called a nano-balance, considering that the sensitivity of modern-day devices is on the order of 1-2 ng/cm2 and could be pushed further, if necessary. More importantly, calling it a balance implies that the Sauerbrey

equation applies strictly, namely that the frequency shift is the sole result of mass loading. It is well known that [in the case of operation of the resonator in a liquid] this is not the case, and the frequency shift observed could more appropriately be expressed by a sum of terms of the form:

where the different terms on the right hand side of this equation represent the effects of mass loading, viscosity and density of the medium in contact with the vibrating crystal, the hydrostatic pressure, the surface roughness, the slippage effect, and the temperature, respectively, and the different contributions can be interdependent. It should be evident from the above arguments that the term quartz crystal microbalance is a misnomer, which could (and indeed has) led to erroneous interpretation of the results obtained by this useful device. It would be helpful to rename it the quartz crystal sensor (QCS) which describes what it really does - it is a sensor that responds to its nearest environment on the nano-scale. However, it may be too late to change the widely used name. The QCM or its analogue in electrochemistry, the EQCM, can each act as a nano-balance under specific conditions, but not in general."

R. Lucklum [17]: "The name microbalance implies that acoustic sensors measure mass or mass changes only. Indeed, in many applications acoustic sensors are used to convert a mass accumulated on the surface into a frequency shift. In chemical and, especially, biochemical applications, however, this basic understanding of the sensor principle can easily lead to misinterpretation of experimental results, especially when working in a liquid environment. It also hinders recognition by the experimenter of the outstanding capabilities of quartz crystal resonators, sensors, and other acoustic devices not available to other sensor principles. In a more general view acoustic sensors enable sensitive probing of changes within films attached to the transducer surface and at solid-solid and solid-liquid interfaces.. ." V. Mecea [29]: "This article reveals that the local mass sensitivity of the quartz crystal microbalance (QCM) depends on the local intensity of the inertial field developed on the crystal surface during crystal vibration. ...The maximum intensity of this field in the center of the quartz resonator is a million times higher than the intensity of the gravitational field on the Earth. Experimental results reveal that the product of the minimum detectable mass and the intensity of the field acting on that mass is a constant for both QCM and beam balances, explaining thus why QCM is

more sensitive than conventional analytical balances. It is shown that the apparent effect of liquid viscosity on the frequency response of a quartz crystal resonator is, in fact, the result of the field intensity dependency of the mass sensitivity, being thus clear that QCM is really a mass sensor."

M. Thompson [30]: "For some considerable period of time it was assumed that the TSM structure employed in water simply responded to added or lost material on the sensor surface.. . In recent times it become clear that the device is exquisitely sensitive to changes in interfacial conditions. This observation is of great significance when biochemical interactions are detected using the TSM device. This technology involves the placement of biochemical receptors, such as antibody or nucleic acid species, on surfaces where the coupling processes described in the present paper are prevalent. Accordingly, modulation of such coupling by biochemical interactions instigated at the sensor surface constitute a new and highly sensitive detection strategy." 7. PRACTICAL DETAILS 7.1. Calibration For a film interacting with a gas, equation (14) provides a quantitative means of determining if any calibration is needed for the mass measurement. The numerator of the factor outside the brackets can be expressed as rndAf, the mass per unit area of the film. If the second term inside the brackets is << 1, then the Sauerbrey equation is obeyed and the QCM is a true gravimetric device. The correction factor is proportional to the film density, the film's shear compliance, and the square of the film thickness. Because many polymers at room temperature are below their glass-transition temperatures at frequencies of 5 MHz, the time-temperature superposition principle [31,321 can be used to estimate the magnitude of S and thus the thickness range over which the Sauerbrey model yields accurate results. The impact of the film's compliance on the useful range of sample thickness is dramatic. At 20 "C, for a glassy polymer such as polystyrene, the mass correction factor is 1% for a thickness of 45 pm. For a rubbery polymer such as polyisobutylene, the correction is 1% for a film 4 pm thick. For QCMs interacting with liquid films, the problem is much more complex. Tsionsky et al. [21] discuss thoroughly the various cases in which it is still possible to apply the Sauerbrey equation, but they are much more limited than in the case of fildgas interaction.

7.2. Comparison of gravimetric and Sauerbrey masses As long as the correction factor in equation (14), relating frequency shift to change in mass per unit area, is small, the QCM functions as a true balance with nanogram sensitivity. To prove this, Masscal Scientific Instruments determined the resonant frequencies of twenty uncoated QCM crystals, then weighed the crystals with a 5-place analytical balance (Mettler 261). Each crystal was spray-coated uniformly with a multi-component polymer coating used in the electronics industry. The resulting film thickness was 1 pm with a standard deviation of 30%. Each crystal was dried at 200°C for 2 minutes to drive off residual solvents. The coated crystals were reweighed, and each mass difference was divided by the area of the coated crystal, 5.06 cm2,to determine a gravimetricfilrn mass per unit area for each sample. From the measured frequency of each coated crystal and equation (2), the Sauerbreyfilm mass per unit area was determined. Comparison of the two sets of results is shown in Figure 13. The mean difference in mass between the gravimetric and the Sauerbrey masses was 2.4 4.6 %, which is well within the experimental error of the less-accurate gravimetric mass determination. Non-uniformity in film coating and incomplete drying also may contribute to the scatter in the data.

-

*

Mass of Film Sprayed Uniformly on OCM 2 1.8 1.6

ri.4

2 1.2

1

m

0.e 0.6 0.4

0.2

0

Permt DMerence behveeffinvlmshic and Sauerbrey Mass 12

70 8 6

%U

8

2 0

-2 -4 -6 -8

Figure 13. Comparison of the gravimetric and the Sauerbrey masses for a QCM.

7.3. Sample preparation In order for both mass and heat-flow sensors to operate, the thin-film sample must adhere to the top surface of the QCM and be of uniform thickness. The mechanical behaviour of films on the quartz microbalance has been modeled by Kanazawa(l2), who examined the amplitude of the shear displacement in the quartz crystal and in the overlying film for several cases. For a 1 volt peak RF applied voltage typical of the Stanford Research systems* oscillator driver, the amplitude of the shear wave of a bare crystal is 132 nm. Mecea [29] has calculated the inertial acceleration at the centre of a similar quartz resonator, and finds that it is roughly lo6 g, where g is the gravitational constant. At these extremely high accelerations, powder or polycrystalline samples do not follow the transverse motion of the QCM surface and cannot be used without being physically bound to the surface with a thin adhesive layer. Many methods have been used to prepare thin film samples for the QCM. For metallic and inorganic materials, vacuum or electrochemical deposition has been traditionally used, but methods such as sol-gel formation of films can also be employed. Uniform films of polymers can be made by dip coating, drop coating, spray coating, or spin coating. To achieve sample homogeneity and uniformity, the best of these methods is spin-coating, assuming the film material is amenable to this treatment. Self-assembled monolayer (SAM) chemistry [33] can be used to construct planar micro- and nano-structures. The effect of non-uniform thickness of films on the operation of the QCM has been treated by several authors [21,341. Because the measured frequency difference, Af =Acrystal + film) -Acrystal), is proportional to the mass per unit area of the film, the total sample mass is obtained from equation (2) as (AmIA)Af where Af is the area of the film exposed to the gas.

8. CHEMICAL AND BIOLOGICAL APPLICATIONS OF THE QCM 8.1. Film-thickness monitors in vacuum deposition QCMs have been used as film-thickness monitors in vacuum deposition of metals and inorganic solids since the 1970s. The monograph by Lu and Czanderna [35], while over 20 years old, is still a very good summary of early applications of the quartz crystal microbalance in physics and engineering, as well as applications as thickness monitors in the vacuum deposition industry. Well before the full ' see "Commercially Available QCM Systems" below

acoustical models of the past fifteen years were developed, it was realized that for thick deposited films there are deviations from the Sauerbrey equation, which depend upon the acoustical transmission properties of the deposited film. The most successhl method of analyzing frequency shifts of thick films is the Z-match method, reviewed by Benes [36]. The working equation is:

where r n and ~ ??tQare the masses per unit area of the applied film and of the quartz resonator, ZF and ZQ are the acoustical impedances of the film and the quartz,f is the resonant frequency of the composite quartz + film, and fQ is the frequency of the bare quartz resonator. Benes gives a table of acoustical impedances of 53 metals and simple inorganic salts, the hardest of which is iridium and the softest graphite. From equation (14) it can be seen that the acoustical impedance is a measure of the shear-storage compliance, S,of the film. 8.2. The metaYsolution interface in electrochemical cells The first subdiscipline of chemistry in which the QCM was widely applied was electrochemistry. In 1992 Buttry and Ward published a review entitled "Measurement of interfacial processes at electrode surfaces with the electrochemical quartz crystal microbalance", with 133 references [S]. This is the most widely cited paper on quartz crystal microbalances. After presenting the basic principles of AT-cut quartz resonators, the authors discuss the experimental aspects and relation of electrochemical parameters to QCM frequency changes. In their review of the investigation of thin films, they discuss electrodeposition of metals, dissolution of metal films, electrovalency measurements of anion adsorption, hydrogen absorption in metal films, bubble formation, and self-assembled monolayers. The review concludes with a brief section on redox and conducting polymer films. In 2003, Tsionsky, Daikhin, Urbakh, and Gileadi [21] published a very thorough treatment of the metal/solution interface as examined by the electrochemical quartz crystal microbalance, with emphasis on the misinterpretations of data that can occur if the basic physics and chemistry at the interface are not understood. Topics covered include the electrical double-layer/electrostaticadsorption, the adsorption of organic and inorganic species, metal deposition, and the influence of roughness on the response of the QCM in liquids.

8.3. Faraday Society Discussion No. 107,1997 This entire volume has articles devoted to the theory and application of acoustic sensors. The quartz crystal microbalance is discussed in the following articles: L. Daikhin, M. Urbakh: Influence of surface roughness on the quartz crystal microbalance response in a solution - New configuration for QCM studies. pp; 2738. K.K. Kanazawa: Mechanical behaviour of films on the quartz microbalance. Pp. 77-90. 0. WovJ E. Seydel, D. Johannsmann: Viscoelastic properties of thin films studied with quartz crystal resonators. Pp. 91-104. H.L. Bandey, A.R. Hillman, MJ. Brown, S.J. Martin: Visoelastic characterization of electroactive polymer films at the electrode/solution interface. Pp. 105-21. R. Lucklum, P. Hauptmann: Determination of polymer shear modulus with quartz crystal resonators. Pp. 123-40. B.A. Cavic, F.L. Chu, L.M Furtado, S. Ghafouri, G.L. Hayward, D.P. Mack, ME. McGovern, H. Su, M Thompson: Acoustic waves and the real-time study of biochemical macromolecules at the liquidlsolid interface. Pp. 159-76. E.J. Calvo, R. Etchenique, P.N. Bartlett, K. Singhal, C. Santamaria: Quartz crystal impedance studies at 10 MHz of viscoelastic liquids and films. Pp. 141-57. J.K Grate, S.N. Kaganove, V.R. Bhethanabotla: Examination of mass and modulus contributions to thickness shear mode and surface acoustic wave vapour sensor responses using partition coefficients. Pp. 259-83. K-H. Stelllnberger, M. Wolpers, T. Fili, C. Reinartz, T. Paul, M. Stratmann: Electrochemical quartz crystal microbalance in modem corrosion research. Pp. 307-22. V. Tsionsky, E. Gileadi, L. Daikhin, G. Zilberman: Response of the EQCM for electrostatic and specific adsorption on gold and silver electrodes. Pp. 337-50.

8.4. Determination of shear and loss modulus at QCM frequencies The determination of shear storage and loss moduli of thin viscoelastic films with TSM resonators has been reviewed [20,37,38]. Even though the basic physics of damped TSM resonators is well understood, the effort to determine Gf' and Gf" from measurements of frequency shifts and motional resistance changes has been fraught with problems. For very thin, rigid films, the frequency shift contains no information on either Gf' or Gf" because the Sauerbrey limit (equation (2)) is reached. For thicker andlor lossier films the frequency shift and motional

resistance depend on Gf' and G;' in a complex manner that is not obvious by examining the equations. There are some examples of the determination of G' and G" from QCM measurements, however. Lee, Hinsberg, and Kanazawa [26] have used a QCM to measure the motional resistance of a film of the rubbery polymer poly(n-butyl acrylate) (PBA) (T, = -64°C) in contact with both air and water as a function of film thickness from 0.3 pm to 4.0 pm. They prepared a series of poly (n-butyl acrylate) films of thickness from 6 to 4000 nm and measured the motional resistance of the film in both air and water. Using the data from their Figure 5(a) for motional resistance in air, AR was plotted vs the cube of the film thickness, as suggested by equation (15). The resulting plot is linear with an R~ of 0.9996. The slope is 6.1x10-~ohm/nm3 5%. Using a value 1.08 g/cm3 for the density of poly (n-butyl acrylate), equation (15) gives a value for the loss compliance, S',of 5.0 x pa-'. Lee et al. [26] report values based on the Kanazawa relations for the shear storage modulus, G', and shear viscosity, r], of 3 x 10' Pa and 0.13 Pa s, respectively. With the relationship: G" = 27cf r], equation (13) yields a loss compliance, S' = 4.5 x pa-' from these data, which is in good agreement with the value derived, 5.0 x 10" pa-', from equation (15). Using a QCM, Josse et al. [27] have determined the storage and loss modulus of a commercially available W-curable epoxy resin (SU-8-2002). An in-line monitor of the W photopolymerization of 2-hydroxyethyl methacrylate with a photoinitiator of 1-chloroanthraquinonewas demonstrated by Kim et al. [39]. The process by which an organic coating or finish is converted from its formulation to the final hard product involves drying, curing, and film formation, as discussed by Sliva 1401. Drying is the physical evaporation of volatile components from the applied coating. Curing is the cross-linking of polymeric systems to stiffen the film, which can be produced by oxidative, reactive, or catalytic means. Both drying and curing will change the viscoelastic properties of films. Decrolon is a commercial aerosol spray enamel made by Sherwin-Williams. The resin in Decrolon is a vinyl toluene alkyd, and the volatile organic solvents are acetone (20%), toluene (13%) and light aliphatic naphtha (8%), plus propellants. As long as the alkyd enamel film remains under the nitrogen flow, no oxidative cross-linking can occur. If a QCM is sprayed with a thin film of Decrolon and immediately placed in a sample chamber under flowing nitrogen, the solvent evaporates but the film remains tacky because there is no cross-linking. This state is dried but uncured. The QCM capability of the Masscal Scientific Instruments G1 (described below), was used to measure [41] the change in frequency and in motional resistance of the film during its exposure to various partial pressures of toluene

*

vapour in a nitrogen carrier, generated by a mass-flow controller system. The QCM crystal was then removed and allowed to cure in air for 3 days and the same flow protocol was followed. The motional resistance data for the uncured and cured Decrolon film were converted to loss compliance using equation (15) and are plotted in Figure 14 as a function of the mass of toluene per gram of film. The loss compliance of each film increases with toluene content, indicating that toluene is softening the coating (i.e., acting as a plasticizer), as expected. The loss compliance of the cured film is lower than that of the uncured film for all values of the toluene content of the films. Because lower loss-compliance is correlated with greater stiffness of a polymer film, these data show that curing increases the stiffness of the spray enamel.

Figure 14. Shear loss compliance of Decrolon spray enamel films. 8.5. Chemical sensors and biosensors When a reactant molecule reacts with a metal, inorganic, or polymer film, the magnitude of the mass change at the interface depends on whether the products are small enough to diffuse into the film or whether they remain bound only to the surface. If the film is chemically compatible with the reactant, a solid solution may be formed; for example, the absorption of water by a hydrophilic polymer. In this case, the mass change and the corresponding viscoelastic damping may be large enough to yield large signal-to-noise changes when the reaction occurs on a coated QCM. But to be really useful, a sensor must be both sensitive and selective. It is

much more difficult to devise a surface coating that absorbs selectively only a single chemical species. For this reason, the concept of using solubility interactions to design chemically-selective sorbent coatings for chemical sensors and arrays was proposed and developed [42]. In a more recent review with 208 references, Grate has summarized the development and use of acoustic wave microsensor arrays for vapour sensing [43]. To quote from the introduction to this review: "The advantages that sensor arrays offer over individual sensors are sensitivity to a wider range of analytes, improved selectivity, simultaneous multicomponent analysis, and the capability for analyte recognition rather than mere detection. By analogy with olfaction systems comprising multiple receptors and neuronal pattern recognition, sensor arrays for gas-phase detection are sometimes dubbed 'electronic noses'. " The field of piezoelectric mass-sensing devices as biosensors is well covered in a review published in 2000 with 194 references by Janshoff, Galla, and Steinem [38]. To quote from the abstract of this review, " In the last decade absorption of biomolecules on functionalized surfaces turned into one of the paramount applications of piezoelectric transducers. These applications include the study of the interaction of DNA and RNA with complementary strands, specific recognition of protein ligands by immobilized receptors, the detection of virus capsids, bacteria, mammalian cells, and last but not least the development of complete immunosensors. Piezoelectric transducers allow a label-free detection of molecules; they are more than mere mass sensors since the sensor response is also influenced by interfacial phenomena, viscoelastic properties of the adhered biomaterial, surface charges of absorbed molecules, and surface roughness. These new insights have recently been used to investigate the adhesion of cells, liposomes, and proteins onto surfaces, thus allowing the determination of the morphological changes of cells as a response to pharmacological substances and changes in the water content of biopolymers without employing labor-intensive techniques. However, the future will show whether the quartz crystal microbalance will assert itself against established label-free sensor devices such as surface plasmon resonance spectroscopy and interferometry." Because quartz crystal microbalances are compact and sensitive, a large literature has developed on their use as sensors of both chemical and biomolecular reactions and processes at surfaces. A current review of this literature, with 176 references, is by Marx [44]. To quote from the abstract of this review: " The technique possesses a wide detection range. At the low mass end, it can detect monolayer surface coverage by small molecules or polymer films. At the upper end, it is capable of detecting ...complex arrays of biopolymers and biomacromolecules, even whole cells. Another important and unique feature of the technique is the ability to

measure mass and energy dissipation properties of films while simultaneously carrying out electrochemistry on solution species or upon film systems bound to the upper electrode of the oscillating quartz crystal surface." As examples of chemical reactions, the review discusses in more detail the QCM studies of biopolymer film formation, electropolymerization, micellar-polymer film systems, self-assembled monolayer adsorption, Langmuir-Blodgett monolayer films, molecular imprinted polymer films and chemical sensors, and nano-architecture and layer-by-layer films. A usefil comparison of experimental methods used to interrogate biomolecular interactions has been published by Cooper [45]. To quote from the abstract, " The majority of techniques currently employed to interrogate a biomolecular interaction require some type of radio- or enzymatic- or fluorescent-labeling to report the binding event. However, there is an increasing awareness of novel techniques that do not require labeling of the ligand or the receptor, and that allow virtually any complex to be screened with minimal assay development. This review focuses on three major label-free screening platforms: surface plasmon resonance biosensors, acoustic biosensors, and calorimetric biosensors. ... The capabilities and advantages of each technique are compared and key applications involving small molecules, proteins, oligonucleotides, bacteriophage, viruses, bacteria, and cells are reviewed. The role of the interface between the biosensor surface (in the case of SPR and acoustic biosensors ) and the chemical or biological systems to be studied is also covered with attention to the covalent and non-covalent coupling chemistries employed." 8.6. Biological surface science QCM is one of the many experimental methods discussed by Kasemo in his wideranging review of biological surface science [46]. In Kasemo's words: "Biological surface science (BioSS), as defined here is the broad interdisciplinary area where properties and processes at interfaces between synthetic materials and biological environments are investigated and biofunctional surfaces are fabricated. Six examples are used to introduce and discuss the subject: Medical implants in the human body, biosensors and biochips for diagnostics, tissue engineering, bioelectronics, artificial photosynthesis, and biomimetic materials. They are areas of varying maturity, together constituting a strong driving force for the current rapid development of BioSS. The second driving force is the purely scientific challenges and opportunities to explore the mutual interaction between biological components and surfaces.

"Model systems range from the unique water structures at solid surfaces and water shells around proteins and biomembranes, via amino and nucleic acids, proteins, DNA, phospholipid membranes, to cells and living tissue at surfaces. At one end of the spectrum the scientific challenge is to map out the structures, bonding, dynamics and kinetics of biomolecules at surfaces in a similar way as has been done for simple molecules during the past three decades in surface science. At the other end of the complexity spectrum one addresses how biofunctional surfaces participate in and can be designed to constructively participate in the total communication system of cells and tissue." Kasemo's laboratory at Chalmers University of Technology and Goteborg University was the source of the QCM-D technique [lo], now embodied in instrumentation offered by Q-Sense AB. They define the dissipation, D, as the inverse of the quality factor Q of the quartz crystal resonator [47] (see equation (4)). In the Q-Sense instrumentation, the driving RF power to the oscillator, causing it to respond at resonant frequencyA is switched off and the exponentially damped sinusoidal wave decays with a time constant z, where D = ll(nfi).I ~ o r than e 100 recent and archived publications on the QCM-D method are available at the QSense website, www.q-sense.com. The publications stress real-time biointerface characterization with QCM-D. 9. SENSORS 9.1. Acoustic microsensors - the challenge behind microgravimetry Lucklum and Hauptmann, who have made significant contributions to the development of the theory and applications of the QCM, address the future of this field in a recent review with 235 references [17]. To quote: "In this review we give an overview of recent developments in resonant sensors including micromachined devices and also list recent activity relating to the (Bio)chemical interface of acoustic sensors. Major results from theoretical analysis of quartz crystal resonators, descriptive for all acoustic microsensors are summarized and nongravimetric contributions to the sensor signal from viscoelasticity and interfacial effects are discussed."

9.2. Piezoelectric sensors The final reference in this section is to a book not yet published, but in press as of November 2006. It contains current reviews by many of the authors quoted in this chapter. Here is a description of the contents:

Piezoelectric Sensors, Claudia Steinem and Andreas Janshoff (Eds), Springer Series on Chemical Sensors and Biosensors, Volume 5. Available December 2006, ISBN 10-3-540-36567-2 Part I: Physical Aspects of QCM-Measurements: Ralf Lucklum, Frank Eichelbaum: Interface Circuits for QCM Sensors. Diethelm Johannsmann: Studies of Viscoelasticity with the QCM. Michael Urbakh, Vladimir Tsionsky, Eliezer Gileadi, Leonid Daikhin: Probing the Solid/Liquid Interface with the Quartz Crystal Microbalance. Diethelm Johannsmann: Studies of Contact Mechanics with the QCM. Part 11: Chemical and Biological Applications of the QCM: Franz L. Dickert, Peter, A. Lieberzeit: Imprinted Polymers in Chemical Recognition for Mass-Sensitive Devices. Maria Minnuni, Sara Tombelli, Marco Mascini: Analytical Applications of QCMBased Nucleic Acid Biosensors. Robert D. Vaughan, George G. Guilbault: Piezoelectric Immunosensors. Claudia Steinem, Andreas Janshoff: Specific Adsorption of Proteins on Solid Supported Membranes. Vanessa Heitmann, Bjorn Rein, Joachim Wegener: The Quartz Crystal Microbalance in Cell Biology: Basics and Applications. Part 111: Applications Based on Advanced QCM-Techniques: Yoshio Okahata, Toshiaki Mori, Hiroyuki Furusawa: Enzyme Reactions on a 27 MHz QuartzCrystal Microbalance. Kenneth A. Marx: The Quartz Crystal Microbalance and the Electrochemical QCM: Applications to Studies of Thin Polymer Films, Electron Transfer Systems, Biological Macromolecules, Biosensors, and Cells. Fredrik Hook, Bengt Kasemo: The QCM-D Technique for Probing Biomacromolecular Recognition Reactions. Matthew A. Cooper: Resonant Acoustic profiling (RAPTM) and Rupture Event Scanning (REVSTM). This volume has articles by the world's leaders in QCM theory and application, and it should be valuable for anyone wishing to obtain current and in-depth coverage of the applications of the QCM.

10. THE QUARTZ CRYSTAL MICROBALANCEMEAT CONDUCTION CALORIMETER 10.1. Introduction A recent report entitled "Chemical Industry Roadmap for Nanomaterials by Design" [48] states: "The largest barrier to rational design and controlled synthesis of nanomaterials with predefined properties is the lack of fundamental understanding of thermodynamic and kinetic processes at the nanoscale. ...Bulk material properties are not size-dependent, but the properties of nanomaterials are a function of size. The underlying principles governing the properties at all lengths, organization complexity, and structural and property stability over time must be understood to enable the nanoscale materials by design approach." Quartz crystal microbalanceiheat conduction calorimetry, (QCM/HCC) [49] is a new measurement technology that permits high sensitivity measurements in realtime of three properties of a nanoscale coating or film undergoing chemical reaction: the mass change (to *lo nanograms), the heat generated (to *1 microwatt), and the change in loss compliance of the film. These sensitivities are sufficient to examine the energetics of the formation of a self-assembled monolayer, as well as the thermodynamics of the chemical processes in nanoscale polymer coatings. The development of this measurement technology is described and a number of applications are given that illustrate its potential to be a key instrumental method of measuring thermodynamic and kinetic processes at the nanoscale. 10.2. Beginnings of QCMMCC In 1997 the author spent a six-month sabbatical leave from Drexel University in the Department of Thermochemistry of Lund University, hosted by Professor Gerd Olofsson and Professor Ingemar Wadso. The need to measure tiny mass changes in a sample of low volatility inside a heat flow calorimeter was discussed. The goal was to determine the enthalpy of sublimation of the material. Heat-flow signals from the subliming material were quite high, but the need to remove the sample container and reweigh it on an analytical balance introduced substantial errors in the measurement of the mass of material evaporated. What was needed was to build a small, sensitive balance into the sample chamber itself. Shortly thereafter,

work began on a sensor that would combine two mature technologies: transverse shear resonators (quartz crystal microbalances) [35] and the heat flow sensors widely used in isothermal microcalorimetry [50]. Both have been employed in research laboratories and in commercial instrumentation for many years. Quartz crystal microbalances have been widely used since the 1970s as thickness monitors of thin films deposited in vacuum. Peltier heat-flow sensors are the basic sensor in many calorimeters and thermal analysis instrumentation, including those designed and built at Lund University. The goal was to develop and use a sensor that simultaneously measures the mass change and the heat generated or absorbed when a chemical process occurs on the surface of a coated sensor. The quartz crystal microbalance can measure mass changes at the sub-monolayer coverage because of its nanogram sensitivity. With suitable thermal shielding and temperature control, heat conduction calorimetry can continuously measure sub-microwatt heat flows to or from a reacting sample. If a method could be devised to combine these two technologies in a single sensor, then chemical processes in a thin-film sample on the sensor surface could be followed in real time by measuring both the heat and mass changes induced by the chemical or biological process. Because heat and mass are the two fundamental extensive variables needed to determine thermodynamic state variables such as enthalpy, entropy, and Gibbs free energy, such a massheat-flow sensor could be used to determine the thermodynamic quantities associated with the reactions of gases or liquids with thin films. The sensitivity in mass and heat detection needed to measure the energy of binding per unit mass of a monolayer of water on the gold surface of a QCM crystal can be estimated as follows. The area of the gold electrode exposed to the sample vapor is 2 cm2. The monolayer is assumed to have the density of water itself and a thickness of 0.3 nm, the length of the water molecule. There are thus 2 x 1015water molecules in this monolayer, and their mass is 60 ng. If the binding energy of the water to the gold is taken to be the enthalpy of condensation of water, 44 kJ mol-', then the heat liberated at the surface in forming this monolayer is 0.15 mJ. The mass resolution of the QCM depends on the integration time of the frequency counter used to measure the QCM resonant frequency, but nanogram sensitivity has been widely reported in the literature. Detecting a heat evolution of 150 microjoules in a heat conduction calorimeter is more challenging, but it has been achieved. By June of 1997 there was a working massheat-flow sensor in the Lund laboratory based on the principles described above, and by 2001 the first of three U.S. patents was issued on this technology.

10.3. Development of QCMMCC Upon returning to Drexel in the summer of 1997, the author worked with graduate student Hamid Shirazi to develop a quartz crystal microbalancelheat conduction calorimeter based on this massheat-flow sensor. The development, testing, and initial uses of the QCM/HCC are documented in Shirazi's PhD thesis, available online [5 11. Shirazi performed the following experiments with this new technology: (i) Solvent vapor sorption into an aliphatic polyurethane film (TecoflexTM)[52]. Solvents employed were ethanol, carbon tetrachloride, chloroform, toluene, acetone, and hexane. The sorption enthalpy and sorption isotherm of each solvent vapor in Tecoflex was determined. (ii) Hydrogen sorption in thin palladium films [49]. Both the sorption isotherm and sorption enthalpy of hydrogen in palladium were measured. The response of the massheat-flow sensor during the palladium-catalyzed hydrogenation of ethylene was also measured. (iii) Detection of the formation of a self-assembled monolayer of nonylthiol on gold [53]. This experiment was done to test the sensitivity of the apparatus, and it showed that both the mass change and the heat release upon forming a selfassembled monolayer could be detected with the QCMIHCC. (iv) Hydration and dehydration of a thin film of the protein lysozyme [5 I].

The further contributions to the development of the QCM/HCC by Rose Mulligan are described in a PhD thesis, also available online [54]: (v) Thermodynamic and rheological properties of Tecoflex upon ethanol vapor and water vapor sorption [55]. Sorption isotherms, sorption enthalpies, and diffusion coefficients of the two gases were determined. (vi) Hydration studies of protein films [56]. Water sorption isotherms, sorption enthalpies and diffusion coefficients were determined in thin films of both lysozyme and myoglobin. The PhD thesis [57] of Jun Tian, the final Drexel graduate student to work on the QCM/HCC contains the following applications: (vii) Solvent vapour sorption by C60and C60-piperazinefilms [58]. Solvents used were water, 1,3-dichlorobenzene, carbon tetrachloride, methylene chloride, and benzene. (viii) Water vapour sorption by pharmaceutical film coatings.

All three of these PhD theses contain sections on the theory of operation of each component of the massheat-flow sensor and experimental details such as block diagrams of the apparatus, sample preparation, data acquisition and control, calibration, and data analysis. 10.4. Biological applications Galit Zilberman joined the research group at Drexel in November 2003. Her previous research at Tel Aviv University with QCM measurements at the solution/electrode interface [59] led us to explore the use of the QCM/HCC as a detector of protein-ligand interactions in aqueous solution [60], and as a detector of the growth of E Coli bacteria on thin film of nutrient medium deposited on the QCM [61]. Zilberman also studied the growth kinetics of alkyl- and carboxylic acid self-assembled monolayers (SAMs) on gold and the EDC-catalyzed amide bond formation on a carboxylic acid-terminated SAM. 10.5. The Masscal Scientific Instruments G1 MicrobalanceICalorimeter In 2001, the author founded Masscal Corporation to commercialize the developments achieved in the laboratory at Drexel. The goal was to make this technology more widely available to the research community. The first commercial product, the MasscalTMG1, was introduced at the 2004 PITTCON. Its capabilities are well described in company literature, but here are some key specifications: Operating temperature from ambient to 100°C;temperature stability of k.005 "C without the use of water baths. Ambient operating pressure. Sensitivity of 10 ng in mass measurement and 500 nW sensitivity in heat flow measurements with a time constant of 12 seconds (5 datapoints per minute). Continuous measurement of motional resistance of the coated quartz crystal, a quantity that can be used to determine the loss compliance of the coating. Provisions for the software control of external mass-flow controllers to provide a versatile programme of gas composition vs time in the sample chamber. Provision for collection of an external analogue signal from a detector such as a relative-humidity meter. With funding from the Department of Energy's SBIR program at Masscal Scientific Instruments, it was shown that the GI could measure both the catalyzed rate of reaction and the mass build-up or depletion at a platinum or palladium surface when H2reacted with C2H4 [62].

10.6. Recent applications The following applications of the QCMJHCCtechnique have been performed with the Masscal G1: Moisture sorption, transport, and hydrolytic degradation in polylactide films [63]. Monitoring the drying and curing of an alkyd spray enamel [41]. Gravimetric analysis of the non-volatile residue from an evaporated droplet, using the quartz crystal microbalancekeat conduction calorimeter [64]. Sorption isotherms, sorption enthalpies, diffusion coefficients and ermeabilities of water in a multilayer PEOIPAA polymer film [65]. Energetics of a self-assembled monolayer (SAM) of butylthiol on gold (in preparation). Sorption isotherms, sorption enthalpies, and viscoelastic damping produced by water absorption in pharmaceutical film coat materials (in preparation). EDC-catalyzed amide bond formation on a carboxylic acid-terminated SAM (in preparation). Growth kinetics of alkyl and carboxylic acid SAMs (in preparation). 10.7. Conclusion Because many of the materials now being made and characterized in nanotechnology are ultra-thin films of thickness < lpm, their thermodynamic and kinetic properties must be measured with methods more sensitive than the normally employed calorimetric or gravimetric techniques. The QCMIHCC technology is ideally suited for such studies, as the references in this article will attest. One key question of importance to nanotechnology is the long-term stability of materials with nanostmctures in challenging environments, such as high temperature and high humidity, or in the presence of oxidizing agents or solvent vapours. With atomic compositions varying systematically at the nanometer scale, these nanomaterials contain many more contacts between different functional groups and molecular subunits than do typical materials. How stable are these nanomaterials to moisture, to oxidative degradation? Answers to these questions will determine the ultimate usefulness of the many extraordinary new nanomaterials being synthesized today. Knowledge of the thermodynamics and kinetics of these thin-film materials is thus essential in assessing their performance. Suppliers of QCM equipment are given in Table 3.

166

Table 3. Vendors of QCM ystems.

11. REFERENCES 1. C.M. Beck 11, Anal. Chem., 66 (1994) 224A. 2. W.G. Cady, Piezoelectricity, McGraw-Hill, New York and London, 1946. 3. J. Curie and P. Curie, Mull. Soc. Min. Paris, (1880) 90. 4. G. Lippman, An. Chim. Phys., 5 (1881) 145. 5. D.B. Sullivan, Time and Frequency Measurement at NIST: the First Hundred Years. IEEE International Frequency Control Symposium and PDA Exhibition, (2001) 4. 6. V.E. Bottom, A History of the Quartz Crystal Industry in the USA, IEEE Proceedings of the 35th Annual Frequency Control Symposium, 35 (198 1) 3. 7. G. Sauerbrey, Z. Physik, 155 (1959) 206. 8. D.A. Buttry and M.D. Ward, Chem. Rev., 92 (1992) 1355. 9. F. Eichelbaum, R. Borngraber, J. Schroder, R. Lucklurn and P Hauptmann, Rev. Sci. Instrum., 70 (1999) 2537. 10. M. Rodahl, F. Hook, A. Krozer, P. Brzezinski and B. Kasemo, Rev. Sci. Instr., 66 (1995) 3924. 11. K.K. Kanazawa, J.G. Gordon, Anal. Chem., 57 (1985) 1770. 12. K.K. Kanazawa, Faraday Discussions, 107 (1997) 77.

13. S.J. Martin, V.E. Granstaff and G.C. Frye, Anal. Chem., 63 (1991) 2272-81. 14. S.J. Martin, H.L. Bandey, R.W. Cernosek, A.R. Hillman and M.J. Brown, Anal. Chem., 72 (2000) 141. 15. C. Behling, R. Lucklum and P. Hauptmann, Meas. Sci. Technol., 9 (1998) 1886. 16. R. Lucklum and P. Hauptmann, Meas. Sci. Technol., 14 (2003) 1854. 17. R. Lucklurn and P. Hauptmann, Anal. Bioanal. Chem., 384 (2006) 667-82. 18. M.V. Voinova, M. Jonson and B. Kasemo, Biosensors & Bioelectronics, 17 (2002) 835. 19. D. Johannsmann, Macromol. Chem. Phys., 200 (1999) 501. 20. D. Johannsmann, Springer Series on Chemical Sensors and Biosensors, Springer-Verlag,Berlin, Heidelberg, 2006, p. 1-61. 2 1. V. Tsionsky, L. Daikhin, M. Urbakh and E. Gileadi in Electroanalytical Chemistry, (Eds A.J. Bard and I. Rubenstein), Marcel Dekker, New York, Basel, 2003, p. 1-99. 22. A.E. Arnau, Piezoelectric Transducers and Applications, Springer, Berlin, 2004. 23. R. Lucklum, P. Hauptmann and R.W. Cernosek, Thin Film Material Properties Analysis with Quartz Crystal Resonators, IEEE International Frequency Control Symposium, (2001) 542-50. 24. R. Borngraeber, J. Schroeder, R. Lucklum and P. Hauptmann, IEEE Frequency Control Symposium (2001) 443-48. 25. R. Lucklum and P. Hauptmann, Faraday Discussions,l07 (1997) 123. 26. S-W. Lee, W.D. Hinsberg and K.K. Kanazawa, Anal. Chem., 74 (2002) 125-31. 27. J. Hossenlopp, L.H. Jiang, R. Cernosek and F. Josse, J. Polym. Sci., Part B-Polym. Phys., 42 (2004) 2373. 28. V. Tsionsky, L. Daikhin, D. Zagidulin, M. Urbakh and E. Gileadi, J. Phys.Chem. B, 107 (2003) 12485. 29. V.M. Mecea, Sensors and Actuators A, 128 (2006) 270. 30. J.S. Ellis and M. Thompson, Phys. Chem. Chem. Phys., 6 (2004) 4928. 3 1. J.D. Ferry, Viscoelastic Properties of Polymers, Wiley, New York, 1980. 32. J.D. Ferry, J. Polym. Sci., Part B Polym. Phys., 37 (1999) 620. 33. J.C. Love, L.A. Estroff, J.K. Kriebel, R.G. Nuzzo and G.M. Whitesides, Chem. Rev., 105 (2005) 1103. 34. A.C. Hillier and D.A. Ward, Anal. Chem., 64 (1992) 2539.

35. C-S. Lu and A.W. Czanderna, Applications of Piezoelectric Quartz Crystal Microbalances, Elsevier, New York, 1984. 36. E. Benes, J. Appl. Phys., 56 (1984) 608. 37. R. Lucklum, P. Hauptmann, Thin Film Shear Modulus Determination with Quartz Crystal Resonators; A Review, IEEE International Frequency Control Symposium, (2001) 408-17. 38. A. Janshoff, H-J. Galla and C. Steinem, Angew. Chem. Int. Ed. 39 (2000) 4004. 39. H.S. Choi, Y.S. Kim and S.M. Shin, J. Polym. Sci. A, 44 (2006) 2428. 40. T.J. Sliva in Paint and Coating Testing Manual, (Ed. J.V.Koleske), ASTM, Philadelphia, 1995, p. 439-46. 41. A.L. Smith in New Developments in Coating Technology. (Eds P. Zarras, B. Richey, T. Wood and B. Benicewicz), ACS Symposium Series, Washington DC, March 2007, Chap. 17 42. J.W. Grate and M.H. Abraham, Sensors and Actuators B, 3 (1991) 85. 43. J.W. Grate, Chem. Rev., 100 (2000) 2627. 44. K.A. Marx, Biomacromolecules, 4 (2003) 1099. 45. M.A. Cooper, Anal. Bioanal. Chem., 377 (2003) 834. 46. B. Kasemo, Surface Sci., 500 (2002) 656. 47. M. Rodahl and B. Kasemo, Sensors and Actuators A, 54 (1996) 448. 48. J. Solomon, Steering Team Chair, Chemical Industry Roadmap for Nanomaterials by Design: from Fundamentals to Function in Chemical Industry Vision 2020,2003, p 25 49. A.L. Smith and H.M. Shirazi, Thermochim. Acta., 432 (2005) 202. 50. I. Wadso, Chem. Soc. Rev., (1997) 79. 5 1. H.M. Shirazi, Quartz Crystal MicrobalanceiHeat Conduction Calorimetry (QCM/HCC), a new technology capable of isothermal, high sensitivity, mass and heat flow measurements at a solidlgas interface., PhD Thesis, Drexel University, Philadelphia, PA, 2000. 52. A.L. Smith and H.M. Shirazi, J. Thermal Anal. Calorim., 59 (2000) 171. 53. A.L. Smith, S.R. Mulligan, J. Tian, H. Shirazi and J.C. Riggs, A Massmeat Flow Sensor Combining Shear Mode Resonators with Thermoelectrics: Principles and Applications, IEEE Frequency Control Symposium. IEEE, Tampa, FL, 2003,1062-1065 54. S.R. Mulligan, The QCM/HCC and Applications in Studying the Thermodynamic and Rheological Properties of Polymer and Protein Thin Films at a GasISolid interface, PhD Thesis, Drexel University, Philadelphia, 2002.

55. A.L. Smith, S.R. Mulligan and H.M. Shirazi, J. Polym. Sci. Part B, Polym. Phys., 42 (2004) 3893. 56. A.L. Smith, H.M. Shirazi and S.R. Mulligan, Biochim. Biophys. Acta (BBA)-Protein Structure and Molecular Enzymology, 1594 (2002) 150. 57. J. Tian, Comparative Solubility Studies of C60 and C60- Piperazine and Applications of the Quartz Crystal MicrobalanceJHeat Conduction Calorimeter, PhD Thesis, Drexel University, Philadelphia, 2002. 58. J. Tian and A.L. Smith in PV 2002-12 Fullerenes: The Exciting World of Nanocages and Nanotubes, (Eds P.V. Kamat, D.M. Guldi and K.M. Kadish), Electrochemical Society, Philadelphia, 2002, p. 255-69. 59. L. Daikhin, E. Gileadi, V. Tsionsky, M. Urbakh and G. Zilberman, Electrochim. Acta, 45 (2000) 3615. 60. G. Zilberman and A.L. Smith, Analyst, 130 (2005) 1483. 61. A.L. Smith and G. Zilberman, Detection of vital bacteria and protein ligand binding using the QCMJHCC, in 32nd Annual Conference of the North American Thermal Analysis Society, (Ed. M. Rich), NATAS - CD-ROM, Williamsburg, VA, 2004. 62. A.L. Smith, H.M. Shirazi and F.C. Smith, Catal. Lett., 104 (2005) 199. 63. R.A. Cairncross, J.G. Becker, S. Ramaswamy and R.O'Connor, Appl. Biochem. Biotech., 131 (2006) 774. 64. A.L. Smith, Gravimetric Analysis of the Non-volatile Residue from an Evaporated Droplet, using the Quartz Crystal MicrobalanceIHeat Conduction Calorimeter, J. ASTM. Intl3, issue 6, paper ID JAI13894 (2006). 65. A.L. Smith, J.N. Ashcraft and P.T. Hammond, Thermochim. Acta. 450 (2006) 118-125.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 6 HEATING STAGE SPECTROSCOPY: INFRARED, RAMAN, ENERGY DISPERSIVE X-RAY AND X-RAY PHOTOELECTRON SPECTROSCOPY Ray L. Frost and J. Theo Kloprogge Inorganic Materials Research Program, School of Physical and Chemical Sciences, Queensland University of Technology, GPO Box 2434, Brisbane, Qld 400 1, Australia

1. INFRARED EMISSION SPECTROSCOPY 1.1. Introduction Spectroscopy and especially infrared spectroscopy has been a widely used technique in the industry for the structural and compositional analysis of organic, organometallic, metalorganic, inorganic and polymeric materials, in addition to quality control of raw materials and commercial products. In general, spectroscopic techniques are concerned with the interaction between radiation of some sort and matter, using radiation with an energy or wavelength appropriate to the distance and time scales relevant to the microscopic study of atoms, molecules and solids. In this regard a large number of radiation types can be used including:

I. electromagnetic radiation 11. elementary particles 111. nuclei IV. ions V. ultrasonic waves Infrared and Raman spectroscopy are vibrational spectroscopic techniques and belong to the first category and will be discussed in more detail below followed by discussion of spectroscopy based on interactions between X-rays and electrons and a sample. Vibrational spectroscopy involves the use of light to probe the vibrational behaviour of molecular systems, usually via absorption, emission or light

scattering experiments. For molecules and crystals the vibrational energy involved ranges from approximately 0 to 60 kJ mol-' equivalent to wavenumbers, v , of 0-5000 cm-'. Both infrared and Raman spectroscopy give rise to a set of absorption or scattering peaks as a function of energy, forming a vibrational spectrum. Individual peaks in the spectrum correspond to energies of vibrational transitions within the sample, or to the frequencies of the vibrational modes. Because a vibrational spectrum is dependent on the interatomic forces in a molecule or crystal it is a very sensitive probe of the microscopic structure and bonding within the molecule or crystal. Therefore, the positions, symmetries and relative intensities of the observed peaks in a spectrum and any changes in these peaks with changes in variables such as temperature, pressure and composition will provide useful structural information about the sample. The standard thermal analytical techniques such as thermogravimetry (TG), differential thermal analysis (DTA) and differential scanning calorimetry (DSC), have been extensively used to study phase changes of materials and reactions such as dehydration, dehydroxylation, decarbonisation, etc. More recently new techniques have become available to study these changes in samples in-situ, such as heating stage X-ray diffraction, heating stage scanning electron microscopy, heating stage transmission electron microscopy and heating stage X-ray photoelectron spectroscopy. In-situ spectroscopic methods such as heating stage Raman spectroscopy and infrared emission spectroscopy have only been used to a limited extent. Infrared emission spectroscopy forms a valuable technique that can be applied in situ during the heat treatment. The technique of measurement of discrete vibrational frequencies emitted by thermally excited molecules, known as Fourier transform infrared emission spectroscopy (FTIR ES, or shortly IES) has not been widely used for the study of materials. The major advantages of IES are that the samples are analyzed in situ at increasing temperatures and IES requires no sample treatment other than that the sample should be of submicron particle size. Further, the technique removes the difficulties of heating the sample to temperatures where reactions take place with subsequent quenching prior to the measurement, because IES measures the process as it is actually taking place. IES has been shown for almost 40 years to be an useful method for obtaining information at increased temperatures of a large variety of materials including coal, minerals and catalysts, polymers and even liquids or melts [I, 21. Probably one of the first papers on the theory IES may be attributed to Max Planck [3] in which he formulated the distribution law for emitted radiation as a fimction of temperature and wavelength. For quite some time this technique received basically no attention because it was thought that no interesting information

could be obtained. Coblenz was one of the few who reported the emission spectra of semitransparent minerals at increased temperatures [4, 51. In 1945 Kapff [6, 71 defined a number of parameters influencing the intensity of emissions from a sample. In 1950 McMahon [8] proposed a theory for infrared emission through a semitransparent material, which was later refined by Gordon to include hemispherical emitted radiation from the surface [9]. In 1965 Low used Fourier Transform infrared spectroscopy (FTIR) to study the infrared emission spectra of mineral surfaces [lo]. The development of FTIR with its high sensitivity to low energy sampling has been very important for the growing interest in infrared emission spectroscopy for the study of solids like minerals and catalysts [I, 11-15]. Jonson et al. [13] used this technique to follow the reduction of various metal oxides such as Vz05,Co304 and CuO by toluene, while Mink and Keresztury [15] described the adsorption and desorption of organic species on the surface of oxide catalysts and modified oxide catalyst supports, such as [Re(C0)30H]4 on A1203 and (C5H5)Fe(C0)21 on SiO2. Vassallo et al. [14] described a number of phase changes observed by IES, such as the a-quartz-P-quartz transition and the formation of cristobalite, the transformation of calcite to aragonite and of kaolinite via metakaolinite to mullite. 1.2. The theory behind infrared emission spectroscopy (IES) The normal single-beam reference spectrum obtained during the recording of a infrared spectrum is actually the emission spectrum of the instrument source, modified by the instrument response function. Transmission spectra are obtained by ratioing the signals in the absence and presence of the sample. Likewise, emission spectra may be obtained by ratioing the emission signal of a sample to that of a reference, normally a blackbody source that emits radiation according to Planck's radiation formula (i.e. a continuum). The energy of a blackbody is a function of the frequency and temperature according to the equation:

where p(v, T ) is the density of radiation of frequency v, at temperature T. v(max) increases and shifts to higher values with temperature, so that at 300 OC a blackbody is emitting energy over the whole of the mid-infrared range of 400 to 4000 cm-'. According to Kirchhoff s law, for a body in thermal equilibrium with its surroundings the absorbed and emitted energies are equal, and because a perfect blackbody will absorb all incident radiation, it will also be the perfect emitter. Real samples therefore will always emit less energy than a blackbody. At a

given temperature T, the ratio of the energy emitted by the sample (E,) to that of a blackbody (Eb)at wavelength ( v ) has been defined as the emissivity $ v ) :

The emissivity of a sample will therefore always be less than or equal to one. It is not always necessary to ratio to a blackbody radiator in order to obtain a useful emission spectrum. A body for which $ v ) is constant and less than 1 for all frequencies is known as a greybody. The energy distribution as function of the frequency is the same as for a blackbody at the same temperature. However, in many materials, the emittance varies strongly with the frequency, because materials absorb and therefore emit strongly at their characteristic vibrational frequencies. For an electromagnetic wave incident on a non-opaque body in isothermal equilibrium, a part of the radiation will be reflected off the surface, another part is absorbed while travelling through the body, and the remainder is transmitted. A balance of energy results in the following relations:

and Kirchhoff s law states E,

= cr,

so: r v + c V + t v =I where rv = reflectance, a,,= absorptance, tv = transmittance and E, = emittance. There are several interrelated constants that describe the absorption of radiation by a material. The fraction of the incident radiation that is transmitted through a body with thickness x is given by e-k" where k is known as the absorption coefficient. A related constant is the extinction coefficient yrc, the imaginary part of the refractive index 17 : 17 = y ( l

+ ilc)

which is related to k as

k = (4?U'/K)/A The reflectivity is expressed as a function of y and K as

and

A model for the unidirectional emission normal to the front surface of a semitransparent slab was developed by McMahon [g]. The total emittance was found to be:

Later, Hvistendhal et al. [16] derived an expression for the emittance of a material in contact with a highly reflective surface:

In these models the emittance is a function of both the transmittance and the reflectance of the sample. The emittance spectrum of an opaque sample (transmittance = 0) is fully determined by the changes in surface reflectance with frequency. In semitransparent samples, strong emission bands can exhibit some distortions or false splittings because the reflectance as well as the transmittance is varying strongly where the absorption coefficient is large. These reflectance effects can be removed by ratioing the emission to the emission of a thick sample of the same material rather than to that of a blackbody. When very thin layers are measured (approximately 1 pm), reflectance effects are rarely a problem. Furthermore, excessive temperature gradients are also avoided by using very thin samples, thus reducing the re-adsorption of intense frequencies by cooler, outer layers. As the sample thickness is increased past a certain point, the spectral contrast between an actual band and the background noise decreases. Both reflectance and re-absorption effects result in attenuation of the stronger bands. Thus, the best spectra are obtained with the thinnest samples. When comparing emission spectra with absorption spectra, the difference in scales should be kept in mind. Absorbance is measured as -loglO(transmittance) and varies from 0 to co as transmittance and emittance vary between 0 and 1. Consequently, weaker bands appear to be enhanced in emission spectra relative to absorbance spectra. Emittance values should be converted to -log(l-G) in order to obtain units that are linearly proportional to concentration. Care should

be taken, however, in quantitative applications because of the reflectance and reabsorption effects. Most of the IES carried out in our laboratory has utilized a Digilab FTS-60A spectrometer, while more recently this spectrometer has been replaced by a Nicolet Nexus 870 FTIR spectrometer (Figure 1). Both were modified by replacing the IR source with an emission cell. Approximately 0.2 mg of the sample was spread as a thin layer on a 6 rnrn diameter platinum surface and held in an inert atmosphere within a nitrogen-purged cell during heating. The infixed emission cell consists of a modified atomic absorption graphite rod furnace, which is driven by a thyristor-controlled AC power supply capable of delivering up to 150 amps at 12 volts. A platinum disk acts as a hot plate to heat the sample and is placed on the graphite rod. An insulated 125-pm type R thermocouple was embedded inside the platinum plate in such a way that the thermocouple junction was <0.2 mm below the surface of the platinum. Temperature control of f 2°C at the operating temperature of the sample was achieved by using an Eurotherm Model 808 proportional temperature controller, coupled to the thermocouple.

Figure 1. Infrared spectrometer with IES attachment on the right. The design of the IES facility is based on an off-axis paraboloidal mirror with a focal length of 25 mm mounted above the heater, which captures the infrared radiation and directs the radiation into the spectrometer. The assembly of the heating block and platinum hot-plate is located such that the surface of the platinum is slightly above the focal point of the off-axis paraboloidal mirror. By

this means the geometry is such that approximately 3 mm diameter area is sampled by the spectrometer. The spectrometer has been modified by the removal of the source assembly and mounting a gold-coated mirror, which was drilled through the centre to allow the passage of the laser beam. The mirror was mounted at 45*, which enables the IR radiation to be directed into the FTIR spectrometer (Figure 2).

Figure 2. Heating stage with the gold-coated mirror on top. The emission spectra were collected at suitable intervals, mostly 50 "C or 25

OC, over the range 200 - 800 OC. The time between scans (while the temperature

was raised to the next hold point) was approximately 100 seconds. Initial experiments have shown that this was sufficient time for the heating block and the powdered sample to reach temperature equilibrium. The spectra were acquired by co-addition of 64 scans for the whole temperature range (approximate scanning time 45 seconds), with a nominal resolution of 4 cm". Good quality spectra can be obtained provided that the sample thickness is not too large. If too large a sample is used, then the spectra become diff~cultto

interpret because of self absorption and the possible presence of combination and overtone bands. In the normal course of events, three sets of spectra are obtained: firstly the black-body radiation (actually grey-body radiation) over the temperature range selected at the various temperatures (Figure 3a); secondly the platinum plate radiation is obtained at the same temperatures (Figure 3b), and thirdly the spectra from the platinum plate covered with the sample (Figure 3). Normally only one set of black-body and platinum radiation is required. The emittance spectrum at a particular temperature was calculated by subtraction of the single beam spectrum of the platinum backplate from that of the platinum + sample, and the result ratioed to the single beam spectrum of an approximate blackbody (graphite). This spectral manipulation is carried out after all the spectral data have been collected.

Figure 3. (a) black body spectra, (b) Pt holder spectra and (c) final sample spectra after blackbody and Pt holder spectral corrections

1.3. Infrared emission spectroscopy of alunite Alunites are a group of minerals which form part of the alunite supergroup. The general formula is given by DG3(T04)2(OH,H20)6where the D sites are occupied by monovalent cations such as K+, ~ a ' , NH;, ~ ~ and 0 others, ' divalent cations such as ca2+,Ba2+,sr2+,pb2+,trivalent cations for example Bi3+; and G is either ~ 1 or~~ ' e ; and ~ +T is s6+, AS" or p5+.Alunites can be divided into alunites and jarosites, depending on whether the concentration of A1 is >Fe (alunites) or Fe> A1 (jarosites). Solid solution formation can exist across a wide range of concentrations and substitutions. Common members of the alunite group are alunite KA13(S04)2(OH)6, natroalunite N~iAl~(S04)~(0H)~, ammonioalunite NH4A13(S04)2(OH)6, schlossmacherite and ( H ~ o + , c ~ ~ + ) A ~ ~ ( s o ~ )The ~ ( ostructure H ) ~ . of alunites is trigonal with a = 6.990 A, c = 16.905 A, space group = R 3m, with Z = 3. Interest in the chemistry of alunites stems from the possible discovery of alunites on Mars [17, 181 which implies the presence of water on Mars either at present or at some time in the planetary past [19, 201. Interest in such minerals and their thermal stability rests with the possible identification of these minerals and related dehydrated paragenetically related minerals on planets and on Mars.

Wavenumbers1cm-1

Figure 4. IES spectra of alunite at intervals of 100 "C from 100 to 1000 "C. The IES spectra in Figure 4 clearly show the thermal stability and, importantly, the stability range of alunite. Two OH stretching bands at 100 "C are observed at 3509 and 3482 cm-' (Figure 5). These bands coalesce at 450 "C. Finally all intensity in the bands attributed to OH stretching vibrations is lost at 600 "C. This shows that alunite will be stable up to 550 "C and will withstand significantly high temperatures such as may be found on planets.

(Kern County, California)

Wavenumbers/ cm-I

Figure 5. IES spectra of alunite at intervals of 100 "C from 100 to 600 "C in the hydroxyl stretching region between 3000 and 3800 cm-'. Libowitzky [2 11 showed that a regression function can be employed relating the hydroxyl stretching fkequencies to the 0 - 0 distance, with regression coefficients better than 0.96 using infrared spectroscopy. The function is described as:

Thus OH---0 hydrogen bond distances may be calculated using the Libowitzky empirical function. The values of the OH stretching bands lead to an estimation of hydrogen bond distances. For K-alunite the bands at 3509 and 3482 cm-' lead to hydrogen bond distances of 2.909 and 2.872 A. These hydrogen bond distances appear to fall into two groups at around 2.90 A and 2.84-2.87 A. X-ray crystallographic studies give a hydrogen bond distance of 2.92 A which is in

excellent agreement with the values estimated from the infrared OH stretching wavenumbers. Two hydrogen bonds of slightly different strengths are suggested to exist for alunites. With the IES data, the changes in bond lengths of the hydrogen bonds in alunites with temperature can be followed. 2. HEATING STAGE RAMAN SPECTROSCOPY Raman spectroscopy is a relatively old technique and was based on theoretical grounds already predicted in 1923. Sir C.V. Raman was in 1928 the first to show, experimentally, the existence of inelastic light scattering. This effect was later named after him as the "Raman effect". The Raman principle has since been applied to fundamental chemical research, mostly in combination with infrared spectroscopy. When light hits a molecule, a part of the light will be dispersed with the same wavelength as the initial light. This effect is known as elastic or Rayleigh scattering. The intensity I, of this scattering varies strongly with wavelength:

A much smaller part, however, will show a slight shift in wavelength (approximately of the initial light) due to the interaction between the vibrating molecules and the incoming photons. The Raman effect creates a shift in both positive and negative directions compared to the initial wavelength of the incoming light. These shifts give rise to frequencies known as the Stokes and anti-Stokes lines. It is this weak Raman scattering that forms the basis of Raman spectroscopy. Raman scattering is thus mainly determined by the structure of a molecule. Raman spectroscopy is very sensitive to the symmetry of molecules and is very suitable for the determination of the molecular vibrations of symmetric molecules. For a vibration to be allowed in the Raman spectrum there must be a change of polarizability accompanying the vibration. An atom or molecule in an electric field will be distorted and therefore be polarized. This induced dipole moment is proportional to the applied field:

where P and E are vectors indicating the dipole moment and the electric field, respectively, and a is the polarizability. The polarizability is essentially a bond property, related to what might be called the 'looseness' of the electron cloud. The numerical value of the polarizability is given in volume units and is a simple number for atoms, but is directionally dependent for molecules. It can be

completely described by three numbers and visualized by a polarizability ellipsoid. The Raman effect is due to the changes of polarizability that occur during intermolecular vibrations. Not only vibrations, but also rotations, may change the polarizability of a molecule. The polarizability changes with a frequency that is twice the frequency of rotation. A linear molecule for example has a rotation axis, which coincides with the symmetry axis of the molecule. All the resulting Raman lines are therefore situated very close to the Rayleigh line and are very difficult to distinguish. These rotations though are normally absent in the solid state. The vibrational selection rules are the same for Raman spectroscopy as for infrared spectroscopy. In the Stokes process, the intense, monochromatic radiation takes a molecule from the v = 0 state to a virtual state, VO, from which it falls back to the v = 1 state. Similarly, in the anti-Stokes process, the virtual state V1 is involved in the overall transfer of the molecule from the v = 1 to the v = 0 state. The Stokes and anti-Stokes transitions lie on the low and high wavenumber sides, respectively, of the exciting radiation. The intensity of the anti-Stokes line, relative to the Stokes transition is very low because of the lower population of the v = 1 state, compared to that of the v = 0 state. Consequently, Raman spectroscopy uses only the Stokes transitions. Absorption occurs when the energy of an electromagnetic wave (hv) is similar to the vibrational or rotational energy of atoms in a molecule. Therefore the wavelength of the absorbed light is well defined. Absorption can be avoided by choosing a monochromatic light source with a colour near the colour of the compound. In addition to absorption, problems can also be caused by fluorescence. This process takes place when quanta, i.e. photons (of lower energy) are re-emitted with a longer wavelength. If this process occurs during illumination in the Raman spectrometer, fluorescence can completely mask the weak Raman effect. In general, fluorescence is more likely to occur for higher frequencies (green or blue) than for lower frequencies (red). On the other hand though, the Raman effect is stronger for higher frequency sources. For normal Raman microscopy, a small amount of a sample is placed on a polished metal surface on the stage of an Olympus BHSM microscope, which is equipped with lox, 20x, and 50x objectives, for the 300 K measurements. No sample preparation is needed. The microscope is part of a Renishaw 1000 Raman microscope system, which also includes a monochromator, a filter system and a CCD (Figure 6a). Raman spectra were excited by a SpectraPhysics model 127 He-Ne laser (633 nm) at a resolution of 2 cm" in the range between 500 and 1500 cm-'. Repeated acquisitions using the highest magnification are accumulated to improve the signal-to-noise ratio in the spectra. Spectra are calibrated using the 520.5 cm-' line of a silicon wafer.

Spectra at increased temperatures are obtained using a Linkam thermal stage (Scientific Instruments Ltd, Waterfield, Surrey, England). Samples are placed on a circular glass disc, which fits over the silver plate of the thermal stage (Figure 6b/c). Because of the increased optical path, spectra are noisier and require longer accumulation times. Spectra are obtained using 20 second scans for up to 30 minutes using the special short 50X (UWLD)objective. A lower intensity Rarnan signal is obtained using this objective owing to the low numerical aperture of this long working distance objective. This, combined with the spherical aberration of the stage window, results in a decreased signal compared with that run without the thermal stage.

-

I Figure 6a. Raman microscope with heatinglcooling stage. On the left are a temperature controller and liquid nitrogenlnitrogen gas flow controller and liquid nitrogen container.

Figure 6b. Heatingfcooling stage.

Figure 6c. Heating/cooling stage mounted on the Raman microscope.

2.1 Heating stage Raman spectroscopy of weddellite

Phases changes associated with the thermal treatment of weddellite (CaC204.2H20) can be followed by the use of a Raman microscope in combination with a thermal stage. Such a suite of spectra are shown in Figure 7 for weddellite. The spectra clearly show changes in the Raman spectrum between 150 and 200 "C and again between 450 and 500 "C. The CO stretching vibration is observed for oxalate in aqueous solution at 1496 cm-' and for solid potassium oxalate at 1449 cm-'. The band is at 1475 cm-' for weddellite in the 298 K spectrum. The observation of a single band in the 298 K spectrum suggests the equivalence of the CO stretching vibrations. No bands are observed in the infrared spectrum at 298 K. Thus at 298 K, the structure is centrosymmetric. Symmetry loss through thermal treatment would occur if the structure was no longer planar.

1000

950

900

850

800

Raman shift lcm-'

Figure 7. Heating stage Raman spectra of weddelite (CaC204.2H20)between 800 and 1000 cm-' at 100 "C intervals from 50 to 550 "C.

Mild heating to 50 "C causes two bands to appear in the Raman spectrum at 1490 and 1464 cm-', Figure 8. These band positions correspond to those of partially dehydrated weddellite, which is the mineral whewellite (CaC204.H20), calcium oxalate monohydrate. The bands are observed at 1488 and 1462 cm-' in the 150 "C spectrum. Thus there is a slight red-shift with thermal treatment. The band positions are in agreement with previously published data. At temperatures above 150 "C, the bands are observed at 1478 and 1466 cm-'. These bands correspond to the symmetric stretching modes of anhydrous calcium oxalate. At temperatures above 450 "C, only a single band is observed at 1480 "C. This band corresponds to the CO symmetric stretching mode of calcium carbonate.

1800

1700

1600

1500

1400

1300

Raman shift I cm-'

Figure 8. Heating stage Raman spectra of weddelite (CaC204.2Hz0)between 1300 and 1800 cm-'at 100 "C intervalsfiom 50 to 550 "C. The 298 K Raman spectrum shows a single low-intensity band for weddellite at 1628 cm-'. This band is attributed to the antisymmetric stretching vibration,

which for a planar structure should not be observed in the Raman spectrum. However the structure is probably a distorted square antiprism. This distortion results in the observation of the forbidden Raman bands. For aqueous oxalate, the antisymmetric stretching (B2J mode is observed in the infrared spectrum at 1600 cm-I. This band is observed at 1631 cm-I in the 50 "C spectrum and at 1630 cm-' for the 150 "C spectrum. This band corresponds to the CO antisymmetric stretching mode of the calcium oxalate monohydrate. This band was reported to be at 1632 cm-', which agrees well with the data reported in this work. The observation of the antisymmetric stretching vibration suggests that the thermal treatment of the weddellite causes non-planarity in the structure. Thermal treatment above 150 "C shows a shift in the band positions. The two bands at 1490 and 1464 cm" shift to 1465 and 1478 cm-', whilst the band at 1631 shifts to 1647 cm-'. The position of this band corresponds with that of anhydrous calcium oxalate. Thermal analysis shows that water is lost over the 100 to 114 "C temperature range. Thus the shift in the CO stretching bands must be associated with water loss. At temperatures above 450 "C, the two bands at 1465 and 1478 cm-' become a single band observed at 1478 cm-I. The band at 1647 cm-' disappears and is replaced with a very broad band centred at 1618 cm-I. Thermal analysis shows that there is a mass loss step around 430 "C in which carbon dioxide is the evolved gas. A further mass loss step occurs around 595 "C in which further carbon dioxide is evolved. Such a temperature is above the upper temperature limit of the Raman thermal stage. It is suggested that above 450 "C, calcium carbonate is formed and above 590 "C carbon dioxide is lost and calcium oxide formed.

3. THERMAL STUDIES OF MATERIALS USING HEATING AND COOLING STAGE SCANNING ELECTRON MICROSCOPY AND ENERGY DISPERSIVE X-RAY ANALYSIS 3.1. Apparatus The environmental scanning electron microscope (ESEM) was developed as a special type of scanning electron microscope during the 1980s [22,23]. It might have been more appropriate to call it the Variable Environment SEM because the primary advantages are due to the fact that the environmental conditions around a sample can be changed through a range of pressures, temperatures and gas compositions, while retaining all the advantages of normal SEM. By maintaining a much lower vacuum in the specimen chamber than in the electron column (up to a maximum of 20 Torr or 2670 Pa), it potentially allows studies of samples that are wet, oily or non-conductive, such as biological specimens, gels, cements, and certain inorganic materials including minerals.

With water vapour as the chamber gas, it is possible to control the relative humidity in the chamber by adjusting the chamber pressure and specimen temperature to a suitable point on the pressure-temperature diagram (Figure 9).

Relative Humidity Isobars

Figure 9. Relative humidity isobars of water as function of temperature and pressure. With appropriate parameters, it is then possible to cause water to condense or evaporate fi-om the specimen. This also means that small amounts of gas produced during heating of a sample on a heating stage in the electron microscope chamber do not hinder the observation of a sample. This opens up a whole new area of research allowing not only mineralogists, but also materials scientists, to observe changes taking place in their materials when heated, like hydrationldehydration [24-291 thermal expansion, decomposition, calcination, sintering etc. [30-341. Scanning electron microscope (SEM) images were obtained on a FEI Quanta 200 Environmental Scanning Electron Microscope (FEI Company, USA) operated at an accelerating voltage of 20 kV. The microscope was fitted with an FEI water-cooled 1000 "C specimen heating stage with a high-temperature controller and K-type thermocouple for monitoring the temperature. A high temperature gaseous secondary electron detector with a pressure limiting aperture, mounted directly above the specimen on the heating stage, was used for electron imaging. During heating, the detector-specimen distance was about

8 mm. The specimen chamber pressure was maintained at 2 Torr (267 Pa) of water vapour. A small amount of sample powder was mounted onto a small aluminum cup with a thin layer of carbon paint, to ensure the particles were attached to the surface and to maintain reasonable heat conductance to the sample. No pre-coating with carbon or gold, as is done for standard high vacuum SEM, was required for the ESEM observations. The sample was heated from room temperature to 750 "C, at a rate of about 10 "C/min, however a constant temperature was maintained during image acquisition or microanalysis. Prior to the analysis an equilibration time of approximately 10 minutes was applied. Elemental analysis was carried out using a thin-window EDAX energydispersive X-ray (EDX) detector and microanalysis system (EDAX Inc., USA). It was not possible to analyse the sample at temperatures above 400 "C because the infrared radiation from the heating stage would swamp the X-ray detector. Therefore, for temperatures above this value, the sample temperature was first lowered to 350 "C for the analysis, and then restored to its prior setting. Since any evolved gas was constantly being pumped away and the pressure inside the chamber was only 2 Torr, the chances of reverse reactions taking place were absolutely minimal. Heating the sample and the holder does result in thermal expansion and thus in some movement of the sample, requiring frequent correction of the sample position in the SEM. Furthermore, as the temperature of the sample is increased the image brightness and contrast increase considerably, consequently the amplifier settings and bias voltage of the gaseous secondary-electron detector must be frequently adjusted to compensate. The brightness changes are most probably due to increased cascade amplification of the electron signal, resulting from a complex interaction between emitted secondary and thermal electrons and the gas molecules above the sample, which gain kinetic energy from the thermal radiation. For this research a very small amount of sample was added as a dry powder into a shallow holder specially made for the environmental electron microscope. The microscope was fitted with an FEI Peltier stage with a temperature controller and K-type thermocouple for monitoring the temperature. A gaseous secondary-electron detector with a pressure limiting aperture, mounted directly above the specimen on the heating stage, was used for electron imaging. During the wetting and drying experiments, the detector-specimen distance was about 10 mm. The specimen chamber pressure was adjusted according to the pressure and temperature diagram shown in Figure 9 around 5 "C and 4 to 6 Torr (water vapour).

3.2. Thermal decomposition of weddelite by heating stage SEM and infrared emission spectroscopy (IES) The thermal decompositions of weddelite, CaC204.2H20,and other metal oxalates are well known [35-391. CaC204.2H20is often used for teaching purposes [40], both as an example and as a standard in thermal analysis. The decomposition of weddelite shows three major mass-loss events (Figure 10).

Figure 10. Thermogravimetry of weddelite (CaC204.2H~0) (heating rate 10 OC min-' in air) The first mass-loss, around 120 "C, corresponds to the dehydration:

The second mass-loss, around 425 OC, results from the decomposition of the dehydrated oxalate to carbonate:

In the final reaction, around 600 "C,the carbonate breaks down to the oxide:

In addition to thermal analysis, a limited number of papers have reported on the crystallographic changes based on X-ray diffraction, infrared emission and Raman spectroscopy. The different experimental conditions in the TG and the SEM can result in minor differences in the reaction temperatures because, in the TG, a continuous heating programme is running, while in the SEM heating takes place in steps. However, these differences will be small because the thermal SEM-EDX analyses are done in a static mode at temperatures intermediate between those of the reactions and these temperatures are sufficiently separated. The SEM images at room temperature show that the sample consists of very small crystals up to about 5 pm in length (Figure 11). Heating to 150 and 280 "C did not change the morphology of the crystals, although after heating to 280 "C dehydration is complete. An EDX spectrum indicated a decrease in the oxygen peak, when compared to a spectrum from the sample prior to heating (Figure 12). because only a very thin layer of sample is used, the EDX spectra show the presence of aluminium from the sample holder under the sample. The loss of volatiles is compensated for in the crystals by internal rearrangements without changing the outer faces of the original crystals resulting in a slightly porous material with pores too small to be observed in the SEM [41]. Additional information can be obtained from the infrared emission spectra (IES) with increasing temperature (Figure 13). The spectra at low temperature do not clearly show the presence of water because the water bending mode normally observed at 1645 cm-' is obscured by the v,,isymm(C=O)around 1620 cm-'. This makes it impossible to follow the dehydration step. The oxalate is further characterised by a broad band around 1640 cm-I plus a sharper band around 1325 cm-', which can be ascribed to v,,,(C=O) mode. The relatively sharp band at 788 cm-' can be ascribed to 6(0-C-0) [42]. The IES spectra show significant changes in the temperature range from 300 to 375 "C although there are no changes in the morphology and chemical composition (no change in the EDX and no mass loss in the TG) suggesting a phase change. A new band appears at 762 cm-' above 300 "C and disappears again between 400 and 450 "C when decomposition of the anhydrous calcium oxalate to calcium carbonate takes place. Such a change is well known for calcium oxalate monohydrate, where the dehydrated oxalate a-CaC204converts, between 350 and 400 "C, to P- and even y-CaC204[43].

r

--

Figure 11. Heating stage SEM images of weddelite at 30 "C (top left), 150 "C (top right), 280 "C (middle left), 500 "C (middle right) and 750 "C (bottom left, close-up bottom right).

194

\\Aemf-quanta\Spectra\Theo\CaC2O4 spectra\CaC2O4 30C.spc

\\Aemf-quanta\Spectra\Theo\CaC2O4 spectra\CaC2O4 280C.spc Label A: CaC2O4 280 C

Label A: CaC2O4 30 C

CaK

CaK

CK

CK OK

OK AIK

CaK

0.50

1.00

CaK

AIK NaK

NaK 1.50

2.00 2.50

3.00

3.50

4.00 4.50

5.00

\\Aemf-quanta\Spectra\Theo\CaC2O4 spectra\CaC2O4 after 500C.spc Label A: CaC2O4 after 500 C (analysed at 350 C)

0.50 1.00

1.50

2.00 2.50

NaK 0.50 1.00

3.50

4.00 4.50 5.00

\\Aemf-quanta\Spectra\Theo\CaC2O4 spectra\CaC2O4 after 750C.spc Label A: CaC2O4 after 750 C (analysed at 350 C)

CaK

CaK

CK OK

3.00

CK OK

CaK

AIK

CaK

AIK NaK

1.50

2.00 2.50

3.00

3.50

4.00

4.50 5.00

0.50

1.00

1.50

2.00

2.50

3.00 3.50

4.00 4.50

5.00

Figure 12. EDX analysis of the decomposition of weddelite at 30 °C (top left), 280 °C (top right), 500 °C (bottom left) and, after heating to 750 °C (bottom right) Heating to 500 °C should result in the formation of calcium carbonate, while CO is released. No distinct change in morphology was observed at this temperature. However an EDX spectrum taken at 350 °C, following heating to 500 °C, indicated a significant loss of oxygen and carbon as compared to an initial spectrum at 30 °C, reflecting the conversion to CaCO3 [41]. The corresponding IES spectra show the disappearance of the oxalate modes around 1640, 1325 and 788 cm'1. At the same time new bands become visible around 712, 874, 1413 (with second band centred around 1471 cm"1) and 1794 cm"1. Taking into account a shift of all bands due to the high temperature, these bands agree well with those of CaCO3 [42,44].

IES of Weddelite

Wswnumber

Figure 13. IES spectra of weddelite between 650 and 1800 cm" at 25 OC intervals &om 75 to 650 OC. Finally heating to 750 OC should result in the formation of calcium oxide. The corresponding X-ray spectrum does show a residual carbon peak, but we attribute this to some electron beam scatter onto the underlying carbon paint, as the sample now appeared more porous. At 750 OC imaging was possible using a rapid scan rate, but it proved very difficult to record good quality slow scan images because of signal noise. Since the decomposition reactions of calcium oxalate under the low pressure conditions are irreversible, the sample was cooled to 500 OC to obtain the images shown in Figure 11. The morphology of the crystals changed from massive to a more sponge-like material. The initial oxalate crystal morphology remained intact during this final decomposition step, but arrays of very small oxide particles were formed within the original crystal boundaries of the oxalate crystal. This way the material accommodates the volume loss during the thermal decomposition. This pseudomorphic growth of the oxide particles has been observed previously by transmission electron microscopy of decomposed Mg and Zn oxalates by one of the authors [45], but

in the ESEM it has been proven possible to observe directly the formation of such particles. In contrast to the rather sharp mass loss around 600°C in the TG the IES spectra show a more continuous decrease in intensity of the carbonate bands when the temperature is increased fiom 400 to 650 OC, but due to the fact that the reaction takes some time, it was not possible to observe the spectrum of the final product CaO [42]. 3.3. Sublimation of urea CH4N20 A second example of the application of heating stage ESEM is the observation of a sublimation reaction in situ. Figure 14 shows the sublimation pressuretemperature diagram of urea (modified after [46]). Under normal conditions at high temperature urea decomposes into ammonia and isocyanic acid (see e.g. [471)-

H2N-CO-NH2(melt) + heat -,N&NCO(melt) * NH3(g) + HNCO (g) urea ammonium iso-cyanate ammonia iso-cyanic acid

Figure 14. Sublimation diagram of urea as function of temperature and sublimation pressure (modified afier [46]). Schaber et al. [48] showed that, in the pyrolysis of urea, no significant reaction is observed when heating from room temperature to 133 OC (urea melting point). Urea typically melts with difficulty, and usually a complete melt, within a reasonable time period, is not achieved until 135 OC. Mass loss begins in earnest at 140 "C and is primarily associated with urea vaporization. At 152 OC, urea decomposition begins, as noted by vigorous gas evolution fiom the melt. However, under vacuum conditions at relatively low temperatures, such as those that can be achieved with heating stage ESEM, urea will sublime at lower temperatures rather than melt and vaporise as observed in TG experiments.

Figure 15a and b show the effects of heating urea fiom room temperature to 100 O C . Around 80 O C significant changes were observed associated with the start of the sublimation. The morphology of the initial particles is lost and the surface shows roughening. At 100 "C all the urea has disappeared from the heating stage. However, because the remainder of the vacuum chamber of the ESEM is not heated, crystallisation of the urea was observed on the coolest parts (Figure 1%).

(c>

Figure 15. Heating stage SEM images of urea at 28 "C (a), while heated at 80 "C (b), and recrystallized urea (c).

3.4. Wettingldrying of montmorillonite Figure 16a shows a montmorillonite clay (Swy-2 from the Clay Minerals Society Source Clays Repository) in its dry initial state as small aggregates at 2 "C and 4.3 Torr. Very characteristic here is the open morphology caused by the fast evaporation of the water from between the clay layers. This is the structure often reported in the literature for SEM observations of smectitic clay samples (e.g.[49, 501). Increasing the pressure to 5.2 Torr does not change the morphology of the clay. Some very minor changes around the edges can be observed. The edges of the clay layers seem to become slightly more rounded and less sharp due to the start of very small amounts of water condensing on those edges. This can be explained by the fact that the clay 001 faces are rather hydrophobic compared to the charged sites on the edges formed by for example hydroxyl groups on the unsaturated A1 and Si bonds at the edges of the clay sheets [51]. A further increase of the pressure to 5.3 Torr results in the rapid wetting of the whole particle. Figure 16b shows the wetted sample after decreasing the pressure back to 5.2 Tom. A slow decrease in pressure to 4.5 Torr results in the slow drying of the clay aggregate. There are clear differences in the morphology visible after the drying step, compared to the initial morphology before the wetting. Rapid increase in the pressure to 6 Torr leads to the fast formation of large water droplets wetting the clay aggregates. The water evaporates quickly in the form of bubbles. The montmorillonite, which was completely dispersed in the water droplets, is left behind along the edge of the bubble, forming rounded upstanding rims on the sample holder (Figure 16c). In the next wetting and drying cycle, one of the rims of vertically oriented clay particles formed in the first wetting and drying cycle was observed. Figure 16d shows the initial particles at 2 OC and 5.1 Torr. Increasing the pressure results in a rounding of the edges. This is very similar to the observations for the first wetting and drying cycle for a whole aggregate. Figure 16e shows that complete wetting takes place very slowly after increasing the pressure slightly to 5.2 Torr. It takes about 2 minutes under these conditions to reach complete wetting. EDS analysis after each wetting and drying cycle does not show any changes in the chemical composition. Because no cations or anions were dissolved in the water, no cation change or dissolution was possible at the temperatures employed in this study. XRD results showed an expansion of the basal spacing as a function of the relative humidity, due to the intercalation of water in the interlayers of the smectite. This effect could not be directly observed in the ESEM images.

(dl

(el Figure 16. Wettinddrying of montmorillonite: dry at 2 OC and 4.3 Torr (a), fully wetted after increasing pressure to 5.2 Torr (b), dry clay aggregate formed after droplet evaporated (c), aggregate partially (d) and fully wetted (e).

4. HEATING STAGE PHOTOELECTRON SPECTROSCOPY (XPS) Photoelectron spectroscopy as an analytical tool has only received limited interest in the field of mineral science. Photoelectron spectroscopy, together with Auger electron spectroscopy, gives information about the positions of the energy levels in atoms or molecules. The principle of photoelectron spectroscopy is rather simple: photons with a certain energy are allowed to collide with an atom, molecule or a solid material. These photons can then interact with electrons present in the atoms and one of these electrons can be excited from a low energy orbital, resulting in a situation similar to a free electron plus a positively charged atom or molecule. The kinetic energy of the free electron is determined by the relationship:

where hu represents the energy of the photon (h = Planck's constant, v = frequency of the incident X-rays), Ekinetic the kinetic energy of the free electron and Ebinding the original (negative) binding energy. The ionisation potential of an electron can be defined as I = -Ebinding, SO this technique measures the ionisation potentials through, at a given hu, measuring the kinetic energy of the emitted electrons. Only those electrons with a binding energy lower than the energy of the photons used can be detected. The most common photon sources are A1-Ka and Mg-Ka X-rays, resulting in photons with energies of 1486.6 and 1253.6 eV, respectively. In these cases the technique is known as X-ray Photo-electron Spectroscopy (XPS) or Electron Spectroscopy for Chemical Analysis (ESCA). The second name points to the fact that, by analysing the energy of the photoelectrons and their energy levels, the chemical identity of the atom involved can be determined. XPS is an excellent qualitative analysis tool especially suited for the characterization of material surfaces but the application of heating stage XPS to study thermal changes in minerals and other materials is relatively unexplored area of research. The XPS analyses were performed on a Kratos AXIS Ultra with a monochromatic A1 X-ray source at 150 W. Each analysis started with a survey scan from 0 to 1200 eV with a dwell time of 100 milliseconds, pass energy of 160 eV at steps of 1 eV with 1 sweep. For the high resolution analysis, the number of sweeps was increased, the pass energy was lowered to 20 eV at steps of 100 meV and the dwell time was changed to 250 milliseconds. The spectra were charge corrected using the advantageous C 1s signal at 285 eV. For the heating experiments, a small amount of sample was loaded onto a gold-coated copper stub which could be heated to 600 OC, either in the sample treatment

chamber prior to analysis, or directly in the sample analysis chamber. Temperature control is through a thermocouple in contact with the bottom of the stub. Band component analysis was undertaken using the Jandel 'Peakfit' software package, which enabled the type of fitting function to be selected and allows specific parameters to be fixed or varied accordingly. Band fitting was done using a Lorentz-Gauss cross-product function with the minimum number of component bands used for the fitting process. The Gaussian-Lorentzian ratio was maintained at values greater than 0.7 and fitting was undertaken until reproducible results were obtained with correlations of r2 greater than 0.995. 4.1. Dehydration of calcium oxalate monohydrate CaC204.H20 Heating stage XPS was used to study the dehydration of calcium oxalate monohydrate (whewellite). Figure 17a shows the survey scan of the starting material at room temperature. Characteristic are the 0 Is, Ca 2p and C 1s peaks. Table 1 gives the chemical analyses based on the high resolution XPS data for 0 Is, C 1s and Ca 2p. The carbon content is slightly higher than expected for pure calcium oxalate due to the presence of a small amount of absorbed carbon on the surface. The high resolution C 1s and Ca 2p do not show any changes upon heating to 150 OC while the 0 1s shows a strong decrease in the water band (Figure 17b). After heating to 200 OC, no water is left and only the band associated with the oxalate group was observed. The shift in binding energy for the C 1s associated with the oxalate is similar to that of the C 1s of calcium carbonate and is observed around 289 eV.

Table 1 Changes in the chemical composition of whewellite (CaC204.H20)as a result of dehydration. Peak 0 1s Ca 2p C 1s Formula

Atom % (room temperature) 60.06 12.46 27.48 CaC204.H20

Atom % (150 "C) 58.61 13.13 28.08 CaC204

0 1s

Ca 2p

Ca Auger

0 Auger 0 2s

1200

1000

800

600

400

200

0

Rinding enernlev

Figure 17a. XPS Survey scan of whewellite (CaC204.H20).

Figure 17b. 0 1s high resolution scans of whewellite (CaC204.H20)at 25 OC and 100 OC showing a strong decrease in the 0 1s of water associated with the dehydration. 4.2. Calcination of titania1PVA expanded hectorite Heating stage XPS is an ideal technique to study the preparation of pillared clays. Pillared clays are porous heterogeneous catalysts prepared by propping apart the individual clay layers with the help of a large inorganic complex [52]. Calcination at an appropriate temperature converts the inorganic complex into

an oxide covalently bonded to the clay layers resulting in what might be called a two-dimensional zeolite-type material with acidic and shape-selective properties. Titania is one of the commonly used pillar types, together with alumina. In this example, a titania sol was prepared through hydrolysis of tetraisopropoxytitanium(1V) (TPT). The sol was exchanged with hectorite clay in the presence of a large organic molecule, hydrolysed polyvinyl alcohol (PVA), to increase the distance between the pillars within the interlayer space of the hectorite. The calcination process was studied with standard thermal analysis and heating stage XPS. TG gave a slow mass loss up to about 200 "C, followed by a rapid mass loss in the range 200 to 250 O C and a further slow mass loss up to the dehydroxylation temperature of the hectorite (causing the complete clay structure to collapse). During this process the initially white material turns via gray to black, due to the decomposition of the PVA. Figure 18 shows the survey scan at 100 O C . Easily recognisable peaks are those of the Si, Mg and F of the hectorite clay, together with the Ti of the titania and the high C signal from the PVA. The 0 peak is very strong and contains signals from both the clay layers and the interlayer titania, PVA and water. Upon heating, the clay layers will not change, hence no changes are observed in the F Is, Si 2p and Mg 2p high resolution scans. 8000

10 Is

Mg Auger

0 Auger

1

4 ' i d /

F 1s

Mg 2s

Ti 2p

FAugw

1

Figure 18. XPS survey scan of titania/PVA expanded hectorite clay. For pure polyvinyl alcohol, [-CH2CH(OH)-I,, one would expect two C 1s bands of equal intensity plus a small band associated with the O-C=O endgroup. However there is always some absorbed carbon present on the surface of samples. Figure 19a shows the band component analysis of the C 1s region at

100 "C with the C=O at 289.5 eV and the C-OH band at 286.6 and the CH2band at 285.0 eV in good agreement with published results for PVA films [53]. Figure 19b shows the changes in the intensity of the C Is peak with heating to 500 "C. Upon the decomposition of the PVA, the C-OH band around 286.6 eV decreases in intensity and has completely disappeared around 400 "C. In addition, the C=O band shows a decrease in intensity upon heating.

2P2

2911

2W

3 6

bind in^ emrp(v/eV

(a)

284

2XL

2%

294

2W 2E3 2116 Binding cner&V

292

284

282

(b)

Figure 19. C 1s high resolution scans of the titaniaPVA intercalated hectorite: (a) band component analysis showing the characteristic C-H, C-OH and C=O bands for hydrolysed PVA, (b) effect of heating to 500 OC on the C 1s spectra showing the decomposition of the PVA.

4 7

4 2 BindingenewleV

4 9

452

5%

5%

SM

52

S O

SZn

Binding energvie\'

Figure 20. Heating stage high resolution Ti 2p (a) and 0 1s (b) spectra of titaniaJPVA intercalated hectorite, (c) and (d) show the change in the 0 Is from 25 to 500 OC as a result of the decomposition of the PVA. The Ti 2p signal shows two transitions, the 2p 112 and the 2p 312. With calcination the Ti 2p 1/2 shifts from 458.6 eV to 459.1 eV (Figure 20a) reflecting the change from the hydrolysed titania sol to titanium oxide bonded to the clay layer. A similar shift is observed for the 0 Is high resolution spectra with increasing temperature Figure 20b). At 100 OC the 0 1s contains four bands; two are associated with the hectorite structure at 532.2 eV (0)and 533.2 eV (OH), while the other two are associated with the PVA at 531.5 eV and the

titania sol at 530.7 eV (Figure 20c). Upon heating the sample to 500 OC, the 0 1s signal becomes slightly sharper due to the fact that the band associated with the PVA has disappeared and the titania sol has converted into a strongly bonded form of Ti02 to form a stable titania-pillared hectorite (Figure 20d). 5. CONCLUSIONS

The use of thermal stages with infrared, Raman, energy dispersive X-ray, and X-ray photon spectroscopy strongly complements the results obtained from traditional thermoanalytical techniques. These techniques enable the changes in the composition and molecular structure of minerals and materials to be obtained in situ at elevated temperatures. 6. ACKNOWLEDGEMENTS

The financial and infra-structure support of the Queensland University of Technology Inorganic Materials Research Program of the School of Physical and Chemical Sciences is gratefully acknowledged. The Australian Research Council (ARC) is thanked for funding the Thermal Analysis Facility through an ARC LIEF grant. The authors acknowledge the experimental work undertaken by the following students: Ms Jocelyne Bouzaid, Ms Daria Wain, Mr Matthew Weier and Ms Kristy Erickson. 7. REFERENCES

P. Baraldi, Chimicaoggi, (1990) 53. F.J. Deblase and S. Compton, Appl. Spectrosc., 45 (1991) 61 1. M. Planck, Ann. Phys. 4th series (vierte folge), (1901) 553. W. W. Coblenz, in Supplementary Investigations of Infra-red Spectra, Vol. Part VII, The Carnegie Institute of Washington publications, Washington D.C, 1908. W.W. Coblenz, Bull. Bureau of Standards, 9 (1912) 81. S.F. Kapff, J. Chem. Phys., 16 (1948) 446. S.F. Kapff, Science, 105 (1946) 274. H.O. Mcmahon, J. Opt. Soc. Arner., 410 (1950) 376. R. Gordon, J. Amer. Ceram. Soc., 39 (1956) 278. M.J.D. Low, Nature, (1965) 206. P.R. Griffiths, Appl. Spectrosc., 26 (1972) 73. G. Fabbri and P. Baraldi, Appl. Spectrosc., 26 (1972) 593. B. Jonson, B. Rebenstorf, R. Larsson and M. Primet, Appl. Spectrosc., 40 (1986) 798.

A.M. Vassallo, P.A. Cole-Clarke, L.S.K. Pang and A.J. Palmisano, Appl. Spectrosc., 46 (1992) 73. J. Mink and G. Keresztury, Appl. Spectrosc., 47 (1993) 1446. J. Hvistendahl, E. Rytter and H.A. Ye, Appl. Spectrosc., 37 (1983) 182. M.E.E. Madden, R.J. Bodnar and J.D. Rimstidt, Nature, 431 (2004) 821. D.T. Vaniman, D.L. Bish, S.J. Chipera, C.I. Fialips, J. William Carey and W.C. Feldman, Nature, 43 1 (2004) 663. J.-P. Bibring, Y. Langevin, F. Poulet, A. Gendrin, B. Gondet, M. Berthe, A. Soufflot, P. Drossart, M. Combes, G. Bellucci, V. Moroz, N. Mangold and B. Schmitt, Nature, 428 (2004) 627. B.M. Hynek, Nature, 43 1 (2004) 156. E. Libowitsky, M.hefte Chemie, 130 (1999) 1047. G.D. Danilatos, Microsc. Res. Techn., 25 (1993) 354. G.D. Danilatos, Adv. Electr. Electr. Phys., 71 (1988) 109. J.C. Baker, B. Grabowska-Olszewska and P.J.R. Uwins, Appl. Clay Sci., 9 (1995) 465. J.C. Baker, P.J.R. Uwins and I.D.R. Mackinnon, J. Petroleum Sci. Eng., 11 (1994) 241. J.C. Baker, P.J.R. Uwins and I.D.R. Mackinnon, J. Petroleum Sci. Eng., 9 (1993) 83. P.J.R. Uwins, J.C. Baker and I.D.R. Mackinnon, Microsc. Res. Techn., 25 (1993) 465. J.C. Baker, P.J.R. Uwins and I.D.R. Mackinnon, J. Petroleum Sci. Eng., 6 (1993) 269. J.C. Baker, P.J.R. Uwins and I.D.R. Mackinnon, J. Petroleum Sci. Eng., 18 (1997) 83. S.M. Habesch, S.C. Hogg and H.V. Atkinson, Microstruct. High Temp. Mater., 5 (2001) 147. V.A. Maroni, M. Teplitsky and M.W. Rupich, Physica C: Supercond., 313 (1999) 169. A.H. Foitzik, M.W. Futing, G. Hillrichs and L.-J. Herbst, Scanning, 19 (1997) 119. I. Belenli, M. Ebrahimi and Y.S. Hascicek, Physica C: Supercond., 247 (1995) 371. N.S. Srinavas, Powder Techn., 124 (2002) 40. D. Dollimore, Thermochim. Acta, 117 (1987) 33 1. D. Dollimore, D.L. Griffiths and D. Nicholson, J. Chem. Soc., (1963) 2617. D. Dollimore, T.E. Jones and P. Spooner, J. Chem. Soc. A, (1970) 2809 V. Jordanovska, R. Trojko and N. Galesic, Thermochim. Acta, 198 (1992) 369.

J. Li, G. Zhang and W. Li, Thermochim. Acta, 205 (1992) 145. J.W. Stucki and D.L. Bish, Thermal analysis in clay science, CMS Workshop Lectures Vol. 3, The Clay Minerals Society, Boulder, 1990. J.T. Kloprogge, T.E. Bostroem and M.L. Weier, Amer. Miner., 89 (2004) 245. J.T. Kloprogge and R.L. Frost, The application of vibration spectroscopy to clay minerals and layered double hydroxides, CMS Workshop Lectures, Vol. 13 (2005) 99. R.L. White and J. Ai, Appl. Spectrosc., 46 (1992) 93. J.A. Gadsden, Infrared Spectra of Minerals and Related Inorganic Compounds, Buttenvorths, England, 1975. J.T. Kloprogge and T.P.M. Koster, Confidential TNO report, TPD-AM-RPT-94-142 (1994) 2 1. D. Ferro, G. Barone, G. Della Gatta and V. Piacente, J. Chem. Thermodyn., 19 (1987) 915. J.A. Caton and D.L. Siebers, Combust. Sci. Tech., 65 (1989) 277. P.M. Schaber, J. Colson, S. Higgins, E. Dietz, D. Thielen, B. Anspach and J. Brauer, Amer. Lab., August (1 999) 13. D.L.A. Tessier and J.C. Petit, Appl. Geochem., (1992) Suppl. Issue 1. D. Tessier and G. Pedro, Deverlopments in Sedimentology (l982), 35(Int. Clay Conf. 1981)165-176. C.T. Johnston, in Organic pollutants in the environment, CMS Workshop Lectures, Vol. 8 (Ed. B. L. Sawhney), The Clay Minerals Society, Boulder, 1996, p. 1. J.T. Kloprogge, J. Por. Mater., 5 (1998) 5. P. Stoyanov, S. Akhter and J.M. White, Surf. Interf. Anal., 15 (1990) 509.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors 2008 Elsevier B.V.

Chapter 7

ELECTRIC TECHNIQUES Madalena ~ionisio'and Jolo F. ~ a n o ~ * 1

REQUIMlX/CQFB, Departamento de Quimica, FCT, Universidade Nova de Lisboa, 2829-5 16 Caparica, Portugal, 2 3 ~ 'Research s Group - Biomaterials, Biodegradables and Biomimetics; Department of Polymer Engineering, University of Minho, Campus de Gualtar 4710-053 Braga, Portugal; *e-mail: [email protected]

1. INTRODUCTION Dielectric materials in the presence of static electric fields In electrical or dielectric measurements, the material to be characterized is usually placed between two conducting electrodes, where an electric field can be created within it by application of a voltage. The extent to which a material responds to an applied electric field can be discussed using a parallel-plate capacitor model (see Figure 1). This section is intended to provide a brief introduction to the general dielectric properties of materials and more and detailed information can be found in [I-41) for example. 1.1.

Figure 1. Parallel-plate capacitor in (a) a vacuum and @) filled with a dielectric material.

For a fixed voltage, V, applied in the circuit with a vacuum between the plates (Figure la), and ignoring edge effects, a uniform electric field will be produced: E=VId, where d is the distance between the plates. The charge density produced at the surface of the plates is proportional to the electric field, o=E&, where EO is the ~.cm-').The capacitance of the capacitor permittivity of free space (8.85418~10-'~ in vacuum,Co, is given by Co=o/V. When an isolator material is placed between the plates (Figure lb), the charges within it are redistributed and the dipoles will be preferentially oriented along the electric field. The surfaces of the material will be charged, with a density of charge P, called the polarization. In this case, more charge can be stored in the electrodes, for the same value of V. The capacitance will be now given by C=(o+P)IV. The permittivity of the material is given by the ratio between the capacitances of the capacitors with the dielectric and in vacuum:

Another important quantity is the electric displacement, D, which corresponds to the total charge density in the surfaces of the electrodes (o+P):

When an alternating field is applied to a dielectric containing permanent dipoles, these dipoles might follow the changes in the field's orientation, as long as the frequency is sufficiently low. At sufficiently high frequencies, the dipoles do not have enough time to follow the changes in the field's direction, and start to lag behind such orientional changes. Above such a frequency range, the motions of the dipoles will be not able to follow at all the changes in the direction of the field, and the polarization resulting from the dipole orientation will not contribute to the total polarization. However, at such frequencies, other induced polarizations can still contribute to the total polarization, such as the atomic or electronic polarizations. Atomic polarization is the relative displacement of atoms in a polar covalent bond, whereas electronic polarization is a slight displacement of electrons with respect to the positively charged nucleus. The permittivity due to these induced dipoles is known as the unrelaxed or permittivity at infinite frequency, E,. Therefore, the contribution of the orientation of the dipoles to the polarization is linked to the

difference between the permittivity at low frequency, E,, and at high frequencies, E, (contribution only of the induced polarization): P=Eo(E,-E,)E. As commented before, the orientational contribution for the polarization emerges only at lower frequencies, and therefore it is the slowest mechanism. We will consider that the other contributions may follow the changes in the electric field instantaneously. The temporal change in the electric displacement as a result of a static electric field appearing at t=O is:

The E~ES term corresponds to the "instantaneous" response of the material to the electric field, whereas the E~(E,-E,)@(~)E term is related to the slower response assigned to dipolar polarization, where the dielectric function Q(t) describes the temporal development of the dipole orientation. The decay function, $(t)=l-@(t), accounts for the decrease of polarization after removing the electric field, $(0)=1 and $(co)=O. In the model of Debye the polarization process follows a first-order kinetics, where its time variation is proportional to the equilibrium value:

where P, is the equilibrium polarization (t=co) and z is the relaxation time. Therefore, upon removing the electric field at PO, the orientational polarization will be given by P(t)=Poexp(-tlz), where Pois the value of the polarization at the moment of field removal. Therefore, in the Debye model, the reaction of materials to external electric fields has an exponential nature, i.e., $(t)=exp(-tlz). 1.2. Application of alternating electric fields 1.2.1. General phenomenological behaviour In the situation that the electric field is not static the increment in D(t) resulting from the increment in dE at time t=u can be calculated from Equation (5):

According to the Boltzmann superposition principle [2], the response of a system to a time-dependent external field can be expressed by the superposition of responses of fields at different times. Each response is only dependent on the magnitude of

the fields and on time. By knowing the time dependence of the field, Equation (5) may be integrated, resulting in:

A useful practical case is to consider an alternating electric field, described, in the complex form, by E(t)=Eoexp(iot), where i=.\l(-1), o is the angular frequency (in rad 6') and Eo is the field amplitude. From Equation (2) the permittivity is ~*=D(t)l(~&(t)) and its angular dependency is:

The integral is an imaginary Laplace transform of -d$/dt that gives rise to a complex permittivity, with real and imaginary components: ~*(o)=~'(o)-i~"(o). As commented before, within certain frequency ranges there is a phase lag between the change of the oscillating electric field and the orientation of the dipoles. This phase lag, 6, depends on the frequency being approximately zero both at low frequencies, when the orientational polarization is able to accompany the electric field (the nearly static case), and at sufficiently high frequencies, when orientational polarization is lost. From this perspective, the electric displacement may be written as D(t)=Doexp[i(ot-6(o))] and the frequency dependence of the permittivity is given by:

The real part of the permittivity, ~ ' ( o ) ,is a measure of the polarization and decreases with increasing frequency, from E, down to E,. ~ ' ( w )is also proportional to the energy stored reversibly in the material per cycle, whereas the imaginary

component of the permittivity, E"(w), is proportional to the energy which is dissipated per cycle. The loss factor, tan ~=E'(w)/E"(w) is the ratio between the energy dissipated and the energy stored in the system during one cycle. 1.2.2. Debye behaviour For the case of an exponential decay, $(t)=exp(-tlz), equation (7) takes the form:

where the real and imaginary components are:

log ,,f

log f

Figure 2. Frequency dependence of the real and imaginary permittivities in a simple Debye process. The half-height width of ~"(logj)can be shown to be 1.14. In dielectric relaxation spectroscopy, both components of the permittivity are measured as a function of frequency (see the next section). Figure 2 shows the typical shape of both ~'(logj)and ~"(logj),wherej+~/2.x:is the frequency (in Hz).

The relaxation time is given by r=1/(2nfm,), maximum E".

where f,,

is the frequency at

1.2.3. Complex systems In real systems, typical decay functions are far from exponential, and loss curves, &"(log j), appear much broader than the ones found by the Debye model. A phenomenological approach to deal with this observation is to consider that this enlargement in the response is the result of the existence of a distribution of relaxation times, with a probability density function of p(ln 7):

In the frequency domain, the distribution of relaxation times changes equation (9) to:

It is possible to obtain p(ln r) from experimental data, but this is a difficult problem. To fit the &*(a) data directly, several empirical models have been proposed that enable valuable information to be extracted. The most used equations can be written in the following form:

with x=a, y=l (O
describes an isolated relaxation process with great success. When the frequency window covers more than one process, the experimental data can be fitted with a sum of HN equations, one for each relaxation process detected. Its shape parameters, am and Pm, describe the slopes of the E" peak at frequencies lower and higher than w,, that is the frequency close to the one that corresponds to E",: C&HN = dl0g ~"/dIog(3for 6l << a,, and aHN.PHN = -dog ~"/dlog6l for >> (3,. The characteristic relaxation rate, om, or relaxation time zm,=l/om,, can easily be obtained from zm (the fitted z parameter in equation (13)), usually preferred because it is a model-independent parameter, through the equation [4]:

log(flfmax) Figure 3. Imaginary component of the permittivity, calculated taking into account the KWW model (Equation (15)), obtained at different PKWW values (shown in the figure). Another way to extract information on Q(t) is to assume a stretched exponential function, as described by the Kohlrausch-Williams-Watts (KWW) model 19,101:

where PKWW is a characteristic time and PKWW is a parameter that describes the nonexponential behaviour of the decay function. A decrease of PKWW may be related to a broadening of the distribution of relaxation times, which also present an asymmetric shape. Figure 3 shows several E" curves obtained by transforming equation (1 5) in the frequency domain. Alegria and co-workers [ l l ] published a comparison between the HN and KWW descriptions, and proposed simple empirical correlations between the fitting parameters of the two models. 2. MEASUREMENT TECHNIQUES

2.1. Introduction Nowadays the frequency range covered by dielectric relaxation spectroscopy is something like 18 decades [12], from 1u6up to 1012Hz, more commonly [13] from lo4 to 10". This exceptionally wide frequencyltime window enables us to assign dielectric response from slow to fast molecular events, but the range is not able to be covered by a single apparatus. Thus, a combination of several equipmentslmethods is used. A recent review of the instrumentation for broadband dielectric spectroscopy can be found in [14,15].The various polarization mechanisms can be modeled by equivalent circuits composed of discrete electrical components, where a resistance (R) is taken to represent the dissipative component and a capacitance (C), the storage component of the dielectric response [16].The simplest equivalent circuits are obtained by combining R and C, either in parallel or in series. The experimental techniques designed to measure the equivalent capacitance and resistance at a particular frequency, are generally classified as "lumped circuit" methods covering the frequency range from to 10' Hz [2]. The high limit of most measuring instruments (ac impedance analyzers) attains the MHz region, where the sample geometry is a disc-like capacitor arrangement. At about 1 MHz parasitic inductances of the lines and connectors start to affect measurements of sample impedance. Thus, for higher frequencies, the sample capacitor arrangement is changed and is used as the termination of a precision coaxial line (coaxial line reflectometers from 1 MHz to 1 GHz) [17]. Above 1 GHz, up to 100 GHz, network analyzers are used with cavity resonators or waveguides [18]. From 10'' to 1012Hz quasi-optical interferometers are used (see [15] and references therein).

Alternatively, as seen for the low-frequency range, high-frequency measurements can be performed in the time domain (time domain reflectometry methods [19]) that can be extended to 20 GHz, but with less accuracyrelative to network analysis [141. 2.2. Equivalent circuits As already mentioned, the lumped-circuit technique is the most widely used by the research community [16]. The dielectric response is measured in parallel-plate or coaxial geometry as an association of resistances (R) and capacitances (C) in parallel or series. In a.c. measurements only, the parallel and series electrical circuits can be forced to be equivalent [20] giving for the admittance, Y(o) and impedance Z(o), the following equations:

where Ci and Ri vary with frequency; o is the angular frequency. This equivalence does not apply to d.c. step-function experiments [20]. The complex permittivity is given by &*(a)= C *(a) lCO,analogous to equation (1) for the static case, where C0= A.Eo/~is the vacuum capacitance of a parallelplate capacitor (A=area of the plate, EO the vacuum permittivity and d the plate separation; (Ald) must have dimensions in cm) and C* is the complex capacitance of the same capacitor filled with the material under study. Under the influence of a sinusoidal electric field, the complex permittivity relates to the impedance through:

The simplest circuit describing a relaxation process with a single relaxation time, i.e., the circuit whose R and C combination leads to the Debye equation, is obtained when the resistance R1 is associated in series with the capacitance C1 and the contribution of induced polarization due to atomic and electronic polarization, C,, is associated in parallel with CIRl according to Figure 4 a) [16, 201 where all the components are frequency independent.

Figure 4. Circuit diagrams for a material exhibiting a) a relaxation process with a single relaxation time and induced polarization, b) a relaxation process with a single relaxation time, conduction and induced polarization and c) a distribution of relaxation times and induced polarization. Both references [16] and [20] present several equivalent model electrical circuits for relaxation and conduction. For the circuit presented in Figure 4a) the admittance is:

Writing E~=C,JC,and z R ~ R s Cand s multiplying both numerator and denominator of the above quotient by C,, gives:

- ~ r n , to equation (9) for the Debye case, where separation where C s I C ~ = ~ Oidentical into the real and imaginary components leads to equation (10). Additionally, if translational diffusion of mobile charges occurs, i.e. if the material exhibits conduction (the schematic circuit is presented in Figure 4b) the admittance is:

that, writing as before, E,=CJC, and T R F R ~ Cand ~ multiplying both numerator and denominator of first quocient above by C , gives:

where CAC,=E~-E,. The conduction process appears as a low frequency tail in the plot of E", yielding a value for CoR,=~oloo,being oothe frequency independent specific conductivity. The existence of a distribution of relaxation times can be taken in account by describing the circuit as a parallel association of several RiCi's associated in series (Figure 4c)). 2.3. Time-domain measurements While in lumped-circuit methods the dielectric response is measured in the frequency domain, following the a plication of a sinusoidal alternating electrical field, for frequencies below 10 Hz it is advantageous to carry out the measurements in the time domain because it is less time consuming. The polarization or depolarization current following the application of a step-like electrical field is measured as a function of time. The depolarization current, I(t) (it is advantageous to carry out depolarization experiments because they are not disturbed by d.c. conductivity [15]), is related to the decay function, $(t), through:

P

where Vo is the polarizing voltage (Co is the capacitance of the empty capacitor as already defined). Taking into account equation (7), we can write:

where L[I(t)] is the Laplace transform of the transient current. This establishes a relation between the polarization decay in the time domain, with the complex permittivity in the frequency domain [21]. To obtain a frequency dependent dielectric relaxation spectrum it is necessary to apply Fourier transform techniques (the particular case of the Laplace transform in equation (22) is a Fourier transform) such as the Fast Fourier Transform. Some approximate numerical transformations were earlier proposed as Hamon's transformation [22], which relates the angular frequency with time by the expression ot=0.1.2~,obtaining:

but whose applicability is only reasonable in a limited range -0.5<1og(o/om,)<0, with serious deviations for lower frequency values [9]. A best approximation was proposed by Brather that estimates &"(a) for angular frequencies o=o,.~-' with oo=2/to and 1=1,2..k, when i(t)=I(t)lVo is measured at t=t0.2"' with 1=1,2,...,k+3 1231; this approximation is tested in references [24-251 predicting a frequency of maximum loss nearer to the value estimated by FFT than to the one found by Hamon's method, with the advantage that both &"(a) and &'(a) can be calculated, contrary to the Hamon transformation, which only allows estimation of &"(a). 2.4. Cells

The measurement technique in lumped-circuit methods estimates the equivalent capacitance and resistance that describes the dielectric sample, which is submitted to a sinusoidal voltage applied across two conductive electrodes. It is necessary to know the plate area and spacing precisely in order to get quantitative information. One of the geometrical arrangements that allows the accurate knowledge of both sample area and thickness is the three-terminal arrangement as described in [2,25] (see Figure 5). A similar three-terminal cell for measuring capacitance in liquid solutions is presented in [26,27]. The dielectric material in the form of a flat circular disc of thickness d, is placed between the electrodes to form a parallel-plate capacitor of plate area A. The upper or high electrode (H) is directly connected to the high-voltage terminal of the measuring unit, whereas the lower electrode (L) is similarly connected to the lowvoltage terminal. These two terminals are isolated from the metal screen of the BNC-type connector and cables. The third or guard electrode (C) is held at the same potential as the lower electrode, but is isolated from it by a Teflon guard. The

guard electrode serves to eliminate the fringing of the field between the low and high electrodes and also the effect of surface conduction round the edges of the dielectric sample that affect two-terminal cells. When properly grounded, the guard electrode can further avoid stray capacitances between a lead and the opposite electrode, provided that the low electrode runs at zero potential. Furthermore, the existence of the guard electrode allows a clear and unambiguous definition of the geometry of the electrode system, so that the vacuum capacitance, Co,can be calculated accurately. The three-terminal arrangement ensures that the field distribution in the guarded area is identical, whether the capacitor is filled with the sample or empty. The real permittivity is thus calculated as the direct ratio of the capacitances between the high and low electrodes in the two cases, that is, E' =Csample/Co.

Figure 5. Three-terminal parallel-plate capacitor showing uniform field lines (dashed lines) between the high-voltage (H) and low-voltage (L) electrodes. Fringing fields (distorted dashed lines) are present only between the electrode H and the guard electrode, C. The ensemble is enclosed by a metal screen connected to ground (detailed scheme in the Supplemental Material of reference [28]). Most of the commercial versions of the equipment currently use two- or fourterminal cells connected to two parallel plate electrodes, where the edge effect can be taken in account according to the sample geometry. Usually, the lower electrode has dimensions around 10 to 20 rnrn diameter and the sample thickness is around 50 p. In highly viscous samples it is possible to use the parallel-plate arrangement, instead of a cell designed for liquids, by keeping the distance between lower and upper electrodes fixed through the use of spacers (e.g. silica spacers).

Temperature control is of crucial importance so the temperature sensor (e.g. Pt100) should be located as close as possible to the sample. Some commercially available cells have the temperature sensor incorporated in the lower electrode. However, even in these cells, differences between the sample and sensor temperatures may exist and procedures should still be employed to calibrate the temperature (see next section). 2.5. Temperature calibration in dielectric and electrical measurements In any thermal analysis technique it is essential to identify the correct values of the sample temperature at any stage of the experiment. A classic procedure implemented in differential scanning calorimetric measurements, DSC, is based on the use of high-purity standards, usually metals, with well known melting temperatures. The melting temperature measured by DSC is then compared with the exact one, enabling the temperature axis to be corrected [29]. However, the implementation of temperature calibration strategies in dielectric measurements is almost non-existent and the experimentalist usually considers that the sample temperature is that read by the thermocouple. For isothermal experiments, such an approach can be reasonable, despite the fact that thermal gradients could exist between the sample and the thermocouple even at constant sensor temperature. However, many dielectric experiments are performed under nonisothermal conditions, such as multi-frequency dielectric relaxation spectroscopy temperature scans, or thermally stimulated depolarization current experiments (see later in the text). Three different methods of calibration are suggested below, all based on the use of systems that have a relaxation, or transition, that is well characterized in terms of its position on the temperature axis. 2.5.1.Fusion of high-purity standards

Metals cannot be used directly as standards in dielectric experiments due to their high conductivity. A possible strategy for employing them in temperature calibration procedures is to combine them with isolating materials. An example is shown in Figure 6, where a piece of the metallic standard is placed between two dielectric films (e.g. poly(imide) sheets). This ensemble is mounted between two metallic discs and put between the electrodes of the equipment's cell (situation i in Figure 6). During heating, the metal melts at a certain temperature. At this moment the top steel disc (B in Figure 6) separates from the top electrode under gravity and a new empty layer between B and the top electrode is formed (situation ii in Figure 6). The signal of the apparatus will be modified due to this geometric change. An example is shown in Figure 6, using high-purity indium. A clear change in the real

permittivity is detected, with an onset temperature at 145 "C, which is independent of the frequency of the measurement.

Figure 6. Storage permittivity during a temperature scan in a dielectric relaxation spectrometer using indium as the standard, according to the inset scheme, recorded at 1 OC min-' and measured at two frequencies (squares: 500 Hz, circles: 1000 Hz). Inset: set-up for the calibration of a DRS equipment using metallic standards: Aelectrodes of the equipment, B- metallic plaques, C- polyirnide films, D- metallic standard. The results were adapted from [30].

2.5.2. Transition inferroelectric crystals Triglycine sulfate, TGS, exhibits a second-order ferroelectric phase transition at 49 "C. At this Curie temperature the material shows a pronounced dielectric anomaly that could be used to correct the temperature read by the sensor. Figure 7 shows the storage permittivity on TGS measured at 1 kHz, where the effect of the heating rate, P, is also analyzed. The peaks are well defined, allowing for the determination of the transition temperature as read by the sensor, T,.As observed, for example, by DSC, a clear shift towards higher temperatures of the transition is observed when the scanning rate increases. This relationship is found, for this particular case, to be quite linear (Figure 7). Note that the intercept (44.5 OC), obtained by extrapolating the data to /3+O, is different from the expected value for the studied transition. This demonstrates the necessity to correct the temperature even for isothermal experiments, and that this caIibration can be performed by executing heating experiments at different scanning rates and calculating the tendency found at P O .

TPC p I ~c.rnin-' Figure 7. Left: Storage permittivity of triglycine sulfate,TGS, measured at 1 kHz at different scanning rates, P. Right: Temperature of the maximum of the storage permittivity of TGS (sensor temperature, T,) as a function of the heating rate. Data adapted from [30].

2.5.3. Use of the a-relaxation of amorphouspolymers as a standard A simple method to calibrate the temperature is to compare the loss peaks related to the glass-transition dynamics (a-relaxation) of pure amorphous polymers, obtained in a well calibrated apparatus, with the data obtained in the apparatus in which the temperature is not calibrated. The calibration tests in the uncorrected equipment must be done under similar conditions (e.g. heating rate) to the ones that are to be performed in future experiments using that equipment. This method was tested using polycarbonate films [30],where a global comparison between the different strategies proposed in this section is also discussed. 3. DIELECTRIC SPECTROSCOPY IN MODEL SYSTEMS AND ASSIGNMENT OF MOLECULAR MOTIONS

Cooling a liquid to below its melting point can result in crystallisation with loss of both liquid-like properties and disordered nature, or, in the absence of phase transition, in a gradual slowing down of the molecular motion as the temperature

decreases; the viscosity increases concomitantly and the material retains its liquid structure. Subsequent cooling of this supercooled liquid, to low enough temperatures (T=2/3Tm) [31-341 ultimately forces the system to freeze in a disordered state to become a glass [35,36]. This state enters at the glass-transition temperature, T,, usually characterized by a viscosity of 1013 Poise, implying a relaxation time exceeding 100 s [37]. In the glass, the structure of the supercooled liquid is retained in a non-equilibrium metastable state, and the material is no longer able to reach equilibrium within the experimental time scale. The dynamical behaviour of such glass-forming materials and thus of their relaxational processes, covers a wide range of time scales from local vibronic processes at = 10-l4s above Tm,where viscosities are of the order of lo-' Poise, to highly cooperative motion with time scales of lo6 s or more [34]. The overall relaxation behaviour along the temperature axis coming from above Tm is rather complex, being governed by a single response time in the liquid state, with a temperature dependence reflecting a simple activated process, followed by bifurcation into at least two processes in the supercooled regime: a slow non-exponential and non-linear [38] process due to intermolecular cooperativity [39], characterized by a departure from Arrhenius behaviour, and a faster secondary process that still remains active below T,, with Arrhenius-type temperature dependence. This is the most common scenario in polymers 1401 and glass-forming simple liquids [41,42], but two splitting regions can also be found in some glass formers [43]. Consequently, the temperature dependence of relaxation times of the different re-orientational processes detected in a system does not exhibit a single continuous trace. In the following sections, the dynamical behaviour of glass formers in supercooled and glassy states is analysed and the different regimes of the temperature dependence of relaxation times are explored mainly by dielectric relaxation spectroscopy, where the active dipoles in the material act as a probe of molecular motions. 3.1 Sub-glass mobility 3.1.1. Molecular origin The continuous cooling of a supercooled liquid will lead to a loss of the metastable equilibrium at the glass transition, resulting in a glassy state with very low mobility. In this glassy state, two types of dynamical processes are observed: localized motions giving rise to local fluctuations of the dipole vector that are the origin of the secondary relaxation processes detected by dielectric relaxation spectroscopy, and a very slow dynamic process due to structural recovery as evidenced by physical aging processes. It is expected that these slow dynamics should be controlled by the segmental motions in the glassy system which,

although having very long characteristic times, are not completely frozen [44]. Nevertheless, during the experimental time scale where the detection of secondary processes is carried out, the slow dynamics do not affects the measurements and thus, we only will consider here the remaining local mobility.

frequency /Hz Figure 8. Dielectric loss spectra of tetra-ethyleneglycol dimethacrylate between 114 and -86 "C, in steps of 2 "C, showing two secondary relaxation processes; the high loss values on the low frequency side for the highest temperatures is due to the incoming of the main relaxation process associated with the glass transition (T,= 83 "C). Detailed dielectric characterization is given in [47]. It is usual to classify the detected relaxation processes, at a fixed frequency, in order of decreasing temperature (or in order of increasing frequency for a fixed temperature). Thus, besides the a main relaxation described in the next section, the p relaxation is located at the highest temperatures, in comparison with the other sub-glass processes, respectively, y and 6 relaxations, with even more local mobility and shorter relaxation times. Figure 8 shows the loss spectra of tetra-ethyleneglycol dimethacrylate, TeEGDMA, in the temperature range between -1 14 and -86 "C, with measurements at every 2 "C, showing the two secondary relaxation processes, P and y in the glassy

state (calorimetric T,= -83 "C [45]). Local twisting motions of ethyleneglycol moieties are the origin of the y relaxation, while the P process is associated with hindered rotations of the carboxylic groups [46] (the TeEGDMA structure is given in the inset of Figure 8). In the diglycidyl ether of bisphenol-A (DEGBA) epoxy resins, two secondary relaxations are found: a P process assigned to the molecular moiety containing an hydroxyl group, that could be rather weak as in EPON828 [44] because it is only present in one in ten molecules, and the more intense y process ascribed to fluctuations of the most mobile dipolar unit, i.e. the ether group 1481. In polymeric materials, the secondary processes correspond to either limited inchain movements or hindered rotations of side groups, laterally attached to the main chain (or of its subunits) that can occur independently of the backbone movements, or even conformational changes in cyclic side groups [2]. In poly(alky1 methacrylates), it is clear that there are two secondary relaxation processes, with the y process being much less intense and more separated towards lower temperatures/high frequencies relative to the P process [49]. Generally, the beta process is attributed to hindered rotations of the lateral COOR group around the CC bond that connects the side group to the main chain. In the lower members of the series, the reorientational process of the side group involves some coordination with the main chain. From NMR studies [50,511 it was concluded that the P process in PMMA and PEMA, below but near Tp,is not strictly local, but rather consists of .n flips of the carboxyl group coupled to iistricted rearrangements of the main chain (small angle %25"rocking motion of the main chain [52]). The coordinated mobility in the p process in PEMA is the origin of the designation adopted by Garwe et al. [40] as a 'locally coordinative' process. The secondary y relaxation process has been attributed to either localized motions within the R alkyl group [2], or rotation of the a-methyl groups bound to the main chain [53,54]. Both motions could be involved in this local process as proposed in [55] because the activation energy decreases for the highest members of the series due to a larger available free volume that makes the rotations of both a-methyl and alkyl side groups more favorable. In n-methyl cyclohexyl methacrylates (21n14), the secondary y process was attributed to motions in the cyclohexyl side group because the magnitude of the relaxation depends on the cyclohexyl groups in the moiety. The broadening of the loss peaks was ascribed to two relaxations due to cis and trans cyclohexyl units in lateral chains [56]. Heijboer [57] suggested that the P relaxation time is determined by a local barrier within the molecule. This is a reasonable assumption in polymers. Nevertheless,

the molecular origin of the p relaxation is not completely understood, being observed in a variety of materials other than polymers, including glass-forming liquids made of simple molecules that do not have internal modes of motion [58621 including ionic liquids [63]. It seems then to be a near-universal feature of the amorphous state, as proposed by Goldstein and Johari [58], and thus called the Johari-Goldstein process. Reconciling Johari's island of mobility picture [64] with the concept introduced by Adam and Gibbs of cooperatively rearranging regions, CRR's [65], that establish below the alpha and beta crossover region (To,, see further on), the f3 process is assumed to be a localized motion in the inner, less dense and more mobile part of the CRR (a schematic picture for the densitylmobility situation in glasses at different temperatures relative to To, is given in [49,66]). Theoretical developments [67] within the so-called coupling model introduced by Ngai and co-workers [68,69] seem to point in the direction of a universal slow p relaxation closely connected to the glass transition process; in this context the term 'slow' is used to distinguish from 'fast' P process, predicted by the coupling model theory, ascribed to fast relaxations and observed in a variety of glass formers in the GHz region ([I21 and refs. cited therein). 3.1.2. Temperature dependence Sub-glass relaxations are thermally activated processes. Therefore, the temperature dependence of their relaxation times is Arrhenius-type,

z(T) = z, exp(E, I RT)

(24)

where zmisthe relaxation time at infinite temperature, that, for a Debye process, is of the order of l 0 - ' ~ - 1 0 -s, ' ~ and Ea is the activation energy representing the potential barrier resisting the molecular rearrangement [70] that can arise from either intramolecular (as two different orientations of a polar side-group relative to the main chain) or intermolecular forces. E, usually varies between 20 and 60 kJ mol", depending also on the environment of the group that undergoes the conformational changes. The above values of zmand Ea are associated with strictly local processes due to individual mobility of sub-units, and thus involve independent molecular motion with nearly no activation entropy, according to the Eyring [71] formalism. The y high-frequency secondary relaxation in lower poly(alky1 methacrylates) is an example of such relaxation process with an activation energy around 38 kJ mol-' [2,72]. For the larger members, even lower activation energies are found, as is the case for poly(n-lauryl dimethacrylate) with a value of 28 kJ mor' for the y

relaxation [73]. The related n-methyl cyclohexyl methacrylates (21n14) give activation energies between 42 and 50 kJ mol-' [57]. Similar values were found for the n-ethyleneglycol dimethacrylate monomers with 21n54, where the y process (whose loss curves were shown above in Figure 3.1 for the n=4 monomer) yields an activation energy of 42 kJ mol-' independently of the size of the ethyleneglycol moiety [47]. (The Arrhenius plot is presented in Figure 11). In EPON828, the activation energy of the y relaxation is 24 kJ morl [42] and 28 kJ mol-' [74] , respectively for average molecular weights of 190 and 380 g mol-'. The P process in EPON828 (Mw=380g mol-') has an activation energy of 47 kJ mol-' [74]. Values of z, lower than 10-l4(=: 10-'~-10-'~ s) imply an activation entropy greater than zero, being associated with more complex mobility mechanisms for which the activation energy could be of the order of 160 - 210 kJ mar', as is the case for the p relaxation in polysaccharides [75,76] and in polymers where the conformational mobility of the main chain is severely restricted [77]. For the lowest members of the poly(n-alkyl methacrylates) series, activation energies of between 79 and 96 kJ mol-I were found for the P process that, as already mentioned, is accompanied by restricted rearrangements of the main chain. For longer poly(n-alkyl methacrylates) the relaxation process is more localized with a lower activation energy value (e.g. 21 kJ mol-' for PnLMA [73]). 3.1.3. The shapes and intensities of the loss peaks The dielectric loss peak of secondary relaxations is extremely broad due to the variety of molecular environments (structural heterogeneity) of the relaxing unit, and, consequently, a variety of energy barriers, being more or less symmetrical. The dielectric strength, A&, which is proportional to the area under the loss peak, is much lower for the secondary processes, relative to the a relaxation analysed in the next section. This is a common pattern found in both polymer materials and glass formers. The P secondary process is even more depleted in linear polymers that contain the dipole moment rigidly attached to the main chain, such as polycarbonate [78-801 and poly(viny1 chloride) (the behaviour of this polymer was revisited in ref [81] where the secondary relaxation motions are considered as precursors of the a-relaxation motions). Polymers with flexible polar side-groups, like poly(n-alkyl met ha cry late)^, constitute a special class where the P relaxation is rather intense due to some coupling with main chain motions. The temperature dependence of the dielectric strength for secondary processes is usually linear, increasing with the temperature increase, but with a lower slope (absolute value) when compared with the main a relaxation process (that varies in the opposite manner decreasing with the temperature increase). Smith and Boyd

[82] proposed a model for polymers containing flexible side groups, whose motions are considered to be the origin of the sub-glass relaxation. It is assumed that below T, an isolated-chain model is adequate to calculate the relaxation strength, where the sub-glass motions are described by two states of unequal energy. The temperature dependent population of the states leads to the relaxation strength increasing with the temperature increase. However, the expected increase of the intensity of A s with temperature for secondary relaxations is not always observed as is evidenced by data relative to two epoxy resins: poly[(phenyl glycidyl ether)-co-formaldehyde] (PPGE) and DGEBA, where the dielectric strength of the P process decreases with temperature [74].The additionally detected y secondary process follows the usual pattern. In these systems the y relaxation is more intense than the j3 process, as was also found in nethyleneglycol dimethacrylate monomers with 2SnS4 [47]. For the poly(n-alkyl methacrylates) the A& signal for the y process is weaker than that of the P relaxation. The dielectric strength of both processes decreases with the increase in size of the alkyl group due to a dilution effect of dipole moment over a larger free volume.

Figure 9. Temperature dependence of dielectric strengths for DEGMA evidencing the change of slope for the secondary relaxations around T,. The j3 process is merged under the main alpha process and so is only detected in a limited temperature range (adapted from reference 1471).

A slope change near Tg is sometimes observed in the A& vs 11T plot. Data well above and well below the glass transition can be fitted by two different linear equations, evidencing that secondary relaxations are markedly sensitive to the glass transition [42]. Figure 9 shows a plot of dielectric strength as a function of the reciprocal of the temperature for di-ethyleneglycol dimethacrylate (DeEGDMA), with the change of slope being of the magnitude expected for the temperature dependence of the secondary relaxations. The dielectric strength for the y relaxation increases more rapidly above Tg as is also observed for the y process in EPON828 [42] and other systems [83-861 . Thus, the glass-transition signature can be seen in some systems through the temperature dependence of the dielectric strength. It is argued that local motions can couple with diffusive motions dominating at temperatures above T, [42].

3.2. a Relaxation

frequency /Hz Figure 10. Loss spectra of tetra-ethyleneglycol dimethacrylate (TeEGDMA) in the temperature range from -74 OC to -46 "C, every 2 OC. The high-frequency wing is due to the y secondary process (dielectric characterization in [45,47]). In the glass-transition region the viscosity and, consequently, the relaxation time increase drastically as the temperature decreases. Thus, the molecular dynamics are

characterized by a wide distribution of relaxation times and strong temperature dependence. Consequently, due to this abrupt increase of the relaxation time with the temperature decrease, the logarithmic plot of the relaxation time as a function of the reciprocal of the temperature shows a departure from linearity, presenting some curvature near T,. Figure 10 shows the loss spectra for the a process of TeEGDMA. The respective relaxation map is presented in Figure 11.

Figure 11. Relaxation map for the dipolar processes detected in TeEGDMA evidencing the curvature of the cooperative a process in contrast to linear temperature dependence of the relaxation times of P and y local processes described previously; solid lines - fitting by VFTH (a)and Arrhenius (P and y) laws (adapted from [47]). The relaxation-time dependence could be described by the Vogel-FulcherTammann-Hesse equation [88]:

T = T exp ~

-

(7?;o)

where A y F and D are constants, To is the Vogel temperature, interpreted as the glass-transition temperature of an ideal glass, i.e., a glass obtained with an infinitely slow cooling rate. Although the VFTH dependence of the relaxation time with temperature is, in general, in good agreement with the experimental data, the equation lacks clear physical meaning [89]. However, it has been shown [90] that the temperature dependence of the relaxation time expressed by the VFTH equation can be derived, under certain assumptions, from the Adam-Gibbs (AG) theory which relates transition probabilities for the primary structural relaxation to the configurational entropy and allows linking of the dynamic and thermodynamic behaviour of the glass formers [89]. The deviation from Arrhenius behaviour lies in the origin of the classification of glass formers [91,92] as strong (close to Arrhenius-like temperature dependence) and fragile (pronounced departure from linear behaviour), with the parameter D being a quantitative measure of fragility. The VFTH equation can be rewritten as

which is known as the Williams-Landel-Feny (WLF) equation [93], where To is an arbitrary reference temperature and C1 and C2 are constants of the material, dependent on the value chosen for To, and where .r(T1)and z(To) are the relaxation times at, respectively, Tl and To. This equation is supported by the free-volume theory [94], where it is assumed that the molecular mobility at a specific temperature depends on the free-volume fraction at that particular temperature. The WLF equation for To=Tg (dilatometric) was found to be applicable, for a wide variety of materials, in the temperature range between Tgand Tg+lOO OC, where Cpl and C p 2 have approximate values of 17.44 and 51.6, respectivly [2]. Experimental data obtained at a specific temperature can be superposed to experimental data obtained at a different temperature simply by a horizontal shift along the log t or log o axis - the time-temperature superposition principle. This procedure allows us to obtain so-called master or reduced curves for a specific material as shown in Figure 12 for TeEGDMA.

Figure 12. Normalized loss spectra for TeEGDMA in the temperature region of the a relaxation process, evidencing the validity of the time-temperature superposition principle. The changes in the relaxation time in the glass-transition region could be interpreted as a change in the length scale of the segmental motions that are increasing as the temperature decreases. This increase of the length scale with temperature decrease is associated with the cooperativity enhancement as described by Adam and Gibbs [66]. The Adam and Gibbs paper is considered as the turning point from rare free volume to small configurational entropy as the reason for slow molecular mobility in glass formers [95]. Their theory assumes that the molecular dynamics on approaching T, is controlled by cooperative rearrangements of particles in different regions, CRR (as mentioned before). The number of particles that rearrange cooperatively increases with the temperature decrease. The next section shows how the number of particles involved in cooperative rearrangements, N, designated as cooperativity can be estimated. The CRR size is ~ , = 5 , 3 with , 5, being the characteristic length of the glass transition. Therefore, in glass formers, the spatial aspect of the dynamic heterogeneity of the alpha relaxation may be described as a consequence of a mobility contrast between an island of high mobility [65] (where local mobility persists even below T, as described previously in characterizing the beta process: cage-rattling motions)

assisted by a cooperative shell of low mobility [49,67], that constitutes the CRR. The reorientations of molecules forming the cage, corresponds to the dynamic glass transition being a cooperative process because the fluctuations of the molecules forming the cage cannot be independent from each other [96]. In the time domain, the a relaxation process of amorphous polymers is successfully described by the Kolrausch-Williams-Watts (KWW) relaxation function (equation (15)) where the parameter PKWWthat takes in account the nonDebye character of the time decay leads to an asymmetric broadening of +(t) at short times, which typically vary between 0.2 and 0.5. The disadvantage of this equation is its transformation into the frequency domain through a Fourier transform that cannot be solved analytically (Koizumi and Kita [97] built a numerical table of complex permittivities for PKwwvaluesin the range from 0.3 to 1). Therefore, one of the most popular equations in the literature is the Havriliak-Negami (HN) equation (equation (13)) whose shape parameters can be related with PKWW[l 11 as already mentioned. A microscopic approach to the glass transition process was developed by Gotze [98] through the 'mode coupling theory' (MCT), which explains the sequence of relaxation events that occurs in a supercooled liquid in terms of a non-linear coupling between the density fluctuation modes; at the glass transition, the system undergoes an ergodic-nonergodic transition at a well-defined critical temperature, Tc. 3.3. Crossover region As mentioned before, the relaxation times of the a process show a strong temperature dependence, described by the VFTH equation (equation (25)), while the temperature dependence of the secondary processes follows an Arrhenius-type behaviour. Therefore, in a logarithmic plot of the relaxation time vs. the reciprocal of temperature (activation plot), the a process corresponds to a curved line, while the p process is a straight line (as previously shown in Figure 11). At temperatures well above T, the time scales of both relaxation processes tend to converge, because log(o,) increases faster than log(og) with the temperature increase. Consequently, the two lines in the activation plot come close together and eventually merge in a single process - the a process - or, with decreasing temperature, the a relaxation process bifurcates into two processes, a and P, and the frequency and temperature, Tp, region where the separation occurs is designated by 'ap splitting'. The crossover region is a more general designation due to the detection of multiple secondary processes, and thus the occurrence of more than one merging region, and is defined as the temperature-frequency region where

changes take place in the dynamic behaviour of the system [74]. The temperature interval where crossover effects are felt is about (1.2Tg - 1.4Tg), the different methods used to estimated it for a specific substance have typical uncertainities of + l o K, and the corresponding relaxation times are of the order of s, with uncertainty of about +1 decade, but large variances occur for different substances (see ref [99] where Beiner et al. compiled data for 38 different glasses). In polymers such as polystyrene, poly(viny1 acetate) and poly(viny1 chloride), the ap splitting occurs in a frequency region between 1o7and 10" Hz [loo]. An interesting class of polymers is that of poly(n-alkyl met ha cry late)^, which show time-scale superpositions associated with movements of both large and short length scale, for which the crossover region approaches T, [loll. In this type of polymer, a pronounced decrease of the glass-transition temperature is observed with increasing size of the alkyl group. This is designated as an internal plasticization effect, being attributed to an increase in the distance between adjacent polymer chains that increases the mobility of the main chain. Concomitantly, a deviation of the crossover region to lower temperatures is observed (the deviation of the crossover region to lower temperatures with the increasing length of side group was also observed by phosphorescence and fluorescence in PMMA, PEMA and PnBMA films containing luminescent probes [551). The study of the ap coupling is advantageous in poly(n-alkyl met ha cry late)^ due to the accessibility to the crossover region by both dielectric and mechanical spectroscopies and to the fact that both secondary and main processes are associated with high dielectric strength values, in contrast to a variety of other materials where the p relaxation is much less intense,. Among the poly(n-alkyl met ha cry late)^, the most studied ones are PMMA (poly(methyl methacrylate)) [40, 50-5 1,102- 1041, PEMA (poly(ethy1 methacrylate)) [40, 50-5 1,103-1081 and PnBMA (poly(n-butyl methacrylate)) [40, 5 1, 103-106, 1091, mainly by DRS and NMR (studies previous to 1965 have been reviewed in [2]). The development of the NMR technique and the increase of the frequency window in DRS brought new insights into the molecular dynamics of these materials (see [I101 for details and references therein). Some general features were found, namely that the P relaxation in the lower alkyl methacrylates appears to be an intramolecular process independent of the nature of the side group, and that the a process is strongly determined by the a relaxation. However, the crossover region between the a and p processes presents different profiles. The P dynamics, which follows Arrhenius-type temperature dependence below T,, seems to change by merging with the a process, followed by a further

change above the crossover region. This is clear in PEMA [40,105] (see Figure 13) but a continuous trace in the dynamical behaviour of the P process that remained separated from the a trace in PnBMA, was claimed by some authors [ I l l ] . Nevertheless, recent work [109] on poly(but-1 methacrylate) isomers, using both dielectric and mechanical spectroscopies, found no evidence of a separated onset of the a relaxation. Instead, the two a and P processes superimpose, being no longer resolved above the crossover, and near the merging region the P trace clearly shows a bending. The two behaviours reported here for PnBMA are examples of the scenarios usually found for the crossover: scenario I shows a separate onset (about 1 or 2 frequency decades) for the a process below the continuous trace of the a and p processes, and scenario I1 is where a continuous trace of the a and a processes occurs, making a large angle in the activation plot with the P process running into the main process at the crossover region [74].

Figure 13. Relaxation map for poly-ethylmethacrylate; the dynamics of the relaxation change by merging with the a process [112].

P

Whatever the scenario followed, it is accepted that two distinct processes occur above and below the crossover as shown clearly, e.g. for PHMA by dielectric, shear and calorimetric measurements that established that the cooperativity onset took place in the crossover region [49].

The qualitative idea that cooperativity starts at the crossover region and develops at lower temperatures [113-1161, was able to be quantified with the development of heat-capacity spectroscopy [I171 (HCS or 3 0 method) that tests entropy fluctuations within the sample, from which the cooperativity (number of particles per CRR as defined above) can be calculated using the formula [49,118]: N, =

RT;A(~/ c,,) RT~AC, !a M, ( 6 ~ ) ' M, (~T)'F; -2

where R is the gas constant, A(llc,)-A(llc,) -AcP/Cp the calorimetric intensity of the a process, and 6 p the average temperature fluctuation of CRRs as taken from the width, 6T, of the a peak in c",(o,T). Mo is the molar mass of the monomeric unit and T, is the dynamic glass temperature for a given frequency, 0 . Details for extracting the relevant parameters are given in [119]. Alternatively, N, can be calculated through the equation [95,120]:

where x is a reduced temperature between the onset, To, and Vogel temperature, To. A is a dimensionless constant of the order of 1 to 10 being material-specific (in a set of twelve different glass-formers [120], ten exhibit A values between 2 and 9). Cooperativity is large for PMMA with N,-40 at T,, becoming small and constant (-1) for poly(alkylmethacry1ates) with C>6 [49]. HCS data for polystyrene, polyisobutylene, and a random copolymer (SBR 1500), indicate, by the increase of N,, a cooperativity onset about 100 K above the Vogel temperature for these polymers. Far below the onset, large cooperativities are reached for different systems, with N, attaining values as high as 100 near T, (5=2-4 nm) [40]. For PPGE and DGEBA, N,(Tg) is of the order of 75 and 110, respectively, with cooperativity lengths, 5(Tg), of 3.3 nm (PPGE) and 3.8 ntn (DEGEBA) [74]. Above the crossover region, the a process is described by different VFTH parameters relative to the a relaxation, having higher Vogel temperature [74]. In the molecular picture discussed in [67,121], the material is nearly homogeneous and the mobility that originates the a process is conceived as localized escape motions of a particle from a cage formed by the nearest neighbours. The number of

particles rearranging cooperatively approaches unity and 5 I 1 nm. Such small cooperativity may be explained by an extraordinary concentration of free volume that allows temporarily the opening of a "cage door" of size of about 1 particle. The absence of significant cooperativity above the ap splitting has been observed for a variety of polymers through the AC, jump that tends to zero in that region [122] (the AC, jump is considered as an indicator of cooperativity in the glass transition). Dynamic neutron scattering studies showed by an unequivocal mode, for several polymers, that the dynamics of the a process are dominated by the Rouse modes or R modes of Rouse-Zimm[l23], which are inhibited bellow the crossover i.e. at higher length scales due to entanglements. Therefore, there are different dynamics below and above the ap splitting region. This change in molecular dynamics is enhanced through the analysis proposed for glass formers by Stickel, Fischer and Richert [34] of the derivatives d/dT, d/d(l/T) and d2/dp of log , a that are used to linearize the different VFTH dependence laws z(T) (reducing Arrhenius-type temperature dependence to a constant). The d/d(l/T) derivative takes the form:

By this method it became possible to solve subtle variations of z(T) that are less obvious in the usual representation of log a , vs 1/T, and it was concluded that the temperature dependence of the a relaxation time changes at a particular temperature, TB, which almost coincides with the temperature where the ap coupling occurs [34,124]. The change of molecular dynamics is also observable by the change of the PKWW parameter that increases strongly with decreasing temperature [124], due to the onset of significant cooperativity at temperatures below TB=Tp. This means that the alpha process below the crossover and the a process above the crossover are really independent and distinct processes. The main difference being this steep increase of cooperativity for the alpha process when compared to the weakly decreasing but small cooperativity of the a-process [125]. The Adam Gibbs equation only holds at temperatures moderately above T, but is violated at higher temperatures, above the crossover region, where the configurational entropy concept fails to describe the experimental relaxation times [126]. Corezi et al. [43] performed a direct test of the Adam Gibbs model for the structural relaxation time of PPGE and EPON828 and a deviation from the linear relation: log z vs (s,T)-' (S,

taken as =: sIiquid-pLas) was perceived at a temperature nearly coincident with Tp. Therefore, the a process that arises at temperatures above the ap coalescence is a distinct process and not a simple superposition of a and P relaxations, as already argued by Williams [107-1081, implying the use of different concepts and formalisms to understand, on a molecular level, the dynamics in the regions below and above the up splitting. 3.4. Low frequency processes d.c.-conductivity and charge carriers block effects that should be observable at low frequencies and high temperatures. This is particularly important in inhomogeneous materials such as biological systems [127], emulsions and colloids [128], porous media [129], composite polymers, blends, crystalline and liquid crystalline polymers. Electrets [130], which give rise to abnormally high values of dielectric permittivity and dielectric losses, often partially mask the true reorientational relaxation processes. Due to the measurement of the translational component (dc conductivity) that is observed simultaneously with the most frequently analyzed rotational dynamic component (orientational relaxation) [34,131], dielectric relaxation spectroscopy is still adequate to study conductivity effects such as: i) frequency-dependent charge-carrier motion within the sample, known as electrical conductivity relaxation [132-1331; ii) the trapping of charge carriers at interfaces within the bulk of the sample (interfacial Maxwell-WagnerSillars polarization [4,134]); and iii) the blocking of charge carriers at the interface between the ion-conducting material and the electron-conducting metallic electrode (electrode polarization [135]). d.c. conductivity is due to the relatively free movement of charge carriers as ions in extended trajectories between the electrodes [136]; e.g., in mixtures of sugars and drugs it was suggested that proton hopping through the hydrogen-bonded network (clusters) of water molecules was the mode of charge transport [137], while in aqueous DNA solution it was found that the molecular origin of conductivity lies in the migration of counterions along the DNA surface [138]. The measured conductivity can embrace contributions of both extrinsic migrating charges (e.g. ionic impurities) and intrinsic migrating charges (as proton transfer along hydrogen bonds) [ 161. When conductivity is of purely electronic origin, no contribution arises to E' because there is no storage of charge, while:

where oo is the dc conductivity of the sample, and €0 the vacuum permittivity, which increases linearly with decreasing frequency. The slope of a plot of log E" vs log f should be = -1. The conductivity term should be added to the HN equation to cany out the data fit in all frequency ranges [4]:

where a has dimensions of (HZ)" (radHz)$. The exponent s is used to take into account a low frequency tail that is influenced by either electrode or interfacial polarization (when the conductivity is not pure d.c. s
Figure 14. Real conductivity for fructose in the temperature range fiom 30 "C (open triangles) up to 70 "C (open squares) in steps of 2 "C,exhibiting the typical behaviour found for ionically conducting materials. The plateau region corresponds to the d.c. conductivity (m+O), while the region at higher frequencies after bending is due to a.c. conductivity. The characteristic relaxation time for conductivity can be estimated fiom the oo value as:

A variety of ionically conductive systems [130,138-1401 show, in the real part of the conductivity plot, a transition from a plateau, due to pure d.c. at low frequencies that bends off to frequency dependent conductivity (a.c. conductivity) at higher as proposed by Jonscher frequencies that follows a power law CY(CO)=CS~~+ACO" [141]. Figure 14 shows the conductivity behaviour for fiuctose between 30 and 70 "C. While at high frequencies (2101°~z)the charge carriers are driven by the external electric field over atomic length scales, in the low limit (o+O) that corresponds to d.c. current, charge carriers travel on some percolation paths from one side of the sample to the other, over macroscopic dimensions [139]. The temperature dependence of the d.c. conductivity relaxation-time has been found to follow either VFTH [34, 63, 130,1401 or Arrhenius [137, 142-1441 behaviour. The VFTH behaviour observed in some polymers, such as PET [130] and Nylon [140], at temperatures above T,, was an indication that charge carrier transport was governed by the motion of the polymer chain. Arrhenius dependence is expected when ion-hopping relies on the formation of lattice defects under the action of thermal excitation. Vacancies are created through which ionic motion may proceed under the influence of external electric fields, and the activation energy is the energy of formation of the defect [136]. The simplest model developed by Dyre [139,145] assumes hopping of charge carriers subject to randomly varying energy barriers where the relaxation time is related to the frequency of attempts to overcome the largest barrier determining d.c. conductivity. Similar models have been developed by Stevels [146] and Taylor [I471 -"random potential model" based on specific mean distances between high-potential barriers for ion migration, and Yamamoto-Namikawa [148] -"conducting path model" based on a non-homogeneous distribution of conducting pathways and nonconductive regions, were employed to interpret molecularly the hopping motion of conductive species in ion-conducting glasses and, more recently, in various cellulose-based materials and polysaccharides [144]. In these materials, proton migration leads to a relaxation process, the so called o process, with Arrheniustype temperature dependence [144,149].

Figure 15. 3D plot of loss specha detected in neutralized chitosan (structure indicated where x=0.7 and y=0.3) in the temperature range from -120 to 150°C. The o process is noticeable in the high temperature-low frequency range. The inset shows the Arrhenius-type dependence of relaxation times for the o process (E,= 94+2 kJ mor') (complete dielectric characterization in [149]). Figure 15 shows the 3D plot of E" relative to neutralized chitosan, evidencing the o process in the high temperature-low frequency range whose Arrhenius-type temperature dependence of relaxation times is shown in the inset. The secondary process observed in the low-temperature region is due to local main chain motion via the glycosidic bond, influenced by the amino side group (see [I491 for discussion). An advantageous analysis of electrical conductivity relaxation is undertaken in the complex modulus formalism [I501 adapted for analysis of dielectric processes occurring in materials with large concentrations of mobile charge carriers, such as inorganic salts [I511 and composite polymeric systems [152], that suppress the influence of electrode polarization, allowing the determination of d.c. conductivity.

In analogy to mechanical relaxation, the complex electric modulus M* is defined as the inverse of the complex permittivity, originally given by McCrum et al. [2] according to the equation:

If &*(a) is described by the Debye function, the complex modulus is written as [4,20]:

with M=Mo-M, (
1

2

3

4

5

6

log,,(frequency l Hz) Figure 16. Real and imaginary parts of the complex permittivity of fructose at 90 "C. The region where the slope in E" (on a logarithmic scale) is close to unity is due to d.c. conductivity, as confirmed by the invariance in E' and the Debye peak in M" (the slight asymmetry on the high frequency side is due to the relaxation process that is starting at the edge of the frequency window). At the lowest frequencies, electrode polarization is influencing E', lowering the E" slope while leaving M" unaffected. The modulus formalism is being applied more frequently to fit data of systems largely influenced by conductivity, such as composites [154], semi-crystalline polymers [143], ionic liquids 1631 and biological systems [138]. This allows a better resolution of relaxation processes and leads to similar shape parameters and temperature dependence of relaxation times to those achieved by using complex permittivity [143].

-\\

electrode polarization

log(frequencyl Hz) Figure 17. Evidence of the Maxwell-Wagner-Sillars effect in the real permittivity of the composite system: nematic E7 dispersed over hydroxypropylcellulose-type matrix. The interfacial polarization can be described by a double-layer arrangement. At lower frequencies and higher temperatures, the real permittivity increases further due to electrode polarization. Besides the described d.c. conductivity effect, charge carriers can accumulate in the sample at interfaces occurring between two media having different permittivities andlor conductivities. This storage of charge in the material gives an additional contribution to polarization and thus strongly increases E' with decreasing frequency. This allows phenomena that do not involve charge storage to be distinguished from measurements of direct current conduction. The frequency dependence of this interfacial polarization is often similar to dipolar relaxation, presenting a clearly defined peak (see results for 8CB liquid crystal confined in nanoporous matrix [155], where the MWS effect results from the rigid matrixtliquid crystal interface) or being partially masked by electrode polarization which is observed at even lower frequencies. Figure 17 shows both MWS and electrode polarization effects for an electro-optical system where the liquid crystal, E7 rests on a polymeric matrix (see [156] for E7 lying on a hydroxypropylcellulose matrix). In this latter system, the non-homogeneous structure can be described by

the simplest model, i.e, a double-layer arrangement, where each layer is characterized by its permittivity ~i and its relative conductivity oi:

E7 Polymer matrix

The complex dielectric function is described by an equation similar to the Debye equation but the parameters have different meanings. When the thicknesses of both layers are equal: E,=E~E~/(E~+E~)and BEis given by [4]:

and

The characteristic frequency of MWS reflects the time scale of rearrangements of the charges, which have piled up at the interface. Theories for interfacial polarization in two-, three-, and multi-phase systems are summarized in [157]; several examples in heterogeneous systems are presented in [158].

3.5. Dielectric response in semi-crystalline polymers Many polymers may partially crystallise, and the presence of a crystalline component will greatly influence the physical properties of the material. Boyd has analyzed the different aspects related to the presence of crystallinity in the relaxational behaviour of different polymers [159]. Two major effects may be generically distinguished: the first effect is detected in polymers with low- or medium- degrees of crystallinity (see Section 3.5.1) and the second effect is typically observed in highly-crystalline systems (Section 3.5.2).

3.5.1. Glass transition dynamics in semi-crystallinepolymers The amorphous fraction of the polymer located between the crystalline lamellae, which are formed during crystallisation, will exhibit different behaviour to that which is found in the bulk. This is due to the fact that the motions will be much more restricted, taking place in geometries with thicknesses of just several nanometers. This problem can be included among the general problem dealing with the behaviour of materials which are geometrically confined on a nanometer spatial scale, such as 3D confinement (as in nanoporous glasses or zeolites), 2D confinement (in silicate layers of nanocomposites or in layered block copolymers) or thin polymer films [4, 160, 1611. An important effect of this confinement is a shift in the glass-transition temperature, which can easily be detected from dielectric measurements. This shift, with respect to the bulk material, may be positive or negative, although basically positive shifts are observed for segmental dynamics in the amorphous phase confined between crystalline lamellae. In this particular case, a broadening of the distribution of relaxation times assigned to the a-relaxation may also be observed, as well as a decrease in the dielectric strength with increasing crystallinity. Many examples can be found of the effect of crystallinity in different polymeric systems, with poly(ethy1ene terephthalate) being the most studied one [2,4,159]. In this section, an example will be given of the use of dielectric measurements to monitor the glass-transition dynamics of a polyester during crystallisation. Poly(Llactic acid), PLLA, is a well studied biodegradable and biocompatible linear aliphatic polyester. It is used in a variety of different fields, such as biomedical applications, including in wound closures, prosthetic implants, controlled delivery systems and three-dimensional porous scaffolds for tissue engineering, as well as in environmental applications [162, 1631. PLLA crystallises slowly and the degree of crystallinity will influence many properties, including the degradation profile, the biological response and its mechanical performance. The crystallisation process in polymers has been traditionally performed by monitoring the development of the crystalline structure at different length scales, using techniques such as atomic force microscopy, transmission electron microscopy, small and wide-angle X-ray scattering, small-angle light scattering or differential scanning calorimetry (DSC). Such data can be complemented by looking at the changes occurring within the amorphous fraction upon crystalline development. Such information may be instructive because the amorphous regions will play an important role in the properties of the material, in particular in systems that do not crystallise to very great extents, such as in PLLA. In this context, dielectric relaxation spectroscopy

may be useful to monitor the changes that occur in the relaxational behaviour, especially in the region of the a-relaxation, during the crystallisation of the polymer. As an example, the time-resolved, dielectric-loss spectra at 80 O C of a PLLA sample, obtained at different crystallisation times, t,, are shown in Figure 18. The initial (tc=O) and the fmal scans (t, =6 h) can be assigned to the glass-transition dynamics of the nearly amorphous and the fully semi-crystalline PLLA, respectively. These two situations will be considered as corresponding to the two extremes for the segmental dynamics that occurs in the amorphous chains. In the first situation, all the conforrnational motions occur in the bulk-like amorphous phase, whereas, at the other extreme, all the amorphous phase exists within the spherulites, constrained between the crystalline lamellae or lamellar stacks. The loss peaks of both extreme cases were fitted with a sum of two Havriliak-Negami functions (Equation (13)): a low frequency function that corresponds to the arelaxation and, a high frequency function, merged under the a process corresponding to a P process that was detected at lower temperatures (between -120 and 25 OC) . As expected, for tc =6 h, the a (constrained) relaxation was characterized by a broader distribution of relaxation times and the position of the loss peak was shifted to lower frequencies, with respect to the initial amorphous material (bulk-like a-relaxation). More details and discussion of these results may be found in [1641.

f /Hz f /Hz Figure 18. Lefk Dielectric loss for a PLLA sample crystallised at 80 OC, obtained at different crystallisation times (see more details in [164], from where the data were extracted). Right: detail of an experimental result obtained at tc = 120 min (circles) and the overall curve fitting, corresponding to the sum of three individual Havriliak-Negami curves (see text).

For intermediate values of t,, a progressive evolution of the loss peaks between the two extremes was found (Figure 18). It was assumed that each loss peak could be obtained by simple linear combinations of the three processes: the a-constrained, the a-bulk-like and the P relaxation [164]. The characteristics of the individual processes were obtained from the extreme curves and only their relative intensities were fitted for each loss curve, with the shape parameters and the average relaxation times being fixed. An example is given in Figure 18, where the contributions of the three processes are displayed for the frequency scan obtained at t, = 120 min. As also found for other crystallisation times, the fitted curve agrees very well with the experimental results.

0

5000

10000

15000

20000

tls

Figure 19. Evolution of the relative dielectric strengths of the three processes involved in the overall loss peaks during crystallisation of PLLA (data adapted from [165]). The scheme is intended to represent the evolution of the spherulitic fraction (darker regions) during different crystallisation times. The model employed to deal with the data presupposes that, at any crystallisation stage, the global segmental mobility results f?om the contribution of two independent processes: one resulting from the bulk-like conformational motions,

and the other to the glass-transition dynamics of the amorphous phase confined between the crystalline lamellae, which are located within the sphemlitic structures (see top scheme in Figure 19, that represents the sphemlitic evolution during crystallisation). Figure 19 shows the evolution of the dielectric strength of the three relaxation processes during crystallisation. As expected, the dielectric strength of the aprocess of the non-restricted amorphous phase decreases as the polymeric chains involved gradually crystallise or become confined between the crystals, giving rise to an increase of the AE of the constrained a-process. This evolution was found to be highly correlated with the evolution of the spherulitic structure of the material [164]. This is shown schematically in Figure 19. This finding suggests that the constrained a-process is a result of the segmental mobility of the mobile amorphous fraction existing within the spherulites and, more specifically, between the crystalline lamellae. The general behaviour displayed in Figures 18 and 19 was also observed in PLLA samples with distinct molecular weights and crystallised from both the melt and glassy states 11651. 3.5.2. a, relaxation Polymers with a flexible chain, such as polyethylene (PE), polypropylene, poly(tetrafluorethylene), or poly(methy1ene oxide), exhibit relaxation processes directly related to the presence of their crystalline fraction. For PE, by far the most important system in this context, such processes may be dielectrically active, provided that the sample is "decorated" with a few C-C1 or C=O dipoles, by chlorination or oxidation [166]. The a, relaxation appears between T, and the melting temperature. For PE it is complex and is comprised of at least two relaxation mechanisms, named a1 and a11 (or a and a'), which cover the temperature range between 30 and 120 "C (see discussion and more references in [167]). There are many studies that have tried to correlate the a, relaxation and its origin at the molecular level. General information in this context may be found in many books or review papers (e.g. [2,159]), and here only a brief reference to the subject will be made. Dielectric relaxation spectroscopy and NMR results demonstrate that the a, process is associated with rotational motions within the crystalline lamellae. A 180" flip in a molecular segment occurs at a particular place and propagates in screw-like motions throughout the crystal. This results in the effective translation of a carbon atom of the chain. This mechanism involves displacements within the chain along states that are energetically equivalent and, thus, it could not originate a mechanical

response. However, there is a mechanical effect of the a, relaxation, which results from an additional shear of the amorphous regions. For this, there is the need for chain transport through the crystallites, which occurs by a long-distance solid-state diffusion process, with activation energy, for the case of linear PE, of 105 kJ morl [168]. Such a value is consistent with the activation energies for the a, relaxation, which may vary between 90 and 300 kJ mol-'. Therefore, in the a, process, a relaxation mode within the crystals and another in the amorphous regions are combined. An unsolved issue of the a, relaxation is to know if it is really a thermally activated process. The Arrhenius plots are found to be non-linear [167, 1691, but this can be attributed to the overlapping of different thermally activated processes. Arguments based on the concept of activation entropy led to the suggestion that the a, relaxation is substantially cooperative [169]. This was attributed to the participation of the amorphous region in this process, which will increase the complexity of the motions involved.

Figure 20. Left: real permittivity (full symbols) and loss factor (open symbols) obtained in a P-PVDF film at different frequencies (see labels inside the graphics) (data adapted from [170]). Right : Dynamic mechanical spectra of a P-PVDF film obtained at a frequency of 1 Hz, in tensile mode, performed along the longitudinal (solid lines) and transverse (dashed lines) directions, with respect to the stretch direction used to process the film (data adapted from [171]). The relaxational behaviour of polyvinylidene fluoride (PVDF) is of interest. Its P phase exhibits piezo- and pyroelectric properties and so it is used in various devices, such as sensors, transducers or actuators for different applications. Typically such materials are prepared through a technological process involving stretching and

pulling of extruded thin films, Dielectric relaxation spectroscopy on P-PVDF reveals a low temperature p-relaxation, assigned to the glass-transition dynamics, and the presence of an a, relaxation would be expected at higher temperature. However, instead of that, in this temperature range, a decrease of the dielectric constant is observed (see Figure 20). In fact, in this temperature range, a recovery in the film's geometry is found along the longitudinal direction of the stretch, used to orient the film during its processing. The occurrence of this recovery suggests that it could be linked to this relaxation, because the a, relaxation enables the occurrence of new molecular motions at the crystalline level. The dielectric experiments did not reveal this process. The same behaviour was reported in another study, where the a, relaxation could only be observed in non-stretched PVDF films [172]. However, this relaxation seems to be detectable by dynamic mechanical analysis, but only visible when the deformation is longitudinal to the stretch direction (Figure 20). 4. THERMALLY STIMULATED DEPOLARIZATION CURRENTS

The thermally stimulated depolarisation currents technique (TSDC) belongs to the class of thermally stimulated techniques, in which some property is activated during a controllable heating step, allowing the temperature-dependent characteristics of the materials, namely their phase transitions and molecular mobility, to be studied. In TSDC, the property monitored is the electric current that is generated from the depolarization process in a sample, previously frozen-in after an electric field polarizing process [173]. Therefore the equipment should include a sample holder, comprising two electrodes, between which the sample is placed, and a temperature-controlled chamber (see scheme in Figure 21). A static electric field is applied to the sample, in order to polarize it, and a d.c. micro-ammeter is used to measure the depolarization current as a function of temperature. It has been argued that TSDC has a high resolution power, which is a consequence of its low equivalent frequency. In fact, van Turnhout showed that a TSDC experiment is equivalent to a low-frequency dielectric loss experiment and that the equivalent frequency is given by [174]:

where E, is the activation energy, P the heating rate, R the ideal gas constant and T,,, the temperature at which the TSDC depolarization peak has its maximum. This low equivalent frequency of the TSDC technique leads to an enhancement of the resolution of the different relaxation processes or, otherwise stated, the separation of the TSDC peaks along the temperature axis is increased when compared with higher frequency techniques such as Dielectric Relaxation Spectroscopy. A detailed discussion on the equivalent frequencies associated with TSDC measurements may be found in [175].

Switch

Furnace DC MicroAmmeter

-

Figure 21. Schematic diagram of an apparatus for thermally stimulated depolarization current (TSDC) experiments. In a TSDC global experiment, the sample is polarised by a static electric field at a temperature T,, well above the temperature range of the occurrence of the process that is being studied, for a given time interval t,. The sample is cooled in the presence of the field to a temperature To<< T,, at a cooling rate PI, and, after an isothermal period, with the field off, the sample is heated at a constant rate, P3, to a temperature Tf > T,. During the heating, the electric current released from the depolarisation process, arising from dipoles, trapped electrons or mobile ions, is measured and recorded as a function of temperature. The result of a global TSDC experiment is a global spectrum that contains all the diele~.~ically active relaxations excited by the field between Toand T,. In Figure 22 (left) an example is shown where two processes can be detected in the temperature dependente polarization plot, P(T) plot, identified by sudden decreases in the polarization, that would give rise to peaks in the polarization current (see later). Processes detected in global

experiments are typically very complex (broad an asymmetric), and are the result of relaxations characterized by a broad distribution of relaxation times.

Figure 22. Experimental schemes presenting the thermo-electric procedure for global (left) and thermal sampling (right) experiments. The time interval during which the constant voltage is applied is indicated by the horizontal line. The insets shows the temperature dependence of the polarization of the sample during the final heating step. The thermal sampling technique, TS, allows narrow segments of the complex thennocurrent to be polarised and, thus, it enables complex global peaks to be resolved into their 'individual' components. The thermo-electric treatment of the sample during a TS experiment is schematically shown in Figure 22 (right). The electric field, Eo, is applied at Tp during tp and the sample is cooled (at a rate PI), in the presence of the field, to a temperature Td, where T, - Td = T, is typically 2-3 "C. With the field off, and after an isothermal period (during t,) the sample is cooled further (at a rate Pz) to To << Tp. At that temperature, the remaining polarisation corresponds to the dipolar activation in the temperature region between Tp - T, and T,. The thermal release of that polarisation is recorded during heating, at a constant rate P3 = dT/dt, to a frnal temperature TpT,. Each peak obtained from a TS experiment may be considered to arise from the dipolar relaxation of a nearly elementary-like process characterized by a single relaxation time, with some temperature dependence, ~ ( 2 " ) . From a practical point of view, a series of TS experiments are performed at different values of T,, covering the temperature region of interest. The ensemble of TS curves obtained can be analyzed independently and provides more insights about the general features of the complex peaks recorded during global experiments.

Each TS curve can be assumed, as a first approximation, to arise from the depolarization of a simple process that may be modelled using the Debye equation (equation (4)). The electric current released by the depolarization of the system during the heating step (with no field on) is:

However, the majority of real systems cannot be quantified according to an unique elementary process. Generally, TSDC results are interpreted formally in terms of a distribution of parallel (uncoupled) activated Debye-like processes. Therefore, considering, as discussed in the first section, a distribution of activation times with a normalized probability function p(ln r), the polarization is given by:

where P(r,t) corresponds to the polarization of the elementary process with a relaxation time r. The corresponding electric discharge current released by the sample, J(t), can also be obtained by this weighted summation over all the elementary discharge contributions, J(r,t):

Assuming different forms of p(ln z), it is possible to simulate and predict different kind of situations that can be found in global or TS experiments. Several examples may be found elsewhere [176]. An example with real experimental results is shown in Figure 23, obtained for a side-chain liquid-crystalline polymer (see more details in [177, 1781). In the global experiment, three peaks were clearly observed. Then, TS experiments were performed at different polarization temperatures, T,, that enabled different curves to be extracted (Figure 23). The ensemble of such curves exhibits a contour consistent with the global spectra. Assuming Debye behaviour for each TS curve, the temperature dependence of the relaxation times may be obtained from equation (39):

0

-140 -120 -100 -80 -60 -40 -20

0

20

40

T",, I "C

Figure 23. Left: Global and TS experiments obtained for a side-chain liquidcrystalline polysiloxane (data adapted from [177, 1781, where more details are given), where at least three distinct relaxation processes can be found. Right: Activation energy as a function of the temperature of the maximum of TS peaks obtained for the same polymer at different polarization temperatures, using T,= 2 "C. The activation energy was calculated from the activation enthalpy, as reported in [177]. Equation (42) may be then used to obtain z(T) from the experimental data, J(T). Different models may be used to extract information from this data. Assuming a simple thermally activated process, the Arrhenius equation may be used (equation (24)). The experimental TS results from Figure 23 were treated using this model, by linear regression of ln(z) versus 1/T, z being calculated from equation (42) along the rising section of the TS peaks. The activation energies of the different peaks are also represented as a function of T, the temperatures of the peaks' maxima. The high values of E, observed for the lower temperature process are typical of the glass-transition dynamics that also occur around the calorimetric glass-transition temperature of the polymer. This process can then be attributed to the typical arelaxation found by dielectric relaxation spectroscopy. This a-relaxation could also be observed by dynamic mechanical analysis, DMA (tan 6 peak at -5 "C, for a frequency of 1 Hz), which is consistent with the TSDC data [178].

The TSDC peak at -8' C should be attributed to specific motions that take place in this peculiar system, involving conformational changes in the mesogenic groups laterally attached to the polymer backbone through flexible spacers (see scheme in Figure 24). In many studies, two major relaxations have been detected in liquidcrystalline polymers by dielectric relaxation spectroscopy. Besides the a relaxation, a 6 process can be found at higher temperatures (lower frequencies) and attributed to mobility modes of the longitudinal component of the dipolar moment of the mesogenic group (see scheme in Figure 24). It is not yet clear if such motions are linked to the dynamics of the main chain. In fact, the dynamics tend to adopt a VFTH behaviour when the temperature approaches T,,and sudden changes in the mechanism occur in the transition between the liquid-crystalline phase and the isotropic liquid.

log #Hz= -2

-40

-20

0

20

-1

0

40

T I "C Figure 24. Comparison between TSDC (solid line) and dielectric relaxation spectroscopy (dotted lines) data for a side-chain liquid-crystalline polysyloxane. For the dielectric relaxation spectrum, the loss factor is plotted for different frequencies. The graphics were adapted from [179]. The left scheme shows the structure of a side-chain LCP, consisting of the main chain, the spacers and the mesogenic groups (elliptic shapes). The molecular origins of the a and 6 relaxations are also represented. The 6 relaxation, although being associated with special motions in the dipolar groups, is also mechanically active and characterized by a narrow distribution of relaxation times [178]. This process is not observable by DSC, because the changes

in entropy are not sufficient to yield a significant change of heat capacity [178]. Despite that, the a-relaxation exhibits all the characteristic of a dynamic glass transition. It was suggested that it could be somehow linked to the occurrence of the fastest motions in the mesogenic groups, including fluctuations of the transverse component of the dipolar moment [178]. Figure 24 compares the global TSDC spectrum of the analyzed polymer with the imaginary permittivity read at different frequencies. The results of dielectric relaxation spectroscopy clearly reveal the existence of two peaks, that tend to converge, for lower frequencies, to the position of the -10 "C and 8 "C peaks detected by TSDC. Such comparison also allows the two peaks found by dielectric relaxation to be attributed to the a and 6 relaxations [179]. The TSDC peak at -30 "C would correspond to a non-dipolar process (motions of spatial charges), being consistent with the conductivity tail found by dielectric relaxation spectroscopy [ 1791.

5. CONCLUSIONS Electrical methods, among which dielectric relaxation spectroscopy (DRS) and thermally stimulated depolarisation current (TSDS) techniques play a major role, have been briefly described, emphasizing their relevance as tools to explore molecular mobility. DRS, where re-orientational motions of permanent dipoles act as a probe of molecular dynamics, allows a frequency range from to 10" Hz (or even 2 decades further on in both low and upper limits) to be covered, enabling molecular relaxational processes (from slow to fast) top be studied. The dynamic behaviour of glass formers, both low molecular weight materials and polymers, can be described below and above the glass transition. In the glassy state, localized motions give rise to local fluctuations of the dipole vector that are the origin of the secondary relaxation processes detected by dielectric relaxation spectroscopy. These processes exhibit an Arrhenian temperature dependence of the respective relaxation times. Hovever, above but near the glass transition, cooperative motions result in a relaxation process (the a-relaxation) with a non-Arrhenian temperature dependence. Both type of processes merge at temperatures well above the glass transition, known as the crossover region. The changes in the relaxation time reflect a change in the length scale of the molecular motions that are the origin of the relaxation process. Thus, by assuming that the molecular dynamics on approaching the glass transition are controlled by cooperative motions, the number of particles that rearrange cooperatively increases as the temperature decreases. Above the

crossover region, no significant cooperativity exists, i.e., the number of particles rearranging in a cooperative motion tends to approach unity. DRS is also valuable for studying the translational motion of charge carriers, i.e., conductivity effects such as: frequency-dependent charge-carrier motion within the sample, the trapping of charge carriers at interfaces within the bulk of the sample (interfacial Maxwell-Wagner-Sillars polarization) and the blocking of charge carriers at the interface between the ion-conducting material and the electronconducting metallic electrode (electrode polarization). These effects are particularly important in inhomogeneous materials such as biological systems, emulsions and colloids, porous media, composite polymers, blends, crystalline and liquid crystalline polymers and electrets. Temperature is a major parameter in electric/dielectric measurements. Different procedures have been suggested that permit the measured temperature in such experiments to be corrected. Some less-conventional dielectric tests have been presented, including the possibility of following crystallization by monitoring the changes in the loss peak. There experiments also allow more information to be obtained about the segmental mobility in semi-crystalline polymers, especially the restricted motions within the spherulitic structures. Complementary experiments may be also obtained by TSDC. This technique provides a simple way to probe the mobility of dipoles and electric charges over a large temperature range. The use of thermal sampling methodologies enables a better insight on the complexity in relaxation characterised by broad time distributions to be obtained 6. REFERENCES 1. H. Frohlich, Theory of Dielectrics, Oxford University Press, 1949. 2. N.G. McCrum, B.E. Read and G. Williams, Anelastic and dielectric effects in polymeric solids, Wiley, New York, 1967 (reprinted by Dover, New York, 1991). 3. C.J.F. Bottcher, Theory of electric polarization, Elsevier ,Amsterdam, 1973. 4. A. Schonhals and F. Kremer (Eds) Broadband dielectric spectroscopy, Springer-Verlag, Berlin, 2003. 5. G. Williams, Chem. Soc. Rev., 7 (1978) 89. 6. K.S. Cole and R.H. Cole, J. Chem. Phys., 9 (1941) 341. 7. D.W. Davidson and R.H. Cole, J. Chem. Phys., 18 (1950)1417.

8. S. Havriliak and S. Negami, J. Polym. Sci., C14 (1966) 99; recently revised in S. Havriliak Jr. and S. Havriliak, Dielectric and Mechanical Relaxation in Materials, Hanser Publ., Munich, 1997. 9. G. Williams and D.C. Watts, Trans. Faraday Soc., 66 (1970) 80. 10. R. Kohlrausch, Ann. Phys., 12 (1847) 393. 11. F. Alvarez, A. Alegria and J. Colmenero, Phys. Rev. B, 44 (1991) 7306; F. Alvarez, A. Alegria and J. Colmenero, Phys. Rev. B, 47 (1993) 125; A. Alegria, E. Guerrica-Echevarria,L. Goitiandia, I. Telleria and J. Colmenero, Macromolecules, 28 (1995) 1516. 12. U. Schneider, P. Lunkenheimer, R. Brand and A. Loidl, Phys. Rev. E, 59 (1999) 6924. 13. G. Williams in Keynote Lectures in Selected Topics of Polymer Science, (Ed. E. Riande), CSIC, Madrid, 1995, Ch. 1. 14. F. Kremer and M. Arndt in Dielectric Spectroscopy of Polymeric Materials, (Eds J. P. Runt and J. J. Fitzgerald), Amer. Chem. Soc., Washington DC, 1997, Ch. 2. 15. A. Schonhals and F. Kremer, (Eds) Broadband Dielectric Spectroscopy, Springer-Verlag, Berlin 2003, Ch. 2. 16. J. Mijovic and B. D. Fitz, Dielectric Spectroscopy of Reactive Polymers, Application Note Dielectrics 2, Novocontrol GmbH, 1998. 17. F. Kremer, D. Boese, G. Maier and E. W. Fischer, Prog. Colloid. Polym. Sci., 80 (1989) 129. 18. A. Von Hippel, Dielectric Materials and Applications, Wiley, New York, 1954, cited in [2]. 19. R. Nozaki and T. K. Bose, IEEE Trans. Instr. Measurements, 39 (1999) 945. 20. G. Williams and D. K. Thomas, Phenomenological and Molecular Theories of Dielectric and Electrical Relaxation of Materials, Application Note Dielectrics 3, Novocontrol GmbH, 1998. 21. R. Nozaki and S. Mashimo, J. Chem. Phys., 87 (1987) 2271 and S. Yagihara, R. Nozaki, S. Takeishi and S. Mashimo, J. Chem. Phys., 79 (1983) 2419. 22. B. V. Hamon, Proc. Inst. Electr. Eng., 99 (1952) 27. 23. A. Brather, Colloid. Polym. Sci., 257 (1979) 787. 24. J. L. Gomez Ribelles and R. Dias Calleja, J. Polym. Sci: Polym. Phys. Ed., 23 (1985) 1505. 25. M. Dionisio, Ph.D. Thesis, Instituto Superior TCcnico, Lisbon, Portugal, 1993. 26. R.G. Heydon, IEEE Trans. Electrical Insulation, 24 (1 989) 649. 27. E. M. Breitung, W. E. Vaughan and R. J. McMahon, Rev. Sci. Instrum., 71 (2000) 224.

28. M. Dionisio, H. P. Diogo, J. P. S. Farinha and J. J. Moura-Ramos, J. Chem. Edu., 32 (2005) 1335. 29. B. Wunderlich, Thermal Analysis of Polymeric Materials, Springer-Verlag, Heidelberg, 2005. 30. J.F. Mano, S. Lanceros-MCndez, A.M. Nunes and M. Dionisio, J. Therm. Anal. Calorim., 65 (2001) 37. 3 1. E. Donth, Relaxation and Thermodynamics in Polymers, Akademie, Berlin, 1992, cited in [34] 32. S. R. Elliot, Physics of Amorphous Solids, Longman, London, 1983, cited in [34] 33. S. Brawer, Relaxation in Viscous Liquids and Glasses, Am. Ceram. Soc., Columbus, 1985. 34. F. Stickel, E. W. Fischer and R. Richert, J. Chem. Phys., 104 (1996), 2043. 35. F. Stickel, E. W. Fischer and R. Richert, J. Chem. Phys., 102 (1995), 6251. 36. J. Jackle, Rep. Prog. Phys., 49 (1986) 171, cited in [35]. 37. R. Richert and A. Blumen (Eds), Disorder Effects on Relaxational Processes, Springer, Berlin, 1994. 38. K.L. Ngai, IEEE Trans. Dielectrics Electrical Insulation, 8 (2001) 329; J. Non-Cryst.Solids, 275 (2000) 7. 39. I-S. Park, K. Saruta and S. Kojima, J. Phys. Soc. Japan,, 67 (1998) 4131. 40. F. Garwe, A. Schonhals, H. Lockwenz, M. Beiner, K. Schrijter and E. Donth, Macromolecules, 29 (1996) 247. 41. G. P. Johari and M. Goldstein, J. Chem. Phys., 55 (1971) 4245. 42. R. Casalini, D. Fioretto, A. Livi, M. Lucchesi and P. A. Rolla, Phys. Rev. B, 56 (1997) 3016. 43. S. Corezzi, D. Fioretto, S. C. Santucci, S. Capaccioli, R. Casalini, M. Lucchesi, E. Hempel and M. Beiner, Philos. Mag. B, 82 (2002) 339. 44. A. Alegria, E. Guerrica-Echevarria, L. Goitiandia, I. Telleria and J. Colmenero, Macromolecules, 28 (1995) 15 16. 45. M. T. Viciosa and M. Dionisio, J. Non.-Cryst. Solids, 341 (2004) 60. 46. M. T. Viciosa, N. RouzC, M. Dionisio and J. L. Gdmez Ribelles, Eur. Polyrn. Sci., in press. 47. M. T. Viciosa, J. L. Gdmez Ribelles and M. Dionisio, to be submitted. 48. S Corezzi, S Capaccioli, G Gallone, A Livi and P A Rolla, J. Phys.: Condens. Matter, 9 (1997) 6 199. 49. M. Beiner, Macromol. Rapid Comrnun., 22 (2001) 869. 50. S. C. Kuebler, D. J. Schaefer, C. Boeffel, U. Pawelzik and H. W. Spiess, Macromolecules, 30 (1997) 6597.

51. A.S. Kulik, H.W. Beckham, K. Schmidt-Rohr, D. Radloff, U. Pawelzik, C. Boeffel and W. Spiess, Macromolecules, 27 (1994) 4746. 52. E. R. deAzevedo, W. G. Hu, T. J. Bonagamba and K. Schmidt-Rohr, J. Am. Chem. Soc., 121 (1999) 8411. 53. A. C. Sommersall, E. Dan and J. E. Guillet, Macromolecules, 7 (1974) 233. 54. H. Itagaki, K. Horie and I. Mita, Prog. Polym. Sci., 15 (1990) 361. 55. M. Christoff and T. D. Z. Atvars, Macromolecules, 32 (1999) 6093. 56. G. Dominguez-Espinosa, M. J. Snachis, R. Diaz-Calleja, C. Pagueguy, L. Gargallo and D. RadiC, Polymer, 46 (2005) 8028. 57. J. Heijboer, Ann. NY Acad. Sci., 279 (1976) 104. 58. G. P. Johari and M. Goldstein, J. Chem. Phys., 53 (1970) 2372. 59. G. P. Johari and M. Goldstein, J. Chem. Phys., 55 (1972) 4245. 60. G. P. Johari, J. Chem. Phys., 58 (1973) 1766. 61. G. P. Johari and K. Pathmanathan, J. Chem. Phys., 85 (1986) 681 1. 62. S. H. Chung, G. P. Johari and K. Pathmanathan, J. Polyrn. Sci. B , 24 (1986) 2655. 63. A. Rivera, and E.A. Rossler, Phys. Rev. B, 73 (2006) 212201. 64. G. P. Johari and E. Whalley, Faraday Symp. Chem. Soc., 6 (1972) 42. 65. G. Adams, J. H. Gibbs, J. Chem. Phys., 43 (1965) 139. 66. S. Kahle, E. Hempel, M. Beiner, R. Unger, K. Schroter and E. Donth, J. Molec. Struct., 479 (1999), 149. 67. K. L. Ngai, Phys. Rev. E ,57 (1998) 7346. 68. K.L.Ngai and S. Capaccioli, Phys. Rev. E, 69 (2004) 31501. 69. K.L. Ngai and M. Paluch, J. Chem. Phys., 120 (2004) 857. 70. K. L. Ngai, J. Non-Cryst. Solids, 275 (2000) 7. 71. H. Eyring, J. Chem. Phys., 4 (1936) 183. 72. The activation energy of 38 kJ morl was also found for the series PMMA,

PEMA and PnBMA in ongoing studies in MD research group, in the aim of Silvia Soreto Master Thesis, but with PMMA having a higher z, value in the Arrhenius plot. 73. G. Floudas, P. Placke, P. StcPAnek, W. Brown, G. Fytas, and K. L. Ngai, Macromolecules, 28 (1995) 6799. 74. S. Corezzi, M. Beiner, H. Huth, K. Schroter, S. Capaccioli, R. Casalini, D. Fioretto and E. Donth, J. Chem. Phys., 117 (2002) 2435. 75. J. Einfeldt, D. Mairjner and A. Kwasniewski, Prog. Polym. Sci. 26 (2001) 1419. 76. M. Viciosa, M. Dionisio, R.M. Silva, R. L. Reis and J. F. Mano, Biomacromolecules, 5 (2004) 2073.

77. R. Diaz-Calleja, E. Saiz, E. Riande, L. Gargallo and D. Radic, Macromolecules, 26 (1993) 3795. 78. S. Arrese-Igor, A. Arbe, A. Alegria, J. Colmenero and B. Frick, Appl. Phys. A, 74 (2002) S454. 79. H. W. Spiess, Colloid. Polym. Sci., 261 (1983) 193. 80. P.I.T. Inglefield, R. M. Amici, J. F. O'Gara, C. C. Hung and A.A. Jones, Macromolecules, 16 (1983) 1552. 81. A. Dubault, L. Bokobza, E. Gandin and J.L. Halary, Polym. Int., 52 (2003) 1108. 82. G. D. Smith and R.H. Boyd, Macromolecules, 24 (1991) 273 1. 83. C. Hansen, F. Stickel, T. Berger, R. Richert and E. W. Fischer, J. Chem. Phys., 107 (1997) 1086. 84. K. L. Ngai and M. Paluch, J. Chem. Phys,. 120 (2004) 857. 85. H. Wagner and R. Richert,. J. Phys. Chem. B, 103 (1999) 407 1. 86. A. Doss, M.Paluch, H. Sillescu and G. Hinze, Phys. Rev. Lett., 88 (2002) 095701. 87. P.W. Anderson, Science, 267 (1995) 1615 and C. A. Angel, Science, 267 (1995) 1924. 88. H. Vogel, Phys. Z., 22 (1921) 645, G. Tammann, W. Hesse, Z. Anorg. Allg. Chem., 156 (1926) 245. and G. S. Fulcher, J. Am. Ceram. Soc., 8 (6) (1925) 339 (originally the rate at which the relaxation time increases with cooling towards vitrification was quantified according the equation loglo(q)=A+B/(T-To)). 89. G. A. Schwartz, E. Tellechea, J. Colmenero and A. Alegria, J. Non-Cryst. Solids, 351 (2005) 2616. 90. R. Richert and C.A. Angell, J. Chem. Phys.,l08 (1998) 9016. 91. The T, scaled viscosity plot that was the the origin of the fragility concept was originally given by W.T. Laughlin and D.R. Uhlmann, J. Phys.Chem., 76 (1972) 23 17 and revised in K.L. Ngai, J.Non-Cryst.Solids, 275 (2000) 7. 92. C. A. Angell in Relaxations in Complex Systems, (Eds K.L. Ngai and G. B. Wright ,National Technical Information Service, US Dept. Commerce, Springfield, VA, 1985, and R. Bohmer, K.L.Ngai, C.A. Angell and D. J. Plazek, J. Chem. Phys., 99 (1993) 4201. 93. J. D. Ferry, Viscoelastic Properties of Polymers ,Wiley, New York, 1970. 94. A. K. Doolitle, J. Appl. Phys., 22 (1951) 1471. 95. H. Huth, M. Beiner and E. Donth, Phys. Rev. B, 61 (2000) 15092. 96. F. Kremer and A.Schonhals in Broadband Dielectric Spectroscopy, (Eds F. Kremer and A. Schonhals), Springer, Germany, 2002, Ch. 4.

97. N. Koizumi and Y. Kita, Bull. Inst. Chem. Res., Kyoto Univ., 56 (1978) 300. 98. U. Bengtzelius, W. Gotze and A. Sjolander, J. Phys. C, 17 (1984) 5915; E. Leutheuser, Phys. Rev. A, 29 (1984) 2765; W. Gotze, Z. Phys. B, 60 (1985) 195 ; W. Gotze and L. Sjogren, Rep. Prog. Phys., 55 (1992) 241. 99. M. Beiner, H. Huth and K. Schroter, J. Non-Cryst. Solids, 279 (2001) 126. 100. M. Beiner, J. Koms and E. Donth, Macromolecules, 30 (1997) 8420. 101. E. Hempel, S. Kahle, R. Unger and E. Donth, Thermochim. Acta, 329 (1999) 97. 102. M. Dionisio, J.J. Moura Ramos and G. Williams, Polym. Int., 32 (1993) 145. 103. T. Tetsutani, M. Kakizaki and T. Hideshima, Polym. J., 14 (1982) 305. 104. T. Hideshima, T. Tetsutani and M. Kakizaki, Polym. J., 14 (1982) 471. 105. K. Schroter, R. Unger, S. Reissig, F. Garwe, S. Kahle, M. Beiner and E. Donth, Macromolecules, 3 1 (1998) 8966. 106. M. Dionisio, J.J. Moura Ramos and G. Williams, Polymer, 35 (1994) 1705. 107. G. Williams, Trans. Faraday Soc.,, 62 (1966) 2091. 108. G. Williams and D.A. Edwards, Trans. Faraday Soc., 62 (1966) 1329. 109. C. Menissez, B. Sixou, L. David and G. Vigier, J. Non-Cryst. Solids, 35 1 (2005) 595-603. 110. M. Dionisio, N. M. Alves and J F. Mano, epolymers, 2004, no.044 111. F. Garwe, A. Schonhals, M. Beiner, K. Schrroter and E. Donth, J. Phys.: Condens. Matter, 6 (1994) 6941. 112. Silvia Soreto, Master Thesis, Universidade de Aveiro, Portugal, 2006. 113. E. Donth, J. Non-Cryst. Solids, 53 (1982) 325. 114. K. L. Ngai, Phys. Rev. E, 57 (1998) 7346. 115. M. Beiner, S. Kahle, E. Hempel and K. Schroter, Macromolecules, 31 (1998) 8973. 116. A. Schonhals, F. Kremer, A. Hofmann, E. W. Fischer and E. Schlosser, Phys. Rev. Lett., 70 (1993) 3459. 117. N. 0.Birge and S. R. Nagel, Phys. Rev. Lett., 54 (1985) 2674. 118. E. Donth, J. Non-Cryst. Solids, 53 (1982) 325. 119. E. Hempel, G. Hempel, A. Hensel, C. Schick and E. Donth, J. Phys. Chem. B, 104 (2000) 2460. 120. J. Korus, E. Hempel, M. Beinec S. Kahle and E. Donth, Acta Polym., 48 (1997) 369. 121. E. Donth, The Glass Transition. Relaxation Dynamics in Liquids and Disorder Materials, Springer, Berlin, 200 1. 122. E. Hempel, M. Beiner, T.Renner and E. Donth, Acta Polym., 47 (1996) 525; E. Donth, S. Kahle, J. Korus and M. Beiner, J. Phys. I (France), 7 (1997) 581.

123. D. Richter, L. Willner, A. Zirkel, B. Fargo, L.J. Fetters and J.S. Huang, Phys. Rev. Lett., 71 (1993) 4158. 124. C. LCon and K.L. Ngai, J. Phys. Chem. B, 103 (1999) 4045. 125. E. Donth, H. Huth and M. Beiner, J. Phys.:Condens. Matter, 13 (2001) L45 1. 126. R. Richert and C. A. Angell, J. Chem. Phys., 108 (1998) 9016. 127. S. Tahashima, Electrical Properties of Biopolymers and Membranes , Hilger, Bristol, 1989. 128. Y. Feldman, T. Skodvin and J. Sjoblom in Encyclopedic Handbook of Emulsion Technology, (Ed. J. Sjoblom),Dekker, New York, 2000. 129. R. Hilfer, Transport and relaxation phenomena in porous media, in Advances in Chemical Physics, Vo1.92, (Eds I. Prigogine and S. A. Rice ), Wiley, New York, 1996. 130. E. Neagu, P. Pissis, L. Apekis and J. L. Gomez Ribelles, J. Phys. D: Appl. Phys., 30 (1997) 1551. 131. F. Stickel, F. Kremer and E. W. Fischer, Physica A, 20 1 (1993) 3 18. 132. K. Pathmanathan and G. P. Johari, J. Polym. Sci., Part B : Polym. Phys., 3 1 (1993) 265. 133. I. M. Hodge and C. A. Angell, J. Chem.. Phys., 67 (1977) 1647. 134. J.C. Maxwell, Treatise on Electricity and Magnetism, Dover reprint, New York, 1892; K.W.Wagner, Arch. Elektrotech., 2 (1914) 371; R. W. Sillars, J. Inst. Electr. Eng., 80 (1937) 378. 135. P.A. Steeman and F. H. J. Maurer, Polymer, 33 (1992) 4236. 136. A. K. Jonscher, J. Phys. D: Appl. Phys., 32 (1999) R57. 137. R. El Moznine, G. Smith, E. Polygalov ,P.M. Suherman and J. Broadhead, J. Phys. D: Appl. Phys., 36 (2003) 330. 138. M. Sun, S. Pejanovic' and J. Mijovic', Macromolecules, 38 (2995) 9854. 139. F. Kremer and S. A. Rbzahski in Broadband Dielectric Spectroscopy, (Eds F. Kremer and A. Schonhals), Springer, 2002, Ch.12. 140. H. Lu, X. Zhang and H. Zhang, J. Appl. Phys., 100 (2006) 054104. 141. A. K. Jonscher, Nature, 267 (1977) 673. 142. A. G. Chametskaya, G. Polizos, V. I. Shtompel, E. G. Privalko, Yu. Yu. Percha and P. Pissis, Eur. Polym. J., 39 (2003) 2167. 143. A. Bello, E. Laredo and M. Grimau, Comparison of Analysis of Dielectric Spectra of PCL in the E" and M" Formalism, P.9, presented in the IDS&DRP 2006, Pbznan, Poland. 144. J. Einfeldt ,D. Meiljner and A. Kwasniewski, J. Non-Cryst. Solids, 320 (2003) 40. 145. J.C. Dyre and T.B. Schrchder, Rev. Mod. Physics, 72 (2000) 873.

146. J.M. Stevels, J. Non-Cryst. Solids, 40 (1980) 69. 147. H.E. Taylor, Trans. Faraday Soc., 52 (1956) 873. 148. K. Yamamoto, H. Namikawa, Japan J. Appl. Phys., 27 (1988) 1845. 149. M.T. Viciosa, M. Dionisio, R.M. Silva, R.L. Reis and J.F. Mano, Biomacromolecules, 5 (2004) 207. 150. K. Pathmanathan, G.P. Johari, J. Chem. Phys., 95 (1991) 599. 15 1. W.C. Hasz, C.T. Moynihan and P.A. Tick, J. Non-Cryst. Solids, 142-144 (1994) 1363. 152. G.M. Tsangaris, G.C. Psarras, and N. Kouloumbi, J. Mater. Sci., 33 (1998) 2027; G.M. Tsangaris and G.C. Psarras, Mater. Sci.,34 (1999) 2151. 153. E. Schlosser, A. Schonhals, H.E. Carius and H. Goering, Macromolecules, 26 (1993) 6027. 154. Y-H. Lee, A.J. Bur, S.C. Roth and P.R. Start, Macromolecules, 38 (2005) 3828. 155. F.M. Aliev, Z. Nazario, and G.P. Sinha, J. Non-Cryst. Solids, 305 (2002) 218. 156. M.T. Viciosa, A.M. Nunes, A. Fernandes, P.L. Almeida, M.H. Godinho and M. Dionisio, Liq. Cryst., 29 (2002) 429. 157. K. Asami, Prog. Polym. Sci., 27 (2002) 1617. 158. P.A.M. Steeman and J. Van Turnhout in Broadband Dielectric Spectroscopy, (Eds F. Kremer and A. Schonhals), Springer, 2002, Chap. 13. 159. R.H. Boyd, Polymer, 26 (1985) 323; 1123. 160. B. Frick, R. Zorn and H. Buttner (Eds), 1st International Workshop on Dynamics in Confinement, J. Phys., IV (2000), 10, Pr7. 161. B. Frick, M. Koza and R. Zorn, 2nd International Workshop on Dynamics in Confinement, Europ. Phys. J., E, (2003) 12. 162. Y. Ikada and H. Tsuji, Macromol. Rapid. Commun., 21 (2000) 117. 163. A. Sodegard and M. Stolt, Prog. Polym. Sci., 27 (2002) 1123. 164. M. Dionisio, M.T. Viciosa, Y. Wang and J.F. Mano, Macromol. Rapid Commun., 26 (2005) 1423. 165. A.R. Bras, M.T. Viciosa, Y. Wang, M. Dionisio and J.F. Mano, Macromolecules, 39 (2006) 65 13. 166. M.S. Graff and R.H. Boyd, Polymer, 35 (1994) 1797. 167. J.F. Mano, R.A. Sousa, R.L. Reis, A.M. Cunha and M.J. Bevis, Polymer, 42 (200 1) 6 187. 168. K. Schmidt-Rohr and H.W. Spiess, Macromolecules, 24 (1991) 5288. 169. J.F. Mano, Macromolecules, 34 (2001) 8825. 170. S. Lanceros-Mendez, M.V. Moreira, J.F. Mano, V.H. Schmidt and G. Bohannan, Ferroelectrics, 273 (2002) 15.

171. J.F. Mano, V. Sencadas, A. Mello Costa and S. Lanceros-Mendez, Mat. Sci. Eng. A, 370 (2004) 336. 172. R. Singh, J. Kumar, R.K. Singh, A. Kaur, R.D.P. Sinha and N.P. Gupta, Polymer, 47 (2006) 59 19. 173. J. van Turnhout, Thermally Stimulated Discharge of Polymer Electrets, Elsevier, Amsterdam, 1975 174. J. van Turnhout, Polym. J., 2 (1971) 173. 175. J.F. Mano, Thermochim. Acta, 332 (1999) 161. 176. J.F. Mano, J. Phys. D: Appl. Phys., 31 (1998) 2898. 177. J.F. Mano, N.T. Correia, J.J Moura Ramos, S. Andrews and G. Williams, Liq. Cryst., 20 (1996) 201 178. J.F. Mano and J.L. G6mez Ribelles, Macromolecules, 36 (2003) 2816. 179. J.F. Mano, J. Macromol. Sci. Phys., 42 (2003) 1125.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors 2008 Elsevier B.V.

Chapter 8

BENEFITS AND POTENTIALS OF HIGH PERFORMANCE DIFFERENTIAL SCANNING CALORIMETRY (HPer DSC) Vincent B.F.

ath hot"'^, Geert Vanden Poelc and Thijs F.J. pijpersb

"SciTe B.V., Ridder Vosstraat 6,6162 AX Geleen, The Netherlands. ([email protected]) b~atholiekeUniversiteit Leuven, Celestijnenlaan 200F, B-300 1, LeuvenHeverlee, Belgium 'DSM Research, P.O. Box 18,6160 MD, Geleen, The Netherlands

1. INTRODUCTION Most commercial 'Standard' Differential Scanning Calorimeters (Standard DSCs) are able to perform controlled, constant heating-rate experiments at typically 10 "Clmin while some of these nowadays can operate a bit faster or even at much higher heating rates. However, almost all calorimeters have problems in achieving high controlled, constant rates in cooling, and this poses a technical challenge to instrument suppliers. In this chapter, the emphasis will be on the need to accomplish a spectrum of rates - including both low and high rates - during heating as well as during cooling. Subsequently, solutions are discussed for controlled, fast cooling and heating by way of high-speed calorimetry. High-speed calorimetry has received a great deal of attention in recent years. The reasons are that firstly, in practice, most processes occur at much higher rates than are realizable using Standard DSC; and, secondly, most materials and substances, including polymers and pharmaceuticals, are in metastable states. Thermal history - specifically cooling and heating rates - and sample/product treatment can change their behaviour drastically, including their end properties. Phenomena related to metastability are well known to thermal analysts; daily they encounter supercooling, amorphization, 'hot7 crystallization (from the melt), cold crystallization (from the glass state), recrystallization (after melting), annealing, etc. In daily practice, researchers and product developers aim to connect the results of analytical research - in this case thermal analysis - to improvements in product properties. To be effective, researchers need the capability to change the

measurement conditions, in order to study and influence the metastability present. This chapter aims to open new horizons for research by way of using the capabilities of high-speed calorimetry. Examples of metastability will be given and the use of a commercially available version of high-speed calorimetry: High Performance DSC - in short HPer DSC - is elucidated to provide insight into metastability phenomena and to solve some of the related problems. In addition, the many advantages of HPer DSC will be discussed: increased sensitivity; access to quantitative measurements, including heat capacity measurements and determination of crystallinity as function of temperature; increased production; improved way of studying of kinetics; studying of reorganization phenomena; hindering of undesired phenomena like cold- and recrystallization; solid-solid transformations; matching rates occurring in daily practice; mimicking of rates during processing, like extrusion and injection moulding; measuring of minute amounts - from milligram down to the microgram level - of materials, such as impurities, coatings, yields of fractionations and combinatorial chemistry experiments; the possibility of running experiments at low to high scan rates using the same device. Though the examples are given for polymers, the findings are also of use for pharmaceutical raw materials and products. 2. MAJOR CHALLENGES 2.1. Introduction The challenges of performing measurements under conditions of practice and of studying metastability [I-41 and the connected reorganization are discussed in this section.

2.2. Measuring under realistic conditions Laboratory measurements on polymer raw materials and products are often carried out (quasi-) isothermally or at relatively slow cooling and heating rates. Such settings differ greatly from those occurring during processing and during the product's use in real life. A first major challenge therefore is to mimic the realistic conditions that occur in practice, such as the high cooling rates [5-91 applied in polymer processing [lo- 121, which influence vitrification [13] and crystallization [10,14] strongly. There is likewise a great need for equipment permitting the use of high heating rates [1,5,15-251 to imitate practical conditions. 2.3. The study of metastability and reorganization Metastability - a common feature in both macromolecular and pharmaceutical systems - is a second major challenge for the coming years. For a thorough understanding of the kinetics of all kinds of temperature- and time-dependent processes related to metastability, there is now an urgent need for new and better-matching thermal analysis and calorimetry techniques. Specifically, as related to metastability, linking melting behaviour to prior crystallization [14] is an important issue for understanding structure - property relations. Especially of relevance for industry is the possibility of deducing a sample's thermal history, using an appropriate heating scan, instead of blurring its history by a badly chosen scan-rate. Such linking of heating measurements to the sample thermal history may seem trivial, but it is not, and most experiments reported so far in literature fail to establish a direct and unequivocal link. The reason for this failure is the fact that, in most cases, the relationship between melting and crystallization and the resulting morphology is obscured by all kinds of reorganization processes [4,14,23,26] through which the metastable system strives towards thermodynamic equilibrium during the heating run. In principle, it should be possible to avoid such reorganization by heating at a high rate, so that the relationship between crystallization, the resulting morphology, and subsequent melting becomes transparent. An example [14] of reorganization occurring, reflecting that the polymer under study is in a metastable state, is given in Figure 1. Using Standard DSC, narrow molar mass fractions of linear polyethylenes (LPEs) were measured with regard to their crystallization and subsequent melting behaviour [4,14]. Crystallization was realized in cooling at different rates: 0.3 1, 5 and 40 'Clmin, while melting was done at 5 Olmin. These rates are typical of the capabilities of Standard DSC. The left side of the plot shows that the crystallization behaviour is markedly dependent on the molar mass and, in addition, to the cooling rate. After an initial increase of the crystallization peak temperature, T,, with increasing molar mass

Caused by sxdensim rsorgeniurfi~n dunnu c w l i n g and healing

Reorgmrznlim is suppressed by

Figure 1. Crystallization and melting at 5 OCImin of narrow molar mass linear polyethylenes (and, in addition, crystallization at 0.31 and 40 "Clmin) and low-density polyethylenes as function of molar mass. (see data at 5 Olmin), a maximum is reached, after which T, drops and levels off. It is known that the initial rise of Tc is connected to the diminishing influence of end-groups. At the same time, extensive chain folding takes over from the paraffin-type of crystallization. The decreasing part is ascribed to an increasing hindering of crystallization by entanglement of the long chain molecules. If the cooling rate is slower, 0.31 "Imin, the hindering is relaxed, while at faster cooling, 40 O / ~ i nhindering , is more pronounced. The leveling off of Tc with increasing molar mass, clearly seen for 0.31 and 5 "/min, is connected to the fact that the longer the chains become, the more chain segments are incorporated at different places in the same or different crystallites. Obviously, if the chain becomes very long, the behaviour of the segments becomes unrelated and, as a result, Tc is invariant to the molar mass. The important issue here is that, after cooling at 5 OC/min and subsequent heating at the same rate, T, shows a very different behaviour compared to T, with increasing molar mass, from approximately 20 kglmol onwards. The fact that, at low molar masses, T, increases with the molar mass is again connected to the diminishing influence of end-groups, and it follows the behaviour of the crystallization temperatures, as expected. However, so far, researchers have not been aware of the increasing discrepancy between Tc(M) and T,,,(M) with

increasing molar mass above approximately 20 kglmol. The explanation is that, during cooling, metastable crystallites are formed, which on heating are vulnerable to reorganization into more perfect and/or bigger crystallites. One of the well-known mechanisms for this is sliding diffusion of chains through the crystallites. Whatever the cause(s), the crystallites that are dealt with are by no means stable ones: they can adjust themselves according to the thermal treatment given. In contrast, see the right side of the plot, low density polyethylene, as produced by a high-pressure process, shows the expected correspondence of T, and T,, because the abundant presence of side groups in LDPE effectively hinders sliding diffusion of the branched chains through the crystallites and, because of that much less reorganization occurs. Obviously, in the case of LPE, the link between the crystallization and melting behaviour is obscured. How could the link be re-established? One could think about ways of hindering sliding diffusion by crosslinking the chains in the amorphous phase. However, that is a difficult, laborious, and sometimes tricky approach. A very interesting way would be fast heating, giving the molecules no time for reorganization. However, because the sliding difhsion process is extremely fast, the chance of success is low for LPE, and it is very difficult or impossible to cancel reorganization fully. Another example of excessive reorganization is represented in Figure 2 for a homogenous ethylene-propylene copolymer. Irrespective of the cooling rate here 1 and 150 "Ctrnin have been used - the melting behaviour is unchanged, and the same has been found for an ethylene-1-octene copolymer. This is quite interesting because, firstly, one would expect that methyl and hexyl branches, as resulting fiom incorporation in the chain of propylene and octene co-monomers, respectively, would hinder reorganization. Secondly, heating at 150 OCtmin obviously does not help to hinder reorganization like in the LDPE case. The difference is, of course, that LDPE has short to long branches (long means lengths in the order of the length of the chain itself). It would mean that there is a kind of 'minimum' length of the side branch where hindering of reorganization starts. Like in the LPE case, the result implies that, fiom a heating curve, it is difficult and sometimes impossible to interpret it on the basis of the preceding cooling rate, and in general on the thermal history. This is a very disturbing result, because it means that the link between crystallization and melting is lost during measurements, even in for a 150 OC/min heating rate, let alone at a standard rate of 10 OCImin. One has to realize that heating at 10 OC/min is even mostly too slow for suppressing recrystallization, see further on, while recrystallization within the "class" of reorganization processes must be judged as a relatively "slow" process. It also

Cooling rate 150 "C/min

w

If

V

Cooling nrte 1 "Chin

ssmpla-: 0.S7 mg

0

I0

20

30

50

40

Temperature ('C) 60 70

i 2 JI(s0C)

Cooling rates

- 1* C h i n

u

- - - 150 "c/min

i I-

-u a

Heating fete 150 "Urnin

rn

S@ mru: 097mg

-80

-60

4 0

-20

0

20

40

Temperature ('C) 60 80 100

120

Figure 2. Crystallization of a homogenous ethylene-propylene copolymer by cooling (top) at 1 and 150 "Clmin and subsequent melting (bottom) by heating at 150 OCImin [27]. implies that, for some polymers, it may impossible to relate the morphology at room temperature, obtained after cooling, by techniques like X-ray difiaction, microscopy, atomic force microscopy, etc., to the melting characteristics, e.g. the melting peak temperature. This is exactly what many researchers suppose to be possible when they apply the so-called Gibbs-Thomson relation, relating crystallite dimensions to melting temperatures. Thus, important correlations reported in the literature are not justified. Therefore, many of the reported

studies have to be redone in such a way - e.g. by applying much faster heating that reorganization is avoided. Obviously, for LPE and the ethylene-propylene copolymer much higher heating rates are needed to avoid or even hinder reorganization. Therefore, experiments at extremely high rates have been carried out using a thin-film (chip) calorimeter or nanocalorimeter, see also next section. Figure 3 shows results obtained by the thin-film chip calorimeter on Polyamide 6 (PA6), and on PA6 dispersed as sub-micrometer droplets in an amorphous matrix of polystyrene and a compatibilizer (PSlSMA2) [28]. Because of the very small droplets, heterogeneous nucleation is impossible, due to the lack of nuclei in the droplets, and the PA6 can be crystallized homogeneously at a temperature as low as 85 "C. This very low temperature can also be reached, without any crystallization, at rates of 30 000 "Clmin and higher. With increasing heating rates, a lower 'real' melting peak results, which is thought to be directly related to the crystallization temperature and time: it represents successfil suppression of reorganization. With increasing heating rate, the 'real' melting peak shifts to higher temperatures because of superheating. The higher melting peak, which is thought to still result from extensive reorganization, shifts to lower temperatures, indicating that reorganization of the crystallites formed at 85 "C is increasingly hindered - but not totally suppressed - by increasing the heating rate. Again, the reorganization process is extremely fast: even a heating rate of 300 000 "Clmin is not high enough to filly prevent this reorganization of PA6. In conclusion, for understanding metastability phenomena, one would like to be able to choose combinations of cooling and heating scans dedicated to the problem at hand. For most processes in practice, the present range of scan rates of Standard DSC is too limited to influence the metastability of the systems studied and hence to provide understanding. HPer DSC - see next section - is very useful to chart the problem and, in some cases, to cure it. For the specific purpose of hindering reorganization, it is evident that, for some polymers (extremely) high heating-rates are needed and, clearly, the rates realized recently are not high enough to cancel reorganization filly. However, the development and use of high-speed calorimetry has just started, and it is expected that, with the ongoing efforts to study the various reorganization processes in polymers using HPer DSC and chip calorimetry and to chart their timescales, and with ongoing technical developments, much more will be learned in the coming years. Moreover, there are many other phenomena and topics, like hot-, cold- and recrystallization, solid phase changes, high throughput combinatorial chemistry, etc., where the present capabilities of HPer DSC and chip calorimetry are more than satisfactory, as will be illustrated.

Temperature I "C Temperature dependence of the specific heat capacities Figure 3. of a Polyamide 6 and of a (PSISMA2)PAG (62/13)/25 blend, measured by thin-film chip calorimetry at different heating rates, after crystallization fiom the melt at 85 "C. 3. HIGH-SPEED CALORIMETRY

3.1. Instrumental aspects Because it is important in industry to be able to vary cooling and heating rates drastically, a project was initiated in the late 1990s at DSM Research (Geleen, The Netherlands) to address this issue, specifically directed to the field of thermal analysis. It was realized that increasing rates demand decreasing sizes of relevant components of the measuring equipment - especially the measuring cell - in order to improve the thermal conductivity paths. Decreasing the sample mass to values as low as 1 pg also seemed to be necessary, if required. Therefore, a commercially available DSC with a very small furnace was chosen. This calorimeter also offered the advantage of the power compensation design, providing direct heat flow rate measurements and an excellent sample temperature control for optimum results, see Figure 4. Currently only one type of calorimeter is available that is capable of measuring at controlled, constant, and appreciable high heating and cooling rates: the power-compensation Pyris 1

or Diamond DSCs of PerkinElmer. Due to their small furnaces - the Pyris 1 and Diamond DSC use two furnaces of approximately 1 gram - strictly controlled cooling at e.g. 300 "Clmin (and even up to 500 "Clmin for well-chosen conditions) and heating up to 500 OCImin are possible, depending on the temperature range of the measurements and the cooling accessory used. Of particular importance for high-speed calorimeters are the influences of the heating and cooling rates and sample mass on temperature calibration, including specific attention to calibration in cooling. The resulting experimental set up was called High Performance DSC (HPer DSC) [29-321 which has been commercialized in the mean time by PerkinElmer and trademarked with HyperDSC [33]. Nowadays, see Figure 4, it is possible to realize extremely high heating rates, see Allen et al. [22,25,34)], even as high as 6 . 1 0 ~"Clmin using the high-speed pulse-calorimeter [35]. Important progress has been made with the development of the ultrafast chip calorimeter by Schick et al. [36-391. This calorimeter uses sub-nanogram samples of material and achieves constant heating and cooling rates in between 1000 and 60 000 "Clmin, while rates of fast cooling and controlled, constant heating up to 600 000 "Clmin are possible, and even 1.2.10~ "Clmin [40] with a recently developed sensor [41]. However, these ultra-fast calorimeters are not commercially available and are not very user-friendly. These calorimeters are, therefore, at present, not suitable for daily practice in industry, but are very well suited for scientific research purposes.

3.2. Temperature calibration The variation of low to high scan rates used in HPer DSC requires an appropriate calibration procedure, both in heating and in cooling. Illers [20] illustrated the need for a proper calibration for thermal lag in order to arrive at the determination of the 'true' melting point of the crystals present in a polymer sample by using different heating rates and a constant sample mass. A linear onset) as a function of the square relation was reported for AT (= Tpeak- Textrapolated root of the heating rate (up to 36 "Clmin). However, this linearity cannot be extended to higher heating rates up to 500 OCImin [30], see Figure 6. For such high rates using HPer DSC, calibrants such as indium, lead, zinc, adamantane, azoxyanisole, etc., still retain peak shapes similar to those recorded at low rates, provided that the sample mass has been adjusted to lower (sub) milligram values. As a first approximation, if the scan rate is increased by a factor of x, the sample mass should be decreased by the same factor x [29]. Furthermore, for optimization of HPer DSC measurements, the thermal conductivity between holder and sample is improved drastically by using aluminium foil for full sample encapsulation instead of a (relatively heavy) pan as a sample container 1291.

Low

#

High-speed Calorimetry

(Standard DSC?

J

Key-success factors: good conductivities and low sample mass

,

'----i

w Standard furnace

'ultre fast ca~orimetryusing chip t e c h n o / o g Y ' ~ constent cooling and heating rates up to 60 000 "&Inin and even to 6 109 "C/min(just

Figure 4. The present evolution of Standard DSC towards a range of low- to high-speed calorimeters [32]. Commercial instruments like heat-flux and power-compensation 'Standard' DSCs work typically at scan rates of 0.1 to 60 OCImin; High Performance DSC (HPer DSC), using a modified PerkinElmer power-compensation Pyris 1 or Diamond DSC, covers the range 0.1 to 500 OCImin; thin-film (chip) calorimeters have scan rates fiom 1000 to 1.2.1o7 OCImin; and rates as high as 6.10~"Clmin are attainable using the high-speed pulsecalorimeter (all numbers are approximate indications).

An important advantage of using a low-mass of aluminium foil (or even a pan made out of aluminium [33]) is that the sample in the foil is wrapped fully and subsequently flattened, thereby creating a 3-dimensional heat-flow path instead of the usual 1-dimensional path from sensor to sample. In case the sample mass has not been adjusted, an increase in thermal lag will be found as caused by increasing the heating rate (top: symbol h) and the cooling rate (bottom: symbol c ) as can be seen for an azoxyanisole sample (Figure 5), and by increasing the sample mass, see e.g. indium (Figure 6). Therefore, there is a need for determining the influences of scan rate and sample mass on the temperature calibration of HPer DSCs. In [30], calibration matrices for the extrapolated onset temperatures and for peak temperatures as function of

the heating rate (from 1 to 500 OCImin) and sample mass (from 0.5 mg to 8 mg) are provided, on the basis of experimental measurements on indium. Azoxyanisole

eio

Temperature ("C) Azoxyanisole

exo

120

124

128

132

136

Temperature ("C)

Figure 5. DSC heating (top) and cooling (bottom) curves for azoxyanisole with different heating (h) and cooling (c) rates.

Indium 14

The influence of the sample mass on AT (= TpeakTextrapolated onset) of indium as a function of the heating rate, Sh.The

Figure 6.

linear relation as reported by Illers holds to approximately 36 "C/min. Reprinted from [30] with permission from Elsevier. Table 1 gives an example of the corrections to be applied for the extrapolated onset temperature for heating experiments, in comparison with a reference condition of 1 mg and 10 "Clmin. In addition, formulas describing the correction factors for such calibration are given in [30]. Correction for the thermal lag of the peak temperatures (see Table 4 and the formulas given in [30]) is a difficult topic, because these corrections are dependent on the specific heat flow rate at the temperature of the maximum in the DSC curve of the polymer or substance measured. Such a heat flow rate will in general be different from the one measured for indium. Therefore, any correction for the peak temperatures should only be considered as a very approximate indication. It is better to use the correction tables and related formulas as guides in choosing the optimum conditions for measurement.

In addition to primary calibration standards like indium, usable for calibration in heating, three secondary liquid crystalline substances [30,43,44], M24, HP-53 Table 1. Calibration matrix for the correction factors for heating rate, Shyand sample mass for extrapolated onset temperatures from indium, reprinted from [30] with permission from Elsevier. Correction factors for the extrapolated onset: CFT,Eo("C) Sample mass (mg)

and BCH-52, which are substances that show no or very small supercooling, were investigated and found to be suitable for calibration in both the heating and cooling modes of HPer DSC. An important outcome reported in [30] was that the 'symmetry' of the HPer DSC used was found to be satisfactory, see Figure 7. Symmetry of the DSC means that it is possible to use the same formulas, describing the indium correction factors, for the calibration in the heating and in the cooling modes. Asymmetry implies that a separate determination of a calibration matrix in the cooling mode has to be performed by using one of the mentioned secondary standards (M24, HP-53 or BCH-52). The calibration procedure developed for HPer DSC allows the right choices to be made to minimize the thermal lag with respect to the sample mass and scan rates at the start of the measurement, instead of just making corrections afterwards. For this purpose, information like that presented in Table 1 is very useful. Of course,

after determining the optimum experimental conditions, the corrections have to be applied to the results found.

Peak 3

/

.......................

.......

/

S,= 10 'Clmin S,= 100 %/mi

S = 200 %/mi

6 Jl(QK)

I

.....

,,

3

..

#1 : :

I 1

f

::

:;

-

peak2

f 0

20

40

f'

i

--.

f

Sc= 10 'Clmin Sc= 100 'Clmin -So= 200 'Clmin -20

;

.. -.-------------.

I(

P

f

.. .. ..

\

3

.........................

i:

Peak 3

SO

SO

Temperature ('C)

Figure 7. HPer DSC curves (left side) of M24 at different cooling and heating rates (S,, Sh: 10, 100 and 200 "Clmin) and peak temperatures (right side) of the phase transitions at a variety of cooling and heating rates. All samples have masses in between 0.9 and 1.1 mg. Reprinted from [30] with permission from Elsevier.

3.3. Constancy of the scan rate Another major topic in high-speed calorimetry is the linearity of temperature with time, that is the constancy of the scan rate. In heating, a rate of a hundred "Clmin does not seem spectacular, because any power-compensation DSC and most heat-flux DSCs can achieve such a rate. However, HPer DSC makes it possible to achieve even higher controlled and constant scan rates, such as the rate of approximately 450 "Clmin in heating, see Figure 8. The spikes present are caused by the cooling accessory. In cooling, Figure 9, scan rates up to approximately 450 "Clmin are possible too. Such cooling rates are in general difficult to realize, and not feasible at all with the present heat-flux DSCs.

Ernpty Holder, heating

Temperature ("C)

Figure 8. HPer DSC: achieved scan rates from sensor temperatures in

Ernpty Holder, cooling 0 -50

-

Purge gas: HelNe (I0190)

c10

c50

Temperature ("C)

Figure 9. HPer DSC: achieved scan rates from sensor temperatures in cooling [45].

Figure 9 illustrates the typical performance of a PerkinElmer HyperDSC in maintaining a constant cooling rate across a temperature range. For cooling rates of 10 to 200 "Clmin, the Figure shows that these rates do not pose a problem down to sub-ambient temperatures and room temperature respectively. At 200 "Clmin, the temperature range ("the window") where the instrument maintains the programmed cooling rate as measured by the sensor, is approximately 400 "C. For 300 "C/min, the window is approximately 300 "C, while at 500 "Clmin the window is approximately 150 "C, while the rate has become effectively 450 "Clmin. The related heat flow rate curves (not shown here) clearly reflect the temperatures where the control of the scan rate breaks down. Thus, while rates of 300 OCImin have been reported earlier by us as feasible in cooling, by making a proper choice of the range over which cooling is to be done, in combination with choosing an optimum choice of the purge gas(es), and by using a deepcooling accessory (like that provided by liquid nitrogen), even higher cooling rates are possible. With respect to the purge gas(es), it has been reported in [29] that a mixture of 10% helium and 90% neon is suited for a temperature range of measurement of -176 to 585 "C. Both inert gases do not condense under cooling by liquid nitrogen, and the mixture offers a useful compromise with regard to heat conduction issues around the sample containers. However, the use of nitrogen only as purge gas also works well.

3.4. Linking experiment with practice and processing As an example of an industry-relevant experiment, in Figure 10 the drastic decrease of the crystallization peak temperatures at increasing cooling rates is shown for polypropylene (PP). The curves are not corrected for the cooling rate, as could have been done by using correction factors as given in Table 1. Note that the areas of the peaks are decreasing, meaning that the crystallinities of the samples are lowered as well. Using the various cooling rates available, most rates that occur during processing can be mimicked, see Figure 11, with the exception of the skin of an injection-moulded product, because usually the cooling rate in the contact area with the mould is too high to mimic using HPer DSC. In such a case, a chip calorimeter should be used. However, the exact information on the actual cooling rates as a function of the depth (from skin to core) of an injectionmoulded part during processing is still missing, because calculation by modeling requires the right experiments, like the ones shown here. The original - not corrected - DSC cooling curves, see Figure 10, underlie the crystallization peak temperatures, T,, which are given in Figure 11. These peak temperatures are also influenced by thermal lag as related to the cooling rates applied and the sample masses used. It is clear that the lower the sample mass

the higher the T,, which is expected. Proper correction of the crystallization peak temperatures for the extra thermal lag for a specific cooling rate should bring the values for the different sample masses to one value for T,. As remarked before, Polypropylene endo 5 J/(g°C)

1

exo

70

80

90

100

110

120

130

140

Temperature ("C)

Figure 10. Polypropylene (PP) curves (uncorrected) at various cooling rates (c) and their influence on the crystallization behaviour. this is a difficult topic and the result of such a correction can only be indicative. Nevertheless, it is very informative to perform such correction and Figure 12 gives the result. These corrections bring the different Tc values resulting from the different sample masses closer to each other [46]. Now, it is seen that the Tc values level off at the highest cooling rates, and such a behavior has been noticed by others too [37]. However, the decrease of Tc is still linear with the logarithmic of the increasing cooling rate, Sc. Nevertheless; it could indicate an increasing difficulty of the instrument with sample and container to realize the good thermal conductivities needed. In industry, despite the shortcomings discussed, this kind of information is critical because of, e.g., the high expense of the processing equipment used. Thus, in processing, as in the case of injection moulding, optimization of the process - and specifically shortening of the cycle time - is of major importance.

For processing of polymers, shortening of the cycle time usually means that the crystallization of the material should be speeded up thus realizing a faster

125

1 120

-

115

-

110

-

0

!7 F 105 -

o

Q

A

I;

r

v

: : , :1

I

I

,

,

,

0.225 rng 0.579 mg 0.589 rng 0.804 mg 0.878 rng 0.917 mg 1.023 rng 1.295 mg

S, (OC/min) 90 100

50

0

T

xoldin x

t

I

w

150

s

i

2M)

250

core - injection film blowing molding o K

Figure 11. Uncorrected crystallization peak temperatures from cooling curves at various cooling rates, S,, and different sample masses for PP. Cooling rates of various processing techniques are indicated. release from the mould. However, sometimes it is of advantage to slow down the crystallization process to achieve good surface properties. Realistic information concerning the crystallization temperatures and heat changes involved are also vital to realizing fast iteration within the "cycle of knowledge" [47], in order to speed up the development of materials of new grades and to optimize existing grades. A striking example of the benefit of using a realistic cooling rate instead of a 'standard laboratory' cooling rate is presented in Figure 13 [32]. At 10 OCImin cooling - as in the case of Standard DSC - two peaks were observed for crystallization of a low-density polyethylene 1 Ziegler-Natta linear low-density polyethylene (LDPEJLLDPE) blend. Such a result suggests a risk of segregation by crystallization during film blowing, which could result in numerous problems (i.e., optical and mechanical) in the film. HPer DSC measurements at

(controlled, constant) 150 "Clmin cooling show that, most probably, no problem will occur at all because only one crystallization peak is observed.

Polypropylene

A -8

0

-

A

corr. 0.094 mg corr. 0.579 mg corr. 1.023 mg 0.094 mg 0.579 mg

A

1.023 mg

u

-

-

6

8e B

-

H

a

-

8 ¤

a

s

H

8 H

-

A I

I

I

I

A I

Figure 12. The crystallization peak temperatures of PP with different sample masses as a function of the cooling rate. The filled symbols are the T, values without correction, and the open symbols those corrected for the sample masses and the cooling rates. Extensive co-crystallization of molecules of different origin probably takes place at this cooling rate, and segregation is hindered effectively. This needs to be verified for the even higher cooling rates that occur during film blowing, which could lead to even more co-crystallization. However, this is on the premise that the various types of molecules present in the blend: the less-highlybranched ones of the LLDPE and the highly-branched LDPE molecules, which cause the low-high- and low-temperature crystallization peaks respectively at standard cooling rates, do not shift very differently with respect to temperature

with increasing cooling rate. Without proper equipment and, specifically, the option to choose the optimum cooling rate out of a spectrum of scan rates using either Standard DSC or HPer DSC or chip calorimetry, the researcherlthermal

LDPL'LLDPE blend 0

e

-150 "C/min

9

Sampfe mass:

cooling

t 0

u

96 "C

-

5u

&

-wng ------------------.*------------------..-*---- ------

c

W

-lO°C/min Sample mass: 4.215 mg

98.9 "c 107.9"C

Figure 13. Cooling an LLDPELDPE blend at a standard rate of 10 "Clmin causes segregation by crystallization. Cooling at 150 OCImin gives co-crystallization instead. analyst could easily steer important and costly developments in the wrong direction. Recrystallization is a specific kind of reorganization. Figure 14 presents an example [46], which most experimenters have come across. During heating at 10 OCImin (as in case of Standard DSC) after cooling at high rates (e.g. such as during processing, or in a fast-cooling experiment in a DSC, like that presented here), one often measures two melting peaks instead of one. After a thorough study of the dynamic behaviour of the system, such a curve could be used as a kind of fingerprint to estimate (approximately) at what rate the material (of one's company or that of a competitor) has been cooled at. More often than not, these peaks are not caused by different molecular structures resulting in different crystallites, as is frequently the interpretation.

Instead, these peaks are frequently caused by extensive recrystallization: crystallites that melt, recrystallize and remelt. The chance of early melting happening increases for crystallites that are more imperfect or unstable, and fast cooling is one of the obvious ways to produce such kinds of crystallites. Polypropylene endo

10 JI(g°C)

t

exo

120

130

140

150

160

170

180

Temperature ("C)

Figure 14. Melting during heating at 10 OCImin, after cooling at various rates for a PP showing recrystallization behaviour. Recrystallization can result in quite stable crystallites having high melting temperatures if the recrystallization occurs at a very high temperature. Such a high temperature is very difficult to realize by slow cooling (if one has the opportunity to do so, which is not possible if one has to study an as-received processed sample) because, during cooling, appreciable supercooling may occur due to kinetic barriers towards nucleation. All these aspects discussed are noticeable in Figure 14. The double peaks start to occur when the cooling rate is a little bit higher (here from 20 OCImin on), than the heating rate, which is held constant at 10 'Clmin. The faster the cooling, the lower the first melting peak occurs, reflecting the lowering of the crystallization temperatures, see Figure 12. Also, the faster the cooling rate the

more recrystallization and remelting occur, as shown by the areas under both melting peaks. After recrystallization, remelting occurs at higher temperatures than after (the relatively slow) cooling at 10 'Clmin. The question is whether it is possible to hinder, or even fully suppress, recrystallization. It turns out that suppression of recrystallization is indeed possible for most of the polymers studied so far using HPer DSC. Evidently, recrystallization is a slow process compared to the reorganization processes discussed earlier. In this specific example [46] heating at a higher rate of 300 "Clmin (Figure 15) totally suppresses recrystallization. During heating at this rate just one melting peak results, which is clearly the former low-temperature one. Thus, recrystallization is effectively eliminated, which turns out be possible even at lower heating rates than the 300 "Clmin applied. It does not mean that the extensive reorganization - as discussed at Figures 1 to 3 - is absent, because Polypropylene endo 10 J1(g°C)

I

exo

120

130

140

150

160

170

180

Temperature ("C)

Figure 15. (Corrected) heating curves at 300 OCImin after cooling at various rates for PP showing that recrystallization is effectively eliminated. the lowering of the first melting peak temperature with preceding cooling rate does not follow the decrease of the crystallization peak temperature. Thus, during heating, reorganization still takes place, by which the first melting peak is

situated at much higher temperatures than expected on the basis of the crystallization peak temperatures. 3.5. Quantitative measurements A fast measurement takes minimal time, and instrumental drift can be negligible. Thus, in principle, it is possible to work quantitatively even at high rates, which is why the name High Performance DSC has been coined. HPer DSC provides more than just the ability to make measurements at high speeds. The examples in [29] show that heat capacity measurements at rates as high as 100 "Clmin are achievable. Recently [27,32] an extreme heating experiment was performed on a 0.39 mg high-density polyethylene (HDPE) sample from -175 "C to 200 "C at 150 "Clmin giving a continuous heat capacity measurement [48] across an tremendously wide temperature range in just a few minutes, see Figure 16. As can be seen, the experimental specific heat capacity of the semi-crystalline HDPE at low temperatures lies in between the reference functions of the extreme states in which polyethylene can be: fully amorphous, or fully crystalline. With increasing temperatures, the heat capacity increases because of melting. This can be seen better in Figure 17 in the plots of the integrated enthalpy - functions. The specific enthalpy curve, h(T), based on the experimental specific heat capacity, is seen to be situated in between the specific enthalpy curves for the fully amorphous and crystalline states. This provides a way of calculating the crystallinity, see Figure 17, as function of temperature, on the premise of the two-phase model. The initial crystallinity at the lowest temperatures is approximately 70%, and it steadily decreases with heating at 150 "Clmin until it vanishes completely. Performing heat capacity measurements is state-of-the-art calorimetry, and provides a way to quantifl research and to compare the results with those from other techniques like X-ray diffraction [49], NMR [50] etc., if performed quantitatively as well. In addition, it permits one to decide whether the simple but powerful two-phase model breaks down. If so, one should explore the possible use of a three-phase model [48]. However, without a sound basis provided by quantitative experimentation it is impossible to arrive at the right judgment.

292

sm,dsrn~

3

+

t

u

- 7 -

- 6

-

5

-

4

-

3

-

2

s2

o 98 me

x $ ua

Heating rate 150 OC/rnin

.

\

Temperature ('C)

1 -1 50 -125 -1 00 -75 I

-50

-25

0

25

50

75

100

125

150 175

Figure 16. The experimental specific heat capacity curve, c,(T), on heating at 150 OClmin across a wide temperature range on an HDPE sample. c,,(T) and c,(T) are the specific heat capacity reference h c t i o n s for 100% amorphous and 100 % crystalline LPE respectively.

_

900

-

800

-3

90 quenched

E t

- 700

-

_

600 500

Figure 17. The specific enthalpy curve, h(T); the specific enthalpy curves for the hlly amorphous and crystalline states; and the percentage crystallinity curve, W(T), a11 based on the specific heat capacity curves shown in Figure 16.

3.6. Higher sensitivity; working on minute amounts of material A high heating rate increases the sensitivity of the DSC because the instrument measures the change in heat flow (dQ) of a sample in comparison to the reference per unit of time (dt) - dQldt (W) - and, if the change in heat flow as caused by e.g. crystallization or melting is the same, it takes place in a shorter time span. However, the thermal lag also increases and, thus, smaller samples are required in order to keep the thermal lag acceptable. The ability to obtain quantitative data on small samples is a new and great advantage, because it facilitates research on minute amounts of material. How low can a sample mass be? Measurements on samples down to 400 ng can be done [30], but is working with a few micrograms still representative of the product? If the material is homogeneous (modern materials are produced with improved purity and decreased contamination during processing) there is no problem. If contaminants, gels, discolored matter, irregular surface parts, etc., are present in a sample or product then these can now be studied down to pg level by means of HPer DSC. A good example of the use of the low-sample mass capability of HPer DSC is discussed next. Removal by evaporation of the eluent used by a size-exclusion chromatograph (SEC) provides a way to deposit and spread the remaining polymer sample on a germanium disk as a function of molar mass. The sample is fractionated according to molar mass. The polymer mark left can then be measured by FTIR to determine the short chain branching content (if present) as a function of molar mass. Removal of fractions from the disk enables measurement by HPer DSC. Since the amount of starting material for SEC (typically 800 pg of polymer) has decreased over the years (20 years ago it was typically 5 mg), HPer DSC is an excellent technique capable of measuring (de)vitrification, hot-, cold-, recrystallization, melting, etc., in detail on such minute amounts. It provides information on the distribution of the short chain branches (SCB) as a function of molar mass on the basis of the crystallization and melting behaviour, which then can be combined with the average content from FTIR. Recently, the authors were able to apply this method to a heterogeneous, metallocene-based ethylene- 1-pentene copolymer [5 11. This new type of copolymer crystallizes and melts across extremely wide temperature ranges due to its broad ethylene sequence length distribution, but these transitions can still be measured effectively, as is seen for a 10-pg fraction of the copolymer shown in Figure 18. As can be seen, during operating the HPer DSC at 150 OCImin in cooling, crystallization occurs all the way fi-om approximately 110 "C down to approximately -20 "C. On heating at the same rate melting occurs up to approximately 130 OC. Though some noise is obvious, the heat flow rate can be measured and even the areas under the peaks can be estimated. Obviously, this

-F r m

C

B

,

-0

s

a

~

~

:

foPg

150 "C/min

Temperature (OC) 20

40

60

80

100

120

140

Figure 18. HPer DSC on a molar mass fiaction (M, = 8-1 1 kgtmol; M, = 15-20 kglmol) obtained by SEC fiom a whole, heterogeneous, metallocene-based ethylene-1-pentene copolymer (M, = 7.6 kglmol; M, = 35 kg/mol; 9.1 mole0/0 of 1-pentene). can be very usefil, for instance, if only a small amount of material is available fiom polymerization. Interesting and useful areas of research are possible when only minute amounts of material are available: (sub)milligram-scale synthesis, fractions, explosives, nanostructures, microelectronics, additives, contaminants, multilayers, coatings, thin films, skin-core problems in the case of chemicals, materials, products, but also in challenging areas like e.g. forensic studies. Degradation is another interesting topic: it can be hindered or even avoided by fast heating, in the same way that chemical changes during heating can be prevented, by simply spending much less time in a critical temperature range. Also 'self -seeding (nucleation by entities reminiscent of incompletely molten crystals, because of a too low temperature of the melt and/or a too short time in the melt) can now be charted much better, avoided or be used on the contrary. In addition, the fast operation of HPer DSC has been proven to enable highthroughput experimentation: a typical number of 100 cooling and 100 heating curves in 8 hours is feasible.

4. CONCLUSIONS

Researchers who are willing to explore high-speed calorimetry instrumentation will become accustomed to measuring at different rates. As such, the intention of this chapter is not to promote the use of high rates only, but rather to present the option of choosing an optimal rate or spectrum of rates, depending on the sample and on the question posed. This enables mimicking of real-life processes; studying of the kinetics of processes of e.g. polymers, pharmaceuticals, including hot-, cold-, recrystallization, melting, reorganization, annealing, (de)vitrification and measuring on minute amounts of substances, materials and products with high sensitivity in short time. As High Performance DSC (HPer DSC), commercially available as HyperDSC, is increasingly applied, more and more researchers will learn how to benefit fiom this exciting development.

5. REFERENCES B. Wunderlich, Macromolecular Physics, Academic Press, New York, 1976, Vol. 2, Crystal Nucleation, Growth, Annealing, 461 pages. B. Wunderlich, Macromolecular Physics, Academic Press, New York, 1980, Vol. 3, Crystal Melting, 361 pages. A. Keller, Macromol. Symp., 98 (1995) 1. V.B.F. Mathot, R.L. Scherrenberg and T.F.J. Pijpers, Polymer, 39 (1998) 4541. A. M. Filonov, 0. S. Novikova and Yu.V. Tsekhanskaya, Zh. Fiz. Khim., 48 (1974) 1597. V. Brucato, G. Crippa, S. Piccarolo and G. Titomanlio, Polym. Eng. Sci., 31 (1991) 1411. S. Piccarolo, M. Saiu, V. Brucato and G.J. Titomanlio, Appl. Polym. Sci., 46 (1992) 625. E. Ratasjski and H. Janeschitz-Kriegl, Colloid Polym. Sci., 274 (1996) 938. H. Janeschitz-Kriegl, E. Ratasjski and H. Wippel, Colloid Polym. Sci., 277 (1996) 217. G. Eder and H. Janeschitz-Kriegl, in H.E.H. Meijer Ed., Material Science and Technology, Processing of Polymers, R.W. Cahn, P. Haasen, E.J. Kramer, Eds. Volume 18, VCH Verlagsgesellschafi mbH, Weinheim, 1997, Crystallization, Chapter 5, p. 296-344. H. Zuidema, Eindhoven, Technische Universiteit Eindhoven, 2000, ISBN 90-386-3021-2, Thesis, Flow Induced Crystallization of Polymers, Application to Injection Moulding, 126 pages. C.M. Hsiung and M. Cakmak, Polym. Eng. Sci., 3 1 (1991) 1372.

13. M.J. Richardson, in Calorimetry and Thermal Analysis of Polymers, (Ed. V.B.F. Mathot), Hanser Publishers, Munich, Vienna, New York, 1994, The Glass Transition Region, Ch. 6, p. 169. 14. V.B.F. Mathot, in Calorimetry and Thermal Analysis of Polymers, (Ed. V.B.F. Mathot), Hanser Publishers, Munich, Vienna, New York, 1994, The Crystallization and Melting Region, Ch. 9, p. 23 1. 15. N.E. Hager, Jr., Rev. Sci. Instrum., 35 (1964) 618. 16. E. Hellmuth and B.J. Wunderlich, Appl. Phys., 36 (1965) 3039. 17. N.E. Hager, Jr., Rev. Sci. Instrum., 43 (1972) 1116. 18. O.F. Shlenskii and G.E. Vishnevskii, Dokl. Akad. Nauk SSSR, 279 (1984) 105. 19. B. Frochte, Y. Khan and E. Kneller, Rev. Sci. Instrum., 61 (1990) 1954. 20. S.L. Lai, G. Rarnanath, L.H. Allen, P. Infante and Z. Ma, Appl. Phys. Lett., 67 (1995) 1229. 2 1. R. Domszy, Proceedings Metallocenes '96, Diisseldorf 1996,25 1. 22. S.L. Lai, G. Ramanath and L.H. Allen, Appl. Phys. Lett., 70 (1997) 43. 23. V.B.F. Mathot, Thermochim. Acta, 355 (2000) 1. 24. V.B.F. Mathot, J. Thermal Anal. Calorim., 64 (2001) 15. 25. A.T. Kwan, M.Yu. Efremov, E.A. Olson, F. Schiettekatte, M. Zhang, P.H. Geil and L.H Allen,. J. Polym. Sci., 39 (2001) 1237. 26. B. Goderis, M. Peeters, V.B.F. Mathot, M.H.J. Koch, W. Bras, A.J. Ryan and H. J. Reynaers, Polym. Sci., Part B: Polym. Phys., 38 (2000) 1975. 27. M.F.J. Pijpers and V.B.F. Mathot, unpublished results. 28. R.T. Tol, A.A. Minakov, S.A. Adamovsky, V.B.F. Mathot and C. Schick, Polymer, 47 (2006) 2 172. 29. T.F.J. Pijpers, V.B.F. Mathot, B. Goderis, R.L. Scherrenberg and E.W. van der Vegte, Macromolecules, 35 (2002) 3601. 30. G. Vanden Poel and V.B.F. Mathot, Thermochim. Acta, 446 (2006) 41. 3 1. V.B.F. Mathot, G. Vanden Poel and T.F.J. Pijpers, American Laboratory, 38(14) (2006) 21: downloadable for free from the website of SciTe B.V.: www.scite.eu. 32. See www.scite.eu and also a webcast by V.B.F. Mathot through this website. 33. See the website of PerkinElmer: www.hyperdsc.com. 34. M.Y. Efremov, E.A. Olson, M. Zhang and L.H. Allen, Thermochim. Acta, 403 (2003) 37. 35. C. Cagran and G. Pottlacher, in this book; B. Wilthan, C. Cagran, and G. Pottlacher, International Journal of Thermophysics, 26 (2005) 1017. 36. A.A. Minakov, D.A. Mordvintsev and C. Schick, Polymer, 45 (2004) 3755. 37. F. De Santis, S. Adamovsky, G. Titomanlio and C. Schick, Macromolecules, 39 (2006) 2562.

38. A.A. Minakov, S.A. Adamovsky and C. Schick, Thermochim. Acta, 432 (2005) 177. 39. A.W. Herwaarden, Thermochim. Acta, 432(2) (2005) 192. 40. A.A. Minakov, A.W. van Herwaarden, W. Wien, A. Wurm and C. Schick, Advanced nonadiabatic ultrafast nanocalorimetry and superheating phenomenon in linear polymers, submitted to Thermochim Acta. 41. Ultra fast chip XEN-3973, see technical data available on the website of Xensor Integration B.V.: http://www.xensor.nl. 42. K-H. Illers, Eur. Polym. J. 10 (1974) 9 11. 43. S.M. Sarge, G.W.H. Hohne, H.K. Cammenga, W. Eysel and E. Gmelin, Thermochim. Acta 36 1 (2000) 1. 44. S. Neuenfeld and C. Schick, Thermochim. Acta, 446 (2006) 55. 45. G. Vanden Poel and V.B.F. Mathot, unpublished results. 46. G. Vanden Poel and V.B.F. Mathot, to be published. 47. V.B.F. Mathot, J. Thermal Anal. Calorim., 64 (2001) 15. 48. V.B.F. Mathot, in Calorimetry and Thermal Analysis of Polymers, (Ed. V.B.F. Mathot), Hanser Publishers, Munich, Germany, Vienna, New York, 1994, Thermal Characterization of States of Matter, Ch. 5, p. 105. 49. B. Goderis, H. Reynaers, R. Scherrenberg, V.B.F. Mathot and M.H.J. Koch, Macromolecules, 34 (200 1) 1779. 50. V.M. Litvinov and V.B.F. Mathot, Solid State Nuclear Magnetic Resonance, 22 (2002) 2 18. 5 1. N. Luruli, T. Pijpers, R. Briill, V. Grumel, H. Pasch and V. Mathot, Polym. Sci., Part B: Polym. Phys., in print.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 9 DYNAMIC PULSE CALORIMETRY - THERMOPHYSICAL PROPERTIES OF SOLID AND LIQUID METALS AND ALLOYS C. Cagran and G. Pottlacher Graz University of Technology, Austria 1. INTRODUCTION - THERMOPHYSICAL PROPERTIES

Even though the term 'thermophysical properties' is commonly used in scientific and metallurgical circles and some books, summaries, and articles are in circulation (see, e.g., [l-3]), no distinct definition can be found, leaving some margin for adding (or even dropping) some properties from time to time. Because there are many possible ways to define 'thermophysical properties', we propose to define thermophysical properties as a selection of mechanical, electrical, optical, and thermal material properties of metals and alloys (and their temperature dependencies) that are relevant to industrial, scientific, and metallurgical applications. Although this definition still covers a wide range of different material properties obtained by numerous different measurement techniques and approaches, the focus within the following chapter will be exclusively on thermophysical properties that are accessible through dynamic pulse calorimetry1, namely resistive self-heating or pulse-heating. Other techniques for investigation of material properties, like levitation, laser heating, chemical flames, shock waves, solar heating, fission/fusion, and electron/neutron heating, have been developed but will be excluded (with the exception of levitation), because they are all considered to be 'non-calorimetric' and lack information about the exact energy input to the test specimen. This generally also holds for levitation but because technologically important properties like viscosity and surface tension cannot be obtained by dynamic pulse calorimetry, levitation will

' According to the general aim of the book, the focus is on calorimetric measurements, which is, by definition, measurements of heat-energy relationships' [4].

also be mentioned2. The list of dynamic pulse calorimetric properties consists of (properties obtained by levitation are denoted by *): property specific enthalpy heat of fusion isobaric/isochoric heat capacity density electrical resistivity thermal conductivity thermal diffusivity phase transition temperatures melting temperature (pure metals) solidus/liquidus temperature (alloys) refractive index extinction coefficient normal spectral emittance hemispherical total emittance critical pressure critical volume critical temperature equation-of-state (EOS) parameters sveed of sound viscosity' surface tension*

symbol identifier H

rn

cdc"

'E

Thermal, electrical, and mechanical properties

a various T, TdTL n k Optical properties &a Eh

PC Vc Critical parameters TC various EOS C.

Levitation

rl

With the development of tools and techniques for computer-assisted calculations and simulations, knowledge of thermophysical properties at elevated temperatures up into the liquid phase has become even more important for the metal-working industry and related fields. Advances in computer-based simulations allow simulations of heat transport, solidification shrinkage, residual stress, or even predictions of microstructures, to name a few. In the list of thermophysical properties, some properties are of greater importance for use in industrial applications, while others are of more scientific interest for different applications. The properties most relevant to casting simulations3 are heat of fusion, heat capacity, electrical resistivity, density, thermal conductivity and difhsivity, thermal expansion, hemispherical emittance, viscosity and surface tension [ 5 ] .

'This means no restriction of any kind, because knowledge of the actual energy input is unnecessary for viscosity and surface tension measurements using levitation techniques. Assuming that the melting temperature (solidus/liquidus temperature for alloys) is already known. Otherwise these temperatures need to be added to the list.

As technology advances and specific needs for applications also change over time, the list of accessible thermophysical properties cannot be expected to remain static. Therefore, the following chapter can only be seen as a snapshot of the current state-of-the-art capabilities in the measurement of thermophysical properties by dynamic pulse calorimetry.

2. DYNAMIC PULSE CALORIMETRY (PULSE-HEATING) 2.1. Historical development and brief description of pulse-heating The basic principle of dynamic pulse calorimetry or pulse-heating is simple but effective at the same time. A large current pulse is passed over an electrically conducting sample, resulting in rapid resistive or ohmic self-heating. This procedure can be compared to an uncontrolled and thus destructive version of a light-bulb. In a light-bulb, the filament is also heated by an electrical current and starts to emit light, but is handled under vacuum or a protective gas atmosphere to prevent oxidation and thus breakage of the sensitive filament. By recording the energy deposited into the specimen a number of different properties can be measured, as will be shown later in this chapter. The first resistive self-heating experiment reported in the literature was performed by Nairne in 1773, when he demonstrated, in front of the Royal Society in London, an experiment in which he literally exploded an iron wire by passing the current from 64 charged Leyden jars (the predecessor of today's capacitors) over the test sample [6] which was used as a current estimating device. It took until the mid 1950s for the pulse-heating technique to become technologically of interest. During this time Chace and Moore published a series of books with the title 'Exploding Wires' discussing and presenting different applications of this technique [7]. Due to the increasing demand in the 1960s for data on the thermophysical properties of materials under extreme conditions and at high temperatures on one hand, and the rapid advances in fast electronics, such as electrical pulse generation, data acquisition equipment, etc., on the other hand, dynamic pulse calorimetry became more popular and more commonly used, i.e., Cezairliyan 1969 [8], and Lebedev et al. 1971 [9]. Conventional steady-state and quasi steady-state techniques for measuring thermophysical properties of solid and liquid materials are generally limited to temperatures below 2500 K. These limitations are mainly due to severe difficulties in handling samples at higher temperatures, such as chemical reactions and contamination, heat transfer, evaporation, loss of mechanical strength and electric insulation, etc. due to the long exposure duration of the specimen to high temperature, ranging from a few seconds to hours. This is where dynamic pulse calorimetry comes into play, because it allows such effects

to be minimized such effects and temperature ranges up to 10 000 K can be reached by passing a large electrical current over the sample during a short period of time. Among the numerous other advantages of these techniques, which are discussed later in this chapter, such methods are particularly relevant because the applied energy is imparted uniformly to the entire volume of the specimen in less than one second on the one hand, and yet permits the time dependence of some physical phenomena to be determined on the other hand. 2.2. Classification of pulse-heating systems and existing systems The main classification for pulse-heating systems is according to the experimental duration andlor the heating rates used. It is common to distinguish three different time regimes: - (1) millisecond experiments (heating rates of about 1o4 K S-I) - (2) microsecond experiments (heating rates of about 1o6 K S-I to 10' K S-I) - (3) submicrosecond experiments (heating rates of up to 10'' K s-') Experiments in each of the three time-regimes have their advantages, as well as their limitations, which makes it necessary to pick the time resolution depending on the scientific interest. Millisecond experiments can be used to determine thermal and electrical properties of metals and alloys in the solid phase up to the melting point at around 3000 K. This moderately fast heating rate allows, e.g., the determination of phase transition temperatures. Nevertheless, millisecond experiments are limited to the solid state as the rate of heating is still slow compared to the gravitational collapse of a liquid sample under gravitational forces once it becomes molten. Another limitation from the moderate heating rate is the need for heat loss corrections at elevated temperatures. Besides measurements of thermophysical properties, millisecond experiments are the tools of choice for metrological investigations. They have been used by , NRLM, Japan; and lately by HIT, China, for NIST, USA; I N R ~ M ~Italy; measurements dealing with radiance temperatures, radiometry, and total hemispherical emittance measurements [lo, 111. Known existing experiments: The first fully functional millisecond setup was built by Cezairliyan et al. [12] in 1970 at National Institute of Standards and Technology (NIST) in Gaithersburg, MD, USA. Some other existing millisecond apparatuses were reported by Petrova [13] at IVTAN, Moscow; Kaschnitz and Reiter [14] (2002) in Leoben, Austria; Maglib [15] (2003) in Belgrade, Serbia; Righini et al. [16] (2000) in Torino, Italy; Matsumoto and Ono [17] (2001) in Tsukuba, Japan; and most recently Dai et al. [IS] and Fan et al. [1 1] (2003), Harbin, China. Formerly known as IMGC.

Microsecond experiments are the most common resistive self-heating devices and have been constantly improved over the last 30 years. Because of the significantly higher heating rates (compared to millisecond setups), microsecond experiments allow access to a seamless temperature range from about 2000 K to about 10 000 K, thus gaining access to the liquid state of material under investigation. A short experimental duration of typically < 100 ps is enough to maintain the samples' initial geometry even after the material is already liquid, allowing investigations in the liquid state. A wire-shaped specimen will be observed as a free standing liquid column for the experimental duration, before slower gravitational forces disintegrate the sample5. Another advantage of this intermediate heating rate (of all pulse calorimetric techniques) is that the current pulses are typically t < 150 ps. Therefore, these pulses are still slow enough to keep the sample near a thermodynamic equilibrium (which can be seen by comparing pulse measurements with results from steady-state techniques [19]), to guarantee uniform heating of the entire sample, and to keep contributions from skin-effects negligibly small. When microsecond pulse-heating became popular, the main interest was in liquid refractory metals such as Nb, Mo, Re, Ta, and W. As pyrometry advanced over time and technological interests changed, the focus shifted more and more towards lower-melting metals and alloys, such as noble metals [20] or titaniumaluminides [21,221. Known existing experiments: The first users of a microsecond device for reporting the thennophysical properties of liquid metals were Lebedev et al. in 1971 [9] from the Russian Institute of High Temperature. Similar experiments have been constructed and used throughout the world by different groups, namely: Martynyuk and Gerrero 1972 [23] at Patrice Lumumba University in Moscow, Russia, Henry et al. 1972 [24] and later continued by Gathers et al. [25] at Lawrence Livermore National Laboratory, USA, and U. Seydel et al. [26] at the University in Kiel, Germany. Later, in the 1980s, a few new experiments were set up in the USA and in Europe, i.e., by Hixson et al. (1986) [27] at Los Alamos National Laboratory, USA, by Berthaud et al. (1986) [28] at Commisariat 21 1'Energie Atomique CEA, France, by Kaschnitz et al. (1992) [29] at Graz University of Technology (TUG), Austria, and by Sheindlin et. al. (1987) [30] at IVTAN, Moscow. Techniques with submicrosecond resolution were the eponyms for the earlier name 'wire explosion experiments' (see [7]). Samples are heated to the end of the stable liquid state on a timescale of a few microseconds or less (hence the AS a result, samples can only be used once, whereas specimens in millisecond systems can be used several times, which is of particular interest for metrology, when tests for reproducibility are of great interest.

name) and undergo rapid evaporation (the sample literally 'explodes'6) usually at the boiling point, but at the latest when passing the spinodal line. Besides allowing access to extremely high temperatures of 10 000 K and above, submicrosecond setups can be used for pressure generation up to several GPa, although the experiments are operated at atmospheric pressure. Such an experiment can be used to explore the critical points of materials of interest as proposed in the 1970s by Martynyuk [31] (1974) and again in 1997, but not pursued until 2002 by Rakhel et al. [32]. Extremely fast heating imposes the biggest limitation on this technique, because the energy can be unevenly deposited in the sample due to strong contributions from the skin-effect. The inability to balance pressure differences on a microsecond timescale commonly leads to strong pressure gradients within the sample and the strong electrical forces can cause severe electrical and hydrodynamic instabilities. Before submicrosecond setups were found to be of use for measurements of thermophysical properties, they were used to provide pulsed light sources with intense light output, or as plasma sources with unusually high material density [71. Known existing experiments: Although not all of the submicrosecond experiments are still operational, such setups and the results thereof have been reported in the literature by Seydel et al. [26] at the University of Kiel, Germany which were later continued by Gallob et al. (1986) [33] at TUG, Austria; by Savvatimskii (1996) [34] at IVTAN, Moscow, Russia; by Kloss et al. (1996) [35] at the University Greifswald, Germany; by Kuskova et al. (1998) [36] at the Institute of Pulse Research and Engineering, Ukraine; by Korobenko and Rakhel (1999) [37] at the High Density Research Centre, Moscow, Russia; and by DeSilva and Katsouros (1999) [38] at the University of Maryland, USA. 3. EXPERIMENTAL DESCRIPTION 3.1. General information about pulse-heating As mentioned before, a dynamic calorimetric pulse experiment uses resistive self-heating of an electrical conductor - typically wire-shaped samples (the diameters ranging from a few hundred micrometers up to a few millimeters), rectangular shaped7 samples, foils, or tubes - by passing a large current pulse The solid-liquid and liquid-gaseous phase transitions are accompanied by a spontaneous volume increase leading to the 'explosion' of the sample at the boiling point, due to internal yressure of the expanding gas. Rectangular shaped samples have to be used if the material cannot be drawn into wires and test samples have to be cut (e.g., by electron beam) from already cut thin plates or blocks of raw material.

over the sample. As a result of its resistivity, the test specimen can be heated to its boiling point in a fraction of a second. Independently of the actual timescale of the experiment (see the 'Introduction') the following parts are common for all pulse experiments: energy storage with charging unit, main switching unit (i.e., high-voltage mercury vapour ignition tubes), experimental chamber with windows for optical diagnostics and the ability to maintain a controlled ambient atmosphere8, and data recording equipment. Besides some modifications within the experimental chamber, the biggest difference between the setups can be found within the manner of energy storage. For slow (millisecond) experiments, the energy is typically stored within a battery bank, whereas capacitor banks are typically used for faster experiments (micro- and sub-micro-second).

SHIELDED ROOM

Figure 1. Schematic arrangements of components in a pulse-heating experiment. The quantities typically recorded and/or accessible (as a function of experimental duration, t ) during such an experiment are: - Current through the sample, l(t) - Voltage drop across the sample, U(t) - Temperature: either by contact thermometry (millisecond experiments only) or surface radiation for optical thermometry, J(t) - Optical quantities: thermal expansion of the sample, r(t), sound velocity measurements, c,, polarization information for optical properties measurements, So-S3(all optional) Pulse-heating experiments are commonly conducted in an inert ambient atmosphere, e.g., nitrogen or argon at ambient pressure or in vacuum.

Based upon the initially measured quantities, all thermophysical properties (with the exception of viscosity and surface tension) listed at the beginning of this chapter can be determined assuming that the mass, m, or the combination of density and the volume (to calculate the mass therefrom) of the sample at room temperature is known. Depending on the specific details of each pulse experiment, the routines to analyze the measured data and to obtain thermophysical properties may be slightly different and manifold. A s a result, we decided to present the microsecond experiment in the Subsecond Thermophysics Laboratory at Graz University of Technology in more detail. This setup qualifies as an example, as quite a few of the other worldwide existing pulse-calorimetric systems are closely related and have similarities to the setup in Graz. Most of the thermophysical properties given in the Introduction can now be calculated, based upon the initially measured quantities: current, voltage drop, surface radiation and other optional optical measurements. A flow diagram showing how the thermophysical properties are related to each other is given in Figure 2. The mathematical formulae for the actual calculations are summarized in the following compilation and are explained thereafter: property formula UC

H

d~(t) uc( t )= u ( t )- Lxy .dt

description/comment Voltage correction for inductive components due to self inductance, L,& of the circuit

1 ' AH = H ( T ) - ~ ( 2 9 8=)- S l ( t ) . ~ ( t )dt.

Specific enthalpy referenced to 298 K

mb

Difference between H a t endbeginning of melting transition Isobaric heat capacity derived from the temperature variation of the enthalpy Isochoric heat capacity derived using cp, the isobaric expansion coefficient, g, and the sound velocity, cs Time propagation of the sample volume based on the change in diameter Time propagation of sample density based on the initial density Resistivity with initial geometry (IG), assuming cylindrical sample geometry

L.T

p,o

LE

A'(T) =

a

L(T) L.T a ( ~ =) A = cp .d ( ~ ) c p .P,G .

Tm Tfl~

T(t) = c2 .[A. ln[s(t). [e&

-*]+

n

k

-1

Resistivity considering volume expansion (VE) Estimate for thermal conductivity based on the Wiedemann-Franz relation with the Lorenz number, L Estimated thermal difisivity based on the thermal conductivity or the electrical resistivity Modified Planck's law using the emittance T,, to optically and the obtain radiance the temperature temperature, Optical constants determined from ellipsometric parameters (p = tan(yr).exp(iA)) and the angle of incidence, O Normal spectral emittance from optical constants

Eh

da

PC

V,

da

Tc

CS

6r,

6ir(8,p,t) cc Y, (8, p). cos(m,t). e-';I

Y

y=-

5'

3m .m,Z 32n 3m.T V=- 207r.r

Hemispherical emittance is estimated from the equilibrium between energy input and emitted radiation Critical parameters are directly obtained from vressure measurements. expansion, and electrical signals Sound speed consists of longitudinal (I) and shear (t) components (shear components can be neglected in the Radius oscillations of a levitated sample by spherical harmonics Y, which are function of angles (8,p), time, angular frequency, m, and damping, r Surface tension of a levitated, spherical sample of radius r and mass m Viscosity of the levitated sample

308

Base Quantities (measured)

Calculated Quantities derived from Base Quantities

U(t)

H(T)

Voltage-Drop

H(t)

Enthalpy vs. Temperature

Enthalpy

I(t) Current Intensity

rIG(t) φ(t)

V V0(t)

J(t)

Volume Expansion

Surface Radiation

n(t) S0

S1

S2

S3

4 Independent Stokes-Vectors

Corrected Electrical Resistivity vs. Temperature

T(t) k(t) Optical Constants

Temperature

e(t) Normal Spectral Emittance

Specific Heat Capacity

a(T)

Uncorrected Electrical Resistivity

Specimen Diameter

cp

rVE(T)

rVE(H)

Thermal Diffusivity vs. Temperature

Corrected Electrical Resistivity vs. Enthalpy l E(T)

V (H) V0

d(H)

Volume Expansion Density vs. Enthalpy vs. Enthalpy

V (T) V0

Thermal Conductivity vs. Temperature

d(T)

Volume Expansion Density vs. Temperature vs. Temperature

Grüneisen Gamma Specific Heat Ratio vs. Density vs. Density

g G(d)

g (d)

cs(T,d)

Longitudinal Sound Velocity

Critical Pressure

KT(d)

KS(d)

PCRIT

Isothermal Compressibility vs. Density

Adiabatic Compressibility vs. Density

pSTAT

Static Ambient Pressure

r(t)

Specimen Radius

dr(T) Radius Oscillations vs. Angles (J,j) and Time

TCRIT

VCRIT

Critical Temperature

Critical Volume

g (T) Surface Tension vs. Temperature

m

h (T)

Specimen Mass

Viscosity vs. Temperature

J(t)

T(t)

Surface Radiation

Temperature

Pulse - Heating (typical)

Levitation (typical)

Figure 2. Pulse-heating properties: Flow diagram to identify how the different properties are related to each other. 3.2. Experiment - Basic electrical quantities As mentioned earlier, the basic electrical quantities, current and voltage drop, are measured directly during a pulse experiment. With respect to the timescale and the resulting electrical pulses during the discharge, different precautions have to be made to avoid stray pick-up. Therefore, all signal lines are

constructed in a double-coaxial fashion with a coaxial cable shielded by an additional copper tube. All measurement devices (amplifiers, voltage dividers, pyrometers, etc.) are isolated so that electrically active pick-up loops are avoided and the data acquisition equipment is kept in an electrically shielded room (Faraday cage). To facilitate data recording and storage, oscilloscopes with storage capabilities have been replaced by personal computers with AID conversion-based data acquisition cards to record all transient measurement signals. Because accurate knowledge of the timescale (experimental duration) during all dynamic pulse-techniques is essential, great care has to be taken when it comes to the starting and terminating of an experiment. Different switches (e.g., solid state switches, mercury vapour ignition tubes (Ignitrons), etc.) are used and operated by a common trigger signal, which also starts the data acquisition process. The electrical current in such an experiment cannot be measured directly but a voltage signal proportional to the current is obtainable, either by measuring the voltage drop at a shunt-resistor of known resistance, or by an inductance coil (Rogowski coil [39]) wrapped around the straight input conductor. The former adds an additional resistance in series to the discharge circuit, is suceptible to stray pick-up, and its resistance has to be well known. The latter needs to be connected to an electrical integrator circuit in order to provide an output signal that is proportional to current and may also need to be separately shielded from stray pick-up. Due to their easier handling, inductive coils are commonly used, especially for ps and sub-ps experiments. Each current measurement technique used for dynamic pulse calorimetry needs to be calibrated separately and tested for short rise-times and linearity to currents of several kA. To obtain the voltage during a pulse-heating experiment, either the total voltage drop of the entire test specimen or the voltage drop at a known fraction of the sample can be measured. Although the total voltage drop is experimentally easier to access, its limitations outbalance this advantage, as the resistivity at the contacts between clamps and sample and the actual length of the sample are unknown. This is usually overcome by using two separate voltage probes along the sample whose distance can be measured before the actual experiment. The voltage probes are usually made of a thin wire or a thin strip of metal and tightly wrapped around or pressed onto the sample. A contribution arising from contact resistances can still not be eliminated by this technique, but it can be expected to be small, because both voltage probes are made of the same material and are mounted identically (which ideally leads to the same contact contribution for both probes). During an experiment, the signals from both voltage probes are recorded (against ground potential) and the difference is calculated afterwards to obtain the voltage drop across the known

sample distance. Assuming the same conditions for both contacts, the contact resistance is also eliminated by calculating this difference. In general, the measured voltage signal consists of three contributing terms: the voltage drop itself, an inductive component due to the self inductance of the circuit, and the change in the sample inductance due to expansion. Whereas the third term is negligible, the self inductive component needs to be evaluated and subtracted [29].

3.3. Experiment - Derived thermophysical properties

Specific enthalpy, heat of fusion (H, M) Definition: The total internal and external energy content of a system. [4] SI based unit: J kg-' or J mol-' Enthalpy is calculated by time-integration of the voltage and current assuming that the experiment is almost isobaric. The result is then divided by the mass of the sample and referenced to 298 K (25 "C) to yield the specific enthalpy. The absolute difference in enthalpy at the beginninglend of the melting transition yields the latent heat of fusion or (seen fiom the opposite side) heat of solidification, a very important property if the amount of energy necessary for melting of a material or the amount of energy deposited into a coolant during solidification has to be known. Literature data: [l] Thermophysical Properties for solid materials, [40] SGTE Data for Pure Elements.

(Specific) Isobaric and isochoric heat capacity (cpand cJ Definition: The amount of heat required to raise the temperature by one degree per unit mass of a material, system, or entity. [4 11 SI based unit: J kg*' K-' or ~mol-'K-' The isobaric heat capacity is obtained mathematically by differentiation of the specific enthalpy. Graphically, cp is given by the slope of the enthalpy versus temperature trace leading to an individually constant number for isobaric heat capacity in the liquid state and also for the solid state, if no phase transitions occur. This result for the solid state is in contradiction with measurement data obtained by techniques such as differential scanning calorimetry (DSC) but has been proved to be correct for the liquid state. Dynamic pulse-heating rates are usually too fast and small phase transitions, which are detectable by DSC or similar instruments, become undetectable. As a result, pulse-heating data for cp

in the solid may only be considered as an estimate at the very end of the solid state before the melting transition. Isochoric heat capacity is somewhat more complicated to calculate than .c, As has been shown for pulse-heating experiments [42-441, c, may be obtained if ,c, the isobaric expansion coefficient and the speed of sound of the test material are known. Literature data: [45] solid metallic elements and alloys, [46] solid and liquid elements, [15] group VA elements.

Volume expansion and/or density Definition of expansion: The characteristic of most metals to increase dimensions as they increase in temperature. [4] Definition of density: The mass of unit volume of material. [4] SI based unit (expansion): K-' or %. SI based unit (density): kg m" The volume of the sample is of great interest for thermophysical properties, because - besides being an important property itself - volume and its thermal expansion are needed for calculating electrical resistivity, thermal conductivity and thermal difhsivity. Once the temperature dependence of the volume of a material is known, the density can be obtained from that information in a straightforward manner, assuming that the density at room temperature (RT) is known. In general, different techniques for volume-, expansion-, and densitymeasurement have been developed over time. The highest accuracy is obtained with push-rod dilatometers, which have been developed by different companies and are commercially available. Newer devices are also suited for measurements of molten materials and up to a temperature of about 3000 K. Most of the available literature data for the solid state has been measured using such dilatometers, e.g., [47]. Another measurement technique used for density measurements on molten materials is the maximum bubble-pressure method. More information about this technique can be found in [48]. For expansion measurements in a pulse calorimeter, two other groups of measurement techniques have been developed: interferometric techniques and optical imaging techniques. Different interferometers have been developed by INRiM in Italy to measure the longitudinal expansion of a sample with a laser interferometer [49,50] and by NIST in USA to record thermal expansion using a modified Double-Michelson-Interferometer [5 11. A expansion measurement setup using a similar device to NIST was recently built at the Austrian Foundry

Institute [52]. Although interferometric techniques may lead to higher measurement accuracy, the instruments are mostly restricted to the solid state and to millisecond experiments9. The second group of optical imaging methods includes shadowgraph techniques [33], Schlieren photography using rotating mirrors [53], photography using Kerr cells as shutters [54], streak cameras [55,22], and fast framing CCDcameras [56, 571. With the development of fast electronics and fast CCD-based camera systems, the first three mentioned techniques have become obsolete. All the techniques mentioned for determining thermal expansion under pulse heating conditions do not image the entire specimen during an experiment, but are limited to a small spatial fraction thereof, yielding only the time propagation of the sample radiustdiameter. Because mounting constraints limit the possible degrees of sample expansion, it is usually assumed that expansion in the solid state occurs in both radial and axial directions but only in radial direction during the liquid state [25, 58, 591. Controversial statements about the existence of axial expansion during the melting transition exist, see [59] and[58]. The validity of these assumptions and the uniformity of the expansion were experimentally confirmed [60, 241 and the contribution of the axial expansion during the solid state (and possibly during the melting transition) is commonly neglected as being only a small overall contribution. Therefore, the volume is proportional to the square of the expanded diameter. Independent experiments at the microsecond facility at TUG were able to confirm these assumptions, which can be seen from subfigures (c) and (d) of Figure 3. For further data evaluations, the volume ration of V(t)/Vo has to be calculated. Pure metals without any major phase transitions in the solid state show an almost linear or slowly increasing volume expansion up to the melting point10, undergo a sudden volume increase during the melting transition due to the solidliquid phase transition, and continue to increase their volume during the liquid state". As rules-of-thumbI2, volume increases of 2-4 % during the solid state and 3-5% during the melting transition are often observed. Literature data: [47] solid metals and alloys, [48]: liquid metals, [64] review of measurement techniques. The limitations are mainly caused by the loss in geometrical integrity of the sample at melting and in the liquid state and the duration of the experiment, because higher heating rates can decrease the accuracy and they obstruct the full formation of phase transitions. lo This does not hold for all materials, because solid state phase transitions forcing a decrease in volume do exist in some materials, e.g., plutonium. " Because the mass of the sample is constant during an experiment, the opposite behaviour is detected for the density. l2 Applies to polycrystalline or cubic materials only.

Electrical resistivity Definition: Resistance to the passage of electricity. It is the reciprocal of conduction measured. [4] SI based unit: pi2 cm (!).

Electrical resistivity comes in two different 'flavours': resistivity at initial geometry (subscript IG, pIG),and resistivity including thermal volume expansion (subscript VE, pVE).The former can be directly determined from the electric signals by rationing the voltage and the current, whereas information about the volume expansion is needed to obtain the latter, by multiplying V/Vo and PIG. Physically, electrical resistivity with volume expansion considered is of greater interest because it can be directly compared with data from other measurement techniques, e.g., 4-point measurements. Due to a lack of reliable expansion data at elevated temperatures in the solid and especially in the liquid states, resistivity at initial geometry is often reported in the literature for pulse-heating measurements. Literature data: [65] precious metals, [66] selected elements.

Figure 3. Expansion and density: Typical thermal expansion and density measurements with a pulse calorimeter. a) Volume expansion of iridium from literature sources. Solid (solid circles): Wimber [61], liquid (solid trianglesmeasurement, open triangles-extrapolation): Ishikawa et al. [62]. Density values are calculated from expansion assuming a density of 22560 kg-m" [63] for iridium (open circles: solid state; stars: liquid state). b) Thermal expansion experiment with a fast framing CCD-camera at TUG. A small section of the sample is imaged every 10.13 ps (vertical lines separate different images) during the experiment; # 1 - sample at RT, # 6 - sample after 5 x 10.13 ms = 50.65ms = end of the stable liquid phase. c) CCD image of a pulse-heated copper sample in the liquid state shortly after the melting transition. Sample has been heated too slowly and undergoes longitudinal expansion and thus deformation. d) A similar experiment to that in c), but with correct heating rate. Note that the liquid

copper has maintained its original vertical position and has only expanded radially. Thermal conductivity Definition: The measure of the ability of a material to transfer heat. [4] SI based unit: W m-' K-'. There is a close relationship between electrical and thermal conductivity. From the simple free-electron model for metals, the ratio of the thermal conductivity and the electrical conductivity (reciprocal of resistivity) for metals is directly proportional to the temperature. This is called the Wiedemann-Franz-Lorenz (WFL) relation and the constant of proportionality yields the theoretical (Sommerfeld) Lorenz number, L = n2/3.(kBle)2= 2.45 x lo-' W 0 K-' [67], which was predicted to be independent of temperature (for temperatures significantly larger than the Debye temperature) and of the material. Assuming a known and/or constant value of L, the WFL relation can be used to obtain the thermal conductivity from pulse-heating data. To examine the validity of the WFL relation, values of the ratio of the experimental to theoretical values for the Lorenz hnction (L/Lo), for temperatures close to the melting point, were compared and L/Lo was found to be close to unity for most metals and the small deviations may be due to measurement uncertainties [68]. At lower temperatures, and for some metals, significant deviations from the theoretical Lorenz number were found and attempts to modify the WFL relation were made [69, 701. Even larger discrepancies occurred when using the WFL relation for alloys, because electron-electron interactions, electron-phonon interactions, as well as lattice contributions, need to be considered [71]. These limiting effects vanish at melting because the crystal structure is destroyed and the WFL relation becomes a reasonable tool for determining thermal conductivities for liquid metals and alloys. Literature data: [68] elements, [71] metals and alloys. Thermal diffusivity Definition: A quantity of which indicates the ability of a material to equalize temperature differences occurring within it. [4 11 SI based unit: m2s-' Closely related to thermal conductivity, diffusivity can be calculated based on the thermal conductivity (thus also relying on the validity of the WFL relation), the density, and the isobaric heat capacity. Due to their individual uncertainties,

thermal diffusivity usually has the largest uncertainty of all the pulse-heating properties. Besides pulse calorimetry, the laser flash (LF) technique is an established tool for directly measuring the thermal diffusivity by penetrating a sample with a short laser pulse on one side and monitoring the change in temperature on the opposite site. More details about pulse methods for diffusivity measurements can be found in, e.g., [72]. Literature data: [73] solid metals and alloys. 1741 metals and alloys (solid and around melting transition). Temperature Definition: The degree of hotness or coldness. [41] A measure of the average (inordinate) kinetic energy of the particles in a sample of matter, expressed in terms of units or degrees designated on a standard scale. SI based unit: K Temperature is undeniably the most important property for all calorimetric measurements, because it is the common denominator. Two different techniques for temperature measurements are used for pulse calorimetry: contact thermometry (e.g. thermocouples) and radiation thermometry or pyrometry. Because pulse calorimetry is often used to handle and measure liquid materials, non-contact radiation thermometry is far more common in pulse-heating than contact thermometry. Other reasons for non-contact temperature measurement methods include the fast heating rates and temperature gradients (inertia of the thermocouples), difficulties mounting the contact thermometers (good thermal contact needed), and stray pick-up in the thermocouple signal because the sample is electrically self-heated. Contact thermometry is successfully used for experiments in the millisecond to second range, [75], where tiny t h e r m o c ~ u ~ l eare s ' ~ spot-welded to the sample and the additional voltage drop between the thermoelectrodes and the thermocouple junction is compensated for. The majority of pulse calorimetric measurements use pyrometry, which is noncontact (optical) measurement of the thermal radiation emitted from any heated body or substance according to Planck's radiation law for black body radiation. Planck's law describes the spectral distribution of black body radiance which provides the basis for the International Temperature Scale (ITS-90) [76], especially above the freezing point of silver [77]. Because Planck's law is only

l 3 To

avoid physical and thermal perturbations caused by the sensor.

valid for black bodies14 which cannot be realized in practice, the (spectral) emittance15has to be introduced, defined as the ratio of the spectral (self-exitant) radiance to that of a black body, both at the same temperature and viewed in an identical manner. For a freely radiating target, Wien's law (applicable in many cases of practical interest, where 1.T I 2898 pm K ) :

relates the true temperature and spectral radiance (or radiation) temperatures, T and T,, respectively, from knowledge of the spectral emittance, EL, and the mean effective wavelength [76]. From the definition, the radiance (or radiation) temperature is the temperature of a black body having the same specific emission as the radiating body in a given range of wavelengths. Thus, T, is always smaller than the true temperature T. Also mentioned in the above definition, the very complex and complicated concept of the mean effective wavelength deals with accurate calibration of pyrometers and the effective wavelength of the optical system that has to be used therein. Because this topic requires far more background information, it is mentioned but will not be explained any further within this chapter. For details and implementation see [79, 801. Based on these fundamental considerations, high-speed pyrometry for use in pulse calorimetry has been developed and considerably improved over the last decades, assisted and driven by metrological institutions such as NIST and INRiM [lo] among others. One of the main issues is the need to accurately calibrate the device to convert the electrical input signal from the detector (e.g., photodiode or photomultiplier tube) to (radiance) temperature. In a second step T, is converted into the true temperature T of the sample, given the emittance at the corresponding wavelength and temperature are known. Various reference sources like black-body radiators, fixed-point cells, tungsten ribbon lamps, etc., are used for appropriate calibration of pyrometers. More difficult has proved to be the knowledge of emittance and its temperature- and wavelength-dependent behaviour . Different techniques and approximations have been developed and used in pulse calorimetry to overcome these limitations. Special tubular samples with small holes simulating a near-black body cavity, together with a scanning pyrometer, were used at NIST for millisecond l4 One which absorbs completely any heat or light radiation reaching it and reflects none. It remains in equilibrium with the radiation reaching and leaving it, and at a given steady temperature emits radiation (black body radiation) with a flux density and spectral energy distribution which are characteristic of that temperature and is described by Planck's radiation formula 1411. l5 IUPAC recommends the term 'emittance' instead of the widely-used 'emissivity' because the latter term is often reserved for radiant power divided by volume and by solid angle [78].

experiments. The spatial scanning pyrometer was used to measure spectral radiance temperatures along a 25 mm long portion of the sample, which also included a small rectangular hole approximating a black body cavity. Besides the advantage of the true specimen temperature, the measurement of spectral radiance temperature of the specimen surface as well as the true specimen temperature enable determination of the normal spectral emittance of the surface by ratioing the two radiances according to Planck's law [81]. The same method has later been implemented at OGI, Austria [82] and demonstrated most recently the limitation of this black-body cavity by using finite-element simulation methods [83]. Another approach was chosen at INRiM, using an integrating sphere reflectometry technique in which the sample reflectivity was measured relative to a reference material of known reflectivity. By applying Kirchoff s law to the result, the normal spectral emittance can be determined [84, 101. Both the NIST and INRiM methods are limited to the solid state due to collapse of the black-body hole for the former, and the contamination of the sphere reflectometer and the reference sample for the latter method. A recent comparison and evaluation of both methods can be found in [85]. For measurements using molten materials, other techniques to estimate the emittance and thus to retrieve the temperature have to be used. One of the approaches is multi-wavelength pyrometry in which several radiance signals at different wavelengths, from the same sample, are recorded simultaneously. This method once again splits into two subgroups: the wavelengths of the different channels are either separated by only a few nanometers and a wavelengthindependent emittance is assumed for the wavelength range covering the channels for evaluation (e.g., two wavelength pyrometry [86]), or a mathematical model (e.g., linear dependence) for the wavelength dependence of emittance is assumed and the temperature is retrieved by modelling the signals [87-891. A recent development combines reflectometry with a multi-wavelength pyrometry but, once again, is limited to the solid state due to the use of an integrating sphere [l 11. A common approach for high temperature pyrometry of liquid materials includes 'self-calibration' of the pyrometer used at one given measurement point of known temperature and deduction of all other temperatures using an assumption of the emittance behaviour , e.g., in the molten state [90, 21. Typically, the temperature at melting is used as the point of self-calibration16,

l6 In principle, every clearly visible point of known temperature, including phase transitions, boiling points, etc., could be used as self-calibration points from the recorded radiance trace, but recorded phase transition temperatures in pulse calorimetry (microsecond and sub-ps experiments used to obtain liquid state data) tend towards higher temperatures due to the high

because the melting transition usually is clearly indicated by a plateau in the radiance-versus-time trace and accurate figures for melting temperatures are available, e.g., [91, 771. Using this approach, the temperature is calculated from the ratio of the recorded sample radiance at the self-calibration point, J,, and the radiance at any given time, J(t), by

where the melting temperature (subscript m) was assumed to be the point of self-calibration. The limitation of the unknown emittance behaviour still remains. The common assumption that emittance changes at the melting tran~ition'~,but remains constant at the same (explicitly unknown) value thereafier throughout the liquid state, permits E ~ ( ~ ) / E ~to, , be set = 1 and provides a simple way of calculating the sample temperature (at least for the liquid state). This approach cannot be used if no fixed-point temperature is visible (e.g., if the melting point is not recorded) or is somewhat limited and leads to higher uncertainties if (a) temperatures in the solid are calculated from recorded solidstate radiances (emittances in the solid are different) and if (b) emittance is not constant in the liquid state (see emittance section). The most accurate temperature for pulse-heated liquid samples can be achieved if the emittance is known, e.g., measured with ellipsometry (see emittance section) in the temperature range of interest. Provided that the emittance is known, the output of a calibrated pyrometer can be converted to temperature using a modification of Planck's law as listed in the formulae compilation. Although this approach is considered to be the most accurate one, it is often substituted by one of the other techniques mentioned before, because only a few feasible techniques and facilities for emittance measurements in the liquid state exist, only little changes in emittance can be observed [92,93], or the measured emittance only leads to a small change in temperature which is well inside the uncertainty limits of the self-calibrated temperature (see Figure 4).

heating rates and the short experimental duration in pulse calorimetry and boiling points are mostly undetectable due to the so-called phase explosion as the sample becomes gaseous. " As the sample becomes liquid at the melting transition, the rough surface smoothens due to the surface tension and therefore changes its reflectivity/emissivity.

Figure 4. Temperatures: (a) Comparison of the three temperatures (bottom to top), radiance temperature, T,, self-calibrated temperature (with an uncertainty of 4%, k = 2), T,,& and the temperature using the measured emittance, T, for the same experiment on cobalt. The beginning and end of the melting transition are marked by vertical dashed lines. (b) Enthalpy of tantalum scaled with two different temperatures: self-calibrated temperature (using the melting temperature and assuming a constant emittance in the liquid state) and temperature using directly measured emittance. Differences in the liquid state are due to the non-constant emittance (with respect to melting point). For more information on temperature measurements, a general review article concerning high-temperature thermometry (not necessarily limited to use in fast dynamic pulse techniques) is given in [94]. Literature data: [77] definition of the temperature scale, [76] radiation thermometry. Optical constants and emittance Definition (optical constants): The refractive index, n, and the extinction coefficient, k, which together determine the complex refractive index N = n - ik of an absorbing medium. [4 11 Definition (emittance): The ratio of the emissive power of a sample to that of a black body at the same temperature. [41] SI based unit (optical constants): dimensionless SI based unit (emittance): dimensionless

Although ellipsometry is one of the standard measurement methods for emittance, application to pulse-heating seemed at first difficult, because no rotating device^'^ can be used, due to the short timescale. In the 1980s Azzam [96, 971 published the concept of a pure optical-based ellipsometer, a so-called Division-of-Amplitude Photopolarimeter (DOAP), without any moving devices, based on the Stokes-Formalism for polarized light. Adopting this concept, Krishnan [98] constructed an instrument for use with levitation experiments and pulse-heating techniques, which later became commercially available. In conformity with all ellipsometers, a DOAP detects the change in polarization of an initially polarized laser beam (of well-defined and known polarization) after reflection from the surface of interest, by dividing the total signal into four polarization components. After an elaborate calibration routine [99], the optical constants of the sample were obtained from the Fresnel equations, provided that the angle of incidence and the refractive index of the ambient medium, e.g. air, were known. The whole instrument was originally designed to measure emittance and the optical parameters result as a by-product from this process, see e.g., [loo] obtained with a ps-DOAP. Once the optical parameters are known, the spectral reflectivity can directly calculated, which, in turn, can be used to determine the normal spectral emittance19 for opaque materials (transmittance = 0) by using Kirchhoff s law of thermal radiation. All of the properties (optical constants, reflectivity, and emittance) obtained with this technique are measured at one given wavelength only, the wavelength of the laser beam. The laser wavelength has to match the wavelength of the pyrometer used for temperature measurements as closely as possible, in order to minimize uncertainties arising from wavelength mismatch. Because the emittance of a solid sample is mainly influenced by its surface roughness, structure and possibly obstructing surface layers, the DOAP techniques work best at the melting transition and in the liquid state. At melting, the surface of the sample smoothens due to the surface tension of the liquid material and forms an almost perfectly smooth and reproducible surface. This behaviour is, in general, observed as a sudden decrease2' in emittance at melting because smoother surfaces tend to have higher reflectivities and thus lower emittances. In the solid state, surface conditions (and thus emittance) are subject

l 8 Common ellipsometers consist of a polarizer setting the polarization of the incident beam and an moveable analyzer (a linear polarizer and a quarter-wave plate, both on rotation stages and adjustable with respect to each other) to analyze the reflected beam [95]. l9 The identifier 'normal spectral' emissivity indicates emittance 'normal' to the surface and at a single wavelength (spectral). 20 For some materials, e.g., platinum, the opposite behaviour was observed and reported [92].

to sample preparation and pre-experimental treatments2' and can usually not be reproduced. The data recorded in the solid state have informative character, but are almost irreproducible. Recent investigations of the behaviour of emittance in the liquid state after the melting transition, have discovered three different cases: increasing, constant, and decreasing emittances [92, 931. Although this finding is contrary to the former assumption of a constant emittance in the liquid state to enable temperature evaluation, calculations show, that the small contributions caused by changing emittances are negligible for most materials. Independent of all the efforts to implement ellipsometry to pulse-heating systems, a second branch of measurements has been jointly developed at NRLM-NMIJ [17] and NIST [ l o l l to obtain the total hemispherical emittance. The principle of this technique is to interrupt the heating process of a pulse calorimeter and to create a short steady-state temperature condition by using the pyrometer as a feedback device. The energy input to maintain this steady-state equals the total radiative losses and thus can be related to the total hemispherical emittance, assuming that no convection occurs during the short steady-state time. Literature data: [102] optical constants for numerous materials, [103], [104] thermal radiative properties of solids, [lo51 NMI intercomparison.

Critical parameters Definition: The point of a fluid, characterized by pressure and temperature, at which the gaseous and liquid phases become identical and form only one phase. [411 SI based unit (pressure): Pa = N m-2 SI based unit (temperature): K SI based unit (volume): m3 A pulse-heating system can also be used to obtain critical parameters, if the system is able to handle high ambient pressures up to some lo3 bar and is equipped with a means of expansion measurement. Such an experiment was designed and built at TUG allowing pressures up to 7 kbar by using water as the pressurized ambient medium [106, 1071. The sample is pulse-heated in the pressurized ambient atmosphere and the temperature (using a pyrometer) as well as the thermal expansion are recorded. When the sample reaches the liquid/gaseous phase transition (boiling point) it suddenly increases in volume (discontinuity in the liquid expansion behaviour). This effect disappears if the ambient pressure matches (or is larger than) the critical pressure of the material. Such as polishing with abrasive paper, washing with acetone, etc.

The critical volume and temperature are later obtained from the expansion data and the pyrometer measurement, respectively. Although this method is not of high accuracy, it is one of few techniques permitting access to critical parameters and results obtained are in reasonable agreement with theoretically estimated values [2]. Literature data: [I081 review of critical point measurements using pulse-heating.

Speed of sound Definition: The speed at which a sound wave propagates through a medium. [411 SI based unit: m s-' Besides the intrinsic value of speed of sound data, such measurements are of great importance if the interest is in equation-of-state (EOS) parameters such as bulk moduli, compressibilities, Griineisen parameters and specific-heat ratios [2]. All of these EOS parameters depend on techniques for the determination of the speed of sound, which have been developed and used in only a few laboratories. Different techniques can be found in literature, such as displacement interferometry [109], longitudinal run-time measurements of a sound wave fiom an impacting projectile [29], or measuring the run-time of a laser-generated stress wave across the sample [27, 110, 1111. The latter method proved to be the most stable one and is still used to obtain speed of sound data [I 121. A laser-generated stress wave is induced by focusing a pulsed laser beam on one side of the sample. The stress wave diverges spherically and degrades to a acoustic wave, which propagates through the sample and creates small surface motions and thus disturbances in the surrounding gaseous atmosphere as it reaches the opposite side of the sample. These disturbances can be detected as density waves in the ambient atmosphere, when the sample is optically monitored, e.g. with a streak camera, for expansion measurements. The streak record yields the transition time of the wave, which is further used to calculate the speed of sound from the measured diameter of the sample. This technique proved to be more stable than interferometric techniques, because arbitrary surface motions of the liquid sample can cause a displacement of the reflected laser beam used for interferometry. For more information on the speed of sound see [112]. Literature data: [2] speed of sound measurement to obtain EOS parameters, [I 121 speed of sound measurement.

3.4. Experiment - Levitation Levitation (from Latin levis, light) is the process by which an object, here a sample, is suspended against gravity, in a stable position, by a force, without physical contact. All that is required on earth for levitation is a force vertically upwards equal to the object's weight. This can be achieved by many different means, for example, electric, magnetic, acoustic, optical, electrostatic and aerodynamic forces, as well as by microgravity levitation in microgravity (parabolic flights or in space). A summary and description of different levitation techniques is given in [I131 and reviews can be found in [114, 1151. Levitation techniques also provide containerless conditions for investigations of materials, eliminating most interactions between the sample and its environment. As stated earlier, the main difference between dynamic pulseheating and levitation is the fact, that the actual energy input during a levitation experiment cannot be determined. Heating of the levitated sample is accomplished by means of induction, a laser beam, or an incandescent radiator. On the other hand, levitation is not necessarily limited to a subsecond timescale and is required to measure thermophysical properties such as viscosity and surface tension of liquid samples, which are inaccessible to pulse-heating. These measurements are possible by digital image processing and frequency analysis of induced surface waves of a levitated liquid sample.

Surface tension and viscosity Definition (surface tension): A measure of the free energy of a surface per unit area. [41] Definition (viscosity): The resistance to fluid flow, set up by shear stresses within the flowing fluid. [41] SI based unit (surface tension): N m-' SI based unit (viscosity): kg m-' s-' The radius of a levitated drop undergoes spherical harmonic oscillations, which can be detected and recorded by digital imaging and processing. The angular frequency of these oscillations can be related to the surface tension, while the damping of such oscillations is due to the viscosity of the sample material. Data evaluation is drastically simplified if the equilibrium shape of the droplet is spherical, allowing usage of the formulae of Rayleigh and Kelvin to determine the surface tension and the viscosity, provided that the mass of the sample and its radius are known. The main limitation is that a spherical shape can only be obtained under force-free conditions, a condition that can only be satisfied in microgravity but not under terrestrial conditions. Therefore corrections (e.g., see [I161 for electromagnetic levitation) have to be applied. Literature data: [I171 viscosities of gases and liquids, [I181 surface tension data.

4. EXPERIMENTAL DATA - IRIDIUM

After all the explanations of how thermophysical properties can be determined from experiments, this section is intended to visually demonstrate the capabilities of pulse calorimetry with data from real measurements. Iridium was chosen as the material of interest because the results were recently obtained and are a good example of the procedure. Wire-shaped annealed iridium samples of 99.9% purity, purchased from Alfa Aesar (reference 19587) were investigated. The specimens were 0.5185 mm in diameter (measured with a laser micrometer) and about 35 mm in activez2 length. The RT density of iridium is 22560 kg.m4 [63] and its melting and radiance temperatures at melting at 656 nm are T, = 2719 K [119, 631 and T,. = 2380 K [120], respectively. The load voltage of the capacitors was about 5.8 kV and the maximum current during the experiment after 28 ps reached 8.9 kA, the first 40 ys were of use for data analysis. After this time, the samples reached about 3600 K and caused overload of the pyrometer detector due to saturation (which can be seen as a horizontal pyrometer signal in subfigure (a) of Figure 5). The experimental signals for such an experiment with iridium are shown in Figure 5 (a), which shows time dependent traces of the current, the two voltage probes and the pyrometer output. The investigated sample reaches the melting transition after 32 ps into the experiment which is indicated by a stable 'plateau' of about 4 ys in the pyrometer trace. Besides the pyrometer signal, the beginning and end of melting can also be identified from kinks in the voltage traces because the resistance of the sample changes at the beginninglend of the melting transition. Unlike the current signal, both voltage traces do not start at zero but start at a value > 0 due to the self-inductivity of the circuit, which needs to be compensated for later during the evaluation. Although the dynamic range of the pyrometer only spans from about 27 ps to 40 psZ3,electrical signals are recorded from RT to 4000 K in about 45 ps. At this time the experiment is intentionally ended by 'crowbar-switching' the energy from the capacitors. All of the results presented are mean values of 9 individual experiments conducted with iridium. The enthalpy (Figure 5 (b)) is calculated from the electrical signals and literature data [I19 (A), 121 (V)] for comparison are also shown in the plot. The melting temperature is indicated by a vertical dashed line. All data are fitted The active length is measured between the two voltage probes and not the overall length of the wire inside the clamps. 23 Depending on the emittance of the sample and the wavelength of the pyrometer used, the lower onset temperature is typically limited to from around 1100 K to 1900 K. The upper temperature limit is once again set by the sample's radiance and its emittance at this temperature. 22

with linear least-squares fits (also shown) and, as explained earlier (see enthalpy section), the slopes of the enthalpy versus temperature traces yield the isobaric heat capacity. For iridium the heat capacity for the solid (before the onset of melting) and liquid states are found to be (almost) identical at 232 J kg-' K-', a finding that is not typical for metals. The electrical resistivities (initial geometry and including volume expansion) for iridium were calculated using the respective formulae, and volume expansion values from the literature (see Figure in volume expansion section) were used to obtain PYE. Unfortunately, the volume expansion data taken from [62] are limited to a maximum temperature of 3000 K, but show a linear behaviour, which was assumed for higher temperatures too and used to extrapolate the data. The plotted literature data can be found in [I21 (V), 122 (All. As explained earlier, thermal conductivity and diffusivity (subfigures d, e) are calculated using the WFL law, assuming the validity and temperature ~ the independence of the theoretical Lorentz number L = 2.45 x lo-*W SZ K - and isobaric heat capacity obtained. Literature data can be found in [I23 (V), 68 (All. The normal spectral emittance, as shown in subfigure f, was measured with the ps-DOAP and this is a good example of a typical emittance result obtained under pulse-heating conditions. The initial sample exhibits an arbitrary emittance, due to its rough and structured surface, which immediately starts to decrease at the onset of melting, due to the increased reflectivity of the liquid material. A slight increase in emittance is observed for iridium in the liquid state as the temperature increases. The comparison value at melting is taken from [120]. Unlike the other properties, emittance is shown as a function of radiance temperature (of the pyrometer used) at 650 nrn. A plot of emittance versus true temperature could be misleading, because emittance itself would have to be used for converting the radiance temperature to the true temperature and thus the emittance would be shown as a function of a property containing emittance itself. Further details and explanations discussing the results found for solid and molten iridium are available in [124]. Besides the graphical presentation of the data, it has proved to be helpful to also tabulate the linear least-squares fit results. The summarized fit coefficients for iridium are given in Table 1.

Table 1. Iridium data: Summarized linear least-squares fit results for iridium. Fits are given in the form a + b-T, where a & b are the fit parameters and T is temperature in K. Note: emittance is given as a function of the radiance temperature, T,.

Figure 5. Iridium: Set of thermophysical properties obtained from a single pulsecalorimetric experiment on an iridium sample: a) basic electrical quantities and the pyrometer signal as a function of experimental duration; b) specific enthalpy as a h c t i o n of temperature; c) electrical resistivity, at initial geometry and with volume expansion, as a function of temperature; d) thermal conductivity as a function of temperature; e) thermal difhsivity as a function of temperature; f)

normal spectral emittance at 684.5 nm as a function of the radiance temperature. = 2365 K. Note: T, = 2719 K but TradSm

5. RECENTLY DEVELOPED (SPECIAL) APPLICATIONS OF PULSE CALORIMETRY 5.1. Extended temperature range by a pulse-calorimeter/DSC combination One of the major limitations of optical temperature measurements with fast pyrometers in pulse calorimeters is the limited sensibility at low temperatures. The maximum of the radiance versus wavelength functions, according to Planck's law, shifts from the IR to shorter (visible) wavelengths with increasing temperatures24.Pyrometers are operated in a narrow wavelength band and are therefore limited to a certain sensitive temperature range which is determined by the wavelength of operation, the sensitivity of the detector and the emittance of the test material. Typical onset temperatures for a pyrometer at 650 nm (Sidetector) are about 1700 K and for a 1570 nm (InGaAs-detector) pyrometer at about 1200 K, respectively. A technique to bridge parts of the temperature gap between room temperature and the pyrometers' onset temperatures, for materials without any significant phase transitions in the solid state, was developed at TUG [125, 1261. The method involves a combination of differential scanning calorimetry (DSC) and pulse-heating measurements, based on the fact that the recorded electrical signals (current and voltage drop) of a pulse-heating experiment start at room temperature and only the pyrometer signal is limited in temperature. This finding can be used to present electrical resistivity as a function of enthalpy starting at room temperature. The second part of the measurement is performed with a commercial DSC, in which a sample of the same material under test is investigated, yielding the isobaric heat capacity versus temperature as the main result. Thermodynamically (see heat capacity section), heat capacity is the temperature derivative of the specific enthalpy and thus can, in turn, be used to obtain the enthalpy by integration over a temperature range. Such a correlation between enthalpy and temperature can either be used to extend the enthalpy results from pulse-heating to lower temperatures, or further to assign temperatures to the enthalpy values in the aforementioned p versus H relation. In both cases this method enables access to enthalpy and resistivity data in the DSC temperature range of operation, typically 500 K to 1500 K. However, because the heating rates are totally different for both techniques (some 10 Ws for DSC and aboutlog K for pulse 24

This behaviour can also be observed in daily life as metals start to appear 'reddish' at about

600°C.Below this temperature, the surface is heated, but still looks metal-grey to the human eye because the human eye is insensitive to IR wavelengths.

calorimetry), phase transitions in the solid state may lead to discrepancies in the results. For pure metals, e.g. platinum, iron, and nickel, this technique proved capable of almost closing the gap between 500 K and the onset temperature of the pyrometer, and the enthalpy results from both techniques are in good agreement, within the stated uncertainties. However, data in the solid state obtained with millisecond experiments are more accurate than data fiom this combined ps/DSC-experiment. Nevertheless, this combination was designed for using the existing instrumentation at TUG.

Figure 6. The DSCIpulse-heating combination: Schematic flow diagram to demonstrate how the different properties are related to each other, using a DSCIpulse-heating combination technique. 5.2. Mechanical properties with a Kolsky bar apparatus To extend pulse calorimetry to selected mechanical properties, NIST has developed a new apparatus called the pulse-heated Kolsky bar, or splitHopkinson pressure bar apparatus [127, 1281. The combination of the two techniques produces high-rate dynamic loading, while simultaneously pulse heating the specimen with electrical current. The Kolsky bar test consists of holding a small sample of the material to be tested between two long steel bars. The first bar (called the incident bar) is impacted on the end with a striker bar fired fiom a small airgun. This impact

produces a strain wave that moves down the incident bar, through the sample, and then into the second bar (called the transmitted bar). The deformations in the sample material affect the strain pulse in the transmitted bar, as well as the strain pulse reflected back into the incident bar. Recorded signals from strain gauges mounted in the centre of the incident and transmitted bars are used to calculate stress-strain functions for the test material. Traditionally, the Kolsky bar test is done at room temperature, but to determine stress-strain functions for actual process modeling, high temperature material properties are needed. Therefore, the NIST Kolsky bar apparatus is integrated into the existing pulse-heating facility, where samples can either be measured in transient mode (experiment during the heating process) or a brief steady-state temperature mode (see emittance section). This setup has been developed to meet the interest in analysis of machining processes such as prediction and measurement of the rapidly changing temperatures near the cutting tool tip. Modem analytical methods, such as finite-element analysis (FEA), are used to model the machining process in an attempt to provide a predictive capability for parameters such as cutting temperature and forces. However, efforts to use the FEA method to model such machining processes have been hampered by the lack of adequate material properties for high strain rates and the rapid heating encountered in modern machining processes.

5.3. Pulse-heating1 laser flash combination A combination of two proven techniques, namely pulse-heating and laser-flash, was recently presented [129] by a group from National Institute of Advanced Industrial Science and Technology (AIST), Japan. The aim of this hybrid electrical-optical pulse-heating method is to obtain high-temperature thermal diffusivity data of conducting samples, which are essential for heat transfer calculations. Commercial laser flash (LFA) systems are limited to temperatures of typically 1800 K (maximum 2300 K), depending on the heater used, see e.g. [130, 1311~'. Although the LFA technique is most common for temperature dependent thermal diffusivity measurements, it is not only limited by the maximum achievable temperatures, but also by heating rates of about 50 Ws. The latter leads to substantial heating (and cooling) times and the overall duration of a measurement may also cause contamination and evaporation of the sample at elevated temperatures.

Certain commercial equipment, instruments, or materials are identified in this chapter to specify the experimental setup and procedure adequately. Such identification is not intended to imply recommendation or endorsement by the authors, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.

25

A brief (some 100 ms) steady-state in temperature can be achieved with AIST's feedback-controlled pulse-heating system by using the pyrometer as a feedback control and switching the input current accordingly with field-effect transistors (FET). So far, this equals the technique used for hemispherical emittance measurements (see section emittance). As a novelty, an Nd:Glass laser with a pulse duration of about 0.3 ms was added and the laser-induced temperature increase can be measured with the pyrometer as a superposition to the basic surface radiance. Some conditions need to be fulfilled in order to successhlly evaluate LFA data: the sample heating is adiabatic, the energy pulse is negligibly short, the sample surface is heated uniformly, the sample needs to be homogenous and opaque. Most conditions can be satisfied with this combined pulse-heating1LFA experiment except that a correction is needed [I321 for the adiabatic heating. Molybdenum was chosen by AIST as a test material for this technique and the first results, recently presented, for thermal difhsivity data in the range from 2000 K to 2600 K indicate an improvement to the scarce and scattered literature data. 5.4. Pulse-heating microcalorimetry An entirely new field of applications for resistive heating was discovered in the mid- 1990s by Allen et al. [I331 when they presented a heater for use in the rapid thermal annealing of thin films, which is resistively self heated and enables heating rates from lo3 K to lo6 K to be applied. The basic principle is simple but elegant: a thin-film layer of the test material is deposited onto a silicon substrate, which acts as a heater and a temperature sensor (thermopile) at the same time. To prevent the applied heater current from shunting through the thin film, an electrical isolating buffer layer of, e.g., silicon dioxide is sandwiched between the film and the substrate. A four-point probing technique is used to measure the response of the thin film to the change in temperature. Only a year after the first discovery, a first version of a high-speed (lo4 Ws) scanning microcalorimeter with monolayer sensitivity was published by the same group [134]. Other versions of such microcalorimeters were specifically designed for different measurements, including AC calorimeters capable of covering three orders of magnitude in frequency [135], scanning microcalorimeters allowing high heating and cooling rates [136], thin-film AC nanocalorimeters for low temperatures and high magnetic fields [137], or special U-shaped calorimeters to analyze samples in the sub-pg range [138]. Newer microcalorimeters have been produced by modern semiconductor microfabrication techniques to produce an extremely small addenda heat capacity, thus allowing submicro-Joule precision for ultra-thin film heat capacity measurements [139], which can be applied to a wide range of materials. Over

the last decade, this method has been successfully used for different applications such as glass-transition measurements in ultrathin films [135], in situ resistivity measurements of A1 and TiISi thin films [133], or melting and crystallization measurements of poly(buty1ene terephthalate) [140]. 6. UNCERTAINTIES

A common standard for evaluating and presenting uncertainties in measurements was set by the 'Guide to the expression of uncertainty in measurement (GUM)' compiled by the International Standardization Organization [141, 1421. According to this guide, all reported uncertainties have to be reported as expanded uncertainties with a coverage factor of k = 2 (defined as a confidence interval of twice the standard deviation for a normal or Gaussian distribution). Due to the lack of specific knowledge and details of each mentioned technique reported within this chapter, no general statement about the uncertainty can be given. Even for the pulse calorimeter at TUG, an exact uncertainty evaluation takes certain experimental parameter into account that may change for different materials. Therefore the uncertainties reported here for iridium can be seen as typical uncertainties for the ps-pulse heating experiment at TUG but may change for other specimen materials. For the thermophysical data of solid and liquid iridium, the following uncertainties should be applied: temperatures below 2400 K, 4%, temperatures above 2400 K, 1.7 %, normal spectral emittance, 6%, enthalpy in the solid state, 4%, enthalpy in the liquid state, 2.5%, heat of fusion, 8%, isobaric heat capacity, 8%, electrical resistivity at initial geometry: solid, 4%, liquid, 3.5%, electrical resistivity including volume expansion: solid and liquid, 6%, thermal conductivity, 12%, and finally thermal difisivity, 16%. For statements concerning the uncertainties of other techniques like millisecond and submicrosecond experiments, levitation, speed of sound measurement, microcalorimetry, etc., study of the original literature is advised. 6. FURTHER READING

Some literature sources for further information are mentioned, knowing that such a list will never be complete (but this is intended as a starting point). Over the years, several reviews of the research conducted with pulse-calorimeters have been published, which are good sources for more and detailed information about the historical developments and, in particular, for specific details of different pulse calorimeters. The first reviews, after the 'Exploding wires' series of books by Chace [7], were presented by Cezairliyan, one of the originators of

pulse-heating as a calorimetric measurement technique [8, 1431 and later on by Gathers [2]. Further reviews and summaries concerning pulse-heating and levitation have been compiled by Pottlacher [144, 1451 and Egry [113]. Suggestions regarding issues and future directions in subsecond research and especially pyrometry have been given by Cezairliyan et al. [146] and Cezairliyan and Righini [CezRig96]. Recent advances in dynamic heating techniques have been presented by Boivineau and Pottlacher [112] and with a focus on millisecond pulse calorimetry by Righini [147]. Summaries and descriptions of other techniques and measurement methods for thermophysical properties are given by MagliC et a1.[72, 1481 or [149]. For readers interested in collection of thermophysical properties data may want to have a look at [150-1541 for pure metals, for commercial alloys at [74], or consult the literature sources mentioned in each section. 7. CONCLUSIONS

The development of today's pulse calorimeters started more than 35 years ago, when scientists used exploding wires to generate dense plasmas, or as bright light sources, and began to realized the potential of this method for measurement of liquid-state properties. Ever since, improvements, such as better and more accurate electronics and instrumentation, or computer assisted data recording and evaluation, have helped to constantly improve and establish pulse calorimetry as a tool for temperatures inaccessible to more common techniques. The fundamental principles and relations for pulse calorimetry have been presented and discussed, as well as demonstrated by data for pure iridium in the solid and liquid states. Although it might seem that pulse-heating has already started aging a bit after such a long time of usage, the developments of new applications and devices, as demonstrated in the section on special applications, is vivid proof of the applicability and versatility of this simple but elegant calorimetric technique. 8. ACKNOWLEDGEMENTS The authors acknowledge financial support of the Osterreichische Forschungsgesellschaft mbH (FFG) and Osterreichisches Weltraumprogramm (ASAP), Vienna, Austria. The authors also acknowledge continuous support from the Subsecond Thermophysics workgroup at the Institute of Experimental Physics at TUG.

9. REFERENCES A. Goldsmith, T.E. Waterman and H.J. Hirschhorn, Handbook of Thermophyscial Properties of Solid Materials, Volume I - Elements, Pergamon Press, Oxford, 1961 G.R. Gathers, Rep. Prog. Phys., 49 (1986) 341. A. Cezairliyan, Int. J. Thermophys., 5 (1984) 177. C.D. Brown, Dictionary of Metallurgy, John Wiley & Sons, Chichester, 1998. E. Kaschnitz and G. Pottlacher, Giesserei-Praxis, 1 (2002) 1. E. Nairne, Phil. Trans. Roy. Soc. London, 64 (1774) 79. W.G. Chace and H.K. Moore, Exploding Wires, Vol. 1-4, Plenum Press, Inc., New York, 1959-1968. A. Cezairliyan, High Temp. - High Press., 1 (1969) 517. S.V. Lebedev, A.I. Sawatimskii and Yu. B. Smirnov, High Temp., 9 (1971) 578. A. Cezairliyan and F. Righini, Metrologia, 33 (1996) 299. Fan Yi, Sun Xiaogang, and F. Righini, Int. J. Thermophys., 24 (2003) 849. A. Cezairliyan, M.S. Morse, H.A. Berman and C.W. Beckett, J. Res. Nat. Bur. Stand., A 74A (1970) 65. 1.1. Petrova and V.Ya. Chekhovskii, High Temp., 26 (1988) 271. E. Kaschnitz and P. Reiter, Int. J. Thermophys., 23 (2002) 1339. K.D. MagliC, Int. J. Thermophys., 24 (2003) 489. F. Righini, G.C. Bussolino and J. Spigiak, Thermochim. Acta, 347 (2000) 93. T. Matsumoto and A. Ono, High Temp.- High Press., 32 (2000) 67. J.M. Dai, Y. Fan and Z.X. Chu, Int. J. Thermophys., 23 (2002) 1401. A. Seifter, G. Pottlacher, H. Jager, G. Groboth and E. Kaschnitz, Ber. Bunsenges. Phys. Chem., 102 (1998) 1266. C. Cagran, B. Wilthan and G. Pottlacher, Thermochim. Acta, 445 (2006) 104. C. Cagran, B. Wilthan, G. Pottlacher, B. Roebuck, M. Wickens and R.A. Harding, Intermetallics, 11 (2003) 1327. M. Boivineau, C. Cagran, D. Doytier, V. Eyraud, M.-H. Nadal, B. Wilthan and G. Pottlacher, Int. J. Thermophys., 27 (2006) 1. M.M. Martynyuk and G.E. Gerrero, Sov. Phys. Tech. Phys., 17 (1972) 148. K.W. Henry, D.R. Stephens, D.J. Steinberg and E.B. Royce, Rev. Sci. Instr., 43 (1972) 1777.

G.R. Gathers, J.W. Shaner and D.A. Young, Phys. Rev. Lett., 33 (1974) 70. U. Seydel, R. Schofer and H. Jager, Z. Naturforsch., 30A (1975) 1166. R.S. Hixson, M.A. Winkler and J.W. Shaner, Physica B+C, 140B+C (1986) 893. A. Berthault, L. Arlks and J. Matricon, Int. J. Thermophys., 7 (1986) 167. E. Kaschnitz, G. Pottlacher and H. Jager, Int. J. Thermophys., 13 (1992) 699. M.A. Sheindlin and V.N. Senchenko, High Temp., 25 (1987) 281. M.M. Martynyuk , Sov. Phys. Tech. Phys., 19 (1974) 793. A.D. Rakhel, A. Kloss and H. Hess, Int. J. Thermophys., 23 (2002) 1369. R. Gallob, H. Jager and G. Pottlacher, Int. J. Thermophys., 7 (1986) 139. A.I. Savvatimskii, Int. J. Thermophys., 17 (1996) 495. A. Kloss, T. Motzke, R. Grossjohann and H. Hess, Phys. Rev. E, 54 (1996) 5851. N.I. Kuskova, S.I. Tkachenko and S.V. Koval, Int. J. Thermophys., 19 (1998) 341. V.N. Korobenko and A.D. Rakhel, Int. J. Thermophys., 20 (1999) 1257. A.W. DeSilva and J.D. Katsouros, Int. J. Thermophys., 20 (1999) 1267. W. Rogowski and W. Steinhaus, Archiv ftir Elektrotechnik, 1 (1912) 4141. A.T. Dinsdale, CALPHAD, 15 (1991) 3 17. J. Thewlis, Concise Dictionary of Physics and Related Subjects, Pergamon Press, Oxford, 1973 H. Hess, E. Kaschnitz and G. Pottlacher, High Pressure Research, 12 (1994) 29. R.S. Hixson, M.A. Winkler and M.L. Hodgson, Phys. Rev. B, 42 (1990) 6485. R.S. Hixson and M.A. Winkler, Int. J. Thermophys., 1 1 (1990) 709. Y.S. Touloukian and E H. Buyco, Specific Heat - Metallic Elements and Alloys, Thermophysical Properties of Matter - Volume 4, IFIPlenum, New York, 1970. R. Hultgren, P.D. Desai, D.T. Hawkings, M. Gleiser, K.K. Kelly and D.D. Wagman, Selected Values of Thermodynamic Properties of Elements, American Society for Metals, Metals Park, 1973. Y.S. Touloukian, R.K. Kirby, R.E. Taylor and P.D. Desai, Thermal Expansion - Metallic Elements and Alloys, Thermophysical Properties of Matter - Volume 12, IFIlPlenum, New York, 1975.

L.-D. Lucas, Viscometry and Densitometry in Physicochemical Measurements in Metals Research, Vol. IV Part 2, (Ed. R. A. Rapp), John Wiley & Sons, Inc., New York, 1970. F. Righini, A. Rosso and A. Cibrario, High Temp.- High Press., 17 (1985) 163. F. Righini, R.B. Roberts, A. Rosso and P.C. Cresto, High Temp.- High Press., 18 (1986) 561. A.P. Miller and A. Cezairliyan, Int. J. Thermophys., 3 (1982) 259. P. Reiter and E. Kaschnitz, Int. J. Thermophys., 23 (2002) 1327. U. Seydel, W. Fucke and W. Wadle, Die Bestimmung thermophysikalischer Daten flussiger hochschmelzender Metalle mit schnellen Pulsaufieizexperimenten, Verlag Dr. Peter Mannhold, Dusseldorf, 1980. H. Jager, W. Neff and G. Pottlacher, Int. J. Thermophys., 13 (1992) 83. R.S. Hixson, M.A. Winkler and J.W. Shaner, High Temp.- High Press., 17 (1985) 267. C. Otter, G. Pottlacher and H. Jager, Int. J. Thermophys., 17 (1996) 987. K. Boboridis, G. Pottlacher and H. Jager, Int. J. Thermophys., 20 (1999) 1289. M. M. Martynyuk, I. Karimkhodzhaev and V. I. Tsapkov, Resistance and enthalpy of refi-actory exploding wires, Sov. Phys. Tech. Phys. 19(11) (1975) 1458-1461 U. Seydel and W. Fucke, Z. Naturforsch., 32a (1977) 994. V.M. Kul'gavchuk and G.A. Novoskol'tseva, Sov. Phys. Tech. Phys., 11 (1966) 406. R.T. Wimber, J. Appl. Phys., 47 (1976) 5 115. T. Ishikawa, P.-F. Paradis, R. Fujii, Y. Saita and S. Yoda, Int. J. Thermophys., 26 (2005) 893. J.W. Arblaster, Platinum Metals Rev., 33 (1989) 14. J.D. James, J.A. Spittle, S.G.R. Brown and R.W. Evans, Meas. Sci. Technol., 12 (2001) Rl. R.A. Matula, J. Phys. Chem. Ref. Data, 8 (1979) 1147. P.D. Desai, T.K. Chu, H.M. James and C.Y. Ho, J. Phys. Chem. Ref. Data, 13 (1984) 1069. C. Kittel, Introduction to Solid State Physics, 3'* Edition, John Wiley & Sons, New York, 1968. K.C. Mills, B.J. Monaghan and B.J. Keene, Thermal Conductivities of Molten Metals, Part 1 - Pure Metals, NPL Report CMMT(A) 53, NPL, Teddington, 1997. L. Binkele, High Temp.-High Press., 18 (1986) 599. R.P. Tye and R.W. Hayden, High Temp.- High Press., 11 (1979) 597.

P.G. Klemens and R. K. Williams, Int. Met. Rev., 31 (1986) 197. K.D. MagliC, A. Cezairliyan and V.E Peletsky, (Eds), Compendium of Thermophysical Property Measurement Methods, Volume 1 - Survey of Measurement Techniques, Plenum Press, New York, 1984. Y.S. Touloukian, R.W. Powell, C.Y. Ho and M.C. Nicolaou, Thermal Diffusivity, Thermophysical Properties of Matter - Volume 10, IFI/Plenum, New York, 1973. K.C. Mills, Recommended Values of Thermophyscial Properties for Selected Commercial Alloys, Woodhead Publishing Limited, Cambridge, 2002. A.S. DobrosavljeviC and K.D. MagliC, High Temp.- High Press., 21 (1989) 41 1. D.P. DeWitt and G.D. Nutter, Theory and Practice of Radiation Thermometry, Wiley, New York, 1988. H. Preston-Thomas, Metrologia 27 (1990) 3, 107. N. Sheppard, H.A. Willis and J.C. Rigg, Pure & Appl. Chem., 57 (1985) 105. H.J. Kostkowski and R.D. Lee in Temperature: Its Measurements and Control in Science and Industry 3(1) (1962) 449-481. P. Saunders, P. Bloembergen and D.R. White, Proceedings of 9th International Symposium on Temperature and Thermal Measurements in Industry and Science (Tempmeko 2004), (Ed. T. ZvizdiC), LPMIFSB, Zagreb, 2005, 1149-1154. A. Cezairliyan and A.P Miiller, Int. J. Thermophys., 14 (1993) 1109. E. Kaschnitz, P. Reiter and J.L. McClure, Int. J. Thermophys., 23 (2002) 267. E. Kaschnitz and P. Supancic, Int. J. Thermophys., 26 (2005) 957. F. Righini, J. Spigiak and G.C. Bussolino, Int. J. Thermophys., 20 (2002) 1095. F. Righini ,Int. J. Thermophys., 27 (2006) 1. B.K. Tsai, R.L. Shoemaker, D.P. DeWitt, B.A. Cowans, Z. Dardas, W.N. Delgass and G.J. Dail, Int. J. Thermophys., 11 (1990) 269. P.B. Coates, Metrologia, 17 (1981) 103. J.L. Gardner, High Temp.- High Press., 12 (1980) 699. G.R. Gathers, Int. J. Thermophys., 13 (1992) 361. J.W. Shaner, G.R: Gathers and C. Minichino, High Temp.- High Press., 9 (1977) 331. R.E. Bedford, G. Bonnier, H. Maas and F. Pavese, Metrologia, 33 (1996) 133. C. Cagran, C. Brunner, A. Seifter and G. Pottlacher, High Temp.- High Press., 34 (2002) 669.

C. Cagran, B. Wilthan and G. Pottlacher, Proceedings of 9thInternational Symposium on Temperature and Thermal Measurements in Industry and Science (Tempmeko 2004), (Ed. T. Zvizdik), LPM/FSB, Zagreb, 2005 1313-1318. R.E.. Bedford, High Temp.- High Press., 4 (1972) 241. R.M.A. Azzam and N.M. Bashara, Ellipsometry and Polarized Light, North-Holland, Amsterdam, 1977. R.M.A. Azzam, Optica Acta, 29 (1982) 685. R.M.A. Azzam, Opt. Lett., 10 (1985) 309. S. Krishnan, J. Opt. Soc. Am. A, 9 (1992) 1615. R.M.A. Azzam and A.G. Lopez, J. Opt. Soc. Am. A, 6 (1989) 1513. C. Cagran and G. Pottlacher, in Temperature: Its Measurement and Control in Science and Industry, 7(2), (Ed. D. C. Ripple), AIP Conference Proceedings, Volume 684, Melville, 2003,765-770 T. Matsumoto and A. Cezairliyan, Int. J. Thermophys., 18 (1997) 1539. E.D. Palik, Handbook of Optical Constants of Solids, Academic Press, Inc., Boston, 1990. J.C. Miller, Phil. Mag., 20 (1969) 1115. Y.S. Touloukian and D.P. DeWitt, Thermal Radiative Properties Metallic Elements and Alloys, Thermophysical Properties of Matter Volume 7, IFIPlenum, New York, 1970. J. Redgrove and M. Batuello, High Temp.- High Press., 27/28 (199511996) 135. G. Pottlacher and H. Jager, Int. J. Thermophys., 11 (1990) 719. K. Boboridis, G. Pottlacher and H. Jager, Int. J. Thermophys., 20 (1999) 1289. G. Pottlacher and H. Jager, J. Non-Cryst. Solids, 205-207 (1996) 265. G.R. Gathers, J.W. Shaner, C.A. Calder and W.W. Wilcox, in Proceedings of the 7thSymposium on Thermophysical Properties, Gaithersburg, MD, ASME, New York, 1977,904. R.S. Hixson, D.A. Boness, J.W. Shaner and J.A. Moriarty, Phys. Rev. Lett., 62 (1989) 637. M. Boivineau, J.B. Charbonnier, J.M. Vermeulen and Th. ThCvenin, High Temp.- High Press., 25 (1993) 3 11. M. Boivineau and G. Pottlacher, Int. J. Materials and Product Technology, 26 (2006) 2 17. I. Egry, Molten Metals in Measurement of the Thermodynamic Properties of Single Phases, (Eds A. R. H. Goodwin, K N. Marsh and W. A. Wakeham), Elsevier, Amsterdam, 2003,504-535. T. Baykara, R.H. Hauge, N. Norem, P. Lee and J. J. Margrave, High Temp. Sci., 32 (1991) 113.

D.M. Herlach, R.F. Cochrane, I. Egry, H.J. Fecht and A.L. Greer, Int. Mat. Rev., 38 (1993) 273. D.L. Cummings and D.A. Blackburn, J. Fluid Mechanics, 224 (1991) 395. Y.S. Touloukian, S.C. Saxena and P. Hestermans, Viscosity, Thermophysical Properties of Matter - Volume 11, IFI/Plenum, New York, 1975. R.F. Brooks, I. Egry, S. Seetharaman and D. Grant, HighTemp.-High Press., 33 (2001) 63 1. J.W. Arblaster, CALPHAD, 19 (1995) 365. J.L. McClure, K. Boboridis and A. Cezairliyan, Int. J. Thermophys., 20 (1999)1137. G.R Gathers, J.W. Shaner, R.S. Hixson and D.A. Young, High Temp.High Press., 11 (1979) 653. M.M. Martynyuk and V.I. Tsapkov, Phys. Met. Metall., 37 (1974) 40. B.V. Vlasov, S.G. Taluts, V.E. Zinov'ev, N.A. Korenoskii and V.P. Polykova, Phys. Met. Metall., 74 (1992) 317. C. Cagran and G. Pottlacher, Physical Properties and Normal Spectral Emissivity of Iridium up to 3500 K, presented at 16' Symposium on Thermophysical Properties, Boulder, Co, USA, July 30 to August 4,2006, to be published in Int. J. Thermophys. B. Wilthan, C. Cagran and G. Pottlacher, Int. J. Thermophys., 25 (2004) 1519. B. Wilthan, C. Cagran and G. Pottlacher, Int. J. Thermophys., 26 (2005) 1017. D. Basak, H.W. Yoon, R. Rhorer and T. Burns, Microsecond Pyrometry during Rapid Resistive Heating of Samples in a Kolsky Bar Apparatus, in Temperature: Its Measurement and Control in Science and Industry 7(2), (Ed. D. C. Ripple), AIP Conference Proceedings, Volume 684, Melville, 2003,753-757. D. Basak, H.W. Yoon, R. Rhorer, T.J. Burns and T. Matsumoto, Int. J. Thermophys., 25 (2004) 56 1. H. Watanabe and T. Baba, Electrical-optical pulse-heating method for high-temperature thermal diffusivity measurement, presented at 16" Symposium on Thermophysical Properties, Boulder, Co, USA, July 30 to August 4,2006, to be publishedin Int. J. Thermophys. http://www.netzsch-thermal-analysis.com/ http://www.linseis.net/ T. Baba and A. Ono, Meas. Sci. Tech., 12 (2001) 2046.

133. L.H. Allen, G. Ramanath, S.L. Lai, Z. Ma, S.Lee, D.D.J. Allman and K.P. Fuchs, Appl. Phys. Lett., 64 (1994) 417. 134. S.L. Lai, G. Ramanath, L.H. Allen, P. Infante and Z. Ma, Appl. Phys. Lett., 67 (1995) 1229. 135. H. Huth, A.A. Minakov and C. Schick, J. Poly. Sci. B, 44 (2006) 2996. 136. S.A. Adamovsky. A.A. Minakov and C. Schick, Thermochim. Acta, 403 (2003) 55. 137. A.A. Minakov, S.B. Roy, Y.V. Bugoslavsky and L.F. Cohen, Rev. Sci. Instr., 76 (2005) 1. 138. A.F. Lopeandia, J. Rodriguez-Viejo, M. Chacbn, M.T. Clavaguera-Mora and F.J. Mufioz, J. Micromech. Microeng., 16 (2006) 965. 139. K.S. Suh, H.J. Kim, Y.D. Park and K.H. Kim, J. Korean Phys. Soc., 49 (2006) 1370. 140. M. Pyda, E. Nowak-Pyda, J. Heeg, H. Huth, A.A. Minakov, M.L.DI Lorenzo, C. Schick and B. Wunderlich, J. Poly. Sci. B, 44 (2006) 1364. 141. Guide to the Expression of Uncertainty in Measurement (GUM), IS0 Guide 98: 1995, International Organisation for Standardisation, Geneva, 1995. 142. Expression of the Uncertainty of Measurement in Calibration, EA-4/02, European co-operation for Accreditation, http://www.europeanaccreditation.org/nl/doc/EA-4-02.pdf, 1999. 143. A. Cezairliyan, High Temp.-High Press., 11 (1979) 9. 144. G. Pottlacher, Thermophyscial Measurements, Subsecond, in Concise Encyclopedia of Materials Characterization, (Eds R.W. Cahn and E. Lifshin), Pergamon Press, Oxford, 1993,529-536. 145. G. Pottlacher, in High Temperatures in Measurement of the Thermodynamic Properties of Single Phases, (Eds A.R.H. Goodwin, K.N. Marsh and W.A. Wakeham), Elsevier, Amsterdam, 2003,488-504. 146. A. Cezairliyan, G.R. Gathers, A M. Malvezzi, A.P. Miller, F. Righini and J.W. Shaner, Int. J. Thermophys., 11 (1990) 819. 147. F. Righini, The World of Subsecond Thermophysics, presented at XVIII IMEKO World Congress - Metrology for a sustainable development, Rio de Janeiro, Brazil, September 17 to 22,2006, prepared for publication 148. K D. Maglik, A. Cezairliyan and V.E Peletsky, (Eds), Compendium of Thennophysical Property Measurement Methods, Volume 2 Recommended Measurement Techniques and Practices, Plenum Press, New York, 1992. 149. 0. Kubaschewski, C.B. Alcock, Metallurgical Thermochemistry, 5thEdn, Pergamon Press, Oxford, 1979.

150. NIST, Selected Publications of the Subsecond Thermophysics Program, US. Department of Commerece, National Institute of Standards and Technology, Gaithersburg, 1993. 151. V.E. Zinov'yev, Metals at High Temperatures - Standard Handbook of Properties, Hemisphere Publishing Corporation, New York, 1990. 152. T. Iida and R.I.L. Guthrie, The Physical Properties of Liquid Metals, Clarendon Press, Oxford, 1988. 153. W. Blanke, Thermophysikalische StoffgroDen, Springer Verlag, Berlin, 1989. ~ 2004-2005, (Ed. 154. CRC Handbook of Chemistry and Physics, 8 5 Edition, D. R. Lide), CRC Press, Boca Raton, 2004.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 10

SURFACE PROPERTIES OF NANOPARTICLES P. Staszczuk

Department of Physicochemistry of Solid Surfaces, Chemistry Faculty, Maria Curie-Sklodowska University, Maria Curie-Sklodowska Sq. 3, 20-03 1 LUBLIN, POLAND 1. INTRODUCTION

1.2. Nanotechnology and nanostructures At present there is much interest in developments concerning nanomaterials and their practical applications [I]. Nanotechnology deals with the production and the studies of properties of very small objects and their applications in practice. Nanotechnology is at present the most dynamically developing discipline of science and technology and over 30 nanotechnological research centres exist at universities. The term nanotechnology was introduced by Norio Taniguchi in 1974 to describe processing with an accuracy smaller than 1 pm, but the quick development and diminution of objects used in practice, as well as manipulation of molecules on this small scale, were predicted by Richard Feynman in 1957. The subject of nanotechnology thus now covers the search for and synthesis of new materials of advanced technology which possess the sizes of nanometres; the determination of their characteristics, and their practical application. Nanostructures are the bridge between individual atoms and molecules, where the laws of quantum mechanics apply, and bulk phases, whose properties usually result from the collective behaviour of billions of atoms. Individual nanostructures may be: clusters, nanomolecules, nanocrystals, so-called quantum points, nanowires and nanotubes. They possess orderly structures and some large molecules can form single nanostructures [2]. The quantum sizes and shapes of nanomolecules affect their mechanical, chemical, electrical, nuclearelectronic, electric-optical and dynamic properties. They may exhibit new, unique physicochemical phenomena, quantitatively different from those of the bulk phase. This leads to the possible control of the action and application of nanostructures

Nanotechnology must satisfy three criteria: 1) Studied structures should have at least one size not larger than 100 nm. Thus nanotechnology is not technology in many cases. 2) In the process of production of nanomaterials, their physicochemical properties should be controllable. 3) There must be a possibility of building larger objects from the advanced materials produced. Devices such as scanning microscopes, scanning tunnelling microscopes and atomic force microscopes are important in the study of nanostructures for obtaining pictures of even single atoms and their dislocations on the surface. Nanomaterials can be produced using the methods: j?om top to bottom and from bottom to top. The from top to bottom method consists in modelling of the surface by addition or removal of some amount of substance. In this way, selfcontained systems with particles slightly broader than 100 nm can be formed. However, in the from bottom to top method, self-arrangement, as in the formation of large structures, is used. Under suitable conditions atoms or molecules form orderly systems spontaneously. In this way nanotubes are formed. So far, advanced materials have found some applications in practice. Fullerenes of 1 nm diameter, prepared for the first time in 1985, can be used as superconductors [3]. Carbon nanotubes, discovered in 1991, were used in 1998 to build diode transistors, transmitters, logical gates, and in 2003 in France to produce indestructible materials [4]. They can also be used as superconductors [5]. In 1999, single organic molecules and intersecting nanowires were used as catalysts for the production of petrol [6]. Non-magnetic layers of a thickness smaller than 1 nm placed between magnetic layers form computer hard discs with high sensitivity and capacity [7]. Drugs placed in lipid bags of a 100 nm diameter act longer because they circulate longer in the blood circulatory system. Biological tests for the presence and activity of searched substances are quicker, more sensitive and flexible when nanomolecules are used as markers [6]. Crystalline nanopowders have improved the chemical, mechanical, optical and magnetic properties of materials. Harder ceramic materials, solar filters and catalysts used in environmental protection have been obtained [8]. In the near hture microscopic robots may revolutionize industrial production [7-91 and may be used in interplanetary travels [lo]. Studies of typical nanomaterials (soil mineral components, adsorbents, silica gels with deposited proteins, so called smart surfaces, latexes, synthetic zeolites modified by ions, MCM-41 molecular sieves) were made earlier by the author of this chapter [ll-171. At present our research focuses on studies of surface properties (e.g. adsorption capacity), total heterogeneity (energetic and geometrical) of surface layers, as well as structures and phase transformations of

fullerenes, carbon nanotubes, active carbons, semiconductors, high-temperature superconductors, modified zeolites and adsorbents with deposited proteins [18261. In this chapter some recent results are presented and discussed.

1.2. Total (energetic and structural) heterogeneity of surfaces Almost all solid surfaces possess heterogeneous properties [16,27,28]. Typical adsorbents such as active carbon, aluminas and silica gels, are characterized by total surface structural heterogeneity, as well as various surface groups, active centres, contamination and irregular heterogeneity. Surface heterogeneity may thus be divided into chemical heterogeneity (different active centres) and geometrical heterogeneity (irregularities of the crystal lattice). Porous solids possess various surface groups and irregularities (surface heterogeneity), as well as fine pores of different sizes and shapes (structural heterogeneity). Their well-developed porous structure is a combination of micropores (< 2 nm),mesopores (2 to 50 nm) and macropores (> 50 nm). Pores are accessible to various adsorbate molecules. Structural surface heterogeneity depends also on the adsorbate and the extent of adsorption. The above heterogeneous properties cause differences in adsorption energy (energetic heterogeneity) and, as a result, in the adsorption capacity. The causes of heterogeneity are different. For minerals found in nature, some of the heterogeneity is of geological origin, produced during formation or deposition, and some results from preliminary treatment and preparation for studies (e.g. grinding). The cause of heterogeneity of synthetic adsorbents, such as active carbon, may be heterogeneity of the substrate for production (e.g. coal) and heterogeneity of the activation process of carbon grains or granules by gaseous substances. To control surface processes, such as separation, adsorption and catalysis, and to optimize the surface (e.g. adhesive and hydrophobic/hydrophilic) properties of solid materials, it is necessary to determine porosities and surface energies and to estimate possible contributions of non-specific and specific interactions in forming the liquid surface films. In relation to energies, surfaces can be divided into two types: homogeneous and heterogeneous. These surface areas possess different energetic minimum depending on the type, thus different adsorption energy Ah"' and different potential barrier for surface diffusion AV". For ideally homogeneous surfaces Ah"' = constant but in practice we have situations presented in Figure 1 [29]. There are two cases: i) when kT > A T - mobile adsorption and molecules can move along the surfaces, ii) when kT < AV" - localized adsorption - strong localization of molecules in the minima.

Figure 1. Adsorption energy of homogeneous surfaces (in theory, left side) and adsorption energy of homogeneous surfaces (in practice, right side). For heterogeneous surfaces, the minima have different depths on the surface, so there are different energetic barriers for diffusion. Heterogeneous surfaces are divided, in relation to the minima, into: i) Periodic heterogeneous surfaces, where the energetic minima exhibit some reproducibility (see Figures 2 and 5(a)). ii) Completely random, where the minima do not exhibit any reproducibility (Figures 3 and 5(b)). iii) Patchwise, where some reproducibility of energetic minima occurs. These are grouped in largepatches (see Figures 4 and 5(c)). Two extreme topographic models are important in describing most of the phenomena concerning adsorption on solid surfaces:

Figure 2. Energetic minima Figure 3. Energetic changes Figure 4. Energetic of the periodic surface of the random surface minima of the patchwise surface On patchwise-type surfaces, it is usually assumed that these patches are so large that in practice they constitute independent adsorption (thermodynamic) centres that are only in thermal and material contact. Exchange of adsorbate molecules is possible between these subsystems. Considering further the interactions between adsorbed molecules (complexes), it is assumed that the states in which two molecules are adsorbed on two different patches make

negligible contribution into the thermodynamic properties of the systems. This can be represented as in Figure 5(c).

Figure 5(a). Topography Figure 5(b). Topography of the periodic surface of the random surface

Figure 5(c). Topography of the patchwise surface

Random type surfaces are illustrated in the scheme in Figure 5(b). Such a completely random distribution of centres means that the probability of finding another centre in the surrounding of any adsorption centre is identical. As a result, the microscopic composition of the phase adsorbed in the surroundings of any centre is the same and identical to the mean composition of the phase on the whole solid surface. In the first case (periodic surface) a high correlation between adsorption energies is observed, i.e. the energies change passing into other adsorption canters but the difference remains unchanged. In the second case (random surface) no correlation is observed between adsorption energies on different centres. Actually existing systems present intermediate properties between these extreme physical situations. It can be assumed that in the case of well formed crystals with a few crystallographic planes, the patchwise model application is more justifiable. Amorphous samples are more suitable for the random model. Even in the case of well formed crystals, that is samples of very high degree of the oxide bulk arrangement, the surface may have a small degree arrangement of surface atoms. As for the correlation between adsorption energies, the a priori assumptions are even more difficult. The strategy of investigations in the present stage consists in the analysis of properties of systems, assuming different models and observations, to determine which of them best describes the experimental system under consideration. Over the past decade, there has been much theoretical and experimental progress, reported in numbers of papers, using automatic sorption apparatuses and calorimetry, chromatography, spectrometry and other methods. However, these methods are time consuming, the apparatus is often very expensive, and the results obtained are not sufficient to explain all of solid surface heterogeneity, liquid-solid interactions, adsorbed film properties and wetting phenomena and their roles in interface processes. Recent theoretical and experimental studies showed that programmed liquid thermodesorption, under quasi-isothermal conditions, could be applied to estimate the total heterogeneous properties of solid surfaces.

1.3. Fractal dimensions of nanoparticles Fractal geometry [30-331 is an extension of classical geometry and can be used for creating accurate models of physical structures. Applications to the structural characterization of complex solids have been successful. The condition that must be satisfied is that self-resemblance of the structure must tend towards infinity as the size of the structure is reduced. Nanomaterials satisfy this condition and can be identified as fractal objects. During the scaling process, the heterogeneity of nanomaterial surfaces or layers forming them, does not change. Mathematical descriptions of the surface should reflect this property and be in agreement with the theoretical structural models. Solid surfaces can be of different kinds. They can be defined as Euclidean, nonporous, completely flat surfaces, with area, S, equal to the square of their linear dimension R [34]:

When there are finite fluctuations within the completely flat surface, the effective surface area is proportional to the linear square of dimension R and a constant c:

When fluctuations depend on the surface size, the effective surface area extends within the third dimension. Then the surface area is defined as:

where: 32Fd22 is the fractal dimension. Many processes and structures that are difficult to describe by means of traditional Euclidean geometry can thus be precisely characterized using fractal geometry, for example the complex and disordered microstructures of advanced materials, adsorbents, polymers and minerals. Recent studies have shown that using fractal dimensions enables the real sizes of pore radii to be determined and pore-size distribution functions to be calculated from the data of programmed thermodesorption of liquids [35].

2. PHYSICOCHEMICAL PROPERTIES OF SELECTED NANOMATERIALS 2.1. Carbon nanotubes Carbon nanotubes are a new allotropic form of carbon and possess interesting physicochemical properties. Their chance discovery was a result of an enormous interest in fullerenes. Carbon nanotubes are built of graphene layers and can assume single- or multi-wallet structures [23,25,35]. Chemical modifications of nanotubes in both open terminated areas and on outer and inner walls create many possibilities. Prospective and present applications of nanotubes depend on their physicochemical properties, such as density, resistance to stretching and bending, thermal and electrical conductivity, field emission, as well as resistance to temperature. Good adsorption properties of nanocarbon materials contribute to their extensive practical application.

2.2. Montmorillonites Knowledge of hydration and dehydration processes of soils, especially montmorillonites, has become increasingly important [16,31,32,33] for various processes, for example, infiltration of rain water through soil, destruction of roads and buildings, exploration of crude oil from boring holes, etc. Moreover, montmorillonites have unique adsorption, catalytic and ion-exchange properties, and also surface and porous properties, which govern their potential applications in many spheres of industry and life. In a water-montmorillonite system free water exists, which fills pores and large spaces between the grains and bound water. Bound water has properties different from those of free water. The layers of bound water consist of interpacket water and water adsorbed on the surface of montmorillonite So far, structural and other properties of adsorbed water on the surface have not been established precisely. Adsorption and surface properties of montmorillonite surfaces have been investigated by many methods, for example gravimetry (McBain balance), gas chromatography, XRD and electron microscopy. However, these methods do not provide complete information about the properties of the water layers and the role it plays in surface wettability. The applications of adsorbents and/or catalysts require knowledge of many surface physicochemical parameters, mainly of their adsorption properties and porosity, as well as of the selectivity, catalytic activity and properties of the surface active sites. For estimation of the nature, quality, localization and energy of the active centres (e.g. Lewis and/or Brijnsted acid type), adsorption and microcalorimetric methods are most frequently used. Lately, a special thermal analysis technique has been succesfully adopted to study the liquidsolid systems, especially for characterization of the total (energetic and geometric) heterogeneous properties of materials. The effects taking place during

thermodesorption of liquids from solid samples are recorded and used in the calculation of the physicochemical parameters. During the thermodesorption process of adsorbed liquid films from solid surfaces, the physical bonds (first of all hydrogen bonds, which are ten times weaker than the chemical bonds) are disrupted. The experimental results obtained so far show that the simple thermogravimetric method of programmed thermodesorption of polar and nonpolar liquids from material surfaces under quasi-isothermal conditions can be applied to study the adsorbed liquid films on montmorillonite surfaces and their heterogeneous properties.

2.3. Zeolites Zeolites, such as alumina oxides and silica gels, are the most popular adsorbents widely applied in industry, environmental protection, agriculture, medicine, optics, microelectronics and many other fields. Their molecular-sieve properties and high resistance to acid action are responsible for the fact that zeolites are used as catalysts. They find applications in the petrochemical industry in cracking, hydrocracking, isomerization and other processes. Zeolite catalysts are characterized by high activity, good selectivity, stability and capability of multiple regeneration. Zeolites are characterized by high ionexchange selectivity towards ' 3 7 ~ and s ' O S ~ as well as other radioactive elements. They are also used for the purification of waste water from heavymetal salts. Ion exchange is the process most frequently applied for removal of these compounds on zeolites. In this process clinoptylolite, which, due to its properties, can be used not only for purification of waste water but also for recovery of these metals found the widest applications. Owing to its large adsorption capacities towards vapours and gases it is used in environmental protection to remove toxic gases, among others, SOz, H2S, NH3, N,O,, C12 from the atmosphere. To use adsorbents in practice, one must be familiar with physicochemical properties of their surfaces. The properties of individual compounds are often examined, but papers dealing with the properties of mixed adsorbents are scarce. Porosity is one of the most important characteristics considered in the determination of the properties of solid surfaces. Specific surface area, average pore radius and total pore volume have a significant effect on physicochemical processes on the solid surface. Surface properties can be modified using high temperature or chemical modification. Thermal analysis is one of the methods used in determination of surface heterogeneity and adsorption properties [34-361. 2.4. Superconductor materials The main interest in superconductor materials, particularly in ceramic hightemperature superconductors, is due to their numerous potential applications

[37-401. The physicochemical properties of superconductor surfaces are known only to a small extent. Due to their well-developed surface (larger specific surface area and presence of various active centres, which results in a larger adsorption capacity), thin superconductor layers are enriched in oxygen and are characterized by higher Curie temperatures, T,. Thin surface layers of various materials possess specific physical properties and for this reason they are applied in microelectronics in passive microwave devices, e.g. as high-quality resonators and low-loss filters. By using YBa2C~307-x and T12Ba2CaCu208+, type materials as thin layer circuits in passive microwave subunits, the conduction losses are dramatically reduced. Attributes such as small size, low mass and low losses make these subunits very desirable, not only for the communication systems used by NASA, but also in the commercial telecommunication industry. Physicochemical parameters such as, adsorption capacity and porosity, amongst others, can change the material's superconductor properties entirely by enhancing chemical reactions and can be responsible for the decomposition of the superconductor material. The results obtained so far using thermogravimetry under quasi-isothermal conditions (Q-TG) show that the are characterized superconductors such as Y13a2Cu307-,and HgBa2Ca2Cu308+x, by specific adsorption properties towards polar and nonpolar liquids (particularly water) and by porosity. The specific surface area, pore size, pore volume and pore-size distribution functions are fundamental parameters for the characterization of solids. Properties such as porosity, strength, hardness, permeability, selectivity, corrosion, thermal stress resistance, etc. can be directly correlated to the porous structure of the materials. These properties can be easily investigated by physisorption techniques, which can be carried out using a sorptomatic apparatus.

3. TECHNIQUES USED 3.1. Q-TG thermogravimetry 3.1.1. The technique Quasi-isothermal (Q-TG) thermogravimetry by means of the the Derivatograph 4-1500 D (MOM, Hungary) [46] involves the programmer raising the temperature of the sample in the usual manner, at a relatively high rate as long as the mass of the sample remains constant, because this period of the experiment is insignificant. When the mass of sample begins to change, the heating programmer system establishes a temperature difference between the sample and the furnace, which ensures that the transformation proceeds at a preselected very small constant rate. The method has a wide range of application, high selectivity and resolution [47,48]. From the experimental data obtained for the programmed thermodesorption of liquids under these quasi-isothermal

conditions, the theoretical basis and methods of adsorption can be determined, including desorption energy distribution functions, pore-size distributions, total heterogeneity of solid surfaces [49,50], fractal coefficients [43,511 and diffusion coefficients [52]. The samples were saturated with liquid vapours in a vacuum desiccator at plp, = 1. The Q-TG mass loss and Q-DTG differential mass loss curves were measured under quasi-isothermal conditions in the temperature range 20-250 "C at a heating rate of 6 "Jmin. 3.1.2. Estimation of the energetic heterogeneity from Q-TG data The method of programmed liquid thermodesorption from the surface of studied solids can be used for calculation of the desorption energy and the desorption energy distribution function. Monomolecular desorption kinetics in the case of unassociated onecomponent layers can be described by the equation [29,50]:

where:

T=To+/3z (5) and: R - the universal gas constant; 8 - the degree of surface coverage; v - the entropy factor; Ed - the desorption energy calculated for each temperature; Toand T - the initial and given temperatures of desorption, respectively; /3 - the heating rate of the sample; z - time. Equation (4) holds for the case when the amount of desorbed substance does not fill the whole surface uniformly, however, desorption takes place in the range of capillary condensation. The above equation can be also used for the analysis of desorption from the multilayer-filled energetically heterogeneous surface of the studied material. Then the desorption rate is described by the integral equation:

The energetic heterogeneity of the solid surface is described by the energy distribution function cp(Ed). This is the density of adsorption centre distribution probability on the surface of the studied sample in relation to the quantity of desorption energy.

The logarithmic form of the initial equation is expressed by the following equation:

The final expression to determine the density function cp(Ed) can be presented as :

3.1.3. Estimation of kinetic and thermodynamic parameters From the experimental mass-loss Q-TG curves and the assumption of an adsorption model, it is possible to determine kinetic and thermodynamic parameters of the processes and heterogeneous properties of solid surfaces. The numerical procedure was developed in order to evaluate the pore-size distribution and desorption energy distribution functions (i.e. total heterogeneity) of preadsorbed liquid on mesoporous solid surface from single thermogravimetry Q-TG and Q-DTG curves recorded under quasi-equilibrium conditions. It is based on the application of a condensation approximation to treat the desorption kinetics under non-isothermal conditions. In this approximation, the desorption energy at each temperature in the Q-TG and QDTG curves was calculated using Redhead's equation. The desorption distribution was calculated from the first derivative of the temperature on the desorption energy, Ed.The pore-size and desorption energy is a function of heat of vaporization of the test liquid, its molar volume and its surface tension. The mesopore-size cumulative and differential distributions can be evaluated from the dependence of the desorption energy on the mesopore volume, V,,,,, and the above desorption distribution, respectively. The approximate desorption energy distribution from the pores for each temperature Ti in the Q-DTG curve is given by equation:

By using the function Ed(rk)we can evaluate the mesopore-size distribution, x(rk), from the relationship:

where rk is the radius of mesopores.

The intensities and shapes of the Q-TG and Q-DTG curves reflect the effects of the pore-size distribution and desorption energy distribution functions of the nanomaterial surfaces and are comparable with those measured by independent methods. Details of the theory and the calculation procedures are given in 13,621.

3.1.4. Calculation offractal dimensionsfrom Q-TG data Calculations of fractal dimensions on the basis of Q-TG curves for the thermodesorption of polar and non-polar liquids can be done using the method presented in [43,51]. Firstly, the Q-TG mass-loss curve was transferred into a relationship between In [(m, - ml) S] as a function of In [(m, - mk)lA, where: m, is initial mass sample, mi is sample mass at temperature Ti, mk is the final sample mass, d is the density of the liquid and S is the specific surface area of the sample. TsImsl

68 66

dTg (Temp)

Tslmsl

:-0 020

dTs (Temp)

34--004

Q-DTG

: -0 030

Q-TG t

40

60

80

100

120

140

160

180

200

40

60

80

T PC1

100

120

140

160

180

XXI

220 T PC1

Figure 6. The mass loss Q-TG and differential Q-DTG curves of n-butanol thermodesorption from Na-montmorillonite (left side) and La-montmorillonite (right side) samples immersed in vacuum desiccator. Point P is the point of intersection of lines a and b and is related to point K on the curve presented in Figure 7 for l-butanol thermodesorption from a Lamontmorillonite surface. Third line is drawn perpendicular to PK and was used for the determination of the value of Fd. Point K is related to the minimum value of differential mass-loss in Q-DTG curve. In our calculations the following equation was used for determination of the Fd value by an analytical method:

where nf

=

fractional part (slope of line c in Figure 7).

Figure 7. Fractal dimension method of calculation on the basis of Q-TG thermodesorption of 1-butanol from La-montmorillonite surface under the quasiisothermal conditions. Fractal dimensions Fd were used for the calculation of the pore volume distribution functions in relation to their radii from equation:

where: V is the pore volume obtained from mass-loss Q-TG curves, and r is the pore radius calculated from the equation:

where: h is the thickness of the liquid films.

3.1.5. Calculation of diffusion coefficients from Q-TG data Adsorption and desorption processes of molecules on the surfaces of adsorbents, on the samples of aluminum oxide, and other materials are significantly affected by diffusion of adsorbate molecules towards the surface in the pores of the adsorbent. The difference in concentration at the gaslsolid interface is the direct cause of diffusion. The fractality of the surface has an influence on the diffusion of molecules through a porous materials [79,81]. Thermal motion of the molecules possessing the kinetic energy of translation is responsible for spontaneous, mutual penetration of various gases and liquids. This phenomenon called diffusion leads to formation of homogeneous mixtures of gases or liquids. Its direct cause is the existence of concentration gradient at the contact boundary of two substances. The larger the concentration gradient is, the quicker the diffusion process proceeds.

3.2. Surface adsorption A large number of physicochemical processes take place and/or are initiated at solidgas or solid/liquid interfaces. Knowledge of the phenomena occurring there is of importance because in many cases they result in changes in the physicochemical properties of the material. One of the most striking properties of a solid surface is its capability of adsorption of vapours and gases. Thus, atmospheric gases such as H20, N2 and C02 may influence significantly the behaviour of materials of advanced technology. Depending on the kind of material and other conditions, one may observe different interactions of physical and/or chemical adsorption. Adsorption properties of the heterostructure and, in particular, the adsorption capacity and adsorbate-adsorbent interactions are necessary for understanding sorption and diffusion mechanisms on nanoparticle surfaces. Porosity properties, e.g. specific surface areas, pore size distribution and pore volume were calculated from low-temperature nitrogen adsorption-desorption isotherms measured by means of a Sorptomat ASAP 2405 V1 .O1 with a special program for the preparation of the isotherms (Micrometrics Co., USA) and a Porosimeter 4000 (Carlo Erba Instruments).

3.3. Porosimetry Mercury porosimetry is a technique for probing porous materials at scales which are larger than those relevant in molecular adsorption and which range over more than three orders of magnitude. For rigid and cylindrical pores, the pressure required to force a non-wetting liquid such as mercury into the pore is given by the Washburn equation. The volume of mercury V injected into the sample at pressure P is measured for a sequence of pressures that typically ranges from 0.1 to 120 bars.

Volume, (mm3)

Volume, (mm3)

Fig. 8. The mercury pressure in relation to pore volume of Na-montmorillonite (left) and La-montmorillonite (right) samples.

Pore radius (nm)

Pore radius (nm)

Fig. 9. Pore-size distribution functions of Na- (left) and La-montmorillonite (right) samples from the porosimetry technique. From such measurements, surface areas (normalized cumulative and relative), pore radii (choice of three measuring units), pore volumes (raw, normalized, cumulative and relative) and pore-size distribution functions of samples can calculated. Figure 8 presents the graphs of mercury-penetrated volume versus pressure in pores of Na- and La-montmorillonite samples. Figure 9 shows poresize distribution functions from porosimetry data.

3.4. Calculation of fractal dimensions from sorptometry and porosimetry data The calculation of fractal dimensions from sorptometry data is based on the theory by Frenkel, Halsey and Hill and also of Kisielev [55]. The fractal dimension Fd can be calculated from the following equations: Fd=2+nf Fd = 3 - d[ln a(x)]/d[ln(-lnx)] Fd = 2+ d[ln (-ln x)da]ld[ln (-ln x)] d Vldr = A(r)- r (2 - Fd)

where: nf is a fractional part of the fractal, a is the value of adsorption, x is the segment of experimental isotherm, and V and r are the volume and radius of pores, respectively. The function A(r) is determined from the experimental data of the adsorption hysteresis. Fractal dimensions can be calculated from porosimetry data [56,57] fi-om the equation:

where: Vand P are the volume and pressure of mercury.

358

Hence:

-

log (dVldP) (Fd - 4) log P

(19)

3.5. Atomic force microscopy, (AFM), scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) These conventional techniques can provide additional valuable information (see examples below). To characterize the structural changes caused by modification, an atomic force microscope (AFM) (NanoScope 111, Digital Instruments, USA), was used and scanning electron micrographs were obtained using a JSM-25 JOEL and LEO SEM 1430VP with EDX detector. 3.5.1. The determination of surfacefractal coeficients~om AFM measurements

Atomic force microscopy (AFM) has an important application in fractal analysis because it provides data characterizing the material surface in the scale from a few angstroms to hundreds of micrometers with great precision. In the AFM method each sample is scanned in a few different places so as to obtain a representative view of the whole surface topography. For powders a few particles are chosen and a surface fragment is scanned on each of them. The algorithms most frequently used for calculation of fractal coefficients from the AFM results are [58]: Fourier spectrum integral method, surface-perimeter method, structural function method and variable method. To determine the surface dimension by the Fourier spectrum integral method it is necessary to obtain the picture of the surface 2D FFT generating amplitude and time of the matrix (more detail s are given in paper [58]. Assuming the surface function as f(x,y), the Fourier transform in two-dimensional space can be expressed as [58]:

where: x and y- are the spatial variables, u and v- are the spatial frequencies for x and y. The spatial spectrum of frequencies, F(u,v) is the two-dimensional system consisting of the amplitude signal and the time along each direction. If the surface topography is fractal, its spatial spectrum of frequencies shows a linear distribution in the doubly logarithmic system and the surface fractal dimension can be expressed as:

where: Fd is the surface fractal dimension, p is the angle of straight line inclination. In the Fourier spectrum integral method with the specified direction of spectrum area and the assumption that various frequencies have a common direction, a spectrum and spatial distribution of frequencies are obtained. After finding the double logarithm, the curve of fractal dimension along a specified direction is obtained. By determining the fractal dimensions for each direction it is possible to determine the dependence between F(q) and the angle q. Assuming that the angle variance in the specified range is the same angle, after averaging the spectrum amplitudes, the surface fractal dimensions are presented in the two-polar system of the coordinates for the angles in the range 0-180". The surface fractal dimension characterizes the irregularities of the pore surface: the larger the value, the more irregular and rougher are the pore surfaces. The lower boundary limit of the pore coefficient of 2 corresponds to a flat surface but the upper value 3 corresponds to a strongly corrugated surface, which fills up the whole space. 4. EXAMPLES OF STUDIES ON SELECTED MATERIALS

4.1. Carbon nanotubes 4.1.1. Materials Five carbon nanomaterial samples produced by the firm Sigma-Aldrich were experimentally tested (Table 1). Table 1. The characterization of carbon nanotubes.

In addition, carbon nanotube samples which were grown in a horizontal quartz tube reactor placed in a furnace by the reaction technique using a xyleneferrocene mixture by means of a method described in details in papers [23,24] were examined. In our investigations of adsorbed liquids and surface porosity parameters of nanotubes, we used carbon products obtained by two methods: a DC electric arc was generated between graphite electrodes and hydrocarbon vapour was thermally decomposed in the presence of a catalyst (N-1 sample). This material was sonicated in a waterlethanol mixture for 30 min (N-2 sample).

The fraction that precipitated on the bottom included nanotubes but there were mainly clusters. The N-3 sample was prepared by catalytic decomposition of xylene CsHlo used as a carbon source and ferrocene Fe(C5H5)2as a catalyst precursor. 4.1.2. Q-TG experiments

Figure 10 is an example of the curves of liquid thermodesorption from the surface of the carbon nanotube A5 sample. The Q-DTG curves describe the energetic state of the liquid molecules on the surface of the carbon nanotubes. The energy of interactions between molecules depends on the kind of adsorbate, the properties of the surface on which they are adsorbed, and the porosity of the sample (presence of meso- and micropores), i.e. the heterogeneity of the surface. The differential Q-DTG curves are characterized by the presence of one distinct peak. This is the evidence for continuity of thermodesorption and a monotonic change of adsorption layer properties depending on the distance of the liquid molecule from the surface. Molecules are bonded with the surface by forces of different power and reach. In the subsequent stages of sample desorption, adsorbed liquid layers (including capillary condensation) that remain in the reach of surface forces and form active surface centres are removed. Liquid adsorption with active centres on the surface depends on bonds of the same type. The Q-TG mass loss and the Q-DTG differential mass loss curves of liquids as a function of temperature from the N-1, N-2 and N-3 carbon nanotube surfaces are presented in Figure 11. The characteristic inflections in the Q-DTG curves correspond to the individual stages of thermodesorption of the selected liquids from nanotube surfaces. The Q-DTG curve is a type of spectrum of thermodesorption process, describing the energetic states of polar and nonpolar molecules on the surface. The spectrum indicates long wide peaks with the minima near 70 "C (N-lhenzene), 115 "C (N-2111-octane) and 120 "C (N-3111butanol) and a few other small peaks. The data presented in Table 2 show that the samples are highly sensitive to water vapour because the mechanism of molecular adsorption depends largely on the activated surface centres. Table 2. Adsomtion cavacitv. a. and surface coverage. 8, of tested samvles. I

I n-bntanol I benzene I n-octane 1 a a a 0 0 0 0 (mmoVmZ) (mmoVg) (mmo~m? (mmoVg) (mmo~m" (mmoUg) (mmoVm2) 0.22 2.15 0.03 5.12 0.06 2.79 0.03 0.08 0.08 1.34 0.04 0.24 2.99 2.95 0.13 6.70 0.08 4.50 0.05 2.73 0.03

water

Samples

a

N-1 N-2

(mmoVg) 19.42 8.80

N-3

11.40

-

2, F

IM; ;:I y/! ;Ix[ e

-0.2

50

80 40

0

50

100

T PC1

150

200

e

=so

e60

E

~0.4

--02

4

60

-0.6

40

250

0.8

10

--03

0

50

100

T PC1

150

200

250

-1.2

40

0

50

lW

1M

200

250

T PC1

Figure 10. Thermodesorption of liquids from the surface of the sample A5 under quasi-static conditions.

Figure 11. The Q-TG and Q-DTG mass loss curves of thermodesorption of benzene, n-octane and n-butanol fiom N-1, N-2 and N-3 samples, respectively. Tables 3 and 4 present the ranges of Ed value changes for individual systems. The shape of the curves of the thermodesorption energy distribution functions presented in Figure 12 are similar to Gaussian curves with one distinct peak indicating the existence of one main active centre on the surface of the adsorbents. The increase of desorption energy Edmaxfor polar liquids (in kllmol: 32.29<Edm,<84.59 for water and 24.44<Edm,<60.06 for n-butanol) is evidence of an increase of the adsorbent-adsorbate interaction forces. Broadening of bands on the desorption energy distribution curves indicates the increase of energetic heterogeneity of the studied materials.

Table 3. Comparison of the desorption energy ranges for individual carbon nanotubes.

Table 4. Comparison of desorption energy ranges for individual carbon nanotubes.

The results for selected systems for benzene and n-octane desorption from the surfaces within the temperature range T = 50-180 "C are presented in Figure 10. The E d m a x values range from 28.79 to 43.12 kJ/mol (benzene) and from 25.84 to 41.62 Wmol (n-octane). The high value of the desorption energy of benzene presented in Table 3 indicates great influence of the surface on the adsorbed molecules. The thermodesorption of the above liquids shows that the investigated materials have nonpolar surface properties. The typical Gaussian and bimodal shapes (Figure 13) of the adsorption site distributions associated with the desorption of water from the surface may be observed for the modified N-2 sample, as well as for the adsorption of n-butanol on N-1 and benzene on N-3 pure materials. However, the changes in the distribution of adsorption energy that occur as a result of the chemical treatment appear somewhat complicated. The above treatment resulted in creation of low energy adsorption sites for the adsorbed molecules. For aromatic hydrocarbons, a considerable increase of the value of the energy of desorption occurred in creation of high-energy adsorption sites for liquid molecules (Table 5).

363

Table 5. The minimum, peak and maximium desorption energy values of

This curve exhibits three maxima and this suggests the presence of three types of active site for the adsorption process. Mechanical and chemical modifications of the N-2 nanotube sample thus caused, not only a decrease in its porosity, but also significant changes in its adsorption properties with respect to the adsorption of polar and nonpolar liquids. The decreases in the specific surface area and pore volume of the modified N-2 surface may be attributed to the presence of newly created macropores pores andlor to an internal reorganization of the crystal network [25,351.

4.1.3. Sorptometry measurements The adsorption-desorption nitrogen isotherms at -196 O C plotted from a large number of experimental points are of type I1 according to the BET (BrunauerEmmett-Teller) classification. This type of isotherm describes the process of nitrogen physical adsorption on the adsorbent surface. Because carbon nanotubes have a heterogeneous porous structure, the pores of the smallest diameters micropores are filled at small pressures of the adsorbate. With the increasing concentration of the adsorbate in the gaseous phase, pores of larger and larger diameters are filled and multi-molecular adsorption layer is formed. The presence of the hysteresis loop in the diagrams gives evidence for the presence of open capillaries in the studied adsorbents. During adsorption a liquid surface of a cylindrical shape is formed in the capillaries until they are completely filled. However, in the desorption process, two surfaces of the adsorbed substance are formed in the shape of spherical cone. As a result, the adsorption curve does not overlap the desorption curve in the capillary condensation area. The adsorption capacities determined from the isotherms are characterized by large values. The double-wallet nanotubes have the largest adsorption capacity i.e. 1500 cm31g.For the other nanomaterials the values are in the range 450-950 cm31g. Figure 14 presents the isotherm of nitrogen adsorption and desorption for the sample A5.

To obtain complete information about the textures of the nanomaterials studied, the following parameters were determined: specific surface area, pore volume, pore size and pore distribution according to the size (Table 6). The mean pore radius and the pore volume of the samples were calculated from the desorption isotherms using the Barrett-Joyner-Holend (BJH) method and the specific surface areas from the Brunauer-Emmett-Teller (BET), Langmuir and BJH theories and methods. The values of the specific surface area (SBET)are in the range 178.6 - 294 cm3tg for the samples of multi-wallet nanotubes and form the series A3
A1

60

j (Ed)*10-3[mol/kJ]

benzene n-butanol n-octane water

200

100

A2

benzene n-butanol n-octane water

40

20

0

0

20

30

60

40

50 Ed [kJ/mol]

60

16

benzene n-butanol n-octane water

A3

30

40 50 Ed [kJ/mol]

70

60

70

benzene n-butanol n-octane water

A4

40

j (Ed)*10-3[mol/kJ]

j (Ed)*10-3[mol/kJ]

12

20

8

4

0

0 20

40

60

80

Ed [kJ/mol]

160

benzene n-butanol n-octane water

A5

120 j (Ed)*10-3[mol/kJ]

j (Ed)*10-3[mol/kJ]

300

80

40

0 20

40

60 Ed [kJ/mol]

80

100

10

20

30 40 Ed [kJ/mol]

50

60

0'401 N-21 water

-31 benzene

:-i-

Figure 13. Desorption energy distribution functions for N- lln-butanol, N-2lwater and N-3lbenzene systems.

'"I

+ curve +adsorption desorption curve

Sfrom the adsorption curve

+from the desorption curve

log r, n m

Figure 14. Isotherm of nitrogen adsorption and desorption (left side) and poresize distribution hnctions of sample A5 (right side).

Table 6. The adsorption-structural parameters of carbon nanotubes.

The pore distribution functions in relation to their radii show that the materials studied can be classified as mesoporous adsorbents with the additional presence of micropores. The exception is the sample A2 in which a contribution from micropores on the surface is not observed. From the sorptometry data (Figure 13) the specific surface area (S), total pore volume (V) and pore diameter were calculated. These quantities were determined from the three BrunauerEmmett-Teller (BET), Barrett-Joyner-Holenda (BJH) and Langmuir theories.

(aY)

4.1.4. AFM, SEM and EDXphotographs Figures 15 present, as example the EDX pictures of the surface of the pure and modified alumina oxide samples. Figures 16 and 17 present AFM pictures of the chosen nanomaterials. From the AFM data fractal coefficients were calculated for the tested surfaces using the method described in our paper [40] and commercial program in the NanoScope I11 apparatus and presented in Table 8. Figures 18, 19 and 20 present scanning electron micrographs of the N-1, N-2 and N-3 carbon nanotube samples, respectively [23,25].

.-

t.laq=3 COK X

'""

I.,_ .$id

- 13 m

- hl.70 - .,

00 K X

r A

'JWP-.,:

-

r'?

Fig 15. The EDX pictures of pure alumina (left) and carbon-covered alumina CCA03 (right) surface samples.

a) sample A 1

b) sample A2

c) sample A3

d) sample A4

e) sample AS

Figure 16. AFM photos of carbon nanotube surfaces.

call carbon nanutubes

N-2-rnolri\r,all carbon nanorubes

S-3-multiaall carbon nanotublbes

Figure 17. AFM photos of the some of the carbon nanotubes

Figure 18. Scanning electron micrograph of the N-1 sample.

Figure 19. Scanning electron micrograph of the N-2 sample.

4.2. Montmorillonites 4.2.1. Materials The materials tested were Na- and La-montmorillonites from Lago Pellegrini (Argentina) [38]. The substituted samples were obtained by saturation of the ion-exchange capacities of the water-saturated clay samples with sodium andlor lanthanium chloride (0.5 M). Finally, the Na- and La- samples were air dried. Moreover, thermodesorption of liquids from natural zeolite-clinoptilolite and zeolite-mordenite (from Ukrainian Transcarpathian region) were made. 4.2.2. Porosimetry measurements Figure 21 presents the graphs of mercury-penetrated volume versus pressure in the pores of Na- and La-montmorimllonitamples. Figure 22 shows the pore-size distribution functions calculated from the porosimetry data. The porosimetry measurements showed that the mean pore radius and volume of the samples calculated from the V = f (P) curves to be 220.3 A (Na-sample) and 171.7 (La-sample) and 0.12 cm31g (Na-sample) and 0.14 cm31g (Lasample), respectively. The specific surface areas of the studied materials were 4.68 m21g (Na-sample) and 2.81 m21g and are different from those obtained from sorptometry because destroyed sample measurements and the differences in the techniques.

;:/

,,,,,,,

0 1

10

100

1000

Volume, (mm3)

10000

Volume, (mm3)

Figure 21. The mercury pressure in relation to pore volume of Namontmorillonite (left) and La-montmorillonite (right) samples.

*-.

2.5 2.0 -

? =ar 1.0- 1.50.5 0.0

10

100

1WO

10

10000

100

1000

10000

Pore radius (nm)

Pore radius (nm)

Figure 22. Pore-size distribution functions of Na- (left) and La-montmorillonite (right) samples from the porosimetry technique. 4.2.3. Diffusion coeficients Additionally, calculation of the diffusion coefficient D of an adsorbed layer, based on the analytical solution of Fick's law of diffusion for desorption processes in liquids, in temperature range of 20-200 "C were made for liquids1 montmorillonite-Na and -La samples [38]. The D parameters of the tested materials are presented in Table 8 and they are close to the value presented in the literature. Gay-Duchnal et al. [59] used quasi-elastic incoherent neutron scattering (QINS) to study the diffusion of interlayer water in partially oriented m2 s-' was montmorillonite-Na samples and a diffusion coefficient D = 1.3 x obtained.

Table 7. Diffusion coefficients (in cm2 s") of polar and nonpolar liquids during thermodesorption processes from montrnorillonites-Na and -La. Samples Na-

montmorillonite Lamontmorillonite

Water

Benzene

n-Butanol

6.7~10-l4 I X ~ O - ~ ~ 2.lxl0-~~ 5.4x10-'~ 3.1~10-l2 5.6~10-l2 7 . 8 ~ 1 0 " ~ 1 . 2 ~ 1 0 - ' ~ 2.7~10-l3 3.4~10-l2 3.9~10-l2 1.1~10-"

n-Octane 2.5~10-l3 3.2~10-l2 3.6~10.'~ 4.8~10-l2

4.3. Aluminas 4.3.1. Materials The y-A1203precursors were prepared by impregnation of commercial y-A1203 (SBA1500, Engelhard) with aqueous solutions of sucrose. ARer being dried at 90 "C the precursors were calcined at 600 OC in N2. The final products discussed here were denoted as CCA03 and CCA03-2. To obtain higher carbon content, CCA03 was further impregnated with an aqueous solution of sucrose to give CCA03-2 sample. More information is given in [60,61]. 4.3.2. Q-TG experiments Figure 23 presents the Q-TG mass loss and Q-DTG differential mass loss curves with respect to temperature for thermodesorption of n-octane from

neutral porous A1203(Aluminum Co. of America, AL-COA Centre, PA, USA) saturated with vapour in a desiccator. The curves are characterized by a few inflections (on Q-TG) and peaks (on Q-DTG) resulting from successive stages of evaporation of n-octane from different energetic states on the alumina surfaces. The thickness of the adsorbed layers is controlled by altering the mode of immersion of the samples. Thus, immersion in n-octane vapours in a desiccator with plp, = 1 saturates all adsorption sites, as well as surface and capillary forces. Under such conditions, the surface and capillary forces are compensated as in the McBain balance static adsorption method. Samples prepared in this way do not contain excess bulk liquid, which would affect the kinetics and mechanism of thermodesorption by lateral interactions between molecules. The Q-DTG curves are characteristic 'spectra7 of the thermodesorption process and reflect an energetic state of n-octane molecules on an alumina surface with various heterogeneous properties. The shape of Q-DTG curves result from the discontinuous properties of adsorbed layers and disruption of the adsorbate-adsorbate and adsorbate-adsorbent bonds. The adsorption energies of n-octane result from dispersion interactions. The thermodesorption of n-octane from alumina surfaces can be explained as follows. The first stage is the evaporation of liquid from capillaries (desorption of the liquid film in the region of capillary condensation on the adsorption isotherm). During the next stages, the thermodesorption of liquid takes place from mesopores and, finally, from the surface and active sites of the samples. Using single Q-TG and Q-DTG curves it is possible to determine the adsorption capacity 0 (e.g. thickness of adsorbed film), pore volume (i.e. geometrical heterogeneity), discontinuous change of adsorbed layer properties, mechanism of the wetting process and the energy of interactions between liquid molecules and solid surfaces (i.e. energetic heterogeneity) [5 1,601.

285 280

0.1

Q-TG

300 320 340 360 380 400 420 440 460 480 500 520 540

Temperature [K]

Figure 23. The Q-TG and Q-DTG mass loss curves of thermodesorption of n-octane from the surface of a neutral alumina sample. The Q-TG curve in Figure 23 shows seven distinct peaks, corresponding to values of their maximum temperatures, (in K): 306.2, 352.2, 393.2,431.6,453.2, 485.2, 51 1.2. The numerical analysis of these separate values of T confirms the presence of seven different types of adsorption centres on the surface. Further analysis of the Q-TG and Q-DTG curves was made using the models described in [29]. From the experimental Q-TG and Q-DTG curves, the dependence:

was calculated. The curve obtained, presented in Figure 24, is characterized by great non-linearity, as confirmed by the large extent of surface energetic heterogeneity. Graphical analysis of the dependence (22) by means of the tangents method makes it possible to separate seven intervals with different values of Ed on the curve presented in Figure 25. From the numerical analysis of the dependence given in equation (22), using the method given in [29,50] the following dependences of desorption energy on temperature were obtained: Ed = 0.18T- 38 (kJImol) for the interval 0 < T < 403.2 K Ed = 0.3 1T- 91 (kJImol) for the interval 403.2 < T < 520 K

(23)

7

70

A h 0 3 (neulml) OCTANE

60 -

6-

F

-

.

8 51

e2 0

53 i

w"

4:

1.8

-

40-

30 -

3-

27

50 -

20 -

. , . , . , . , . , . , . 2.0

2.2

2.4 2.6 2.8 11~*10-~

Figure 24. The dependence of -ln(-1/13 d9)IdT on the inverse temperature.

3.0

3.2

A1203 (neutral) OCTANE . . . . . . . . . . . . . . . . . . . . . . 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

e

Figure 25. The curve of the desorption energy on the surface coverage 9.

The dependence of Ed on temperature is shown by the curve composed of two straight lines intersecting at T = 403.2 K. The boiling point of n-octane is Tb= 398.9 K. Equation (23) shows two ranges of dependences Ed = f(T) on the surface of the process of different character. Dependence of the desorption energy on the extent of coverage 9 of the surface is presented in Figure 25: for the interval 0 < T < 403.2 K (24) Ed = 35.5 - 9.24 In 9 Ed=21.5 -45.21 In 9 for the interval 0 < T < 403.2 K Figure 25 indicates that for the thermodynamic system adsorbate-adsorbent the dependence of the energy Ed on the surface coverage 9 is of exponential character. One can see two sections with the intersection point at T = 403.2 K. The boiling point of octane corresponds to the intersection point. Figure 26 presents the diagram of the dependence of Ed on In 9. The dependence of the entropy coefficient v on the desorption energy Ed is given by the following equation [50]: Calculation of the desorption energy distribution function was made using the equation [50]:

The diagram obtained (Figure 27) is qualitatively in agreement with the massloss curve whereby the Q-DTG peaks in Figure 23 are in agreement within the accuracy of the error calculation. Figure 28 presents normalized values of

distribution functions calculated from equation (26) (curve 2) and, for comparison (curve 1) using the method given in [62]. Both distribution functions were calculated from the Q-TG curve (Figure 231. The agreement of the results calculated using the above methods is observed only for the interval of the mesopores because diameter of n-octane molecules.

A1203 (neutral) OCTANE 10 -3.2

-2.8

-2.4

-2.0

-1.6

-1.2

-0.8

-0.4

0.0

Figure 26. The dependence of the desorption energy Ed on the In 8 value.

Figure 27. Energy distribution function from equation (26).

The diameter of the n-octane molecule is 0.75 nm. Calculations give the thickness of the layer, for horizontal positions of the molecules, as 0.49 nm and for the vertical position of the molecules, as 1.33 nm. It seems that, starting from the second layer, the dependence between the kinetic and potential energies of the molecular interactions favours vertical orientation of spherical cylindrical molecules of n-octane. When the point of multilayer saturation of the surface is approached, mobility of the molecules in the layer increases causing "swelling". It was assumed in the calculations that, in the first layer, the molecules are oriented horizontally and in the next ones, vertically. The thickness of the adsorption layer h was calculated. Figure 29 presents the dependence of thickness of the layer h as a function of plpo. The dependence obtained is in agreement with the analogous adsorption of n-hexane on active carbon [27]. Slightly greater values of the quantity t are due to the increase of the crosssectional area, value for n-octane and differences in the orientation of the molecules. The sizes of the pore radii were calculated from equation (27). Volumes of mesopores were calculated from the molar volume of the volumetric liquid. Errors in the size of the molar volume are important only for pores smaller than 7 nm because spherical orientation of n-octane molecules during adsorption

process. Preliminary calculations showed that the increase in the molar volume of n-octane in the neighbourhood of mesopores close to the critical radii for noctane pk=3 nm (it is depend on orientation of n-octane molecule during adsorption) does not exceed 15%, but for pores of diameter larger than 7 nm the change of molar volume can be neglected. Figure 30 shows the dependence of the volumes V of the mesopores on their radii r.

Figure 28. Energy distribution curves calculated from equation (26) - curve A and using methods [5] - curve V.

Figure 29. The plot of the dependence of the thickness of adsorption layer h onplp,.

To confirm the calculations of the radii of pores r, the coefficient of conversion of core volume into volume of pores (V) for cylindrical pores was calculated using the equation [49,50,63]:

The value of V changes from 1.95 for pip, = 0.4 to 1.35 for pip, = 0.8. These values of V are in agreement with the values calculated for n-hexane [27]. The dependence of pore volume on their radii presented in Figure 30 gave:

0.005

AI,O, (neutral) OCTANE

2

0.004 -

0.003 -

G.

L

a %

0.002

-

0.001 -

0

.

0 0

0

0 2

, . 4

,

.

6

,

.

6

~ . ~ . ~ 1 0 1 2 1 4

r lnml

Figure 30. The relationship between pore volume and pore radius

Figure 3 1. Pore-size distribution function

The volume distribution of mesopores in relation to their radii, calculated from equation (27), can be given in the form [49]:

where the surface fractal dimension, Fd = 2.72. The values of the mesopore volumes and radii and the value of the fractal dimension, Fd, are convergent, as for the size order, with the corresponding values obtained during studies of the structure A1203 [64]. Values of pore volume near 2 nm have the highest intensity in this curve. The concentration of mesopores decreases with increase of their radius. The observed shape of the pore-size distribution curve is typical for most industrial mesoporous adsorbents. For example, this shape is similar to that found from low-temperature nitrogen adsorption isotherms on various activated carbons by using the Dollimore-Heal method [63].

.

~

.

-

400

-e-

-e-

adsorption curve desorption curve

adsorption curve desorption curve

-e-

300

adsorption curve desorption curve

6

% 2cQ

2 >

1cQ

103

50

0

0.2

0.4

PIP0

0.6

0.8

1

0

0.2

0.4

0.6 PIP0

0.8

1

0

0.2

0.4

0.6

0.8

1

Figure 32. Nitrogen adsorption isotherms of pure A1203(left), CCA03 (middle) and CCA03-2 (right) samples.

"1

-from +from

adsorption curve desorption curve

'P

-from +from

adsorption curve desorption curve

-from +from

adsorption curve desorption curve

f

Figure 33. Pore-size distribution hctions of pure A1203(left), CCA03 (middle) and CCA03-2 (right) samples. Nitrogen adsorption-desorption isotherms at 77.35 K of pure A1203, CCA03 and CCA03-2 (right) samples are shown in Figure 32. The adsorptiondesorption hysteresis loops appear at relative pressures range from 0.7 to 1. Such shapes of the hysteresis loops arise from open capillaries with various crosssectional configurations (circular, triangular, square, and so on). In such capillaries, condensation sometimes initially occurs along their internal angles until a cylindrical meniscus, the result of the joining of angular menisci, appears. Further condensation takes place until the filling-up of pores is complete, slightly changing the relative pressure. Figure 33 presents differential pore-size distribution functions obtained from the BJH method on the basis of adsorption and desorption data of the above samples. The dVldr = f(r) curves are of Gaussian type and the single sharp peaks on the curves correspond to pore radii of about 5-20 A. It can be noticed that the samples tested have both mesopores and micropores, which causes their high specific surface areas [65].

The Q-TG and Q-DTG curves for the thermodesorption of benzene from an alumina oxide surface are shown in Figure 34. The Q-DTG curve shows one peak and a few inflections, corresponding to evaporation of benzene molecules from pores and active centres of the sample surface. From the data of mass-loss of the sample in relation to time, values of milmawere calculated, where mi is the sample mass after the time of thermodesorption z i, and m, is the initial mass of the sample. A plot of:

is shown in Figure 35. Thermodesorption can be divided into two time intervals [52]: 1. z < 2500 s in which desorption from the sample surface takes place (2500 s is the time corresponding to benzene evaporation at its boiling temperature), and 2. z > 2500 s in which is characteristic of benzene difhsion in the alumina oxide pores. The second stage of thermodesorption is described by the following kinetic equation:

where: Fa= D z/h2, h = plate thickness, pn and Bnare constants taken from [66]. From an analysis of the kinetic curve (Figure 35) it is possible to draw the conclusion [52] that the process of difhsion is described by the first member of the series equation (32): m,lm,

= B,

exp (-:p F,)

(32)

Differentiating equation (32) gives: (llm,)l(dm,ldz)

=

- B, 01:

~ l h exp ~ )(- , D : F,)

(33)

7 [7

Figure 34. The Q-TG and Q-DTG thermodesorption mass loss curves of benzene from the alumina oxide sample made using the programmed quasiisothermal method.

Figure 35. Dependence of ln(milm,) on time.

Figure 36. The Arrhenius diagram of the dependence of the diffusion coefficient on 11T.

Dividing equation (34) into equation (32):

From equation (34) the diffusion coefficient D can be calculated from equation (35):

D = - (h2lPc(,2)(llm,)l(dm,ldtz) (35) From equation (35) the diffusion coefficient can be calculated using h= 10" m, .pc(,2= 2.46. An Arrhenius-type plot of: l n D = f(l/T)

is presented in Figure 36. The value obtained for the diffusion coefficient of benzene in the pores of the alumina oxide sample is 1.16 x 10-lom 2s-' at T = 25 "C.

4.4. Fractal dimensions

-

0 2.4

2.5

2.6

2.7

2.8

2.9

2.2

2.4

Fractal coefficients from the sarptomatic method

2.6

2.8

Fractal coefficients from the AFM method

Figure 37. The dependence of specific surface areas and fractal coefficients calculated from sorptomatic (left side) and AFM methods of some nanomaterials. The fractal dimensions of the nanoparticles described above were calculated and are given in Table 8. The data in Table 8 show that the fractal dimensions of the selected advanced materials, calculated using different and independent measuring methods, are in good agreement. Figures 37 and 38 present relationships between specific surface areas and pore diameters and fractal coefficients [35] calculated from sorptometry and AFM methods for the selected nanoparticles. From the Figures, it appears that bigger fractal coefficients have been obtained for bigger surface areas and smaller pore diameters (micropores) of surfaces studied.

4

2.5 2.6 2.7 2.8 2.9 Fractalcoefficients from the sarptomatic method

, . 2.5, . 2.6, . 2.7, .

2.4

2.8

2.9

Fractal coefficients from the AFM method

Figure 38. The dependence of pore diameters and fractal coefficients calculated from sorptomatic (left side) and AFM (right side) methods of some nanomaterials.

5. SUMMARY

Applications of thermogravimetry (Q-TG) for the investigation of adsorbed liquid layers and porosity parameters used for the quantitative characterisation of the energetic and geometrical (e.g. total) heterogeneities of typical advanced materials have been presented. The technique is very useful for investigating the physicochemical properties of surface liquid films, adsorbate-adsorbent interactions and total surface heterogeneity. The thermodesorption of liquids depends on surface wetting phenomena and the surface properties of the solid surfaces. Comparison of thermogravimetric and other data provides new information about the adsorption and pore structure of the materials studied. Thermodesorption in a continuous manner is evidenced by single inflexions on the Q-TG curves and indicates the occurrence of one main active centre. Applications of Q-TG to the investigation of adsorbed liquid layers allows porosity parameters to be determined and used for the quantitative characterisation of the energetic and geometrical (e.g. total) heterogeneities of the surfaces of carbon nanotubes. The fractal dimensions of the surfaces of the materials studied have been calculated using Q-TG and independent techniques. Q-TG results are in good agreement with the results from sorptometry, porosimetry and AFM techniques and can be used for calculation of the pore-size distribution functions. Relationships between the specific surface areas and pore diameters and fractal coefficients calculated from sorptometry and AFM methods of selected nanoparticles have been found. A new method for the calculation of diffusion coefficients from Q-TG data has been presented. The programmed thermodesorption of polar and nonpolar liquids from aluminium oxide and montmorillonite-Na and -La samples under quasi-isothermal conditions has been studied. The results from above methods were compared with literature data and good correlation was obtained. Studies of low-temperature nitrogen adsorption on carbon nanoparticles and total pore volume confirmed that the values of specific surface area (SBET) depend on the size and structure of the sample. Comparison of the complex QTG, sorptometry and AFM data provides new information about the adsorption and structure of the materials studied.

Table 8. Fractal dimensions of chosen advanced material surfaces calculated on the basis of sorptometry, thermogravimetry Q-TG, porosimetry and AFM data.

nanotubes double-

A single Q-TG experiment can be used for quantitative characterization of the structural and energetic heterogeneities of mesoporous solids. The evaluated energy distribution and pore-size distribution functions of n-octane on alumina oxide surface (e.g. total heterogeneity) agree satisfactorily with the parameters of the porous structure of alumina and with known characteristics of n-alkane adsorption. Studies can be extended to other nanomaterials with different pore sizes, using adsorbates possessing different acidbase properties and wettability.

6. REFERENCES 1. P. Staszczuk, J. Therm. Anal. Cal., 79 (2005) 545. 2. J. Jortner and C.N.R. Rao, Pure Appl. Chem., 74 (2002) 1491. 3. M.S. Dresselhaus, G. Dresselhaus, NanoStructured Materials, 9 (1997) 33. 4. M. Kociak, A.Yu. Kasumov, S. Gueron, B. Reulet, 1.1. Khodos, Yu.B. Gorbatov, V.T. Volkov, L.Vaccarini and H. Bouchiat, Phys. Rev. Letters, 86 (2001) 2416. 5. V.V. Yashchenko, C.I. Vanegas and B.V. Romanovsky, React. Kinet. Catal. Letters, 40 (1989) 235. 6. A.P. Alivisatos, Swiat Nauki, 123 (2001) 57. , (2001) 64. 7. K.E. Drexler, Swiat ~ a u k i123 8. R.E. Smalley, ~ w i a~t a u k i123 , (2001) 66. , (2001) 74. 9. S. Ashley, Swiat ~ a u k i 123 10. G.P. Collins, Swiat Nauki, 123 (2001) 76. 11. A. Waksmundzki, P. Staszczuk and E. Szymanski, Polish J. Soil Sci., 14 (1981) 25. 12. P. Staszczuk and A. Waksmundzki, Problemy Agrofizyki, Ossolineum Wroclaw, 37 (1982) 1-81 (review). 13. S.L. Swartzen-Allen and E. Matijevic, Chem. Rev., 74 (1974) 385. 14. E. Chibowski and P. Staszczuk, Clays Clay Minerals, 36 (1988) 455. 15. P. Staszczuk, M.A. Cabrerizo-Vilchez and R. Hidalgo-Alvarez, Colloid Polym. Sci., 27 1 (1993) 759. 16. P. Staszczuk, J. Them. Anal., 48 (1997) 755. 17. P. Staszczuk, T. Danielkiewicz and J. Klinowski, Adsorption Sci. Tech., 18 (2000) 307. 18. P. Staszczuk, M. Majdan, T. Danielkiewicz and M. Matyjewicz, Prog. Colloid Polym. Sci., 118 (2001) 100. 19. P. Staszczuk, D. Sternik and G.W. Chqdzynski, J. Therm. Anal., 71 (2003) 173. 20. M. Majdan, M. Kowalska-Ternes, S. Pikus, P. Staszczuk, H. Skrzypek and E. Ziqba, J. Mol. Struct., 649 (2003) 279.

21. V. Babiuk, P. Staszczuk, 0. Bodak, Yu. Gorelenko, L. Romka and Yu. Stadnyk, J. Alloys Compounds, 365 (2004) 21. 22. V. Kurpusyuk, N. Semus'o, R. Gladyszewski and P. Staszczuk, J. Alloys Compounds, Annals Pol. Chem. Soc., 2 (2003) 692. 23. P. Staszczuk, M. Matyjewicz, E. Kowalska, J. Radomska, P. Byszewski and M. Kozlowski, Proceedings of SPIE, Nanotechnology, 51 18 (2003) 245. 24. P. Staszczuk and D. Sternik, Proceedings of SPIE, Bioengineered and Bioinspired Systems, 5 119 (2003) 2 10. 25. P. Staszczuk, M. Matyjewicz, E. Kowalska, J. Radomska, P. Byszewski and M. Kozlowski, Rev. Adv. Mater. Sci., 5 (2003) 34. 26. M. Matyjewicz and P. Staszczuk, J. Therm. Anal. Cal., 74 (2003) 413. 27. S.J. Gregg and K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press, London, 1982. 28. W. Rudzinski and D.H. Everett, Adsorption of Gases on Heterogeneous Surfaces, Academic Press, New York 1992. 29. P. Staszczuk, D. Sternik and V.V. Kutarov, J. Them. Anal. Cal., 69 (2002) 23. 30. D. Avnir (Ed.), The Fractal Approach to Heterogeneous Chemistry, Wiley, 1989, New York. 3 1. M. M. Novak, (Ed.), Fractals in the Natural and Applied Sciences, North-Holland, 1994, Amsterdam. 32. B.B. Mandelbrot, The Fractal Geometry of Nature, Freeman, 1983, New York. 33. P. Pfeifer and M. Obert, in The Fractal Approach to Heterogeneous Chemistry, (Ed. D. Avnir), Wiley, Chichester, 1989. 34. A. Provata, P. Falaras and A. Xagass, Chem. Phys. Lett., 297 (1998) 484. 35. D. Sternik, M. Blachnio, P. Staszczuk, G.W. Chqdzynski and E. Kowalska, J. Therm. Anal. Cal., 86 (2006)(1), 101. 36. L. Stoch, Mineraly ilaste (Clays minerals), Wyd. Geologiczne, Warszawa, 1974. 37. E. Chibowski and P. Staszczuk, Clays Clay Minerals, 36 (1988) 455. 38. P. Staszczuk, J.C. Bazan, M. Blachnio, D. Sternik and N.J. Garcia, J. Therm. Anal. Cal., 86 (2006) 57. 39. P. Staszczuk, M. Majdan, T. Danielkiewicz and M. Matyjewicz, Prog. Colloid Polym. Sci., 118 (2001) 100. 40. M. Majdan, S. Pikus, M. Kowalska-Ternes, A. Gladysz-Ploska, P. Staszczuk, L. Fuks and H. Skrzypek, J. Colloid Interface Sci., 262 (2003) 321. 41. M. Majdan, M. Kowalska-Temes, S. Pikus, P. Staszczuk, H. Skrzypek and E. Ziqba, J. Molec. Struct., 649 (2003) 279. 42. P. Staszczuk, G.W. Chqdzybski and D. Sternik, J. Therm. Anal. Cal., 62 (2000) 45 1.

43. P. Staszczuk, D. Stemik and G.W. Chqdzyriski, J. Therm. Anal. Cal., 71 (2003) 173. 44. P. Staszczuk, D. Stemik, G.W. Chqdzybski and V.V. Kutarov, J. Alloys Compounds, 367 (2004) 277. 45. G.W. Chqdzynski, V.V. Kutarov and P. Staszczuk, J. Therm. Anal. Cal., 76 (2004) 633. 46. F. Paulik, Special Trends in Thermal Analysis, J. Wiley & Sons, Chichester, 1995. 47. P. Staszczuk, American Lab., 28 (1996) 21. 48. P. Staszczuk, Themochim. Acta, 308 (1998) 147 (Rewiev). 49. P. Staszczuk, V.V. Kutarov and M. Planda, J. Therm. Anal. Cal., 7 1 (2003) 445. 50. V.I. Bogillo, P. Staszczuk, J. Therm. Anal. Cal., 55 (1999) 493. 51. M. Matyjewicz and P. Staszczuk, J Therm. Anal. Cal., 74 (2003) 413. 52. P. Staszczuk, M. Planda-Czyz, M. Blachnio and V.V. Kutarov, J. Them. Anal. Cal., 86 (2006) 261. 53. D. Rutheven, Principles of Adsorption and Adsorption Process, Wiley, New York, 1984, pp. 124-205. 54. H.S. Carslaw and J.C. Jaeger, Conduction of Heat in Solids, 2nded., Oxford University Press, Oxford, 1959, p. 100. 55. A.B. Kisielev, in The Structure and Properties of Porous Materials, Butterworths, London, 1958, p. 195. 56. W.I. Frisen and R.J. Mikula, J. Colloid Interface Sci., 120 (1987) 263. 57. S.P. Rigby, R.S. Fletcher and S.N. Riley, J. Colloid Interface. Sci., 240, (2001) 190. 58. J. Zhuangde, W. Hairong and F. Bin, Mach. Tools Manuf., 41 (2001) 2183. 59. M. Gay-Duchosal, D.H. Powell, R.E. Lecher and B. Ruffle, Physica B, 276-278, (2000), 234. 60. L. Lin, W. Lin, Y.X. Zhu, B.Y. Zhao, Y.C. Xie, Y. He and Y.F. Zhu, J. Molec. Catalysis A: Chemical, 236 (2005) 46. 61. L. Lin, W. Lin, Y.X. Zhu, B.Y. Zhao and Y.C. Xie, Langmuir, 21 (2005) 5040. 62. V.V. Yashchenko, C.I. Vanegas and B.V. Romanovsky, React. Kinet. Catal. Letters, 40 (1989) 235. 63. V.I. Bogillo, V.P. Shakilev and A. Voelkel, J. Material Chem., 8 (1998) 1953. 64. D. Dollimore and G.R. Heal, J. Colloid Interface Sci., 33 (1970) 508. 65. P. Staszczuk, L. Lin, Y.X. Zhu and M. Blachnio, J. Therm. Anal. Cal., 86 (2006)(1) 5 1. 66. D.P. Timofiejev, Kinetika Adsorpcji, Izd. Akad. Nauk SSSR, Moskwa, 1962.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter I I HETEROGENEOUS CATALYSIS ON SOLIDS Ljiljana ~amjanovic"~ and Aline ~ u r o u x l * '~nstitutde Recherches sur la Catalyse et 17Environnementde Lyon, UMR 5256 CNRS - UCB Lyon 1 , 2 avenue Einstein, 69626 Villeurbanne Cedex, France 2permanent address: Faculty of Physical Chemistry, University of Belgrade, Studentski trg 12-16, 11000 Belgrade, Serbia *corresponding author: [email protected]

1. INTRODUCTION It is a readily accepted fact that surface heterogeneity leads to a heterogeneous population of active sites on the surface of a catalyst. It is very often discussed but very seldom taken into account in practical cases, simply because there are very few tools with which to study the heterogeneity of active sites in catalysis. Most attempts to do so have been based upon the use of selective poisons. However, thanks to the improved sensitivity of calorimeters and to the development of refined data analysis techniques, adsorption calorimetry can make a significant contribution to the characterization of a catalytic surface [I]. In particular, the surface properties of a solid can be conveniently investigated by studying the adsorption of suitably chosen probe molecules. The amount of heat evolved during the adsorption process is closely related to the adsorbate substrate bond strength. Furthermore, the differential heat of adsorption is often dependent on the surface coverage of the adsorbate, due to the lateral adsorbate - adsorbate interactions or due to the surface heterogeneity. So the role of the probe is one of the most decisive parameters in the determination of heats of adsorption. Moreover many catalytic reactions are structure sensitive, and proceed at a rate that depends on the detailed geometric structure of the surface atoms of the catalyst. Norskov et al. [2] have demonstrated that the heat of adsorption of a species is directly related to the local structure of the catalyst, and that the more accessible sites are more active unless poisoned and bind the adsorbate more strongly. Therefore, the determination of the energy profile, i.e. the energy of the adsorbed phase as a function of loading, is an essential component of the characterization of the surface active sites of catalysts and/or supports of

catalysts [3]. Besides the data gathered from traditional methods for the study of solid surfaces such as Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), temperature-programmed oxidation (TPO), temperature-programmed reduction (TPR) or temperature-programmed desorption (TPD) experiments, characterization requires information on the energy (or enthalpy) evolved when an adsorbate or a reactive gas is put in contact with the surface. The objective of this chapter is to show how the adsorption enthalpy profile of a probe molecule can be used to characterize the surface sites. Adsorption microcalorimetry permits an accurate determination of the strength and strength distribution of surface sites based on the heat of adsorption of suitable probe molecules and the differential heat vs. coverage curve. A survey of applications of microcalorimetry to heterogeneous catalysis will be given, with particular emphasis on the determination of the acid-base properties of zeolites, molecular sieves, metal oxides and supported metal oxides. Additional applications to the study of reducible metal oxides and their redox character will be discussed. The heat of adsorption measured experimentally depends upon the imposed conditions, such as the temperature of the adsorption which means the temperature of the calorimeter, the sample pretreatment, the diffusion problem and the quality of the probe, among others. The probe will depend on what has to be determined: the acid-base character of the catalyst, the redox properties or the surface distribution of metal particles on a supported catalyst. The contribution of adsorption calorimetry to the problem of characterizing various types of catalyst surfaces has been established by numerous research articles and extensive reviews which have appeared in the literature [4-141. 2. EXPERIMENTAL

The thermochemical techniques most commonly used to investigate the acidbase or redox character of catalyst surfaces are DTA, TG, DTG, DSC, and calorimetry. These techniques can be used either by themselves, or combined with other techniques (for instance, TG-DSC, calorimetry-volumetry, DSCchromatography, DSC-MS, etc.) [lo]. Techniques such as DTA, DTG or DSC study the thermal behaviour of a catalyst as it undergoes heating at a constant rate (or by steps of temperature) and are particularly adapted for studying the decomposition of catalyst precursors or for desorption studies involving poisoned catalysts. By contrast, in the adsorption microcalorimetry technique, the sample is kept at a constant temperature while a probe molecule adsorbs onto its surface, and a heat-flow detector emits a signal proportional to the amount of heat transferred per unit time.

Values of the enthalpy of adsorption, determined either from the variation of adsorption with temperature (isosteric enthalpy of adsorption) or by direct calorimetric measurements, provide a valuable insight into the mechanism of adsorption. When taken together with the data from adsorption isotherms, they provide information which could not be extracted from either set of data alone. Heats of adsorption and other thermodynamic parameters can be obtained either by direct calorimetric determination, -AH = (where n, = adsorbed amount), or by using the Clausius-Clapeyron equation and the data from isosteric measurements. However, the fact that adsorption is often irreversible in the presence of micropores, frequently makes estimates of adsorption heats obtained from isosteres very unreliable. The average errors in the evaluation of the differential heats of adsorption estimated by Stach et al. [15] are 1-2 % only for the direct measurement and around 5 % for the isosteric measurements. The direct measurement of heats is also more accurate than the computation of energies from TPD data, which quantifies average activation energies of desorption. Whereas calorimetry can yield a detailed picture of the distribution of strong acid sites on the catalyst surface, the TPD of ammonia usually yields only an average value [16], except when appropriate kinetic models are employed. the amount adsorbed (na) In the direct calorimetric determination is calculated either from the variations of the gas pressure in a known volume (volumetric determination) or from variations of the mass of the catalyst sample in a static or continuous-flow apparatus (gravimetric determination). In a static adsorption system, the gas is brought into contact with the catalyst sample in successive doses, whereas in a dynamic apparatus the catalyst is swept by a continuous flow. Comparative calorimetric studies of the acidity of zeolites by measuring ammonia adsorption and desorption using static (calorimetry linked to volumetry) and temperature-programmed (DSC linked to TG) methods can be found in the literature [17]. Another interesting comparison of various thermal analysis techniques, namely adsorption microcalorimetry, TPD and temperature-programmed reaction using constant rate thermal analysis (CRTA), has been performed by Fesenko et al. in order to study the reactivity of zeolites in terms of the adsorption or desorption of base probe molecules [IS]. As an example, CRTA was used to study the desorption of isopropylamine from Na-Y zeolite and its acidic form HY. Calorimetry linked to a volumetric technique is still the most commonly used method to this date [14,19]. The typical experimental procedure is the following. Prior to the adsorption, the catalyst is outgassed at the desired temperature and under high vacuum (- 0.1 mPa) in the calorimetric cell. After cooling to the adsorption temperature and establishing the thermal equilibrium of the

calorimeter, a dose of gaseous probe molecules is brought into contact with the catalyst sample, and both the pressure and heat signal are monitored until equilibrium is reached. Then, successive new doses are added and the new equilibrium pressures are recorded together with the corresponding evolved heats. In order to detect energetic heterogeneity of the surface, which is of primary interest in catalysis, small doses of probe gas (typically < 10 pmol g-' of catalyst) have to be admitted successively on the solid in order to saturate the active sites progressively. The data obtained directly from adsorption calorimetry measurements can be plotted as in Figure la, which represents both the calorimetric signal and the evolution of the pressure for each dose as a function of time. By accumulating these raw data until full coverage of the surface is attained, the adsorption calorimetry data can be summarized on various plots (Figure 1): (i) The amount of gas adsorbed at constant temperature plotted as a function of the equilibrium pressure (adsorption isotherm I, Figure 1b).

Figure 1.

Calorimetric and volumetric data obtained from adsorption calorimetry measurements [4]

In order to accurately determine the chemisorbed amount from the overall adsorption isotherm, the sample can be outgassed at the same temperature to remove the physically adsorbed species, after which a new adsorption procedure is carried out to obtain isotherm 11. The difference between the first and second isotherms gives the extent of irreversible adsorption (Virr) at a given temperature, thus making it possible to distinguish between physical and chemical adsorption. (ii) The corresponding calorimetric isotherms (Qint vs. P) (Figure lc). (iii) The integral heats (Qint) as a function of the adsorbed amount (nu) (Figure Id). This representation leads to the detection of coverage ranges with constant heat of adsorption, for which the evolved heat is a linear function of the coverage. (iv) The differential heat Qdiff= dQintldna (molar adsorption heat for each dose of adsorbate) as a function of nu (Figure 1e). The ratio of the heat evolved for each increment to the number of moles adsorbed (in the same period) is equal to the average value of the differential enthalpy of adsorption in the considered interval of n, values. The curve showing the differential heat variations in relation to the adsorbed amount is therefore traditionally represented by histograms. However, for simplification, the histogram steps are often replaced by a continuous curve connecting the midpoints of the steps. Although it is often difficult to determine the nature of the adsorbed species, or even to distinguish between the different kinds of adsorbed species from the calorimetric data, the variation of the differential heats of adsorption with coverage depicts quite clearly the distribution of surface sites with respect to a given adsorbate and their varying reactivity on given adsorbents. The curves showing the differential heat variations in relation to the adsorbed amounts generally present the following features: 1) an initial region of high heats of adsorption, representing adsorption on the strongest sites, which are usually thought to be of Lewis type. The initial drop in the curve of QdV vs. coverage can be observed even in the case of adsorption on apparently homogeneous surfaces, where it could be ascribed to residual surface heterogeneities. 2) one or more regions of intermediate strength. A region of constant heat in this domain is characteristic of a set of acid sites of homogeneous strength, such as e.g. Bronsted acid sites in the case of zeolites. 3) a region where heats decrease more or less steeply depending on the heterogeneity of the sites. 4) a region at high coverage where the heat of adsorption approaches a nearly constant value characteristic of hydrogen bonding between the probe and the sample or physisorption of the probe. This constant value depends on the nature of the probe. Figure l e and Figure 2

roughly illustrate some of the most common Qwvs. n, curves corresponding to specific conditions. Figure 2a represents a homogenous distribution of sites (a zeolite presenting only Bronsted acid sites, or a well-dispersed supported metal) and Figure 2b represents a heterogenous distribution of sites (an oxide presenting mainly Lewis acid sites, or a sample presenting diffusion limitations either within the adsorbent layer (external diffusion) or in the pores (internal diffusion)). Of course, all types of profiles between those two extreme cases can be observed, including step-shaped curves. It is important to note that the absence of a plateau of constant heat in the differential heat curve can be either the result of molecular interactions between molecules adsorbed at neighboring sites, or a true indication of differences between sites [20]. This matter can be checked by varying the probe size or the site density.

Figure2. a) Homogenous and b) Heterogeneous distribution of sites: differential heat of adsorption as a function of coverage (v) Energy distribution spectra (Figure 1f ) In some cases the variation of the adsorption heats with progressive coverage corresponds to step-shaped curves. Such a behaviour may be associated with the discrete surface heterogeneity due to the existence of several energetic levels [21]. In such cases, to describe the change in the adsorption heats with coverage, another approach is to plot the energy spectra (Figure If): assuming that the variation in the adsorption heats coincides with energy distributions, one may wish to measure the number of sites of a given strength, i.e. sites that give rise to the same differential heat. This is achieved by plotting -dn/dQdiffas a function of Qdlf The area below the curve included between QdZf and Qdiff+ dQdiff represents the population of sites of identical strength estimated via Qd@ (vi) The variation of the thermokinetic parameter as a function of the adsorbed amount of probe.

The output of heat conduction microcalorimetry consists of power versus time curves and these can be analyzed to produce not only thermodynamic but also kinetic data. The kinetics of heat release during adsorption can be monitored by the change in the thermokinetic parameter z [22,23]. Namely, the calorimetric signal decreases exponentially with the adsorption time after the maximum of each adsorption peak (Figure la). This can be approximated by D = Dm e-": where D and D m are the deviation at time t and the maximum deviation of the calorimetric signal, respectively. In this expression, the thermokinetic parameter z, known also as the time constant, can thus be calculated as the reciprocal of the slope of the straight line obtained upon plotting log D as a function of time [22]. This thermokinetic parameter is not constant and varies with coverage. One can then plot the variations of the thermokinetic parameter with the amount of adsorbed probe. For example, when NH3 is used to probe the acidity of zeolites at a given temperature, the time needed to establish thermal equilibrium after each dose at first increases with increasing adsorbed amount, passes through a maximum, then decreases rapidly and, finally, reaches a value close to the time constant of the calorimeter. The establishment of the adsorption equilibrium can be monitored both through the change in the heat signal and by the change of the pressure in the system. The time required to establish equilibrium depends on the quantity of adsorbed probe, on the temperature and on the inertia of the calorimeter. At low temperatures, a slower adsorption is observed in covering the strong adsorption centres than at higher temperature. The long time to establish equilibrium is apparently related to a redistribution of the adsorbed probe on the centres that are energetically more favourable [24]. When the adsorption temperature is increased, there is a decrease in the time required to establish equilibrium and a decrease in the region of coverage in which slow adsorption is observed [24]. When the time to establish thermal equilibrium is determined solely by the inertia of the calorimeter, one can be sure that the adsorption temperature was well chosen. However, the use of probes with a large molecular diameter often leads to other diffusional problems. (vii) Estimated entropy of adsorption, determined from the adsorption equilibrium constants obtained from adsorption data (isotherms and heat of adsorption). The differential molar entropies can be plotted as a function of the coverage. Adsorption is always exothermic and takes place with a decrease in both free energy (AG < 0) and entropy (AS< 0). With respect to the adsorbate, the gassolid interaction results in a decrease in entropy of the system. The cooperative orientation of surface-adsorbate bonds provides a further entropy decrease.

As indicated above, the extent of adsorption can also be measured directly by weighing the catalyst sample together with the adsorbed probe molecule by means of a microbalance set inside the calorimeter. The sample is outgassed as previously described and then contacted with successive doses of probe gas. The adsorbed amount is calculated from the mass gain for each dose [lo]. An alternative method is flow adsorption microcalorimetry, which involves the use of a carrier gas passing continuously through the adsorption cell [25-291. In the pulse flow method, the procedure consists of the injection of a precise and well-defined gas volume (probe molecule + carrier gas) into the stream which flows through the catalyst bed held on the fritted glass of a specially designed calorimetric cell. For each pulse, the calorimetric signal is recorded and the amount of gas which has not been retained by the catalyst is measured by a gas chromatograph (or mass spectrometer) connected on-line to the calorimetric cell. The major disadvantage of this technique is that the weakly chemisorbed portion of the probe gas is not held by the catalyst and gives rise to an endothermic peak of desorption which follows immediately the exothermic peak of adsorption, and thus necessitates peak deconvolution. Finally, calorimetric measurements can also be used to monitor adsorption phenomena on the surface of solid catalysts in contact with a liquid phase (in a solvent). For example, the so-called cal-ad method [30-331 has been used to measure the adsorption heats evolved upon addition of dilute solutions of pyridine in n-hexane to a solid acid catalyst (TS, H-ZSM-5) in a slurry with nhexane. The amount of free base in solution is measured separately using a UVVis spectrophotometer [30,3 11. A similar technique has been used to determine the acidic character of niobium oxide and niobium phosphate catalysts in different solvents [34,35], using aniline and 2-phenyl-ethylamine as probe molecules.

2.1. Some limitations of the technique for characterizing catalytic sites It is often difficult to determine the nature of the adsorbed species, or even to distinguish between the different kinds of adsorbed species from the calorimetric data. In many cases this technique fails to distinguish between cations and protonic sites due to the insufficient selectivity of the adsorption. For example, the differential heats of NH3 adsorption on strong Lewis centres and strong Bronsted sites are relatively close to each other. This can make it difficult in some cases to discriminate Lewis and Bronsted sites solely by adsorption microcalorimetry of basic probe molecules if no complementary techniques are used. Because no exact information can be obtained regarding the nature of the acid centres from the calorimetric measurements, suitable IR, MAS NMR, and/or XPS [36] investigations are necessary to identify these sites. However,

due to the complex nature of the acid strength distribution, it is currently still not possible to make a detailed correlation between sites of different nature and their strength. It is also difficult to find good correlations between the site strength distribution and the activity or selectivity of a catalyst. In catalysis, only a fraction of the energy spectrum of the surface sites may actually be involved during the catalytic reaction. If the critical size of the reactant molecules is commensurate with the pore dimensions, the rate of conversion may be determined by diffusional limitations and not only by the strength of acid sites [37]. The validity of energy distributions derived from heats of adsorption has been examined in the literature [38]. In addition, the interplay of reaction and diffusion depends on the temperature. Initial values representing very small concentrations of the strongest acid sites can be easily missed in the measurement if the gas doses are not small enough. The adsorption of the gas should not be limited by diffusion, neither within the adsorbent layer (external diffusion) nor in the pores (internal diffusion). Should diffusion limitations occur, adsorption on active but less-accessible sites may occur only after better-exposed but less-active sites have interacted. Diffusion may thus cause the "smoothing out" of significant details in the energy spectrum, and the differential heat curves determined under the influence of diffusion phenomena may indicate less surface heterogeneity than actually exists on the adsorbent surface [8]. External difhsion is especially a serious problem with adsorption calorimeters which contain a thick bed of adsorbent. The adsorption of very small doses of gas may help to solve diffusion problems if the number of molecules in the dose is small enough compared to the number of readily accessible reactive sites. The admission of small doses, which prevents the rapid increase of the equilibrium pressure, favours the reaction of the gas with the most active surface sites. However the detailed analysis of the surface sites by means of adsorption calorimetry is then very time-consuming. Microcalorimetry can give erroneous results if adsorption equilibration is too slow, a particularly serious problem at low temperatures. The literature contains some controversial articles on this subject [39]. Generally speaking, the adsorption temperature should not be too low, in order to allow the detection of differences among the sites. Under certain circumstances, the evolved heat measured at low temperature can be merely an average value over various site populations of different strengths. Another important issue is that chemisorption must predominate over physisorption.

2.2. Probe molecules most commonly used to characterize catalytic surfaces Even though, as a general rule, adsorption calorimetry allows the total number of adsorption sites and potentially catalytically active centres to be estimated, the values obtained depend on the nature and size of the probe molecule. Appropriate probe molecules for adsorption microcalorimetry should be stable over time and with temperature. The desired properties of the probe depend on the materials to be studied: for example, in zeolites of the alumino-silicate type, the negatively charged aluminum atoms in the framework generate an electrostatic field, making zeolites capable of interacting strongly with polar molecules. Furthermore, the probe molecules should be small enough to readily penetrate into the intracrystalline space of most zeolites. In general, the probe adsorbed on the catalyst surface should also have sufficient mobility to equilibrate with active sites at the given temperature. For example, the heats of adsorption of COz and NH3 are extremely sensitive to changes in the chemistry and geometry of the surface of the void spaces and channels in zeolites, because the significantly large quadrupole (C02) and dipole (NH3)moments cause a very significant contribution of specific interactions to the total energy of adsorption of these molecules [40].Moreover, repulsive interactions between adsorbate molecules could arise from the ionic repulsion of two positively charged adsorbates, or from changes in a protonic site due to adsorption at adjacent sites [201. Different types of probe molecules can be used depending on the catalyst surface to be characterized: 2.2.1. Acidic probe molecules For basicity measurements, the number of acidic probes able to cover a wide range of strength is rather small [41]. Moreover, a difficulty stems from the fact that some acidic probe molecules may interact simultaneously with cations (such as ~ a ' ) . The ideal probe molecule should be specific to basic sites and should not be amphoteric. It should not interact with unwanted types of basic sites or give rise to chemical reactions [41]. For instance C 0 2 (pKa = 6.37) is a suitable probe to determine and characterize, simultaneously, the surface basicity as well as the Lewis acidity of acidic metal systems. It can form carbonate-like species on the former sites, whereas it can be molecularly coordinated in a linear form at the latter sites [42]. Moreover, the energetic features of the adsorption of C 0 2on various molecular sieves, over a large domain of temperature and pressure, can provide interesting information on the nature of the adsorbate-adsorbent interactions [43]. Similar problems may arise when using SO2 as an acidic probe, despite the fact that SO2 (pKa = 1.89) is more acidic than CO2 and, thus, more likely to probe the total basicity of the surface.

2.2.2. Basic probe molecules Heats of adsorption of strong bases are related to both the intrinsic acidity of the site and the interaction energy between the deprotonated surface of the sample and the protonated base. Ammonia (pKa = 9.24, proton affinity in gas phase = 857.7 kJ mol-') and pyridine (pKa = 5.19, proton affinity in gas phase = 922.2 kJ mol-') are the favoured molecules to probe the overall acidity of acid catalysts, because both Lewis and Bronsted acid sites retain these molecules. As indicated above, in order to distinguish qualitatively and unambiguously between these two types of sites it is necessary to use infrared spectroscopy or to analyze the shifts in the binding energies of N1, core-level lines of adsorbed N-probe molecules observed by XPS. The use of substituted pyridines (2,6-dimethylpyridine) has also been considered in order to probe, specifically, the Bronsted sites [44]. Reviews by Gorte et al. [45,46] deal with the adsorption complexes formed by strong and weak bases with acid sites in zeolites. Among strong bases, these works examine the thermochemistry and adsorption enthalpies of a series of molecules similar to ammonia, i.e. alkylamines, but also pyridines and imines. Concerning weak bases, studies of the adsorption properties of water, alcohols, thiols, olefins, aldehydes, ketones and nitriles are taken into account. The authors report a poor correlation between the differential heats of adsorption on a H-MFI zeolite and enthalpies of protonation in aqueous solutions, but a much better correlation with gas-phase proton affinities [46]. Ammonia is among the smallest strongly basic molecules, and its difhsion is hardly affected by the porous structure, if at all. This makes it the most commonly used probe in calorimetry for testing the acid sites. Ammonia is adsorbed as an ammonium ion, and the corresponding heat of adsorption depends both on the proton mobility and on the affinity of ammonia for the proton. Acetonitrile is also an interesting molecule for probing acid sites in catalysts [39,47-491. It is a weak base, so no protons are abstracted and actual hydroxyl groups can be observed. It also allows the investigation of both Lewis and Bronsted acidities. While it is normally considered to be a weak base, it actually has a moderately high proton affinity (798 kJ m o ~ ' ,compared to 857 kJ mol-' for ammonia and 773 kJ mol-' for methanol). Other nitriles and alcohols have also been used to probe the acid sites of catalysts [50,5 11. 2.2.3. Redox probe molecules NO can be employed either as a probe to identify Lewis acid sites and characterize their density and strength [52], or as a reducing agent. However, NO may disproportionate into N 2 0 and oxygen and is also very likely to form nitrosyl complexes in the presence of transition-metal ions. From a different

point of view, calorimetric measurements of CO, O2 and H2 adsorptions at low temperature also provide a powerful tool for characterizing redox sites or metallic sites at the surface. However, it is known that, in the absence of processes other than a plain surface coordination, carbon monoxide acts as a weak Lewis base and can interact with the strongest surface Lewis acid sites. 2.3. The role and the influence of the probe molecule in determining adsorption heats

2.3.1. Correlation with gas phase affinities The influence of the probe molecule on adsorption heats is particularly important when determining the acid-base character of catalysts. Due to their strong basicity, ammonia and pyridine are adsorbed very strongly even on the weakest sites of acid catalysts. As a consequence, their interactions with surface acid sites are relatively unspecific. Therefore, a significant number of studies using these probe molecules report a homogeneous acid site strength [53,54]. In any case, the size and the strength of the probe molecule are important parameters and have to be carefully considered when performing adsorption studies. As shown by studies on various zeolites (ZSM-5, mordenite, H-Y) [44,55], it is difficult to compare adsorptions of ammonia and pyridine, because of their difference in basicity and the specific interactions of each of these molecules with the host zeolite. The adsorption heats should be correlated, not to the basicity of the probe molecules in liquid phase or in aqueous solutions (where NH3 is more basic than pyridine by about 20 kJ mol-', because the pKa values are 9.3 and 5.2, respectively), but to their basicity in the gas phase, which can be expressed in terms of the proton affinities (PA). In the gas phase, isopropylamine is a stronger base than pyridine and than ammonia: the PA value of isopropylamine ( Ka = 10.7) is about 924 kJ mol-', and that of pyridine is about 922.2 kJ mol- , while that of ammonia is only about 857.7 kJ mol-'. This feature is illustrated in Figure 3, which represents the differential heats of ammonia, pyridine and isopropylamine vs uptake on a H-mordenite zeolite [54]. Other parameters that have to be taken into account are size (the kinetic diameters of pyridine and ammonia are 0.533 and 0.375 nm, respectively) and the secondary interactions between these adsorbates and the zeolite structure (hydrogen bonds may lead to the formation of a monodentate bond with pyridine and a bidentate bond with ammonia), as shown by Parillo et al. [55-571. These differences are reflected in the heats of adsorption, which are usually 20-30 kJ mol-' higher in the case of pyridine, than in the case of ammonia. A similar difference is observed for the chemisorption limit [44,54].

P

.

.I ,

lsopropylamine Pyridine

I

Ammonia

50-C 0

I

200

I

400

I

I

600 800 probe uptakelpmol g-1

I

1000

1200

Figure 3. Differential heats of adsorption of ammonia, pyridine and isopropylamine plotted against uptake on a H-mordenite zeolite 1541 A comparison of the adsorption energies of the complexes formed by ammonia, pyridine and isopropylamine with a simple potential-energy model, which assumes that the heat of adsorption scales linearly with gas-phase proton affinities, suggests that proton transfer dominates the interaction between the adsorbate and the acid site [56]. The microcalorimetric technique has also been applied to investigate the adsorption of acetonitrile, dimethylether, water, pyrrole and ammonia on ferrierite, a small-pore zeolite [58]. This investigation showed that the results of the determination of the site-strength distribution are dependent on the basic strength of the probe, on the acidity and porosity of the acid solid, and on the adsorption temperature. A similar calorimetric investigation of the acidity of dealuminated Y-type zeolites was performed using the same probes [59]. The available data concerning the adsorption of different bases indicate that adsorbed bases interact with zeolites more strongly than hydrocarbons of similar structure and molecular weight [21]. A comparison of the heats of adsorption of various bases, such as ammonia, pyridine and n-butylamine, with those of benzene on A, X, Y and mordenite zeolites, modified by ion exchange and aluminum extraction, has been carried out by Klyachko et al. [21]. Surprisingly, the heats of pyridine adsorption were found to be virtually the same on sodium and on hydrogen zeolites. Furthermore, the sorbed amounts of large molecules such as pyridine and n-butylamine were very limited, due to the finite void

volume of the zeolites. In the case of ammonia, some complications could arise because of penetration of NH3 into the small cages. It is possible to evaluate quantitatively the location of acid sites in zeolites by using different probe molecules, e.g., on one hand, ammonia, which is able to reach all acid sites, and, on the other hand, pyridine, cyclohexane, benzene, etc., probes that can only attach to sites in the main channels. 2.3.2. Pore-size effect Microcalorimetric studies of the adsorption of molecules of different sizes have also been used to assess the micropore network of zeolites (micropore sizes, channels, presence of cavities, etc.) and to determine the topology of a zeolite. The sorption properties and associated enthalpies of many aliphatic and aromatic compounds have been shown to vary in accordance with the size of the sorbate molecules and the pore-size distribution of the sorbant [60-631. For example, a study of the adsorption of benzene and benzene derivatives onto zeolite H-Y at 323 K [60] has revealed that the dH values, corresponding to interaction with acidic Bronsted sites, vary between 66 and 125 kJ mol-' and increase in the order: benzene < ethylbenzene < 1,4 diethylbenzene = 1,3 diethylbenzene. Benzene exhibited a constant adsorption enthalpy over a wide range of sorbate loadings, while the substituted benzenes showed gradual increases in Hods with coverage, indicating mutual interaction between sorbate molecules. Jiinchen et al. [64] have reported that the heats of adsorption of acetonitrile on mesoporous (MCM-41) and microporous (FAU and MFI) molecular sieves are mainly influenced by a specific interaction with the acidic sites, while the adsorption heats of a non-polar molecule like n-hexane are determined by the pore size or density of those materials. However, a pore-size effect, affecting the heats of acetonitrile adsorption on acidic molecular sieves, has to be taken into account when employing those heats as a measurement of acidic strength. The contribution of the pore-size governed dispersion interaction in mesoporous MCM-41 is about 15 kJ mol-' less than that in the narrow channels of MFI. The adsorption of molecules of different sizes (toluene, xylenes, etc.), and the consecutive adsorption of these same molecules, studied by adsorption microcalorimetry together with reaction tests, can provide useful indications of the pore geometry and reactant accessibility of new zeolitic materials such as MCM-22 [65] or ZSM-11, SSZ-24,ZSM-12, H-M and CIT-1 1661. Similarly, insight has been gained into the pore structure of zeolites of type MWW, ITQ-2 and IM-5 through microcalorimetric studies of adsorption of nhexane, toluene, m- and o-xylenes, 1,2,4- and 1,3,5-trimethylbenzene [6 1,671. Another study [68] has shown that the y-cages of H-ZK-5 and K-ZK-5 are the

preferred adsorption sites for propane and n-butane because of the more favourable heats of adsorption.

2.3.3. Structure sensitivity The sensitivity of adsorption of hydrogen, oxygen and carbon monoxide to the structure of supported metal catalysts, such as Pt or Pd on y-A1203,has been investigated by Uner et al. [69] by changing the metal particle-size. No structure dependency was observed for the initial heats of adsorption of hydrogen, carbon monoxide or oxygen. Moreover, while the hydrogen site-energy distribution was found to change with increasing particle size, oxygen and carbon monoxide adsorption site-energy distributions did not change appreciably with the metal particle-size. Water can be used as a probe molecule to provide knowledge of hydration enthalpies, which is of considerable interest when assessing the hydrophilic or hydrophobic character of solid catalysts [70]. In conclusion, the choice of a suitable probe molecule depends very much on the physico-chemical characteristics of the samples studied. 3. ACID-BASE PROPERTIES OF CATALYST SURFACES

The measurement of the acidity of solid acid surfaces has been the focus of a vast number of studies. The most commonly used techniques are Hammett titrations, chemisorption of bases and TPD. Extensive discussions of these methods and their shortcomings are available in the literature [4]. The use of adsorption calorimetry makes it possible to determine quantitatively the surface acidity and the acid-strength distribution of solid acids. Surface acid-base properties of catalytic solids can also be studied by base desorption using TG V11. - he examples given below are mainly on zeolites and oxides, but the acid-base properties of all kinds of catalyst surfaces can be examined using adsorption calorimetry, including, for example, activated carbons [72], which are often used as supports of active phases due to their very high surface-area, oxynitrides [73] or hydrotalcites [74-761, which act as base catalysts [73], or heteropolyacids, which behave as strong acid catalysts [77]. 3.1. Zeolites and related materials Zeolites are crystalline and microporous aluminosilicates whose framework is made up of A10i and Si04 tetrahedra. The general formula of zeolites is Mdn[(A102),(Si02),]-wH20, where A4 is the compensating cation of valence n, x + y is the total number of tetrahedra per unit cell (1 < ylx < 5), w is the number

of water molecules. The most studied zeolites are HY (faujasite), ZSM-5 (MFI), silicalite, mordenite and offretites. The unique catalytic properties of zeolites are mainly attributed to their acidic properties. An important characteristic of zeolites, and other acidic molecular sieves, is that each material contains a well-defined, discrete number of acid sites. However, the acidity of zeolites is difficult to characterize because these materials can contain both Lewis and Bronsted acid sites and may exhibit a heterogeneous distribution of acid-site strengths. The number of studies in which adsorption microcalorimetry has been successfully applied to this end has increased in recent years, especially concerning the determination of the acidic function of molecular sieves, and extensive reviews of the systems investigated using this methodology have been published [1,5-14,19,78-811. In particular, an extensive review [4] summarizes some of the most recently published results concerning applications of microcalorimetry to the study of the acid-base sites of zeolites and mesoporous materials. The efficiency of thermal analysis techniques for the characterization of the acid-base strength of zeolite materials is also discussed, as well as their ability to provide information consistent with catalytic data [4]. The high variability of the acidic properties of these materials is caused by the existence of crystallographically inequivalent sitings and different local environments of the framework atoms. Bronsted acidity plays a very important role in the catalytic activity of zeolites. It is caused by the tetra-coordination of isomorphously substituted A1 in a tetrahedral SiOz framework, which results in a negatively-charged A10i anion: this charge is compensated by cations and particularly by protons, the latter resulting in the Bronsted acidity. Microcalorimetric studies of several zeolites (H-mordenite, USY, H-ZSM-5), treated in such a way as to contain a noticeable amount of extra-framework aluminum, have shown that the distribution of the sites with respect to the differential heats of NH3 adsorption is exponential for the Lewis sites (Freundlich isotherm) and linear for the Bronsted sites (Temkin isotherm) [82,83]. In most cases, the catalytic activity is related to the number of Bronsted sites rather than Lewis acid sites. However, the influence of acidic Lewis sites in catalytic reactions over zeolites is still subject to controversy and cannot be neglected. Energetic surface heterogeneity is, first of all, a consequence of the structural surface heterogeneity, which is one of the fundamental features of real solid surfaces. Therefore, it seems reasonable to assume that there should exist correlations between structural and energetic heterogeneities. Indeed, the adsorption energies on given acidic sites are related to the nature of their nearest neighbors or even next-nearest neighbors [83].

In the determination of acidity by microcalorimetry, several factors play an important role, such as the adsorption temperature, the pretreatment temperature and the choice of the probe molecule. Other factors, more specific to zeolites, are the topology, the Si/Al ratio, the chemical composition, and the modifications to which the samples have been subjected. The shape of the differential heat curves versus coverage demonstrates quite explicitly that the number, the reactivity and the distribution of surface sites are significantly modified when the composition, or the pretreatment, of the samples are changed. However, the two main factors are the zeolite structure and the framework aluminum content. Most of the studies described herein are summarized as reference tables in a review by Cardona-Martinez et al. [6] or fully detailed in [41.

3.1.1.Influence of the zeolite structure and pore diameter Under identical conditions of adsorption on molecular sieves of the same chemical composition but different pore systems, the structure of the pore system and the diameters of the pores of the microporous adsorbents are responsible for the phenomenon of selective adsorption. The intensity of the electrostatic field is also a determining factor. In acid-catalyzed reactions, zeolites often show shape selectivity because of their unique pore structure. Outer surface acidity, however, diminishes this important property, and can even cause pore blocking by coke formation. 3.1.2. Influence of the Si/AI ratio and of de-alumination The Si/Al ratio plays a significant role, because the aluminum atom is directly related to the acidic site and accounts for the formation of carbenium and/or carbonium ions, or possibly cation radicals, inside the zeolite. De-alumination processes can promote modifications of porous structures, which may improve some important properties of zeolites, like thermal and hydrothermal stability, acidity, catalytic activity, resistance to aging and low coking rate. The effect of steaming on the number and strength of acid sites is apparent from a comparison of the differential heat curves for zeolites de-aluminated to various extents. The microcalorimetric curves show that the strength of sites, corresponding to the intermediate plateau region, first increases and then decreases with steaming severity. Steam de-alumination of H-Y zeolite is known to cause a progressive destruction of weak and intermediate sites, while generating new stronger sites. Microcalorimetric measurements of adsorption of ammonia and pyridine have shown that samples containing extra-framework aluminum possess sites with adsorption heats that are much higher than those observed for samples containing only framework A1 [20,84-861. Moreover, the acid sites of steamed samples present a wide distribution of acid strengths. The high initial heats of

adsorption of ammonia or pyridine, that can be observed on steamed zeolites in comparison to unsteamed samples, may be attributed to Lewis acid centres (in particular extra-framework Al) or to a combination of Lewis and Bronsted sites [84,86]. Microcalorimetric studies of de-aluminated mazzite [87,88] zeolites, prepared by steaming and subsequent acid leaching, in order to remove (partially or totally) the extra-framework species generated by steaming, have led to the conclusion that the initial strong sites can be attributed to Lewis acidity (alumina phases or non-framework aluminum). This result agrees with most IR studies, which confirm that some of the Lewis acid sites generated by de-alumination are stronger than the Bronsted acid sites of pure zeolites. The samples which had similar total (chemical analysis) and framework (NMR) SiIA1 ratios (see Figure 4) presented a plateau in their acid-strength distribution, whereas the other samples showed a more heterogeneous distribution. An increase of the initial heat values and of the site-strength heterogeneity was observed for samples presenting many extra-framework aluminum species.

0

Figure 4.

200

400 600 NH3 Uptake / wnol.g4

800

1000

Differential heats of NH3 adsorption vs coverage for de-aluminated mazzite zeolites

3.1.3. Influence of the pretreatment The acidic properties of zeolites are dependent on their pretreatment and particularly on their activation temperature, which plays an important role in the mechanism of generation of acid sites.

Calcination changes the acid site strength distribution, and high-temperature calcination is a method of reducing total acidity via dehydroxylation and dealumination, while increasing the number of Lewis acid sites. Shannon et al. [89] have shown that dehydroxylation of H-Y zeolite at 923 K results in the destruction of most of the acid sites of medium strength (75-140 kJ mol-') and their replacement by a smaller population of stronger sites (150- 180 kJ mor'). 3.I . 4. The effect of pi-oton-exchange level The effect of proton-exchange level, or sodium content, on H-Y zeolites has been the subject of numerous studies [4]. The acid form is mostly obtained by decomposing the ammonium form, obtained from the Na form by cation exchange, so that the acidity varies with the exchange level. Generally, an increase in acidity with increasing proton exchange was measured; however, uncertainty exists as to the strength of the acid sites introduced at different exchange levels. Number of sitedwmd NH, g'

Figure 5 . The number of sites of a given strength as a function of the exchange level. For this reason, the effect of the exchange degree (of sodium cations by protons) on the heterogeneity of acid sites in a ZSM-5 sample has been studied. Figure 5 reports the number of sites of a given strength determined by NH3 adsorption calorimetry, as a function of the exchange level. It shows that the numbers of medium and strong sites increase with decationization, while the number of weak sites decreases slightly. Figure 5 clearly shows that, at low exchange levels, most of the acid sites are rather weak. While this population of weak sites remained almost constant with the exchange level, the population of stronger sites increased progressively up to the point where, for extensively

exchanged samples, the strongest sites became predominant. The population of sites presenting heats of adsorption above 150 kJ mol-' illustrates the dramatic effect of removal of the very last sodium ions on the acid strength, not only for the newly created sites but also for the preexisting ones. 3.1.5. Substitution by other cations

The nature of the exchanged cation is one of the key points that determine acidity in zeolites. The acidic properties of a series of X faujasites exchanged with Li, Na, K, Rb and Cs have been studied by adsorption microcalorimetry, using ammonia as acidic probe. The heats of NH3 adsorption were found to decrease in the sequence from Li to Cs (Figure 6 ) . Li and Na zeolites presented much higher heats of NH3 adsorption, and greater coverage at the same pressure, than the other zeolites. The acid-base properties of alkali-metal ion exchanged X and Y zeolites have also been investigated by ammonia and sulfur dioxide adsorption microcalorimetry, in parallel with the study of a catalytic reaction, viz. 4-methylpentan-2-01 conversion [90].

20

-

0 0

+LiNaX

+NaX I

5

+leKNaX I

4RbNaX

+CsNaX

I

10 15 NH3 uptake /mol u.c?

20

25

Figure 6. Differential heats of NH3 adsorption on alkaline X zeolites vs. coverage [90] 3.1.6. Acidity in fluid cracking catalysts (FCCs) The main components in FCCs are usually de-aluminated Y zeolites (USY), a small quantity of rare-earth elements (Re), a binding matrix, and an acidic

component consisting of a small amount of H-ZSM5 zeolite, in order to enhance the octane number of gasoline. The determination of acidity in fluid cracking catalysts (FCCs), using microcalorimetry of adsorption of probe molecules, has been the subject of a review article by Shen and Auroux [91]. The catalytic activity of such materials is due to the presence of acidic sites and is determined by the zeolite content and by the types of zeolite and matrix in the FCC. The catalytic selectivity is determined by the zeolite type, the nature (Bronsted or Lewis), strength, concentration and distribution of the acid sites, the pore size distribution, the surface area of the matrix and the presence of additives or contaminants, among other factors. Stability is affected by both the composition and the structural characteristics of the catalyst components. Therefore, the acidity of FCCs is designed to meet specific requirements, and a full characterization of the acidity is necessary; this gives a great deal of importance to the information gathered by direct methods such as the monitoring by microcalorimetry or by temperatureprogrammed desorption (TPD) of the adsorption or desorption of gaseous bases, particularly ammonia or pyridine. 3.1.7. Metal-substituted zeolites Metals other than A1 have been successfully introduced in numerous zeolite frameworks. Aluminum substitution by other metals, such as Fe, Ga, Zn, Co or Cu in the zeolite framework results in modified acidity, and subsequently modified catalytic activity, for certain reactions such as selective catalytic reduction of NO, by hydrocarbons. For example, a calorimetric and IR spectroscopic study of the adsorption of N20 and CO at 303 K on Cu(I1)exchanged ZSM-5 zeolites with different copper loadings has been performed by Rakic et al. [92]. The active sites for both N 2 0 and CO are Cu (I) ions, which are present on the surface as a result of the pre-treatment in vacuum at 673 K. The amounts of chemisorbed species adsorbed by the investigated systems and the values of the differential heats of adsorption of both nitrous oxide (between 80 and 30 kJ mol-') and carbon monoxide (between 140 and 40 kJ mol-') demonstrate the dependence of the adsorption properties on the copper content. 3.1.8. Mesoporous materials Since their relatively recent discovery, ordered mesoporous materials have attracted much interest because of their high surface-area and uniform distribution of mesopore diameters. To impart the desired catalytic activity to the inert silicate framework, synthesis conditions have been modified to introduce the desired A1 concentration into the wall of the different nanostructures. The amounts of Bronsted sites and the acidity are lower than in microporous zeolites.

3.2. Bulk, doped, supported and mixed oxides Metal oxide surfaces react with gases or solutions and they can be used as active phases, or as supports for catalysts. The behaviour of metal oxide surfaces is controlled by: (i) coordination - sites of low coordination are in general more reactive than sites of high coordination; (ii) acid-base properties - clean and anhydrous metal oxide surfaces present two different types of active sites, cations and anions (acid-base pairs) which determine reactivity towards gas-phase adsorbates; (iii) the redox mechanism - when the oxide deviates from the stoichiometry due to the presence of defects such as vacancies or adatoms, the oxidation state of surface atoms varies [93]. The reaction medium also plays an important role in adsorption processes. Clean surfaces exist only in dry conditions, when the surface is exposed to gases at low pressure. Under hydrated conditions, when the metal oxide surface is covered with water, the surface sites are not available to other molecules. As a consequence, either the adsorption is strong enough to cause desorption of the water molecules that are directly bound to the clean surface, or attachment occurs directly upon these groups through H-bonds. Molecular and dissociative adsorption can be understood as acid-base processes. Molecules adsorbing without dissociation always bind to one or several metal cations. NH3 and pyridine are the most commonly used probes for determining the acid site strength of oxides. In the case of supported metal oxide catalysts, the role of the support is to disperse the active phase and to create new active surface species by host (active phase) - guest (support) interaction. The dispersion of the active phase plays a fundamental role, and very often a maximum of strength of the active sites is observed when the monolayer coverage is reached. The most frequently used catalyst supports (A1203, Zr02, T i 0 4 cany both basic and acidic Lewis sites on their surface; depending on the probe molecules used (C02 or NH3), these pure oxides can exhibit either acidic or basic character. Excess negative or positive charges can be induced, and therefore acidity (Bronsted or Lewis) or basicity can be generated by mixing oxides. Modifying the surface with a minor anionic, cationic or metallic component enhances or decreases the acidic or basic strength of the sites. For example, the incorporation of chloride, fluoride or sulfate ions increases the acidity of carrier oxides (A1203, ZrOz, Ti02), while alkali cations enhance the basic strength of alumina, silica or zeolites.

3.2.1. Bulk oxides The heats of adsorption of basic molecules (ammonia, n-butylamine, pyridine) or acidic molecules (hexafluoroisopropano1) on silica are low in all cases, indicating that the surface sites on silica are both weakly acidic and weakly basic. Adsorption is mainly due to hydrogen-bonding and Van der Waals interactions. More than 98% of the pyridine adsorbed at 453 K is hydrogen bonded [94]. The differential heats of adsorption of ammonia and sulfur dioxide on amphoteric alumina show the coexistence of strong acid sites and basic sites on its surface [94,95]. The heats of ammonia adsorption on alumina are typical of a strong acidic surface; the initial heat increases and the adsorption capacity decreases with increasing pretreatment temperature [ll]. On the other hand, magnesia, which is a basic oxide, displays only strong basic sites and no acidity. These behaviours are illustrated in Figure 7, which represents the differential heats of NH3 and SO2 adsorption vs probe uptake for Si02, y-A1203and MgO, respectively. MgO 200-

Al2O3

-

7

!E

--3

150-

d

100-

50 -

06

SO,

I

5

4

I

I

I

I

I

I

3

2

1

0

1

2

probe uptake Iprnol rn-2

3 '"3

Figure 7. Differential heats of adsorption of NH3 and SO2on Si02,y-A1203and MgO The surface energies and thermodynamic phase stabilities of nanocrystalline aluminas have been studied by McHale et al. [96]. The results provided conclusive experimental evidence that differences in surface energy can favour the formation of a particular polymorph. Calorimetric measurements of adsorption of C02 at 303 K on different titania samples have provided evidence of their surface heterogeneity, as expected for

-

oxides, with heats of adsorption ranging from 100 kJ mol-I to 30 kJ mol-I. Acidity measurements by ammonia adsorption microcalorimetry on the same samples gave rise to adsorption heats ranging between 150 and 60 kJ mol-' [97]. The literature also contains comparative studies of the differential heats of adsorption of NH3 and C 0 2 on a large number of bulk oxides [95,98]. These oxides were either acidic: Cr203, W03, Nb205, V205/Si02 and Moo3; amphoteric: BeO, Ti02, A1203,Zr02 and ZnO; or basic: Tho2, Nd203, MgO, CaO and La203. Many of the oxides in the amphoteric group adsorbed more NH3 and gave rise to higher adsorption heats (> 200 kJ mol-I) than some of those in the acidic group. The Zr02 sample adsorbed NH3 with a heat of 150 kJ mol-I, comparable to Nb20s and W03. The group of basic oxides adsorbed NH3 very weakly (< 20 kJ mol-I), essentially by physisorption. La203,Nd2O3 and Tho2 adsorbed COz in larger amounts and with higher heats, for instance 200 kJ mol" for Nd2O3,than the other oxides. On the alkaline earth oxides, the heats of adsorption of C 0 2 were only 120-160 kJ mol-I, perhaps due to bidentate adsorption of C02. Ti02, Zr02 and A1203 also adsorbed C 0 2 with heats of adsorption of the order of 100 kJ mol-', but in amounts smaller than on the basic oxides. In another study, information on the acid-base properties of lanthanum and cerium oxides was obtained by adsorption microcalorimetry of ammonia and carbon dioxide and correlated to reaction selectivities in the dehydration of 4methylpentan-2-01. Ceria is more acidic than La203,whereas the interaction with C 0 2 is more pronounced for La203than for Ce02, in terms of both number and strength of the adsorption sites [99]. The calorimetric results obtained on La203and Zr02 have been compared [99]. Both oxides feature a smooth decrease of the differential heat of NH3 adsorption with increasing coverage, with nearly identical differential heat curves. This indicates very similar acidic properties, in terms of both the number and strength of the sites. The initial differential heats of C 0 2 adsorption are very high in both cases (225 and 210 kJ mol-I for La203and Zr02, respectively). After a sudden decrease at very low coverage, a plateau around 150 kJ mol-I is established for both oxides, which seems to indicate that strong basic sites of the same type are present on the two samples. The amount of these sites is higher in the case of Zr02, for which the plateau of differential heats extends over a wider range of CO2 uptake than for La203. The strong Lewis acidity of phase-pure y-Ga203 has been investigated by studying the adsorption of CO at ambient temperature, using adsorption microcalorimetry and in situ FTIR spectroscopy in a combined manner [loo]. The concentration of strong Lewis acid sites turned out to be quite low, but grew rapidly with increasing surface dehydration upon thermal treatment in the 573-773 K range. The CO adsorption process was a rather heterogeneous one,

and the molar adsorption heats (Qdiffrangingbetween 40-45 kJ mol-' and some 25-35 kJ mol-') were comparable with those typical of weak and reversible adsorption processes. The combination of the two employed techniques made it possible to distinguish two main populations of Lewis acid sites, which have been ascribed to coordinatively unsaturated ~ a ions ~ + located in defective (higher vCO) and regular (lower vCO) crystallographic sites, respectively. 3.2.2. Doped oxides Ammonia and s u l k dioxide adsorption microcalorimetry and the catalytic reaction of 2-propanol conversion have been used to study the effects of doping on the acid-base properties of y-alumina, silica or magnesia surfaces loaded with + , 0 or~zr4+ ~ ions ; [loll. The small amounts of ca2+, ~ i ' , ~ d ~ '~, i ~ ~ modification of y-A1203by small amounts of the above ions had only a moderate effect on its amphoteric character. In particular, the curves of differential heats of ammonia adsorption v. coverage showed little change in the acidic properties of alumina upon doping. More substantial changes were observed on magnesia, with the formation of new centres of moderate and weak basic strength. The number of acid-base centres on doped silica was strongly affected by doping, according to the acidity determination performed by NH3 adsorption microcalorimetry. A comparison of the results shows that acid sites of intermediate strength play a central role in 2-propanol dehydration. A classification of the modified catalysts according to an aciditylbasicity scale, defined in terms of the specific effect of ions, has been proposed. The acidity of the catalysts was correlated with the chargelradius ratio and with the generalized electronegativity of the doping ions, while the basicity was correlated with the partial oxygen charge of the corresponding oxides [loll. The acidic properties of catalysts prepared by doping a-Fez03 (mixed with sodium silicate) with small amounts of V, Sb or Cr oxides have been investigated by ammonia adsorption microcalorimetry [102]. The addition of silicate to a-Fe203resulted in an increase in the initial value of Qd@(from 216 kJ mol-' for Fe to 327 kJ mol-' for FeSi), suggesting the presence of a small amount of very strong acid sites. A continuous decrease of Qdg with coverage was observed over the FeSi, FeSiV, FeSiSb and FeSiCr catalysts, indicating the heterogeneity of the adsorption sites. The addition of V, Cr or Sb oxides led to an increase in the total number of acid sites of the catalysts. 3.2.3. Oxides modified by sulfation Enhancing the acidity of a catalyst often leads to higher activities. Sulfation of a bulk oxide has been studied as a technique to achieve this aim. The quantitative and energetic aspects of the room temperature interaction of unsulfated and sulfated model Zr02 systems with CO have been reported in the

literature [103]. The initial heat of CO adsorption for the sulfated system (> 80 kJ mol-') was significantly higher than that measured for the unsulfated system (70 kJ mol-I). This indicates that the major difference between the two samples stems from the presence of surface charge-withdrawing groups. The adsorption of C02 has been studied in order to compare the surface acidbase properties of yttria-stabilized tetragonal Zr02, either plain or sulfated to various extents [104]. The site populations and their energy distributions were studied by microcalorimetry, which evidenced the modification of the basicity of zirconia induced by the sulfation process. Upon sulfation the initial heats of CO2 adsorption decreased from about 120 kJ mol-' to less than 85 kJ mol-', depending on the amount of loaded sulfate. In a recent study [105], the surface chemistry of sulfated zirconia samples with a variable Ga203content (in the 1-15% molar range) has been studied by means of FTIR spectroscopy and adsorption microcalorimetry, using selected probe molecules (CO and 2,6-dimethylpyridine). CO adsorption microcalorimetry showed that the total acid strength decreases considerably when the Ga203 content goes over 5 mol%. The catalytic activity of these materials was found to be greatly dependent on the gallium loading. Lewis acidity is attributed to coordinatively unsaturated zr4+ions located in defective surface sites, whereas Bronsted acidity is associated with surface sulfate groups. Gas-volumetric and microcalorimetric measurements of 2,6-dimethylpyridine adsorption have been carried out on plain tetragonal ZrO2 and on sulfated value of zirconia at 423 K. The non-sulfated sample displayed an initial QdSff about 125 kJ mol-', while sulfated samples exhibited much higher Qdg values (near 175 kJ mol-'). The uptake of 2,6-dimethylpyridine can reveal, although to different extents, the presence of acidic sites that differ either in nature (i.e. Lewis or Bronsted) or in acid strength [106]. Calorimetric measurements of pyridine adsorption have been performed on fresh and deactivated sulfated zirconias, showing that it takes very little carbon (approximately 0.12 mass%) to poison the entire catalyst and that very few sites are covered by coke on the completely deactivated catalysts [107]. The surface acidity and basicity of iron oxide (a-Fe203)catalysts, pure and surface-doped with variable amounts of sulfate groups, have been characterized by microcalorimetry of CO adsorption at room temperature. An appreciable decrease of Qdgwith surface coverage was detected [108]. The surface basicity of iron oxide, tested by adsorption of C02 and monitored by FTIR, which revealed the formation of carbonate-like surface species, is gradually decreased by sulfates, but not suppressed. The vacuum reducibility of iron oxide, which can be spectroscopically evidenced both by a colour change (UV-VIS spectra) and by the formation upon CO adsorption of surface carbonyl-like species with a

pi-back-donation component (IR spectra), turns out to be dramatically hindered by the presence of surface sulfates. Sulfated titania has been investigated much less extensively than sulfated zirconia. Desmartin-Chomel et al. [97] have studied the acidic properties of sulfated titania using ammonia adsorption calorimetry and FTIR spectroscopy. The number of acid sites on the sulfated catalyst was noticeably increased, and dependent on the surface area of the original titania. The dispersion of the initial oxide controls the amount of sulfur retained by the solid and the thermal stability of the resulting sulfate. Ammonia adsorption is commonly used to determine the acidity of sulfated oxides; however, it is also well-known that NH3 is a powerful reductant, and that the acidity of sulfated zirconia is decreased by reduction. At low ammonia coverage, sulfated titanias exhibit a much lower heat of adsorption, and the IR study of NH3 adsorption showed that the first doses of NH3 dissociate at the surface with the formation of OH species. The lower heat of adsorption was then attributed to the contribution of NH3 dissociation to the differential heat of adsorption. This phenomenon has been observed for sulfated aluminas [109].

3.2.4.Supported oxides Adsorption microcalorimetry of pyridine at 473 K has been used to determine the distribution of acidic strength of silica-supported oxides [110]. The acidic strength distributions were found to present several distinct regions of constant heat, while silica had homogeneous surface sites. Supported Ga, A1 and Sc oxides were found to have both Bronsted and Lewis acidity, whereas Zn, Fe and Mg samples showed only Lewis acidity. The sites with the highest heats of pyridine adsorption over these catalysts were found to correspond to strong Lewis acidity, whereas intermediate heats have been attributed to weaker Lewis acid sites or a combination of Lewis and Bronsted sites. These results have been confirmed using IR spectroscopy of adsorbed pyridine [Ill]. Moreover, a study of the activity of these catalysts for isopropanol dehydration has shown that the samples presenting only Lewis acidity were at least one or two orders of magnitude less active than the samples displaying Bronsted acidity. The activities of the latter samples were found to vary in the order A1 > Ga > Sc. The modification of the acid-base properties of a series of oxide sup orts (alumina, magnesia and silica) upon the addition of ions ( ~ i ' , ~ i and ~ SO4 + ) in quantities ranging from 1% to 50% of support surface coverage has been studied by adsorption microcalorimetry, using NH3 and SO2 as acidic and basic probe molecules [109]. The highest guest oxide loading (50% of surface coverage) led to drastic modifications of the surface acid-base behaviour, depending on the support and additive.

E

Zou et al. [I121 have shown that the addition of basic metal oxides, K20 and La203, decreases the surface acidity and increases the basicity of y-A1203,as determined by microcalorimetry of NH3 and C 0 2 adsorptions, respectively. This allows the preparation of supports with low acidity, high surface area at high temperatures, and potentially high stability. While y-A1203 exhibited an initial heat of ammonia adsorption of about 132 kJ mol-', the addition of 6% KzO killed almost all of the acid sites with heats higher than 40 kJ morl. The addition of K 2 0 greatly enhanced the basicity in terms of both the strength (170 kJ mol-' for C 0 2 adsorption) and number of the sites. A sample containing 10% La203 on y-A1203exhibited an initial heat of NH3 adsorption of about 103 kJ mol-', 20 kJ mol-' lower than y-A1203, and a greatly increased basicity compared to the support. Catalysts play a crucial role in environmental applications, in particular concerning the removal of air pollutants produced by stationary and automotive sources, such as NO, and SO2. Supported Cu, Sn and Ga oxides are among the best candidates for applications in environmental catalysis. The differential enthalpies of adsorption of air pollutants, such as SO2,NO2 or NO, on aluminasupported tin and gallium oxides have been measured by calorimetry coupled with isothermal volumetry [113]. Depending on the tin or gallium content, the enthalpies of adsorption of SO2 at low coverage were of the same order of magnitude, or up to 50 kJ mol-' lower, than those of the alumina support. The amount of SO2 adsorbed decreased with increasing Sn02 loading, and increased with increasing Ga203loading. The differential enthalpies of adsorption of NO2 were close to those of the support, regardless of the amount of tin or gallium (around 120-130 kJ mol-I). The adsorption of NO on these samples was found to be reversible. Guimon et al. [114] have characterized alumina- and titania-supported tin dioxide catalysts, active in the selective catalytic reduction of NO by C2H4, using calorimetry and XPS experiments. The SnO21Ti02 samples were found to be markedly more acidic than the Sn02/A1203samples, because the N/(Ti+Sn) molar ratios, measured by XPS of ammonia, were noticeably higher than the corresponding N/(Al+Sn) ratios. The acid sites were shown to be of Lewis type, and their number, determined by NH3 adsorption calorimetry, seemed to increase with the tin content when tin dioxide was well dispersed on the support, with adsorption heats ranging fkom 147 kJ mol-' to 160 kJ mol-'. Sulfur dioxide adsorption led to the formation of three types of species: SO2, sulfites, and sulfates. The basicity of the Sn02/A1203series of samples was weaker than that of the alumina support and passed through a minimum around 12 mass% Sn. By contrast, the basicity of the SnTi series did not seem to depend on the Sn concentration. This could be due to the poor dispersion of the SnO21TiO2 samples.

Supported vanadia is very commonly used in industrial catalysts for processes such as selective oxidation, ammoxidation and selective catalytic reduction (SCR) of nitric oxides. It is usually supported on different carriers for different applications (to provide improved surface area, thermal stability or mechanical strength). The redox and acid-base properties of V205/Ce02catalysts have been studied using TPR and microcalorimetry for the adsorption of NH3 and C 0 2 at 423 K [115]. Variations in the V205 loading and calcination temperature resulted in changes in the surface acidic and redox properties. The results showed that Ce02 was quite acidic, with an initial heat of 142 kJ mol-' for the adsorption of NH3 at 423 K. The addition of V205enhanced the surface acidity: the initial heats were 169 and 177 kJ mol-' respectively for 2.5 % and 5 % V205/Ce02samples [115]. Further increases in the V205loading did not bring about a further increase of the initial heats, but increased the ammonia coverage significantly, indicating the formation of weak acidic sites. These weak acid sites might be due to the formation of crystalline V205,when the loading exceeds the monolayer capacity, because the initial heat of NH3 adsorption on bulk V205has been measured at 137 kJ mol-'. Amphoteric Ce02 exhibited an initial heat of C 0 2 adsorption of 138 kJ mol-', but the addition of even a small amount of V205 (2.5 mass%) greatly decreased the site density for C 0 2adsorption. The acid-base properties of V205/yA1203catalysts have been characterized using ammonia, pyridine and sulfur dioxide adsorption microcalorimetry [116]. For vanadium contents less than 10 mass% V205, the vanadium cations were found to be well dispersed as vanadate compounds over alumina. These vanadate species generated Bronsted and Lewis type acidity, but did not exhibit a basic character, because sulfur dioxide was not chemisorbed on bulk vanadium pentoxide and also appeared not to chemisorb on vanadate species. At low vanadium coverage (< 3 mass% V205) the acidic character of the V2OS/yA12O3 catalysts is associated with vanadium-fkee alumina, whereas at higher vanadium coverage it can be largely attributed to vanadate compounds. SO2 adsorption made it possible to differentiate the vanadate layer from the alumina support, because the differential heats of adsorption decreased steadily with the vanadium loading. The acidity, reducibility and adsorption capacities of nanostructured V205/A1203,V205/Si02,V205/Ti02 and V205/Ti02/Si02prepared by atomic layer deposition have been examined by microcalorimetry and H2 temperature programmed reduction (TPR). The surface reactivity towards ammonia was strongly enhanced by the modification of the surface when depositing highly dispersed vanadia on silica, or on titania-coated silica. The number and strength of the acid-base sites on the surface of the catalysts were directly related to the

V-0-Ti and V-0-Si concentrations, as well as to the dispersion of titania and vanadia on the samples [117,118]. The acidic properties of V205/Si02catalysts, with vanadium loadings varying fiom 0.61 to 10.7 mass%, have been studied using ammonia adsorption calorimetry [119]. Figure 8, which represents the differential heats of NH3 adsorption on these materials vs coverage, evidences the strong influence of vanadia deposition on the surface acidity.

- V,O, ..........

---.

1

-

2

ISiO, V,O, ISiO, V,O, ISiO, V,O, ISiO,

0.61 WhV 2.70 WhV 5.56 WhV 10.7 WhV

3

ammonia uptake I pmol m-2

Figure 8. Differential enthalpies of adsorption of ammonia at 353 K as a function of coverage on V205/Si02catalysts [I191 Wang et al. [120] have characterized Ag/A1203 catalysts, with various Ag contents, using TG, DTA, TPD, NH3 and C 0 2adsorption calorimetry at 473 K, FTIR spectroscopy, and diffuse reflectance spectra. Hydroxyl groups on alumina were found to be closely related to the Ag loading through either ion exchange with the proton of an acidic -OH, or bonding with the oxygen of a basic -OH species. A silver loading of less than 240 pmol g-l decreased the number of Lewis acid sites, but increased the population of Bronsted acid sites with medium NH3 adsorption energies (around 130 kJ mol-I). A greater Ag loading induced interactions between NH3 molecules and Ag oxides, leading to the reduction of the Ag oxides and the dissociation of the adsorbed NH3. The basic sites of alumina in the medium energy range (100-140 kJ mol-') were slightly enhanced by the Ag loading, and silver oxides were stable in the oxidized C02 atmosphere. The acid-base properties were also found to play an important role in de-NO, activity when using propane as the reductant.

The literature contains many other studies of supported oxides by adsorption microcalorimetry, and in particular oxides used for propane or isobutane dehydrogenation such as chromia supported on Zr02 [I211 or y-A1203[122], or Ca-doped chromium oxide catalysts supported on y-A1203[123]. 3.2.5. Mixed oxides Silica-aluminas (amorphous aluminosilicates) are widely used as catalyst supports due to their high acidity and surface area. The behaviour of silicaalumina surfaces is similar to that of zeolites, concerning the initial differential heats of ammonia and pyridine, but the total number of acidic sites varies with the preparation method and the Si/Al ratio. The basicity of silica-alumina surfaces, as determined by CO2 adsorption [94,95], appears to be weaker than that of pure alumina. Auroux et al. [I241 have studied the influence of different preparation procedures on the acidic properties of the solid. The surface properties of the pure oxides, silica and alumina, can be modified by grafting. The surface acidities of the pure oxides and samples, obtained by grafting silica on alumina (SA) or alumina on silica (AS), of both Lewis and Brijnsted type, have been studied by means of microcalorimetry and TPD, using pyridine and 2,6dimethylpyridine as probe molecules. Both techniques indicate that the grafted mixed oxides, SA and AS, have acidic properties different fi-om those of the pure alumina and silica supports used as starting materials. The influence of the acid-base and redox properties of VMgO catalysts on their performance in the oxidative dehydrogenation of propane has been studied [125]. The acid strength of the VMgO samples, as measured by the differential heats of adsorption vs ammonia coverage, increased with the vanadium content up to 14 mass% of V loading. The strong acidity decreased at higher loadings, and weaker acid sites developed at loadings above 25 mass% V. The basicity of the VMgO sample measured by sulfur dioxide adsorption remained unchanged for loadings up to 25 mass% V. The basicity decreased at higher loadings and was found to vanish entirely for a VMgO sample containing 45 mass% V. Solinas et al. [I261 have investigated ceria-zirconia catalysts containing 100, 75,50,25 and 0 mol% of ceria by adsorption calorimetry, using NH3 and C 0 2 as probe molecules. For all the investigated samples, Qdg decreased with the adsorbed amount of ammonia or carbon dioxide, indicating the heterogeneityof the acidic and basic sites. The acidity of the mixed oxides is lower than that of zirconia, but higher than that of ceria. The basic character of the mixed oxides is attenuated when the zirconia content increases up to 80 mol%, but increases again for pure zirconia. In the case of ceria-lanthana catalysts, the initial Qdg values for ammonia adsorption were in the range 130-105 kJ mol-', the highest value corresponded to pure ceria and the lowest one to pure lanthana [127].

Fe203-Ti02 solid acid catalysts have been studied [I281 using ammonia adsorption microcalorimetry. Fe203-Ti02catalysts show an initial value of Qd@ for ammonia adsorption of about 200 kJ mol-', with a continuous decrease with coverage, due to the wide heterogeneity of the adsorption sites. The initial values of Qdiffforpure Ti02 and Fe203are much lower, about 20 kJ mol-' and 50 kJ mof' respectively. The acid site distributions of the investigated samples were calculated fkom the calorimetric data, showing that the addition of Fe to titania significantly increases the total amount of acid sites. 3.2.6. Conclusion A thermodynamic scale of surface acidity and basicity can be constructed by exploring the acid-base properties of numerous solids and comparing the heats of adsorption and the adsorption uptakes of gas-phase probe molecules (NH3, C02, SOz). These solids, varying in their physical and chemical properties, have been selected in order to cover a wide range of acid-base behaviours representative of acidic, amphoteric and basic solids. They can be divided into three main groups according to their adsorption properties towards acidic probes (which interact with basic solids) or basic probe molecules (which adsorb on acidic solids). Amphoteric solids display an adsorption capacity towards both acidic and basic probe molecules. The acidic solids investigated in this study belong to one the following families: 1) bulk oxides [95], 2) doped oxides (chlorinated alumina, doped silica, ...) [loll, 3) supported oxides [log], 4) mixed oxides such as silica-alumina, silica-titania or silica-zirconia, zeolites and clays [ l 1,141, 5) heteropolyanions 6) phosphates (Zr, Al, V, Ti, such as H3PW12040,H4SiW12040 and H5BW12040, B, Sn), 7) superacids such as sulfated oxides or Nafion H (perfluororesin sulfonic polymeric acid). The main families of amphoteric solids are: 1) bulk oxides such as y-A1203, Zr02, Ti02, Ga203,etc., 2) alkali-exchanged zeolites (Li, Na, K, Rb, Cs-X or Y zeolites, ...), 3) doped and supported oxides (V-MgO, ...), 4) oxynitrides (ZrPON, AlPON, AlVON). Finally, the basic solids that were investigated can be classified into: 1) bulk oxides (CaO, La203,ZnO, MgO, Thoz), 2) doped oxides (ca2+-dopedMgO), 3) hydrotalcites (Mg0-A1203). For each of the studied solids, the differential heat of adsorption of the probe molecule has been plotted as a hnction of the coverage. Scales of acidic and basic strength have been established (Figures 9 and lo), based on the average heat at the plateau of the differential heat curve, or at half coverage when no plateau is observed, as was found for most of the oxide samples [129,130]. The scales are based on average heats of adsorption rather than on initial heats, because the latter determination is not always reliable. Indeed, initial heats of

adsorption depend strongly on the interaction with the first dose sent onto the sample, which can vary significantly according to experimental conditions. Acidity (NH3) at 353 K or 423 K

Figure 9. kJ mol"

(

Strength (at half coverage)

H3PWi2040

Mazzite (Si/A1=12)

Cr203

-

')'-A1203 ZrOz wo3

-

H6P2Wi8062 zr3@'04)4

Ga203, ZnO Ti02 (anatase) NbzOs

H-Y (SilAk2.4) H-Beta (SilA1=10)

H-X (Si/A1=1.25) NaA

Ti3@'04)4 BP04

LiX NaX KX

RbX, CsX Silicalite

Figure 10. Basicity (SO2,at 353

Basicity (C02, at 303 K)

I

kJ mol-' Strength (at half coverage)

CaO CeO2 Hydrotalcite (Mg/Al=3) ZnO

- La203 N403, MgO ZrO2 Tho2 Pr.501I

75 -

50 -

Ga203 A1203 Ti02 (anatase)

4. REDOX PROPERTIES OF CATALYST SURFACES 4.1. Metals and supported metals Studies of metals or supported metals by adsorption calorimetry are not as extensive as for metal oxides. Several reviews have been published [6,131]. Many recent studies deal with measurements of integral and differential heats of adsorption of Hz, CO, O2 and hydrocarbons, because these molecules are involved in numerous commercial catalytic processes. Microcalorimetric methods provide an effective means of measuring the strengths of adsorbatesurface interactions, not only on clean metal surfaces, but also on metal surfaces that have been exposed to reaction conditions. Differential heats of adsorption of hydrogen, ethylene and acetylene on Pt powder have been determined at 303 and 173 K [132]. Hydrogen adsorbs dissociatively on platinum powder, with a heat of 90 kJ mol-' and a saturation coverage of 40 pmol g-l, corresponding to 80 pmol g-l of surface platinum sites. Ethylene and acetylene adsorb associatively at 173 K with heats of 120 and 210 kJ mol-', respectively, and saturation coverages of 0.25 monolayers. At 303 K, ethylene adsorbs dissociatively to form ethylidyne species and atomic hydrogen, with a heat of 160 kJ mol-'. The adsorption of acetylene at 303 K gives rise to a heat of 220 kJ mol-', with the formation of di-o/n-bonded acetylene and adsorbed CCH2 species [132]. A microcalorimetric study of CO adsorption at 573 K on Pt clusters supported on L-zeolite [133] has shown that the initial heat of CO adsorption on this material is 175 kJ mol-', and the adsorption heats decrease to 90 kJ mol-' near saturation coverage. An overview of literature data concerning CO and Hz adsorption over supported Pt surfaces, as well as data on oxygen adsorption over various Pt surfaces, has been compiled by Uner et al. [69]. Metals supported on reducible oxides such as Ti02, Nb205, and especially Ce02,present interesting catalytic and chemisorption properties when subjected to high-temperature reduction (> 773 K), due to the effect of strong metalsupport interactions (SMSI). Serrano-Ruiz et al. [134] have studied the effect of adding Sn to Pt/Ce02-AI203 and Pt/A1203 catalysts by adsorption microcalorimetry of CO at room temperature. The investigated catalysts were reduced in situ at 473 K (non-SMSI state) and 773 K (SMSI state). For Pt/A1203, reduction at high temperature caused a decrease in the initial differential heat of adsorption of CO (120 kJ mol-' for Pt/A1203reduced at 773 K, instead of 140 kJ mol-' for the catalyst reduced at 473 K). The calorimetric measurements indicate a higher homogeneity of the surface behaviour for CO adsorption when the catalyst was reduced at 773 K. The addition of cerium caused the appearance of a more heterogeneous distribution of active sites, whereas adding tin led to a

higher homogeneity of the sites. CO chemisorption results indicate that reduction at 773 K decreased the Pt surface area for all the studied catalysts. Uner et al. [I351 have investigated the adsorption of oxygen over Pt/Ti02 surfaces as a function of the metal loading. Oxygen adsorption over pure Ti02 was molecular at all pressure ranges investigated. On the other hand, differential heats of adsorption measured by adsorption calorimetry indicated that oxygen adsorption was dissociative over Pt/Ti02 surfaces until the Pt surface was saturated with oxygen. The saturation coverage of atomic oxygen on Pt was determined as approximately one oxygen atom per Pt. The initial heat of oxygen adsorption was about 375 kJ mol-' over 0.5 mass.% Pt/Ti02. The differential heats of oxygen adsorption, at coverages less than one monolayer of Pt surface, were of the order of 200 kJ mol-', consistent with the oxygen adsorption heats over Pt single crystals reported in the literature. As the pressure increased after the saturation of Pt surface, oxygen adsorption continued in a manner similar to oxygen adsorption over pure TiO2. The isosteric heats of oxygen adsorption over pure Ti02 and Pt/Ti02 samples at coverages greater than 5 pmol g-' of catalyst were determined to be around 10 kJ mol-'. The amounts of oxygen adsorbed over Pt/Ti02 surfaces were found to depend strongly on the pretreatment conditions. The most favourable results were obtained over calcined and reduced samples. The adsorption of CO on silica-, alumina- and titania-supported Pd catalysts (2, 5 and 10 mass% Pd) has been studied by adsorption calorimetry [136]. An increase in CO adsorption with increasing Pd content has been observed and interpreted in terms of the particle size and/or surface structure of the Pd crystallites. CO adsorption on Pd/A1203 and Pd/ Ti02 catalysts was found to decrease significantly upon increasing the reduction temperature, while no changes were detected for PdJSiO2 systems. Two of the support materials (Ti02 and A1203)were active, while Si02 can be considered as inert. The strength of the interaction between the metal and the support was found to increase with the ionic character ( S O 2< A1203 < Ti02)of the support oxide. Guerrero-Ruiz et al. [I371 have studied the adsorption of CO on Pd catalysts supported on A1203, Zr02, Zr02-Si02, and Zr02-La203, using adsorption microcalorimetry and IR spectroscopy. CO was adsorbed on palladium catalysts in three different modes, giving rise to differential heats of adsorption lying in three different ranges (210-170 kJ mol-', 140-120 kJ mol-', and 95-60 kJ mol-' ). The energetic distribution of adsorption sites on the samples was found to depend on the support, the pre-treatment, and the reduction temperature, with all three factors influencing the fraction of sites adsorbing CO with specific heats. Moreover, the addition of Ce02 promoter weakened the adsorption strength of CO on palladium.

Sorption and microcalorimetric investigations of the interaction of hydrogen with Pd on a 1% Pd-graphimet catalyst at room temperature have been carried out by Kirhly et al. [138]. The enthalpy of chemisorption increased from 77 to 102 kJ mol-' upon medium-temperature reduction by H2 at 573 K; hence reduction increased the strength of the Pd-H bond, which may be related to the decreased catalytic activity of Pd-graphimet in hydrogenation reactions. Guerrero-Ruiz et al. [I391 have studied the catalytic behaviour of ruthenium for the conversion of n-hexane when supported on non-reducible carriers such as activated carbon and high-surface-area graphite. The samples were also characterized by microcalorimetry of CO adsorption. The higher initial heat of CO adsorption observed for rutheniudgraphite (135 kJ mol-') compared to rutheniudactivated carbon (115 kJ mol-') indicates an enhanced electron density of the ruthenium particles caused by electron transfer from the graphite. The catalytic results show that ruthenium particles with an increased electron density have a higher activity for n-hexane conversion. In another article [140], Guerrero-Ruiz et al. have investigated carbonsupported Pt and Ru catalysts, using high surface area graphite supports with similar textural characteristics but different surface chemistry. Adsorption microcalorimetry experiments on these catalysts revealed that, within the investigated range, metallic dispersion had no significant effect on the differential heats of CO adsorption. However, the differential heats of CO adsorption were higher for the Pt and Ru samples supported on graphite without oxygen functional groups on its surface, indicating that the presence of oxygen groups at the surface of graphite restrains the metal-support interaction. Microcalorimetric measurements of CO and H2 adsorption at 308 K have been performed on nickel powder, nickel boride (Ni-B) and nickel phosphide (Ni-P) in order to establish the effects of B and P on nickel [141]. The initial values of Qdflfor CO and H2 adsorption on reduced Ni powder were 120 and 85 kJ mol-', respectively. The presence of B or P decreased the initial heats of CO and H2 adsorption (by 20 and 10 kJ mol-' respectively for B, and by 30 and 20 kJ mol-' for P), and also decreased the saturation uptakes of probe molecules per unit surface area. A series of binary (CdZnO and CdAl2O3)and ternary (CdZnOIA1203) copper catalysts employed in methanol synthesis has been investigated by d'Alnoncourt et al. [142]. The results obtained using CO adsorption microcalorimetry showed that CdZnO/A1203had the lowest initial heat of adsorption of 68 kJ mol-' and was the most active catalyst for methanol synthesis. CuIZnO showed a heat of adsorption of 71 kJ mol-' and a lower activity, and Cu/A1203had the highest initial heat of adsorption of 8 1 kJ mol-' and the lowest activity. The decrease of about 10 kJ mol-' in the heat of adsorption of CO induced by the presence of ZnO has been attributed to strong metal-support interactions.

4.2. Oxides and supported oxides The reducibility of V205 catalysts has been studied by temperature programmed measurements, using a microbalance linked to a differential scanning microcalorimeter. Ethane, ethylene and hydrogen were used as reducing agents (with helium as carrier gas), at a heating rate of 5 K min-'. The enthalpies of reduction and re-oxidation were calculated for both bulk V2O5 [I431 and supported V205/y-A1203[144]. The vanadate species of supported V205were less easily reduced by hydrogen than the crystallites of vanadium pentoxide, and ethane and ethylene reduced the vanadate species less thoroughly than hydrogen.

I-0.3 DTG (g.min-')

Figure 11. Heats of reduction of bulk V205in H2, C2Haand C2H4gas flows, and derivatives of the thermogravimetric curves [143].

C2& is a much more efficient reducing agent for bulk V2O5than Hz and C2&. Figure 11 shows the heats of reduction, which are exothermic, and the DTG curves, which are negative because a mass loss occurs during the reduction. The derivative is a better indication of whether the heat evolved is associated with a mass variation than the TG signal itself. The mass-loss data indicated that reduction occurred to a similar extent, which corresponds to the formation of

V203,regardless of whether hydrogen, ethane or ethylene was used as reducing agent. The influence of the support (Si02, Zr02, Ti02, y-A1203, or MgO) on the reduction patterns, the acid-base properties, and the catalytic activity of supported tin dioxide catalysts has been investigated by temperatureprogrammed reductionloxidation, adsorption calorimetry, and reduction of NO, by ethene in an oxygen-rich atmosphere [145]. Two series of Sn02 catalysts with low (approx. 3 mass%) and high (approx. 20 mass%) Sn contents were prepared by impregnation. The dramatic influence of the support on the activity and selectivity of the Sn02surfaces in the NO reduction by C2H4was evidenced. A direct relationship between reducibility and catalytic activity has also been observed. Above monolayer coverage, the molecular structure of SnOz plays an important role. For the 20 Sn mass% series, the reducibility for SnIV--, SnII, based on the temperature at the maximum of the reduction peak, was found to vary in the order SnSi-20 > SnTi-20 > SnAl-20 > SnZr-20 (Figure 12). The results suggest that a relatively strong acidity is necessary for good catalytic performance, but no direct correlation was observed between the number of acid sites and the catalytic activity.

Temperature ("Cl

Figure 12. TPR profiles (heat-flow signal v. temperature) of high-loading supported tin dioxide samples [145]. The main properties of alumina-supported indium oxide (In203)catalysts with In loadings between 2 and 22 mass%, prepared by impregnation, have been

characterized by Perdigon-Melon et al. [146]. The redox character of the dispersed In203phase was studied by performing redox cycles both in a flow apparatus and in a thermobalance coupled with a differential scanning microcalorimeter (TG-DSC). The results were interpreted in terms of In203 surface dispersion or aggregation. The catalysts tested in the reduction of NO, by ethane in an oxygen-rich atmosphere showed an interesting ability to selectively reduce NO, to N2, independently of the In loading.

5. CORRELATION WITH CATALYTIC ACTIVITY One method often used to cast light on the mechanism of catalytic action is to search for correlations between the catalyst activity or selectivity and some other property of its surface. Because contact catalysis necessarily involves the adsorption of at least one of the reactants as a step in the reaction mechanism, the correlation of quantities related to chemisorption of a reactant with the catalytic activity has frequently been attempted. The intensity of the bond between adsorbate and absorbent is obviously a relevant parameter. For this reason, many attempts have been made to correlate heats of adsorption with activities [I]. Moreover, for catalysts with energetically heterogeneous surfaces, it is important to determine which fraction of the surface actually participates in the reaction and to measure the adsorption heats on the corresponding sites. A direct correlation between activity and the bond energy between the surface and a reactant frequently means that adsorption of the reactant is the rate-limiting step in the reaction mechanism. This ought to be checked by studying the kinetics of the reaction. In acid catalysis, it is important to discriminate between the strength of given categories of acid sites and the total acidity. For example, the total acidity must be sufficient for the catalytic process to take place, but, in order to prevent undesired side reactions (such as polymerization), the presence of very strong acid sites must sometimes be avoided. Coke formation also significantly decreases the acidity and acid strength of catalysts. It is not always easy to find direct correlations between the heats of adsorption and the catalytic behaviour (activity and/or selectivity). Various attempts at establishing such correlations have been reported in the literature. For instance, a comparison of microcalorimetric measurements with kinetic studies performed over acidic zeolites for methylamine disproportionation reactions, methanol dehydration, and reactions of methanol and dimethylether with methylamines, suggests that acid sites are required in these reactions for the strong adsorption of ammonia and methylamines, while weak adsorption sites are required to facilitate desorption of adsorbed amine species from the acid sites [147].

NH3 adsorption microcalorimetry has been used to characterize the acid sites of a H-USY zeolite and another USY sample in which the strong Lewis acid sites were poisoned with ammonia. Poisoning of the Lewis acid sites did not affect the rate of deactivation, the cracking activity, or the distribution of cracked products during 2-methylpentane cracking. Thus, strong Lewis acid sites do not seem to play any important role in cracking reactions [148]. The acid-base properties of zeolites or oxides are often studied by measuring the selectivities to the different products in the decomposition of alcohols and particularly isopropanol. The rate of propene formation can very often be correlated to the number of acidic sites determined by ammonia adsorption. A relationship has been found between the strength of the acid sites of bulk oxides, as determined by ammonia adsorption microcalorimetry [95], and the activation energy of dehydration, while the activation energy of dehydrogenation was independent of the strength of the sites [149]. In another study, two series of alkali-metal ion-exchanged zeolites have been investigated in order to analyze the possible correlations between the acidity and basicity of the X and Y zeolite structures and their catalytic properties [90]. The catalytic results for the conversion of 4-methylpentan-2-01 show that the activity and selectivity are both affected to some extent by the acid-base character of the catalysts. The main reaction that takes place is dehydration, giving 4methylpent-1-ene and 4-methylpent-2-ene in variable amounts. Skeletal isomers of C6-alkeneswere also formed in some cases. Simultaneous dehydrogenation to 4-methylpenten-2-one may also occur. Catalytic reactions involving secondary alcohols on metal oxides are thought to proceed through mechanisms involving a cooperative action of acidic and basic sites. For the studied zeolites, a quasi-linear correlation was established between the alkene selectivities and the ratio of basic to acidic sites, as determined by adsorption calorimetry (Figure 13; see also Figure 6). On the basis of calorimetric measurements, an interpretation of the 4methylpentan-2-01 conversion data can be formulated which takes into account the role of the concentration and strength of the sites in governing the competition among the various mechanisms for dehydration and dehydrogenation [127].

Figure 13. Alkene selectivity in 4-methylpentan-2-01 conversion vs ratio of basic to acidic sites [90]. Several zirconium oxide catalysts, differing in their preparation procedures and/or the addition of dopants, have been investigated with the aim of finding a correlation between the catalytic behaviour and the acid-base properties [150]. Ammonia and carbon dioxide were chosen as probe molecules for the microcalorimetric characterization of the samples, which were tested in the dehydration of 4-methylpentan-2-01. Depending on the samples, the initial heats of NH3 adsorption fell within the 210-85 kJ mol-' range, and the Qdg curves showed either a sharp drop or a smooth decrease with increasing coverage. Similar trends were observed for the differential heats of COz adsorption, but in the opposite direction (the most basic sample was the least acidic). The acidbase properties of these catalysts appeared to play a key role in governing the reaction mechanism, as a good balance between the numbers of acidic and basic sites seemed necessary to obtain a good selectivity to the alk-1-ene isomer. Among the various catalysts investigated in the literature, copper-exchanged ZSM-5 zeolite has generated a great deal of interest as a potential catalyst for NO, removal. The differences in the activity of the copper sites have been investigated by studying the adsorption properties of copper-exchanged ZSM-5 and ETS-10 catalysts towards NO and CO probes and the corresponding adsorption energies [151]. Carbon monoxide was found to adsorb strongly and preferentially on Cu' ion species [152], while NO could be more specifically adsorbed on isolated cuZ' species. The samples presenting the best NO adsorption properties were also the most active in the reduction of NO by

hydrocarbons [153]. The energy distribution of the copper sites towards NO adsorption has also been determined using the method of isosteres [154,155]. For supported tin oxide catalysts, the influence of the support on the reducibility and the acid-base properties [I451 has already been discussed above (see Figure 12). This study has also evidenced the primordial influence of the support on the activity and selectivity of the Sn02 surfaces in the reduction of NO by Cz&. A direct relationship between reducibility and catalytic activity has been observed, as illustrated in Figure 14, which represents the competitiveness factor versus the temperature of the reduction peak maximum as determined by TPR, using scanning heat flow calorimetry.

420

440

460

T max red f

480

500

OC

Figure 14. Competitiveness factor in deNO, versus temperature of the reduction peak maximum [145]. The catalytic properties of niobic acid (Nb2O5.nH2O)and niobium phosphate (NbOP04) surfaces have been studied in the reaction of fructose dehydration carried out in water. The reaction was performed in a continuous reactor at different temperatures (363-383 K) and pressures (from 2 to 6 bar). Niobium phosphate exhibited a superior activity and selectivity to 5-hydroxymethyl-2furaldehyde (HMF) compared to niobic acid. The higher catalytic performance of niobium phosphate compared to niobic acid could be related to the higher effective acidity of its surface, as evidenced by the calorimetric study of acidbase titrations carried out in different polar liquids [34,35]. The conversion of methane by carbon dioxide to form carbon monoxide and hydrogen is called dry reforming. Its strategic interest resides in the fact that it provides a C O m ratio adapted to the gas-to-liquid process (methanol and Fischer-Tropsch synthesis of higher hydrocarbons). This reaction takes place on metals such as Rh/Si02, Co/Si02 or nickel-based catalysts. For these catalysts

the interaction between metal and support has been found to strongly influence reducibility and therefore catalytic activity. A direct relationship has been observed between the acid-base properties of the catalysts (in their native form or modified by basic additives such as MgO, La203) as determined by calorimetry and coke deposition on their surface [156]. Hydrotreating is a critical process in petroleum refining. Hydro-demetallation (HDM), hydro-desulfurization (HDS) and hydro-denitrogenation (HDN) are examples of hydrotreating operations involving fixed bed units in which different catalysts are used. The main variations come from the different metal contents and the different physicochemical properties of the supports on which the metals are dispersed. In general, hydrocracking catalysts (HC) are dual function cracking catalysts with an acidic component and a hydrogenationdehydrogenation component. The acidity is provided by large quantities of HYtype zeolites and alumina, while the hydrogenationldehydrogenation function is provided, either by a noble metal such as Pd, or by a combination of Ni and Mo. In contrast, in diesel selective HC, the acidity function is provided mainly by alumina-silica mixtures, or by aluminosilicate gels containing a low amount of HY zeolite, while W/Ni or Mo/Ni combinations act as hydrogenationdehydrogenation centres. Thus by changing the nature and quantity of the acidic component it is possible to change the HC activity and selectivity properties [157]. Consequently, the determination of the number and strength of the active sites provided by adsorption calorimetry is of primordial importance to applications in the petroleum refining chemistry.

6. CONCLUSIONS This chapter has presented a brief overview of various studies in which calorimetry has been used to characterize zeolites, oxides, and metallic catalysts. It is apparent from the results in the literature that microcalorimetry is a very powerful technique even if it does not provide direct information about the molecular nature of the adsorbed species. The introduction of very sensitive flow microcalorimeters has provided the tools needed to measure, in detail, the energy of an adsorption system as a function of coverage, thus making it possible to study surface heterogeneity, whether induced or structural, and to gather information on the strength and distribution of the different sites on the surface of a catalyst. Among the various available surface techniques, adsorption calorimetry has the merit of being doubly quantitative, because the adsorbed amounts and the evolved heats are measured simultaneously, providing a thermochemical picture of surface interactions. This chapter has given some illustrations of successful applications in different fields of catalysis, focusing mainly on redox and acid-base catalysis.

However, the calorimetric technique is still not as widely employed as the importance of energy data would suggest. Further applications of calorimetry will be developed sooner or later in other domains where catalysts play a crucial role, such as fuel cells, hydrogen production or storage, and more generally green and environmental chemistry. 7. REFERENCES

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

P.C. Gravelle, Thermochim. Acta, 96 (1985) 365. J.K. Norskov, T. Bligaard, A. Logadottir, S. Bahn, L.B. Hansen, M. Bollinger, H. Bengaard, B. Hammer, Z. Slivancanin, M. Mavrikakis, Y. Xu, S. Dahl and C.J.H. Jacobsen, J. Catal., 209 (2002) 275. A.L. Myers and F. Siperstein, Colloids and Surfaces A, 187-188 (2001) 73. A. Auroux in Molecular Sieves, Vol. 6, (Eds H.G. Karge and J. Weitkamp), Springer-Verlag, Berlin Heidelberg, 2006, DO1 10.1007/3829~008. W.E. Farneth and R.J. Gorte, Chem. Rev., 95 (1995) 615. N. Cardona-Martinez and J.A. Dumesic, Adv. Catal., 38 (1992) 149. P.J. Anderson and H.H. Kung, Catalysis, Vol. 11, Royal Society of Chemistry, 1995, p. 441. P.C. Gravelle, Adv. Catal., 22 (1972) 191. P.C. Gravelle, Catal. Rev. Sci. Eng., 16 (1977) 37. A. Auroux in Catalyst Characterization: Physical techniques for solid materials, (Eds B. Imelik and J.C. VCdrine), Plenum Press, New York, 1994, p. 611. A. Auroux, Topics in Catal., 4 (1997) 71. B. Fubini and Thermochim. Acta, 135 (1988) 19. V. Solinas and I. Ferino, Catal. Today, 41 (1998) 179. A. Auroux, Topics in Catal., 19 (2002) 205. H. Stach, U. Lohse, H. Thamm and W. Schirmer, Zeolites, 6 (1986) 74. A.L. Klyachko, T.R. Brueva, I.V. Mishin, G.I. Kapustin and A.M. Rubinshtein in Proc. Int. Symp. Zeolites, Szeged, Hungary, Sept 11-14, 1978 (Ed. P. Fejes); Acta Phys. et Chem. Nova Series XXIV p. 183. A. Auroux, Y.S. Jin, J.C. VCdrine and L. Benoist, Appl. Catal., 36 (1988) 323. E.A. Fesenko, P.A. Barnes, G.M.B. Parkes, D.R. Brown and M. Naderi, J. Phys. Chem. B, 105 (2001) 6178. D.J. Parillo and R.J. Gorte, Thermochim. Acta, 3 12 (1998) 125.

20. A.I. Biaglow, D.J. Parrillo, G.T. Kokotailo and R.J. Gorte, J. Catal., 148 (1994) 213. 21. A.L. Klyachko, T.R. Brueva, I.V. Mishin, G.I. Kapustin and A.M. Rubinshtein, Acta Phys. Chem., 24 (1978) 183. 22. A. Auroux, M. Huang and S. Kaliaguine, Langmuir, 12 (1996) 4803. 23. I. Ferino, R. Monaci, E. Rombi and V. Solinas, J. Chem. Soc. Faraday Trans., 94 (1998) 2647. 24. G.I. Kapustin, T.R. Brueva, A.L. Klyachko and A.M. Rubinshtein, Kinet. Katal., 22 (1981) 1247. 25. D.R. Brown and A.J. Groszek, Langmuir, 16 (2000) 4207. 26. A.J. Groszek, Thermochim. Acta 3 12 (1998) 133. 27. A.J. Groszek in Adsorption and its Application in Industry and

Environmental Protection, (Ed. A. Dabrowski), Elsevier, Amsterdam, 1999, p. 143.; Stud. Surf. Sci. Catal., 120A (1999) 143. 28. A.T. Aguayo, A.G. Gayubo, J. Erena, M. Olazar, J.M. Arandes and J. Bilbao, J. Chem. Tech. Biotechnol., 60 (1994) 141. 29. M.Yu. Sinev and V.Yu. Bychkov, Kinet. Catal., 40 (1999) 819. 30. R.S. Drago, S.C. Dias, M. Torrealba and L. de Lima, J. Am. Chem. Soc., 119 (1997) 4444. 3 1. R.S. Drago, S.C. Dias, J.M. Mc Gilvray and A.L.M.L. Mateus, J. Phys. Chem. B, 102 (1998) 1508. 32. S. Savitz, A.L. Myers, R.J. Gorte and D. White, J. Am. Chem. Soc., 120 (1998) 5701. 33. S. Savitz, A.L. Myers, R.J. Gorte and D. White, Microporous Mesoporous Mater., 3 1 (1999) 211. 34. P. Carniti, A. Gervasini, S. Biella and A. Auroux, Catal. Today, 118 (2006) 373. 35. P. Carniti, A. Gervasini, S. Biella and A. Auroux, Chem. Mater., 17 (2005) 6128. 36. C. Guimon and H. Martinez, Recent Research Developments in Catal., 2 (2003) 99. 37. I.V. Mishin, A.L. Klyachko, G.I. Kapustin and H.G. Karge, Kinet. Catal., 34 (1993) 828. 38. Y. Mitani, K. Tsutsumi and H. Takahashi, Colloid Polym. Sci., 264 (1986) 445. 39. S.M. Babitz, B.A. Williams, M.A. Kuehne, H.H. Kung and J.T. Miller, Thermochim. Acta, 312 (1998) 17. 40. S.S. Khvoshchev and E.A. Vasil'eva, Izvest. Akad. Nauk. SSSR Ser. Khim., 5 (1982) 973. 41. D. Barthomeuf, Mater. Chem. Phys., 18 (1988) 553. 42. G. Busca and V. Lorenzelli, Mater. Chem., 7 (1982) 89.

43. P. Cartraud, B. Chauveau, M. Bernard and A. Cointot, J. Thermal. Anal., 11 (1977) 51. 44. A. Boreave, A. Auroux and C. Guimon, Microporous Mater., 1l(1997) 275. 45. R.J. Gorte, D. White, Topics in Catal., 4 (1997) 57. 46. R.J. Gorte, Catal. Lett., 62 (1999) 1. 47. J. Sepa, C.C. Lee, R.J. Gorte and D. White in Proc 11" Congress on Catalysis, (Eds J.W. Hightower, W.N. Delgass, E. Iglesia and A.T. Bell), Baltimore, Maryland, USA, June 30- July 5, 1996, PO 144. 48. J. Jbchen, H. Stach, M. Busio and J.H.M.C. van Wolput, Thermochim. Acta, 312 (1998) 33. 49. J. Kotrla, L. Kubelkovk, C.C. Lee and R.J. Gorte, J. Phys. Chem. B, 102 (1998) 1437. 50. C.C. Lee, R.J. Gorte and W.E. Farneth, J. Phys. Chem. B, 101 (1997) 3811. 51. J. Sepa, C. Lee, R.J. Gorte, D. White, E. Kassab, E.M. Evleth, H. Jessri and M. Allavena, J. Phys. Chem., 100 (1996) 18515. 52. F. Witzel, H.G. Karge and A. Gutsze in Proc 9thInt. Zeolite Conf., (Eds R. von Ballmoos, J.B. Higgins and M.M.J. Treacy), Montreal, Canada, July 5-10, 1992, Butterworth-Heinemann, Boston, 1993, Vol 11, p. 283. 53. D.T. Chen, S.B. Sharma, I. Filimonov and J.A. Dumesic, Catal. Letters, 12 (1992) 201. 54. D.J. Parrillo and R.J. Gorte, J. Phys. Chem., 97 (1993) 8786. 55. C. Lee, D.J. Parillo, R.J. Gorte and W.E. Farneth, J. Am. Chem. Soc., 118 (1996) 3262. 56. D.J. Parrillo and R.J. Gorte, Catal. Letters, 16 (1992) 17. 57. D.J. Parrillo, R.J. Gorte and W.E. Farneth, J. Am. Chem. Soc., 115 (1993) 12441. 58. A. Auroux, in Innovation in Zeolite Materials Science, Proc. Int. Symp., (Eds P.J. Grobet, W.J. Mortier, E.F. Vansant and G. Schulz-Ekloff), Nieuwpoort, Belgium, September 13-17, 1987, Elsevier, Amsterdam, 1988, p. 385; Stud. Surf. Sci. Catal., 37 (1988) 385. 59. A. Auroux, Z.C. Shi, N. Echoufi and Y. Ben Taarit in Zeolites as Catalysts, Sorbents and Detergent Builders, Proc. Int. Symp., (Eds H.G. Karge and J. Weitkamp), Wiirzburg, Germany, September 4-8, 1988, Elsevier, Amsterdam, 1989, p. 377; Stud. Surf. Sci. Catal., 46 (1989) 377. 60. E.N. Coker, C. Jia and H.G. Karge, Langmuir, 16 (2000) 1205. 61. A. Coma, V. Fornes, J.M. Guil, S. Pergher, Th.L.M. Maesen and J.G. Buglass, Microporous Mesoporous Mater., 38 (2000) 301. 62. C. Mellot, M.H. Simonot-Grange, E. Pilverdier, J.P. Bellat and D. Espinat, Langmuir, 11 (1995) 1726.

J.M. Guil, R. Guil-Lopez and J.A. Perdigon-Melon in Proc. 1 2 Int. ~ ~ Congress on Catalysis, (Eds A. Coma, F.V. Melo, S. Mendioroz and J.L.G. Fierro), Granada, Spain, July 9-14,2000, Elsevier, Amsterdam, 2000, p. 2927; Stud. Surf. Sci. Catal., 130 (2000) 2927. J. Jtinchen, H. Stach, M. Busio and J.H.M.C. van Wolput, Thermochim. Acta, 312 (1998) 33. A. Corma, C. Corell, J. Perez-Pariente, J.M. Guil, R. Guil-Lopez, S. Nicolopoulos, J. Gonzalez Calbet and M. Vallet-Regi, Zeolites, 16 (1996) 7. J.M. Guil, R. Guil-Lopez and J.A. Perdigon-Melon, A. Corma, Microporous Mesoporous Mater., 22 (1998) 269. A. Corma, A. Chica, J.M. Guil, F.J. Llopis, G. Mabilon, J.A. PerdigonMelon and S. Valencia, J. Catal., 189 (2000) 382. W.J.M. Van Well, J. Jhchen, J.W. de Haan and R.A. van Santen, J. Phys. Chem. B, 103 (1999) 1841. D. Uner and M. Uner, Thermochim. Acta, 434 (2005) 107. S. Yang and A. Navrotsky, Microporous Mesoporous Mater., 37 (2000) 175. A. Gervasini, P. Carniti and A. Auroux, Thermochim. Acta, 434 (2005) 42. J.A. MenCndez, B. Xia, J. Phillips and Lj.R. Radovic, Langmuir, 13 (1997) 3414. N. Fripiat, R. Conanec, A. Auroux, Y. Laurent and P. Grange, J. Catal., 167 (1997) 543. D. Tichit, M.H. Lhouty, A. Guida, B.H. Chiche, F. Figueras, A. Auroux, D. Bartalini and E. Garrone, J. Catal., 151 (1995) 151. L. Dussault, J.C. Dupin, F. Dumitriu, A. Auroux and C. Guimon, Thermochim. Acta, 434 (2005) 93. H.A. Prescott, Z-J. Li, E. Kemnitz, A. Trunschke, J. Deutsch, H. Lieske and A. Auroux, J. Catal., 234 (2005) 119. Lj. Damjanovic, V. Rakic, U.B. Mioc and A. Auroux, Thermochim. Acta, 434 (2005) 8 1. G. Della Gatta, Thermochim. Acta, 96 (1985) 349. S. Cerny and V. Ponec, Catal. Rev. 2 (1968) 249. B.E. Spiewak and J.A. Dumesic, Thermochim. Acta, 290 (1996) 43. B.E. Spiewak and J.A. Dumesic, Thermochim. Acta, 3 12 (1998) 95. A. Auroux, M. Muscas, D.J. Coster and J.J. Fripiat, Catal. Lett., 28 (1994) 179. W. Rudzinski and D.H. Everett, Adsorption of Gases on Heterogenous Surfaces, Academic Press, London, 1992, p. 608. A. Auroux and Y. Ben Taarit, Thermochim. Acta, 122 (1987) 63.

85. D. Chen, S. Sharma, N. Cardona-Martinez, J.A. Dumesic, V.A. Bell, G.D. Hodge and R.J. Madon, J. Catal., 136 (1999) 392. 86. B.A. Williams, S.M. Babitz, J.T. Miller, R.Q. Snurr and H.H. Kung, Appl. Catal. A: General, 177 (1999) 161. 87. C. Viornery, A. Auroux, F. Fajula and P. Schulz in Catalysis on Solid

88. 89. 90. 91. 92. 93. 94.

95. 96.

Acids and Bases, Proc. of the DGMK Conf., (Eds J. Weitkamp and B. Liicke), Berlin, Germany, March 14-15, 1996, DGMK, Hamburg, p. 41. D. McQueen, B.H. Chiche, F. Fajula, A. Auroux, C. Guimon, F. Fitoussi and P. Schulz, J. Catal., 161 (1996) 587. R.D. Shannon, K. Gardner, R.H. Staley, G. Bergeret, P. Gallezot and A. Auroux, J. Phys. Chem., 89 (1985) 4778. A. Auroux, P. Artizzu, I. Ferino, R. Monaci, E. Rombi and V. Solinas, Micropor. Mat., 11 (1997) 117. Jianyi. Shen, A. Auroux, Stud. Surf. Sci. Catal., 149 (2004) 35. V. Rakic, V. Dondur, S. Gajinov and A. Auroux, Thermochim. Acta, 420 (2004) 5 1. M. Calatayud, A. Markovits and C. Minot, Catal. Today, 89 (2004) 269. A. Auroux and J.C. VCdrine in Catalysis by Acid and Bases, (Eds B. Imelik, C. Naccache, G. Coudurier, Y.Ben Taarit and J.C. VCdrine), Elsevier, Amsterdam, 1985, p. 3 11. A. Auroux and A. Gervasini, J. Phys. Chem., 94 (1990) 6371. J.M. McHale, A. Auroux, A.J. Perrolta and A. Navrotsky, Science, 277

(1997) 788. 97. A. Desmartin-Chomel, J.L. Flores, A. Bourane, J.M. Clacens, F. Figueras,

G. Delahay, A. Giroir-Fendler and C. Lehaut-Burnouf, J. Phys. Chem. B, 110 (2006) 858. 98. A. Gervasini and A. Auroux, J. Thermal. Anal., 37 (1991) 1737. 99. A. Auroux, P. Artizzu, I. Ferino, R. Monaci, E. Rombi, V. Solinas and G. Petrini, J. Chem. Soc., Faraday Trans., 92 (1996) 2619. 100. M. Rodriguez Delgado, C. Morterra, G. Cerrato, G. Magnacca and C. Otero Areh, Langmuir, 18(26) (2002) 10255. 101. A. Gervasini, G. Bellussi, J. Fenyvesi and A. Auroux, J. Phys. Chem., 99 (1995) 5117. 102. E. Rombi, I. Ferino, R. Monaci, C. Picciau, V. Solinas and R. Buzzoni, Appl. Catal. A: General, 266(1) (2004) 73. 103. V. Bolis, G. Magnacca, G. Cerrato and C. Morterra, Thermochim. Acta, 312 (1998) 63. 104. V. Bolis, G. Magnacca, G. Cerrato and C. Morterra, Topics Catal., 19 (2002) 259. 105. G. Cerrato, C. Morterra, M.R. Delgado, C.O. Arean, M. Signoretto, F. Somma and F. Pinna, Microporous Mesoporous Mater., 94 (2006) 40.

106. C. Morterra, G. Meligrana, G. Cerrato, V. Solinas, E. Rombi and M.F. Sani, Langmuir, 19 (2003) 5344. 107. R.M. Marcus, R.D. Gonzalez, E.L. Kugler and A. Auroux, Chem. Eng. Comm., 190 (2003) 1601. 108. G. Magnacca, G. Cerrato, C. Morterra, M. Signoretto, F. Somma and F. Pina, Chem. Mater., 15 (2003) 675. 109. A. Gervasini, J. Fenyvesi and A. Auroux, Langmuir, 12 (1996) 5356. 110. N. Cardona-Martinez and J.A. Dumesic, J. Catal., 127 (1991) 706. 111. G. Connel and J.A. Durnesic, J. Catal., 105 (1987) 285. 112. H. Zou, X. Ge and J.Y. Shen, Thermochim. Acta, 397 (2003) 81. 113. B. Gergely and A. Auroux, Res. Chem. Interned., 25 (1999) 13. 114. C. Guimon, A. Gervasini and A, Auroux, J. Phys. Chem. B, 105(42) (2001) 10316. 115. X. Gu, J. Ge, H. Zhang, A. Auroux and J. Shen, Thermochim. Acta, 45 1 (2006) 84. 116. J. Le Bars, J.C. VCdrine, A. Auroux, S. Trautmann and M. Baerns, Appl. Catal. A: General, 119 (1994) 341. 117. J. Keranen, C. Guimon, E. Iiskola, A. Auroux and L. Niinisto, Catal. Today, 78 (2003) 149. 118. J. Keranen, C. Guimon, E. Iiskola, A. Auroux and L. Niinisto, J. Phys. Chem. B, 107 (2003) 10773. 119. J. Le Bars, J.C. VCdrine and A. Auroux, Appl. Catal. A: General, 88 (1992) 179. 120. Z.M. Wang, M. Yamaguchi, I. Goto and M. Kumagai, Phys. Chem. Chem. Phys., 2 (2000) 3007. 121. M.G. Cutrufello, S. De Rossi, I. Ferino, R. Monaci, E. Rombi and V. Solinas, Thermochim. Acta, 434 (2005) 62. 122. E. Rombi, M.G. Cutrufello, V. Solinas, S. De Rossi, G. Ferraris and A. Pistone, Appl. Catal. A: General, 25 1 (2003) 255. 123. G. Neri, A. Pistone, S. De Rossi, E. Rombi, C. Milone and S. Galvagno, Appl. Catal A: General 260 (2004) 75. 124. A. Auroux, R. Monaci, E. Rombi, V. Solinas, A. Sorrentino and E. Santacesaria, Thermochim. Acta, 379 (2001) 227. 125. A. Pantazidis, A. Auroux, J.-M. Herrmann and C. Mirodatos, Catal. Today, 32 (1996) 81. 126. V. Solinas, E. Rombi, I. Ferino, M.G. Cutrufello, G. Colo6n and J.A. Navio, J. Mol. Catal. A: Chemical, 204-205 (2003) 629. 127. M.G. Cutrufello, I. Ferino, R. Monaci, E. Rombi and V. Solinas, Topics in Catal., 19 (2002) 225. 128. G. Neri, G. Rizzo, S. Galvagno, G. Loiacono, A. Donato, M.G. Musolino, R. Petropaolo and E. Rombi, Appl. Catal. A: General, 274 (2004) 243.

129. A. Auroux, in Proc. of the 29thNorth American Thermal Analysis Society Conf. (Ed. K.J. Kociba), St. Louis, Missouri, U.S.A., September, 24-26, 2001, B&K Publishing, 2001, p.53. 130. A.Auroux, in Physical Chemistry 2004, Proc. of 7thIntern. Conf. on Fundamental and Applied Aspects on Physical Chemistry, (Eds. A.AnticJovanovic, S.Anic), September 2 1-23, Belgrade, Serbia, p. 189. 131. J. Phillips, J. Thenn. Anal., 49 (1997) 541. 132. B.E. Spiewak, R.D. Cortright and J.A. Dumesic, J. Catal., 176 (1998) 405. 133. R.M. Watwe, B.E. Spiewak, R.D. Cortright and J.A. Dumesic, Catal. Lett., 51 (1998) 139. 134. J.C. Serrano-Ruiz, G.W. Huber, M.A. Sanchez-Castillo, J.A. Dumesic, F. Rodriguez-Reinoso and A. Sepulveda-Escribano, J. Catal., 241 (2006) 378. 135. D. Uner, N.A. Tapan, I. Ozen and M. Uner, Appl. Catal. A: General, 25 1 (2003) 225. 136. H. Dropsch and M. Baerns, Appl. Catal. A: General, 158 (1997) 163. 137. A. Guerrero-Ruiz, S. Yang, Q. Xin, A. Maroto-Valiente, M. BenitoGonzalez and I. Rodriguez-Rarnos, Langmuir, 16 (2000) 8 100. 138. Z. Kirhly, A. Mastalir, F. Berger and I. D6khny, Langmuir, 13 (1997) 465. 139. A. Guerrero-Ruiz, B. Bachiller-Baeza and I. Rodriguez-Ramos, Appl. Catal. A: General, 173 (1998) 23 1. 140. A. Guerrero-Ruiz, P. Badnes and I. Rodriguez-Ramos, Appl. Catal. A: General, 173 (1998) 3 13. 141. J. Shen, B.E. Spiewak and J.A. Dumesic, Langmuir, 13 (1997) 2735. 142. R.N. d7Alnoncourt,M. Kurtz, H. Wilmer, E. Loffler, V. Hagen, Jianyi. Shen and M. Muhler, J. Catal., 220 (2003) 249. 143. J. Le Bars, A. Auroux, J.C. VCdrine, B. Pommier and G.M. Pajonk, J. Phys. Chem., 96 (1992) 2217. 144. J. Le Bars, A. Auroux, M. Forissier and J.C. VCdrine, J. Catal., 162 (1996) 250. 145. A. Auroux, D. Sprinceana and A. Gervasini, J. Catal., 195 (2000) 140. 146. J.A. Perdigon-Melon, A. Gervasini and A. Auroux, J. Catal., 234 (2005) 421. 147. D.T. Chen, L. Zhang, J.M. Kobe, Yi. Chen and J.A. Dumesic, J. Mol. Catal., 93 (1994) 337. 148. S.M. Babitz, M.A. Kuehne, H.H. Kung and J.T. Miller, Ind. Eng. Chem. Res., 36 (1997) 3027. 149. A. Gervasini and A. Auroux, J. Catal., 131 (1991) 190.

150. A. Auroux, P. Artizzu, I. Ferino, V. Solinas, G. Leofanti, M. Padovan, G. Messina and R. Mansani, J. Chem. Soc. Faraday Trans., 91 (1995) 3263. 15 1. A. Auroux, C. Picciau and A. Gervasini in Porous Materials in Environmentally Friendly Processes; Proc. 1" Int. FEZA Conf., (I. Kiricsi, G. PA-Borbely, J.B. Nagy and H.G. Karge),Eger, Hungary, September 14, 1999, Elsevier, Amsterdam, 1999, p. 555; Stud. Surf. Sci. Catal. 125 (1999) 555. 152. G.D. Borgard, S. Molvik, P. Balaraman, T.W. Root and J.A. Dumesic, Langmuir, 11 (1995) 2065. 153. A. Gervasini, C. Picciau and A. Auroux, Microporous Mesoporous Mater., 35-36 (2000) 457. 154. A. Gervasini, P. Carniti and A. Auroux, Thermochim. Acta, 379 (2001) 95. 155. P. Carniti, A. Gervasini and A. Auroux, Langmuir, 17 (2001) 6938. 156. R. Bouarab, 0 . Akdim, A. Auroux, 0. Cherifi and C. Mirodatos, Appl. Catal. A: General, 264 (2004) 161. 157. M.L. Occelli, S. Biz, A. Auroux and P.S. Iyer, Appl. Catal. A: General, 179 (1999) 117.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 12

COORDINATION COMPOUNDS AND INORGANICS Stefano Materazzi Department of Chemistry University "La Sapienza" p.le A. Moro 5 - 00 185 ROMA - Italy stefano.materazzi@uniromal .it

1. INTRODUCTION Coordination compounds and inorganics are extensively studied for their use in many different fields, often as starting precursors. Moreover, they represent useful (sometimes simple) models to understand the behaviour of more complex molecules that are involved in biological reactions or that can be of biomedical interest. The thermoanalytical techniques (thermogravimetry TG, differential scanning calorimetry DSC, differential thermal analysis DTA, etc.) allow fundamental characterizing data to be obtained. These techniques have been described in previous volumes of this Handbook (for example, in Volume 1 of this series most of the classic thermoanalytical techniques are described in detail, Seifert's Chapter 10 in Volume 2 extensively describes several techniques applied to general inorganic chemicals and coordination compounds, Temperature modulated DSC kndamentals are covered in Volume 3, and many books on the theory of the thermal analysis are cited therein). Hyphenated techniques have also been reported when applied to inorganic or coordination compounds. TG has long been used as a tool for studying thermally activated processes in a quantitative manner, but the addition of mass spectrometry (TG-MS) or the complementary technique of Fourier transform infrared spectroscopy (TGFTIR) permitted the development of evolved gas analysis (EGA), the identification of gaseous species evolved during thermal processes. The quantitative aspects of the combined techniques are usually taken care of by the thermal analyzer, while the spectroscopic techniques are used primarily for identification purposes. However, if two or more gaseous species are evolved simultaneously, the possibility of quantifying the individual components by

spectroscopy could offer a significant advantage. IR is particularly effective at identifying molecular functional groups, complex gas mixtures, isomers, and low molecular weight species (e.g., H20, CO, C02, NH3). The low molecular weight species are often decomposition products which are more difficult for MS to distinguish from the ubiquitous "air background". This chapter does not pretend to be a comprehensive review, but covers the thermal analysis literature on coordination compounds and inorganics published in the years 2000-2006 and is limited to the most recent representative publications. Vyazovkin reported in his 2006 review on thermal analysis, according to the IS1 Web of Science, the application of only one technique, DSC, to the process of crystallization alone has been reported in over 1100 papers. In this situation, it is not possible to cite all the scientific contributions.

2. REVIEWS Every two years, a fimdamental review on thermal analysis is published in Analytical Chemistry, in which the development of new methods and the main applications to calibration, thermodynamics, kinetics, polymers, inorganics, pharmaceutical, biological, foods, etc. are reported [l-31. Several articles regarding coordination compounds are cited and critically described. An update of the applications of evolved gas analysis (EGA), coupled to the thermoanalytical instruments, is also published every four years, and many studies on coordination compounds are cited [4-71. In 2001, Simon reviewed the theory of temperature-modulated DSC (TMDSC) with several examples of its applications, including many inorganic and coordination compounds [8]. Starink [9] provided a comprehensive review of the application of calorimetry in the analysis of a large variety of processes in aluminum-based alloys. Two comprehensive reviews on phase diagrams, crystal structures and thermodynamic properties of ternary chlorides formed in the systems AC1/LnCl3 (A=Na, K, Rb, Cs) were presented by Seifert [10,11]. The second review continues an earlier review with the same contents on the lanthanides from La to Gd, and, in both papers, the author's own studies, published since 1985, together with original papers from other scientists, were treated. With the three larger cations, compounds of the composition A3LnC16, A2LnC15, ALn2C17 and beginning with holmium Cs3Ln2C19were formed. With sodium the compounds Na3Ln5Cl18(Ln=La to Sm) and NaLnCw(Ln=Eu to Lu) were also reported. The stability of a ternary chloride in a system ACl/LnC13 is given by the 'free enthalpy of synreaction', the formation of a compound from its neighbour compounds in its system. This AGO,,, must be negative. A surprising result is that the highest-melting compounds in the systems, A3LnC16,are formed from

ACl and A2LnC15with a loss of lattice energy, U.They exist as high-temperature compounds due to a sufficiently high gain in entropy at temperatures where the entropy term TAS compensates the endothermic AH.

3. USE OF COORDINATION COMPOUNDS AND INORGANICS TO DEVELOP NEW METHODS Coordination compounds and inorganics are often used as model systems to develop new methodologies. Parkes et al. used the decomposition of basic copper carbonate to illustrate the sensitivity of a new thermal analysis instrument that uses microwaves. Because physical or chemical alterations in a material cause variations in its dielectric properties, the instrument shows a large temperature increase on the formation of the strongly coupling oxide [12]. Allen et al. introduced a scanning calorimeter for use with a single solid or liquid sample having a volume of a few nanoliters. Its use was demonstrated with the melting of few nanoliters of indium, using heating rates from 100 to 1000 KS-I [13]. Tozaki et al. developed a new method (see Figure 1) for simultaneously measuring heat flow and thermal expansion at a temperature resolution of millikelvins. The method has been applied to study the thermal behaviour of a single crystal of BaTi03 that is known to undergo a first-order tetragonal-tocubic transition at around 404 K [14] and of the magnetic effect on the ferroelectric transition of single crystalline KD2P04[15]. A new batch microcalorimeter has been developed for measuring the dissolution of small amounts of easily or slightly soluble solids. The calorimeter has been calibrated by dissolution of potassium chloride and successfully tested by measurements of the enthalpies of dissolution of acetanilide and adenine [161. Binner et al. [17] presented the design and construction of a calorimeter in which the specimen may be heated by microwave radiation, as well as by hot air. The effect of the intensity of microwave radiation was examined by measuring the melting of benzyl and the solid-state phase transition in silver iodide. For the latter, the transition temperature has been found to vary significantly with the intensity of microwave irradiation. Nanocalorimetry has been used by Olson and coworkers to investigate sizedependent melting of Bi nanoparticles. It has been discovered that for particles smaller than 7 nm the measured melting temperatures are about 50 K above the value predicted by the homogeneous melting model. The discrepancy is analyzed in terms of a possible size-dependent crystal structure change and the superheating of the solid phase [ 181.

Figure 1. Schematic drawing of the high resolution and super-sensitive DSC working in a magnetic bore between 120 and 420 K. A, refrigerating head; B, thermal reservoir; C, thermal insulator; D, copper plates connected to the calorimeter, TS 1-TS4; Pt resistance thermometers. Reproduced from reference [15] with permission from Springer.

Paulik et al. determined, by the "simultaneous Q-DTA-Q-TG method, a latent error in determining the decomposition heat of salt hydrates, decomposing congruently or incongruently. The authors reported that this error cannot be revealed by traditional calorimetric or thermoanalytical methods owing to overlapping of the processes taking part in the thermal decomposition, but it can be detected and eliminated by applying the "simultaneous Q-DTA-Q-TG method with its very high resolution and selectivity [19]. The temperature calibration of high performance DSC (HPer DSC) in the heating and cooling mode has been discussed by Vanden Poel and Mathot. Several primary and secondary calibration standards were studied at different sample masses - from 0.4 pg to 10 mg - and various heating and cooling rates from 1 to 500 "Clmin and from 1 to 300 'Clmin, respectively. The experimental onset and peak temperatures of indium samples, with different masses, were measured at different heating rates and the two related correction factors were presented. The symmetry of the HPer DSC with respect to the cooling and the heating modes was found to be good. The liquid crystals M24, HP-53 and BCH52, being substances exhibiting little or no supercooling, are recommended as secondary standards for temperature calibration in both the cooling and heating modes. To verify whether the proposed correction factors for indium could also be used in the cooling mode, the melting behaviour of indium and the phase transition temperatures of the secondary standards, obtained on heating, were compared, and it turned out that the latter are usable as well. The calibration procedure developed for HPer DSC (see Figure 2) was reported to facilitate the right choices for minimizing the thermal lag with respect to the sample mass and scan rates at the start of the measurement, instead of just making corrections afterwards [20]. The potential of simultaneous techniques that enable the qualitative analysis of evolved species, such as TG-MS or TG-FTIR, has been hrther improved by the introduction of pulse thermal analysis (Figure 3) (see also Chapter 4 of this volume.). This method provides a quantitative calibration by relating the mass spectrometric or FTIR signals to the injected quantity of probe gas. The influence of several experimental parameters such as concentration of the analyzed species, temperature and flow rate of the carrier gas on FTIR signals was investigated [21]. The reliability of quantifying FTIR signals was checked by relating them to the amount of evolved gases measured by thermogravimetry.

Indium

155 )

160

(a)

Indlum, Healing Rate ISO'Clmin

-

e

8

;lm 1 165

170 175 180 Temperature (DC)

185

190

R .--4

0 _s c-;,

--j

r

,T

I

T*

Tim

Temowawre

ex0

Temperature("C)

a 3 0 0

I:

3

0

I

155 (b)

160

165

170 175 180 Temperature ("c)

185

190

Figure 2. (left upper) The experimental values of AT (=Tp - TEO)of indium as a hnction of the square root of the sample mass multiplied with the heating rate. R: thermal resistance, Ro: the outside thermal resistance, Rs: thermal resistance within the sample and m: sample mass. (left lower) HPer DSC heating curves for the melting of indium for four different masses at Sh= 150 OCImin (lines). The dotted curves represent the 0.016, 1.010 and 5.354 mg indium curves, multiplied by their respective factors, to equal the curve for 8.043 mg of indium. (right) HPer DSC heating curves of the melting of indium, for samples with (a) a mass of 0.113 mg and (b) 5.354 mg, at various heating rates (Sh from 1 to 500 'Clmin). a: angle, B: heating rate, R: thermal resistance, t: time, 7P: peak temperature, ZEO: extrapolated onset and Cs: heat capacity of the sample. Inset: idealized melting curve of a calibration standard. Reproduced from reference [20] with permission from Elsevier.

s

25 00

e/

tim

50 02

00 0

(A)

0

50

0 1 00

01

0.00 0.01 0.02

Absorbance Units

30

00 3 50

0 40 00

445

2500 2100 2300 2200 2100 2000 1900 1800 1700 1600 1500 1400 1300 1200

Wavenumber / cm–1

Figure 3. (A) 3-D FTIR-diagram of the calibration (first pulse: CO2, second pulse: CO) and the evolved gases during the decomposition of ZnC2O4-nH2O. (B) Normalized CO and CO2 traces obtained during calibration (injection of 1 ml CO and CO2) and decomposition of ZnC2O4-nH2O; the TG curve is shown in the lower part of the plot. Reproduced from reference [21] with permission from Elsevier. 4. INORGANICS The thermal treatment of inorganic substances has a great synthetic potential because it can turn simple compounds into advanced materials such as ceramics, catalysts, glasses, etc. 4.1. Alloys Toda and coworkers applied temperature-modulated DSC to measure the kinetics of the martensitic transformation in Ti-Ni alloy [22]. By analyzing the temperature dependence of the relaxation time, they found that the process has two characteristic relaxation times, apparently related to different processes. The effects of the heating and cooling rates on the transformation characteristics in TiNiCu shape-memory alloy were investigated by Wang et al. [23]. The results showed that the martensitic end transformation temperature (Mf) and reverse end transformation temperature (Af) strongly depend on the DSC scanning rate of the heating-cooling process, with the evidence that Mf decreased and Af increased with increasing cooling/heating rate. However, the martensitic start transformation temperature (Ms) and reverse start transformation temperature (As) are not so sensitive to the scanning rate.

-

* a

;

1;

2 ;

..M, 1 :

;6 1;

A

22

Scanning rale I Wmin

Figure 4. Phase transformation temperatures as a function of the heatinghooling rates, and DSC curves of the TiNiCu sample annealed at 500 OC for (a) 30 min and (b) 3 h and measured with different heatingtcooling rates. Reproduced from reference [23] with permission from Elsevier.

Temperature IK

Ziewiec et al. [24] reported on the preparation, thermal stability and glassforming ability of copper-nickel-phosphorus alloys. They found that, depending on the composition, melt spinning may result in either amorphous or partially crystalline systems, whose thermal behaviour was characterized by DSC, DTA, and XRD. By combining calorimetry with three-dimensional atom probe analysis, Starink and coworkers 1251 examined the aging, at room-temperature, of Al-Cu-Mg-Mn alloys and found that the process is accompanied by a substantial exothermic heat evolution, whereas the only micro-structural change involved the formation of Cu-Mg co-clusters. Tatar and Zengin [26] studied the effects of neutron irradiation on the oxidation behaviour, microstructure and transformation temperatures of CuAlNiMn shape-memory alloy. They showed that irradiation

leads to observable changes of structure, and decreased temperatures of the martensite * austenite phase transformation. An increase in activation energy of the oxidation reaction is observed due to applied irradiation. Kost'al et al. [27] applied the penetration method to measure the temperature dependence of viscosity for C U , ( A S ~ S ~melts ~ ) ~(x ~ ~=- 1, ~ 5, 10, and 20) and found it to follow the Arrhenius law. Both the kinetic and thermodynamic fragility exhibited a similar compositional dependence, indicating that increasing Cu content causes topological changes in the structure of As2Se3 super-cooled melts. Mehta and Kumar studied the crystallization kinetics of two alloys, namely Seso-xTe20Cdx (x=O, 5, 10, 15) and SeskTezoGeX(x=5, 15, 20), by an isothermal method. For this purpose, conductivity measurements were done during isothermal annealing at various temperatures between the glass transition and crystallization temperatures. Avrami's equation was used to calculate the activation energy of crystallization (E,) and the order parameter (n). The authors showed that Avrami's theory of isothermal crystallization correctly describes the crystallization kinetics in the present alloys [28]. Vizcaino et al. developed a method to simultaneously measure hydrogen concentrations in zirconium-based alloys. The measurements provided both the temperature of terminal solid solubility and the heat of hydride dissolution, using a DSC. The hydrogen concentration measured with that technique, and the values obtained with a standard hydrogen gas meter, showed a linear relation with a correlation factor of 0.99 over the entire solubility interval in the aZr phase. The authors suggested that the present method was especially appropriate for alloys where a partition of the overall hydrogen concentration in two phases exists, and is applicable to all hydride forming metals which ideally follow the van't Hoff law [29]. Fe-Mn-Si-based alloys have been extensively studied since the shape memory effect (SME) was discovered in a Fe-30Mn-1Si single crystal in 1982 and later in polycrystals in 1986. Since these alloys exhibit mainly one-way SME, they are expected to be used as a material for tighteners or pipe couplings on a large scale, due to their low cost and good workability. Tatar and Yakuphanoglu reported an experimental investigation of the effect of pressure on the shape memory behaviour of Fe-32%Mn-6%Si-3%Cr alloy, and showed that there are significant differences in the Gibbs energy, dislocation density and transformation temperatures of the alloy due to the applied pressure [30].

Figure 5. The variation of transformation temperatures and enthalpy with applied pressure. Reproduced from reference [30] with permission from Elsevier. Studies on alloy systems related to semi-conducting and soldering materials are of interest from both the scientific and technological points-of-view, and accurate information on the thermodynamic properties and phase diagrams of these systems is essential. Some of the binary systems have been studied extensively, but the experimental data for ternary andlor multi-component systems are not always adequate. The method for predicting the thermodynamic properties for ternary alloys has been developed by model calculations based on the data for the three constituent binary systems. Katayama et al. [3 11 measured the EMF of galvanic cells with zirconia solid electrolytes to determine the activity of gallium in liquid Ga-In-T1 alloys in the temperature range of 9501300 K. A mixture of Ga and Gaz03was used as a reference electrode. Multi-component alloys of gallium and germanium with rare-earths are of high interest due to their possible application in the production of novel magnetic materials. Molten gallium has been used as a solvent for the synthesis of rareearth intermetallides containing germanium and transition metals. The partial enthalpies of mixing of yttrium and the integral enthalpies of mixing of liquid Ga-Ge-Y alloys have been determined by high-temperature mixing calorimetry for five sections with constant ratios of Ga and Ge at 1760 K [32]. The specific heat capacities (c,) of two single-crystalline ferromagnetic shapememory alloys, close to the stoichiometric Ni2MnGa, have been measured by the ac technique in both quasi-isothermal and scanning modes. The measurements were carried out in a temperature range slightly broader than the transformation interval of each alloy. This allows reliable values of the cp difference between parent and martensite phases within the transformation range to be obtained, which is important for the thermodynamic analysis of the thermoelastic martensitic transformation undergone by these alloys [33]. The potential of DSC in suggesting modifications of thermal treatments of A1 and Mg alloys to increase their mechanical properties was described by Riontino

and coworkers. Significant results were obtained fiom calorimetric evolution after a series of annealings, even without a direct observation of the microstructure, and the role of a reference baseline was discussed. The formation, dissolution or transformation of Guinier-Preston (GP) zones was followed, and their relevance for the increasing of microhardness was shown, for an AlZnMg alloy of technical interest. Multi-stage thermal treatments have been confirmed to be beneficial. A secondary precipitation occurs at room temperature after annealing at temperatures at which primary precipitation is almost complete [34]. Moreover, the DSC investigation on a Mg-RE-Y-Zr (RE=rare earth) technical alloy allowed the hardness trend during isothermal treatments to be correlated with the evolution of the calorimetric traces and to the forming of phases with its precursors. Oversaturation of solute elements occurred at temperatures higher than 150 "C, on cooling at room temperature after the annealing. Activation energies, found under non-isothermal conditions on artificially-aged samples, suggest a slow transformation velocity, while the hardness response is relatively fast [35]. Thermal analysis was applied by Campos et al. [36] to determine the temperatures of thermal reactions, phase transformations or melting reactions during continuous heating of steels. These reactions are shown to be a direct response to the steel composition and to the sintering atmosphere. Simultaneous thermal analysis TG-DTA (STA) showed up the sintering behaviour of sintered low pre-alloyed chromium steels and their peculiarities. Given the high oxygen affinity of chromium, graphite additions can modify their thermal reactions, and hence the sintering behaviour of the steel. Evidence was also given of the effect of carbon on the sintering process and the nature of the oxides. 4.2. Arsenates Sharpataya et al. [37] measured the heat capacity of caesium hexafluoroarsenate in the temperature range 300-850 K and detected a solidsolid transformation from the rhombohedra1 to cubic phase, occurring in the range 235-360 K. The transition has been interpreted as an order-disorder type, and its enthalpy and entropy have been determined. The magnetocaloric effect has been discussed by Kalva and Sestak [38], who proposed the use of a quasiparticle formalism to model it. Bi et al. [39] reported an interesting study of six new hetero-poly-compounds in the [ M ~ ( H ~ o ) ~ ( A ~ ~ series w ~ (M ~ o=~ Cu(II), ~ ) ~ ]Mn(II), ~ ~ - Co(II), Ni(II), Zn(II), Cd(I1)). The two copper atoms of the C U ~ Ocluster ] ~ are coordinated by water molecules. The replacement reactions of the coordinated water molecules of this series of hetero-poly-compounds in aqueous solutions and in selected organic solvents were reported. The results showed that [ F ~ ( c N ) ~ ][~F-~, ( c N ) ~ ] ~ , H2NCH2CH2NH2,etc., can replace the coordinated water to form its

characteristic colour in aqueous solutions, while in organic solvents the coordinated water molecules are lost, leaving unshared coordination positions that can be occupied by some organic ligands, such as pyridine, lactic acid, and acetone to restore the octahedral coordination of M~'. 4.3. Borates

Hamilton et al. described the coordination polymeric compound Pb[B(Im)4](N03)(xH20), formed from sodium tetrakis(imidazoly1)borate and lead-(11) nitrate solutions, as a layered material with the metal centres facing the interlayer spacing. As for naturally occurring layered minerals, this compound can readily undergo anion exchange and reversible intercalation of solvent water in the solid state with retention of crystallinity. Changes in solvent intercalation b state NMR (SSNMR) and TG analyses. While the were observed by 2 0 7 ~solid iodide compound can be obtained through facile exchange from the nitrate parent compound, an organic anion benzoate is placed in the interlayer spacing for nitrate under self-assembly conditions [40]. Zhang et al. [41] studied the crystal structure of the compound Ca3La3(B03)5 and, by means of infrared spectra, they confirmed that the boron atoms have three-fold coordination. DTA revealed that reversible decomposition takes place at 1165 "C and that it melts completely above 1300 "C. By using TG-DTA and DSC, Koga and coworkers [42] determined that the thermal dehydration of dipotassium tetraborate tetrahydrate occurs in three steps giving rise to an anhydrous glass that shows a glass transition at 700 K and subsequently crystallizes in two consecutive steps at 770 and 900 K. The final crystallization product, triclinic K2B4O7,melts at 1072 K. Because in the west of China some salt lake brines contain abundant boron and lithium, in which solute-solvent and solute-solute interactions are complex, studies on the thermochemical properties for the systems related with the brines are essential to understand the effects of temperature on excess free energies and solubility, and to build a thermodynamic model that can be applied for prediction of the properties. Yin et al. [43] measured the enthalpies of dilution for aqueous Li2B407solutions from 0.0212 to 2.1530 mollkg at 298.15 K. The relative apparent molar enthalpies and relative partial molar enthalpies of the solvent and solute were also calculated, and the thermodynamic properties of the complex aqueous solutions were represented by a modified Pitzer ioninteraction model. Heat capacity measurements of K3Nb306(B03)2,a borate that exhibits superionic conductivity and ferroelectric properties, have been performed in the 60220 K temperature range. In this work, Maczka et al. documented the presence of two heat capacity anomalies: the first anomaly occurred at 198.3 K and has

been attributed to a first-order phase transition, while the second anomaly suggested the presence of a second-order phase transition at around 80 K [44]. 4.4. Carbonates As an alternative to the traditional physico-geometrical models, Korobov explored a new class of discrete models, based on Dirichlet tessellations, whose application is illustrated by the thermal decomposition of NH4HC03 [45]. Korobov considered the thermal decomposition of ammonium hydrogen carbonate as an example of a discrete kinetic model of reactions in crystals, which simultaneously play the roles of reactant and reaction medium [46]. Gallagher et al. combined simultaneous TGIDSC with high-temperature XRD to investigate reactions occurring in the system SrC03-Fe203on heating to 1300 "C in an atmosphere of C02. While elucidating the reaction mechanisms, the authors do not find any evidence of a Hedvall effect associated with the firstorder phase transformation in SrC03 at 927 OC [47]. Gallagher and Sanders [48] studied the thermal decomposition of CaC03 using simultaneous TG-DSC for two different ranges of particle size from the same source and a physical mixture of each. The difference in kinetic behaviour was as expected qualitatively, but significantly different quantitatively. In addition, the mixture did not behave as a simple combination of its end members. These discrepancies were attributed to the problems associated with mass and thermal transport. The TG data again proved easier to fit than the DSC data. Koga and Tanaka [49], from quantitative analyses of evolved C02 and H 2 0 during the thermal decomposition of sodium hydrogen carbonate, NaHC03, plotted calibration curves, i.e. the amounts of evolved gases vs. the corresponding peak areas of mass chromatograms measured by TG-MS. The accuracy and reliability of quantitative analyses of evolved C02 and H20, based on the calibration curves, were evaluated by applying the calibration curves to the mass chromatograms for the thermal decompositions of copper(I1) and zinc carbonate hydroxides. Li and coworkers [50] studied the zinc hydroxide carbonate precursor, Zn4C03(OH)6.H20,synthesized from zinc sulfate using ammonium carbonate as a precipitating agent. TG), DSC, transmission electronic microscopy (TEM), infrared spectroscopy (IR) and XRD were used to characterize the precursor and the decomposition product, while the non-isothermal kinetics of the thermal decomposition of zinc hydroxide carbonate were studied in nitrogen. The kinetic parameters were obtained using a model-free method and the reaction model was then derived by means of non-linear regression. The results showed that the decomposition of zinc hydroxide carbonate is a two-step reaction: a reversible reaction of two-dimensional diffusion (D2), followed by an irreversible one of

nth-order reaction (Fn), and that the decomposition of ZII&O~(OH)~.H~O to ZnO is accompanied by changes in particle morphology and particle size. Strobe1 et al. [51] prepared barium carbonate nanoparticles (50-100 nm) by flame spray pyrolysis. The rapid quenching during the preparation process resulted in the unprecedented formation of pure monoclinic BaC03. The asprepared material was characterized by electron microscopy, XRD as well as by TG and DSC analyses (Figures 6 and 7). At ambient conditions the metastable monoclinic phase transformed easily into the thermodynamically stable orthorhombic BaC03 (witherite) within a few days.

DSC

-25

,

0

l

200

.

l

400

.

l

600

.

l

800

.

l

1000

.

l

1200

.

1400

temperature I 'C

Figure 6. Thermal analysis and XRD investigations of monoclinic BaC03. (A) Change of the mass (TG) and thermal effects (DSC) recorded during heating of BaC03 in Ar with a rate of 10 Wmin. The samples analyzed later by XRD were collected after separate runs stopped at 600 (2), 261 (3) and 102 "C (4). (B) XRD patterns of BaC03 heated up to the temperatures marked on TG curve. Reproduced from reference [511 with permission from Elsevier.

l , . , . , s , . , . , . , . , . , . , . l 0

10

20

30

40

50

60

70

80

90

100

time I hrs

Figure 7. Degree of the transformation of monoclinic into orthorhombic barium carbonate. Reproduced from reference [511 with permission from Elsevier. The application of the third-law method to decomposition studies [52] supported the basic assumptions underlying the physical approach to interpretation of decomposition kinetics. A good fit of experiment to theory for the ratio of the initial decomposition temperature to the E parameter, the peculiarities of carbonate decomposition in COz and the regularities of solid and melted nitrate decompositions are in complete agreement with the mechanism of dissociative evaporation and consumption of a part z of the condensation energy by the reactant. 4.5. Chromates

Labus and coworkers [53] studied the phase transitions and thermal effects occurring during annealing of CrO, (x > 2.4) in air. The mass changes observed on heating of the materials are accompanied by exothermic effects that change to endothermic ones when the sample mass decreases. This phenomenon is explained by the competition between reconstruction of the crystalline lattice (endothermic effect) and recombination of the evolved atomic oxygen (exothermic effect). 4.6. Iodides Silver iodide, which undergoes a solid state phase transition at 147 "C from the low-temperature P-phase (Wurtzite structure) to the high-temperature, ionically conducting a-phase (body-centred cubic iodide containing a disordered silver ion sub-lattice) was investigated by means of modulated temperature calorimetry, using a novel apparatus that could expose the sample to

conventional heating, microwave heating or any combination of the two. Binner and coworkers showed that a 20 "C decrease in temperature for the P-a phase transition was achieved when the sample was irradiated with microwave energy. Quasi-isothermal studies also suggested that silver iodide could be transformed between states, without changing its temperature, when irradiated with microwaves. This study suggests that a genuine 'microwave effect' is occurring in this material [54,55]. Studies by the same authors, on the a-P phase transition of silver iodide, by conventional and modulated-temperature calorimetry, were reported. The dielectric properties of silver iodide were measured during conventional heating using the cavity perturbation technique [56]. 4.7. Nitrates and nitrites Ammonium nitrate (AN) is both a common commercial fertilizer and an ingredient of explosives. It is typically processed and used in one of several forms that include neat AN, its mixture with fuel oil, and aqueous solutions, which form the basis of ammonium nitrate emulsion explosives. Turcotte et al. performed hazard assessment of these AN systems by using laboratory-scale calorimeters. They studied the effects of sample mass, atmosphere, and formulation on the resulting onset temperatures and proposed a method of extrapolating these results to large-scale inventories [57]. Desensitizing of AN is an important practical problem. Oxley et al. applied thermal analysis to screen various AN formulations in search of a possible desentitizers. The sodium, potassium, ammonium, and calcium salts of sulfate, phosphate, or carbonate, as well as certain high-nitrogen organics (urea, oxalate, formate, guanidinum salts) are considered as promising candidates because they enhance the thermal stability of AN and because they can be used with agricultural products [58]. Singh and Felix [59] demonstrated that transition metal salts of 5-nitro-2,4dihydro-3H-1,2,4-triaole-3-one increase the steady burning rate of an AN-based propellant. TG and DSC experiments showed that these compounds also lower the decomposition temperature of AN and alter the phase-transition temperatures. Drake et al. [60] synthesized three new families of triazole-based salts with nitric, perchloric, and dinitramidic acids. Many of these salts have melting points well below 100 "C, but high temperatures of onset of decomposition, which define them as new, highly energetic members of the well-known class of materials identified as ionic liquids. Sun and coworkers used calorimetry to study the effect of inorganic acids on the decomposition of ammonium nitrate that has been found to accelerate in the presence of the acids. Kinetic data have been collected and evaluated in terms of thermal runaway models [61]. Li and Koseki [62] studied the effect of the contamination by chlorides on the thermal stability of ammonium nitrate. The

results suggested that the addition of a small amount of chloride gives rise to a decrease in the initial decomposition temperature of ammonium nitrate by nearly 100 OC, as well as to a significant increase of the reaction heat. A sample of strontium nitrite has been synthesized and thermally characterized by Tuukkanen and coworkers [63] to investigate its role in the ageing of magnesium-strontium nitrate pyrotechnic compositions in the presence of water vapour. Studies by TG-MS and isothermal microcalorimetry showed that the addition of strontium nitrite to a 50% magnesium-50% strontium nitrate composition eliminated the induction reaction normally observed in closed ampoule studies in air at 50 OC and relative humidities in the range 65-69%. The detailed TG/DTA/EGA-MS studies of thermal decomposition of several hydrates of d-metal nitrates: Cr(N03)3.9H20,Z ~ I ( N O ~ ) ~ - ~ MII(NO~)~.~H~O, H~O, C O ( N O ~ ) ~ . ~ HNi(N03)2.6H20, ~O, Hg(N03)2 , carried out by Malecki and Malecka [64], showed that nitrous oxide N20 is always observed among gaseous products together with HN03, NO, NO2, O2 and H20. The general scheme of thermal decomposition of Me(N03),.qH20, given in Figure 8, should be completed with another way leading to nitrous oxide formation. Formally this way could be described by the following reaction: 2 ~ 0 ~ - = ~ ~ 0 + 2.0 ~ + 0 ~ -

Figure 8. The main routes of thermal decomposition of hydrates of nitrates of delectron metals. Reproduced from reference [64] with permission from Elsevier. The thermal decomposition of gallium nitrate hydrate (Ga(N03)3.xH20)to gallium oxide has been studied by TG/DTG and DSC measurements performed at different heating rates. Berbemi and coworkers [65] concluded that 8 water molecules are present in the hydrate compound. Anhydrous gallium nitrate does not form at any temperature because the reaction consists of coupled dehydration/decomposition processes that occur with a mechanism dependent on heating rate. TG measurements performed with isothermal steps (between 31

and 115 "C) indicate that Ga(OH)2N03forms in the first stage of the reaction. Such a compound undergoes further decomposition to Ga(OH)3 and Ga(N03)0, compounds that then decompose, respectively, to Ga(0H)O and finally to Ga203 and directly to Ga203. Charrier et al. accurately measured the equilibrium temperature of the solidsolid phase transition of high purity caesium nitrate by stepwise heating and by the method of extrapolation to zero heating rate. The authors reported a mean value of 154.3h0.1 OC, that was obtained by using two different heat-flux DSC instruments. Because the transition temperature of caesium nitrate is close to that of the fusion of indium, indium can be used as a single-point calibrant. This approach decreases the error that may arise when the calibration value for the material of interest is determined by interpolation between two-point or multiple-point calibrations [66]. 4.8. Oxalates For many seemingly simple processes, complex kinetic behaviour has been observed, such as the thermal decomposition of CaC03 and the dehydration of CaC204-H20,both of which produce a single gaseous product and, therefore, are often considered as model or standard single-step processes. However, these are reversible processes, and their kinetics are determined not only by the temperature, but also by the equilibrium and actual pressure of the gaseous product. The decomposition of oxalate is the first major thermal event of a series of complex reactions to form strontium metatitanate at high temperature, starting from anhydrous strontium titanyl oxalate. A kinetic study of oxalate decomposition in the temperature range 553-593 K has been carried out by Patra et al. [67]. The results fitted the Zhuravlev equation for almost the entire a-range (0.05-0.92) indicating the occurrence of a diffusion-controlled, threedimensional rate process. The activation energy has been calculated to be 164*10 kJ mol-'. Results from elemental analysis, TG, IR and SEM studies of undecomposed and partially decomposed samples have been used to supplement kinetic observations in formulating the mechanism for oxalate decomposition. Polyoxometalates have become a subject of general interest because of their potential applications in fields as diverse as catalysis, biochemical analysis, medicinal chemistry, and material science. A novel polyoxometalate, which is a layer-like transition metal oxalate templated polyoxovanadium borate, was synthesized under hydrothermal conditions and characterized by EPR, elemental analysis, thermal analysis, single crystal XRD and 2-D IR correlation spectroscopy studies, to show that there is a fragment of {Mn4(C204)[B204(0H)2]2)2fixed in the central cavity as a guest part. The

cluster units are connected to form a two-dimensional framework by octahedral Mn(I1) and ~ a sites + [68]. Shivaiah and Dah [69] reported the characterization of ~ l and ~ + cr3+ polyoxometalate-supported transition metal complexes (Figure 9).

oao

iooao

zooan

~ooao

ilooao

5oo.00

Temp.[%]

Figure 9. TG performed on [AI(OH)6Mo6018{Cu(phen)(H20)2)21 [A1(OH)6M06018{Cu(phen)(H20)C1}2] -5H20= [lc] [la].5H20 Reproduced from reference [69] with permission from ACS. Bayot et al. [70] prepared and characterized ammonium derivatives of molybdenum(V1) complexes of general formula (NJ&)2[MoO(02)2(HxL)]. nH20 and (NH4)2[Mo02(02)(L)]with L=oxalate, citrate, tartrate, glycolate and malate on the basis of elemental and thermal analysis as well as by IR and I3cNMR spectroscopy. . 2H20),, was A phosphate-oxalate compound {Na2[Zn(C204)1.5H2P04] synthesized and characterized [71] using single-crystal XRD and TG. The results revealed that its structure was a new mixed inorganic-organic ligand supramolecular framework built up through multilayer assembly. Mohamed and coworkers [72] reported a comparative investigation of the nonisothermal, solid-state thermal decompositions of the oxalates of six divalent transition metals (cations: manganese, iron, cobalt, nickel, copper and zinc) in alternative flowing atmospheres, inert (N2, CO), reducing (H2) and oxidizing (air). DTG and DSC response peak maxima, providing a measure of reaction temperatures, have been used to determine salt reactivities and thus to characterize the factors that control the relative stabilities of this set of chemically related reactants. Two trends were identified. Trend (1): in the inert and reducing atmospheres, the decomposition temperature (salt stability) increased with rise in enthalpy of formation of the divalent transition metal oxide, MO. It is concluded that the rupture of the cation-oxygen (oxalate) bond is the parameter that determines the stability of salts within this set. Trend (2):

the diminution of decomposition temperatures from values for reactions in inertlreducing atmosphere to those for reactions in an oxidizing atmosphere increased with the difference in formation enthalpy between MO and the other participating oxide. The thermal decomposition in air of NiC2O4-2H20,has been investigated [73] by means of TGDTG-DTA, DSC and XRD. TGDTG-DTA curves showed that the decomposition proceeds through two well-defined steps with DTA peaks closely corresponding to the mass loss obtained. XRD showed that the final solid decomposition product was NiO. Maciejewski et al. [74] convincingly demonstrated the complexity of the ostensibly simple decomposition of CoC204. 2H20, by applying a combination of experimental techniques including TG, DTA, DSC, XRD, TG-MS, and pulse thermal analysis. The process was demonstrated to involve numerous reactions that occur in both the solid and gas phases. To verify the previously assumed [75] decomposition stages of [ C O ( N H ~ ) ~ ] ~ ( C ~ O ~(HACOT), ) ~ . ~ H ~ Ounder different atmospheres (inert, oxidising and reducing), the gaseous products of the decomposition were qualitatively and quantitatively analyzed by MS and FTIR. It was confirmed that the gaseous products of HACOT decomposition, under the studied atmospheres, were H 2 0 (stage I) and NH3, C02 (stage 11). The main gaseous products in the third stage in argon and hydrogen (20 vol.% H2/Ar) were CO and C02, whereas in air (20 vol.% 02/Ar) only C 0 2 was identified. Under oxidizing, as well as reducing atmospheres, the influences of secondary reactions on the compositions of both, solid and gaseous products, were found to be particularly strong during the third stage of the process. These studies of the multi-stage decomposition of HACOT, additionally complicated by many secondary reactions, required application of TA-MS or TA-FT-IR techniques, combined with pulse thermal analysis PTA, which allowed quantification of the spectroscopic signals and investigation of gas-solid and gas-gas reactions in situ [76]. Similar results were reported by Vanhoyland et al. [77], who applied highresolution TG, TG-FTIR, and high-temperature DRIFT and XRD to follow decomposition of La2(C204)310H20.

l

l

100

,

,

200

,

,

300

,

,

:

:

l

400 100 temperature I"C

.

l

200

.

,

300

.

,

l

400

Figure 10. Mass spectrometric signals recorded during decomposition of [ C O ( N H ~ ) ~ ] ~ ( C ~ Oin ~ )thin ~ . ~layers H ~ Oin quartz crucibles in argon, 20 vol.% H2,balance Ar and 20 vol.% 02, balance Ar. Reproduced from reference [76] with permission from Elsevier. From on-line coupled TG-MS and TG-FTIR measurements, in combination with quantitative chemical analysis, the chemical formula for an unknown by bismuth oxalate compound was deduced to be Bi(NH4)(C204)2.3.71(6)H20 Vanhoyland et al. [78]. Solution of the crystallographic structure on the basis of X-ray powder data proved this formula to be correct. Bi is 8-fold coordinated by oxygen from the oxalate anions. Because these BiOs polyhedra do not share any edges or vertexes, an open framework is formed with water and ammonium molecules between. As a result, water can easily be removed, which is clearly indicated by the rapid initial mass loss in the TG curve. Ubaldini and coworkers [79] studied the thermal decomposition of the mixed oxalates (Cel-xGdx)2(C204)3.nH20. The mechanisms of decomposition of Ce and Gd oxalate are different, and mixed oxalates behave in an intermediate way. Their dehydration stages are more similar to those of Gd oxalate, because not all the molecules of water are equivalent, as they are in the cerium oxalate. The decomposition leads to (Cel-xGdx)02-x/2.For x close to 0 or to 1, two solid

solutions exist, while for the central composition, the presence of a biphasic region cannot be excluded. Lamprecht et al. [SO] used DSC, TG and TG-FTIR, and powder XRD to examine the effects of supposedly inert atmospheres of argon and nitrogen on the mechanism of the thermal decomposition of copper(I1) oxalate. The DSC curves in pure argon at 10 " ~ m i n - showed ' a broad endotherm with onset at about 280 "C and maximum at about 295 "C. In mixtures of argon and nitrogen, as the proportion of argon gas was decreased, the endothermic character of the decomposition decreased until, when nitrogen was the main component, the decomposition exhibited a complex broad exothermic character. Powder XRD studies showed that, regardless of the proportions of nitrogen and argon, the DSC residues consisted of mainly copper metal with small amounts of copper(1) oxide (cuprite) and, under some conditions, traces of copper(I1) oxide (tenorite). Various explanations for this behaviour were discussed and a possible answer is given in the disproportionation of C02(g) to form small quantities of 02(g) or monatomic oxygen. The possibility exists that the exothermicity in nitrogen could be explained by reaction of the nitrogen with atomic oxygen to form N20(g), but this product could not be detected using TG-FTIR. Molybdenum-based high-temperature syntheses from the starting elements, with high yields, resulted [81] in chemically inert coordination polymers in which the desired cluster units are connected by bridging ligands into 1-D, 2-D, and 3-D frameworks. Sokolov and coworkers introduced an alternative method of cluster excision from the solids based on the mechanochemical reaction of cluster coordination polymers with an appropriate ligand. TG helped the characterization of these cluster oxalate complexes. 4.9. Oxides The thermal decompositions of inorganic and metallo-organic compounds can provide simpler synthetic routes for many materials. For instance, Nelis et al. [82] applied TG-MS, TG-FT-IR, high-temperature XRD, and diffuse reflectance FTIR to follow the formation of the ferroelectric material, strontium bismuth niobate (SrBi2Nb209),from an aqueous acetate-citrate precursor gel. On heating the amorphous precursor to 425 "C, they detected the formation of an intermediate fluorite phase that converted into the desired ferroelectric perovskite phase at 625 "C. Diez et al. obtained the metastable D-Bi203by decomposing hydrated bismuth oxalate in a CRTA apparatus under air at 270 "C [83]. Aonoand coworkers employed TG-DTA to follow the thermal decomposition of a heteronuclear La-Mn complex that at relatively low temperatures (600-700 "C) yields single-phase nanoparticles of hexagonal LaMn03, which could not be obtained by conventional sintering even at 1200 "C [84]. Tolochko et al. compared citrate and ceramic routes for the preparation

of the La2.,SrxNi04 conducting oxide. The use of citrate mixtures leads to fine powders and dense ceramics and also decreases the temperature by 150-200 "C as measured by TG-DTA [85]. Carp and coworkers started fiom polynuclear coordination compounds containing as ligand the anion of malic acid, to obtain nickel-zinc ferrites via thermal decomposition. They also compared these compounds and the thermal behaviour of the polynuclear coordination compounds. The authors stressed that Fe2Ni,Znl,O4 (x =0.25, 0.5 and 0.75) ferrites with mean particle sizes of 65-85 A and free from other phases are formed after a heating treatment of only one hour at 500 "C [86]. Wilding et al. [87]reported a first-order transition between two liquids in yttrium-aluminate liquids that was interpreted as the nucleation and growth of a low-density phase in a matrix of a higher density liquid when Y2O3-Al2o3 liquids are cooled below the liquidus. DSC in combination with structural data was used to discuss the thermodynamic features of Y2O3-Al2o3liquids that lead to liquid-liquid transitions. Sanders and Gallagher [88,89] studied the kinetics of the oxidation of magnetite (Fe304) to hematite (a-Fe203)in air, using simultaneous TG-DSC. The process was found to be complex and was suggested to involve the metastable spinel, y-Fe203,as an intermediate. The formation of the intermediate was supported by thermomagnetometric evidence. The formation of porous Ti02 thin films fiom commercially available sols of amorphous Ti02 and anatase nanoparticles has been studied by Madarasz et al. [90] by using simultaneous TG-DTA, coupled with MS, which enabled the release of C 0 2 and the crystallization of amorphous Ti02 to anatase to be detected. Both increased crystallinity and improved solar efficacy have been accomplished in the resulting porous material. Thermal decomposition of dried Ti02 gel in air has been monitored by simultaneous TG-DTA-FTIR measurements by Krunks et al. [91], who reported that decomposition occurs in five steps. They detected the release of H 2 0 below 120 "C, followed by acetone, isopropyl acetate, and 1-propanol around 200-300 "C, and finally CO and C02 up to 550 "C. Crystalline Ti02-anatase is formed around 500 "C, overlapping with the exothermic reaction at 410-550 "C caused by combustion of carbon residues. Hardy et al. [92] used thermal and spectroscopic techniques to follow the formation of ferroelectric bismuth titanate (Bi4Ti3OI2)fiom an aqueous metal-chelate gel. The gel appears to remain homogeneous throughout the heat treatment and forms single-phase Bi4Ti3OI2at 625 "C. The Co,Nil,(SeO3).2H20 (x=O, 0.4, 1) family of compounds was hydrothermally synthesized by Larraiiaga et al. under autogeneous pressure and characterized by thermogravimetric and thermodiffiactometric techniques [93]. Lu et al. [94] applied TG-DSC-MS coupled techniques to make a simultaneous characterizing study of the metallic iron-organic coordination compounds

formed by iron ions and the organic solvent, glycol, during the sol-gel process of preparing nanometer films of iron oxide. The positive-ion mass spectra [F~(ocH~cH~)~]+ (mass to charge (m/z)=191) and [F~(ocH~cH~oH)~]+ (m/z=239) of the organic iron coordination compounds were obtained. Mixed-oxide compounds of potential usefulness for fibre coatings were prepared by hybrid sol-gel synthesis and their thermal crystallization was monitored by thermal analysis and XRD [95]. Both the gels convert to the crystalline phase below about 1200 "C, via amorphous intermediates. Lanthanum hexaluminate crystallises sharply at 1230 "C via y-A1203. Ceramic pigments are used for colouring of ceramic glazes, ceramic materials and enamels, and this is possible because of their high thermal stability and chemical resistance. They are also used for colouring of paints, plastic or building materials. Sulcova and coworkers reported in several articles [96-1001 the syntheses of new compounds based on the Ce02-Pro2-Nd203 , Bi203H0203, Bi2-xZr3~403 C00.46Zn0.55(Ti0.064Cr0.91)204 and Li20-Fe203-Ti02 systems, that can be used as pigments for colouring of ceramic glazes. The optimum conditions for the syntheses of these compounds have been estimated and the methods of thermal analysis have provided useful information about the temperature regions of the formation of the investigated pigments. Mixed vanadium-chromium oxide compounds present a wide range of interesting properties; for instance, they have excellent catalytic properties, and recently they were shown to be potential candidates for anodes in lithium-ion batteries. DTA, TG and powder XRD analyses were used [loll to monitor the dehydration/crystallization and phase transitions upon heat treatment of the hydrated vanadates obtained through the reaction of peroxo-polyacids of vanadium and chromium, and to determine the ranges of coexistence of the phases in equilibrium. The enthalpies of reactions between alkaline-earth cuprates M2Cu03(M = Ca, Sr) and hydrochloric acid were measured in a hermetic swinging calorimeter at 298.15 K. The M2Cu03 samples were prepared by solid-phase synthesis from calcium or strontium carbonate and copper oxide. The standard enthalpies of formation obtained for the cuprates were discussed and compared with previous experimental and assessed values by Monayenkova and coworkers [102]. Guo et al. [I031 demonstrated the usefulness of TG-FT-IR and TG-MS techniques by means of the thermal decomposition of Z~[CU(CN)~] to explore the preparation of CdZn nanocomposites. Yariv and Lapides [I041 demonstrated the utility of thermo-XRD analysis in the study of organo-smectite complexes, such as montmorillonite complexes with anilines, fatty acids, alizarinate, and protonated Congo red, and of complexes of other smectites with acridine orange. The technique is very efficient in determining whether the adsorbed organic species penetrates into the 7

interlayer space of the smectites mineral. Xi et al. [I051 used high-resolution TG and XRD to examine the thermal stability of montmorillonite modified with octadecyltrimethylammonium bromide and found that the surfactant remains stable up to 180 OC. Uranium-plutonium mixed oxides (MOX) are used as the fuel in fast breeder reactors, and cerium is one of the major fission products. Owing to its structural similarities, Ce02 readily forms solid solutions with U02 and Pu02, the fuel constituents. Hence, the thermodynamic properties of uranium-cerium mixed oxides were studied by DSC by Krishnana and Nagarajan because they are helpful in estimating the enthalpy changes and heat capacities of MOX fuels [106]. There is also a wide interest in the reprocessing of spent nuclear fuel. A promising way of reprocessing is the use of molten salts and particularly fluoride melts, because of the advantages related to their solvent properties for the dissolution of the fuel, their large electrochemical window and the absence of sensitivity to radiolytic degradation. Lanthanum compounds in fluoride melts were chosen as a model system by Ambrova and Yurisova [107], where the solubility of lanthanum oxide was measured by thermal analysis. The solubility in alkali cryolites is rather high, because of chemical reactions between lanthanum oxide and cryolites. In Li3A1F6-La203,alumina precipitates and in the other systems the mixed oxide LaA103is formed. 4.10. Perchlorates The thermal decomposition characteristics of micron-sized aluminum powder and potassium perchlorate mixtures were studied with thermal analytical techniques by Pourmortazavi and coworkers [108]. The results showed that the reactivity of aluminum powder in air increases as the particle size decreases. Pure aluminum with 5 pm particle size has a fusion temperature of about 647 "C, but for 18 pm powder this temperature is 660 "C. Pure potassium perchlorate has an endothermic peak at 300 OC corresponding to a rhombiccubic transition, a hsion temperature around 590 "C and decomposes at 592 "C. DTA curves for an A15/KC104 (30:70) mixture show a maximum peak temperature for thermal decomposition at 400 "C. Increasing the particle size of aluminum powder increases the ignition temperature of the mixture. The oxidation temperature is increased by increasing the aluminum content of the mixture. Most of the studies on perchlorate compounds are related to their characteristics as energetic compounds, and their applications related to explosives, propellants, etc. are covered in Volume 2 of this Handbook.

4.11. Phosphates A TMA study suggested that this technique could be advantageous in detecting phase transitions in decomposing solids. More specifically, the heating of KH2P04, in which intensive dehydration starts around 180 O C and causes anomalies in various physical properties (e.g., heat absorption in DSC) that are frequently mistaken for the tetragonal-to-monoclinic transition, is described [l09]. Yamashita et al. have carried out a calorimetric study of phase transitions in CsZnP04. They find a heat capacity anomaly due to the IV-111 phase transition around 220 K that becomes larger when the sample is annealed for longer time below the transition temperature. At 584 K, the authors detected the 1-11 transition (Figure 1I), which has a complex nature showing the features of both order-disorder and displacive transitions [110,111].

Figure 11. Three dimensional equi-density surfaces of phase I and DSC traces of CsZnP04 (upper figures). Annealing experiments at 200 and 195 K respectively (lower figures). Reproduced from reference [I101 with permission from APS. By using coupled TG-FTIR, direct evidence of the decomposition of lithium hexafluorophosphate has been obtained by Teng et al. [112]. Their studies showed that LiPF6 is stable under normal temperature when the content of water

is very low. When heated, LiPF6reacts with the trace water in the air and the HF and OPF3 released simultaneously could be detected from the FTIR spectra collected in situ. The peak temperature of the decomposition of LiPF6 varied from 192.93 OC (average heating rate 2.0 OCfmin) to 223.83 OC (average heating rate 20.0 "Clmin). Based on the on-line FTIR spectra, the characteristic absorption bands of PF; in the decomposed substances, which might come from the reaction of trace HF with PF5,were detected.

Figure 12. TG-DTG curves of LiPF6heated from 25 to 500 OC and a 3-D plot of the related FTIR spectra. Reproduced from reference [I121 with permission from Elsevier. Popa and coworkers synthesized barium-zirconium diorthophosphate BaZr(P04)2 and its high temperature behaviour was studied. At room temperature the compound was found to have a monoclinic structure, in agreement with earlier literature reports. A monoclinic-to-hexagonal phase transition at about 733 K was found by DTA and this transition was confirmed by XRD and drop calorimetry [113]. They also reported the high-temperature enthalpy increments of monazite-type EuP04 and SrnP04, measured by drop calorimetry in the temperature range 450-1570 K; the heat capacity was also derived. The excess heat capacity Cex,was calculated by subtracting the lattice heat capacity, interpolated fiom isostructural LaP04 and GdPO compounds. Good agreement was found with values of C,,, calculated from (estimated) crystal field energies. The heat capacity of PrP04 was estimated with this approach [114].

4.12. Stannates The application of thermal analysis in combination with XRD and FTIR showed that the thermal decomposition of a mixture of barium carbonate and tin tetrahydroxide results in the formation barium stannate that has a cubic perovskite structure and can be used as a sensor for the detection of liquefied petroleum gas [I 151. Koferstein and coworkers demonstrated that, starting from

a barium tin 1,2-ethanediolato complex, the initially formed transient barium tin oxycarbonate phase disintegrates into BaC03 and Sn02, reacting subsequently to BaSn03 [116]. 4.13. Sulfides, sulfites and sulfates Some natural or artificial polyalcohols (e.g. sugars or cyclodextrins) can often absorb and transport metal ions in vivo, and the most important chemical factors in these processes are hydrogen-bonding and complexation, i.e. the formation of coordination bond(s) between the OH group and metal ions. For the characterization of these processes, Labadi and coworkers prepared mixed ligand nickel(I1) and cobalt(I1) complexes of different compositions, with water, sulfate ion and 1,2-ethanediol as ligands [117,118]. They observed that:, on decreasing the dielectric constant of the solution by using ethanol or a mixture of ethanol and benzene, the complex formed had a lower water content. When the solution or the solid complex was stored over CaC12 or P2O5in a desiccator, complexes containing both water and 12-ethanediol molecules were formed. In the presence of P2O5, the water content of the complex formed decreased or disappeared. Krunks et al. [I191 studied zinc thiocarbamide chloride as a single-source precursor for obtaining thin films of zinc sulfide by spray pyrolysis. By heating this compound to 1200 "C, they demonstrated that cubic ZnS (sphalerite) forms below 300 "C and stays in this form until 760 "C, when it transforms to hexagonal ZnS (wurtzite). Silva and coworkers [120] reported the thermal characterization of double sulfites with empirical formula Cu2SO3.MSO3.2H2O(where M is Cu, Fe, Mn, or Cd), obtained by saturation with sulfur dioxide gas of an aqueous mixture of M(I1)-sulfate and copper sulfate at room temperature. The thermal behaviour of the double sulfites, evaluated by TG and DSC, showed that these salts are thermally stable up to 200 OC, but the structures of sulfite ion coordination strongly influence the course of the thermal decomposition. The sulfite species coordinated to the metal through the oxygen was more easily oxidized to sulfate than the sulfur-coordinated species. Masset and coworkers studied the thermal decomposition of FeS04-6H20by MS coupled with DTA-TG under an inert atmosphere. The TG measurements suggest that the mechanism of decomposition of FeS04.6H20 can be divided into: i) three dehydration steps:

* Dehydration 1 :

FcS04.6H2Q -

FeS04.4H20t2H20

(Am (exp.)= -X3.8%, Am (th.)= -13.85%)

Dehydration 2:

E 5

FeSOK4i120 FeS04.H20+3N20 (Am (exp.)= -34.1%, Am (th.)= -34.62%) Dehydration 3: FeS04.H20 FeS04+HZ0 (An1 (exp.)= -4 1 %, An1 (&.I= -41.55%)

and ii) two decomposition steps Decomposition 1:

The intermediate compound was identified as Fe2(S0& and the final product as hematite, Fe203[12 11. Because the application of conventional methods requires separation of individual process stages and then their description by complex expressions, Tomaszewicz and Kotfica applied neural networks for the description of the decomposition of cobalt(I1) sulfate(V1) hydrate. Such networks allow description of a given process without the necessity of dividing it into stages; as a consequence, they are easy to handle and result in a very good accuracy of approximating the experimental data [122]. Budmgeac et al. [I231 examined the kinetics of the non-isothermal by employing the methods of Friedman and crystallization of (GeS2)0.3(Sb2S3)0.7 of invariant kinetic parameters and demonstrated that the process can be treated as a single step. A more complex kinetic situation has been encountered by Thomas and Simon in re-crystallization of nickel sulfide from the a- to p-form. Their analysis yielded evidence of at least two steps involved in the overall process [l24]. The thermal behaviour of nanocrystalline CuS particles, obtained by mechanochemical synthesis by high-energy milling in an industrial mill, has

been studied by Godocikova et al. [125]. Thermal stability and structural properties were characterized by thermal analysis and XRD: these techniques revealed the formation of both copper sulfide CuS and copper sulfate CuS04-5H20,and that the thermal stability of the anhydrous CuS04 formed by the thermal decomposition is lower than the thermal stability of non-milled samples. The final product of the thermal decomposition is metallic copper instead of Cu20,which is stable up to 1100 OC. Because of the possible role as a catalyst solvent, Hamma and coworkers studied the physico-chemical properties of the binary system NaHS04-KHS04 by calorimetry and conductivity. The enthalpy of mixing has been measured at 505 K in the full composition range and the phase diagram calculated. The phase diagram has also been constructed from phase-transition temperatures obtained by conductivity for 10 different compositions and by DTA. The phase diagram (Figure 13) is of the simple eutectic type, where the eutectic is found to have the composition X(KHS04) = 0.44 with the melting point at 406 K [126].

Figure 13. Phase diagram of the NaHS04-KHS04 system based on conductivity measurements (circles) and thermal measurements (triangles). Open symbols indicate the solidus line (temperature of fusion of the eutectic). Reproduced from reference [I261 with permission from Elsevier.

5. METAL-ORGANIC FRAMEWORKS: COORDINATION POLYMERS 5.1. Introduction The construction of novel coordination polymers is a current interest in the field of supramolecular chemistry and crystal engineering, stemming from their potential applications as functional materials as well as their intriguing variety of architectures and molecular topologies. In particular, solid materials with either helical or chiral structures are of intense interest in chemistry and material science. Although solid materials with unusual structures are expected to increase in number, the exploration for preparing framework solids with chiral structures still remains a challenge. Thermal analysis is a fundamental tool for the characterization of these new coordination polymeric compounds.

5.2. Bismuth Mansfeld and coworkers [I271 described the synthesis and characterization of the one-to-one complexes of the bismuth trihalides BiX3 (X= C1, Br, I) with the chelating oxygen-coordinating-donor ligand (~~O-P~O)~(O)PCH~P(O)(O~S~-P The compounds were analyzed by single crystal XRD, IR, thermal analysis, NMR, ESI MS and conductivity measurements. The combination of these techniques gives a full description of the structures in the solid state and in solution, and provides information about the use of phosphonic ester complexes as precursors for the formation of bismuth phosphonates. New bioinorganic complexes of aspartic acid with bismuth triiodide were synthesized by a direct solid-solid reaction at room temperature. The formula of . 2.5H20. From the crystal the complex is Bi13[OOCCH2CH(NH2)C0]2.5 structure, the complex was shown to be a dimer with bridge structure and belongs to a triclinic system. The infrared spectra and the thermal analyses were usehl to demonstrate the complex formation between aspartic acid and the bismuth ion [128]. The same authors [I291 synthesized a new solid complex of nitrilotriacetic acid and bismuth trichloride by a solid-phase reaction at room The crystal structure of temperature, with the formula BiC13[N(CH2COOH)3]2.5. the complex belongs to the triclinic system The far-IR spectra showed the bonding between the Bi ion and N atom of nitrilotriacetic acid, while the thermal analysis demonstrated the complex formation between the bismuth ion and nitrilotriacetic acid. The gaseous pyrolysis products and the final residue of the thermal decomposition were determined. 5.3. Cadmium A 3-D diamondoid cadmium-naphtalenedicarboxylate framework has been structurally characterized using single-crystal XRD and thermal analyses [130]. Two cadmium coordination polymers, [Cd(is~nicotinate)~(H~O)].DMF and Cd3(isonicotinate)4(N03)2(4,4'-bipy)2(H20)2 have been synthesized and

characterized by Liao and coworkers. Both compounds adopt 3-D frameworks, built up by two sets of interpenetrated fragments. The TG study indicated that the first compound is porous and that its DMF and coordinated water molecules can be removed upon heating [13 11. The results of TG and of powder XRD reported by Liu et al. [132], showed that compound remains intact the framework rigidity of the [Cd(C12H6N204)-H20]n upon the removal of guest molecules, and maintains the thermal stability up to 440 "C. The second-row transition-metal ions are capable of engaging higher coordination modes because of their atomic sizes and intrinsic electron configurations. The authors underlined that the heptacoordinated cadmium centre plays an important role in the overall framework rigidity and high thermal stability of compound. TG, DTA and DSC curves of the cadmium peroxotitanate complex Cd2[Ti2(02)20(OH)6]-H@were recorded and used to determine the isothermal conditions suitable for obtaining the intermediate samples corresponding to the phases observed during the thermal decomposition. The experimental results were used to propose a mechanism of thermal decomposition of the investigated compound to CdTi03. The aim of this study has been to determine the optimum conditions for obtaining CdTi03 with well-defined crystallinity [133]. 5.4. Cobalt Eslami studied the solid-state thermal isomerization of [CO(NH~)~(ONO)]C~~ (nitrito isomer) to [ C O ( N H ~ ) ~ ( N O ~(nitro ) ] C ~isomer) ~ and the reverse reaction, using DSC [134]. The isomerization was shown to be essentially an equilibrium process and the interconversions are accelerated at above 65 "C, to reach the equilibrium state at about 155 "C. After establishment of the equilibrium, the relative amounts of the two isomers at any temperature are governed by a Gibbs energy relationship. H~O], Two Co(I1) coordination polymers with formulas [ C O ( O ~ ~ ) ( H ~ O ) ~ .and [Co(oda)(H20).H20], (H20da=oxydiacetic acid) were characterized by single H ~ aOcovalently ], linked 1-D chain crystal XRD and TG. [ C O ( O ~ ~ ) ( H ~ O ) ~ .has structure, while [Co(oda)(H20).H20], has a covalently linked 3-D chiral network with channels. The structures showed an unusual example of topological isomerism, and the structural interconversion between [ C O ( O ~ ~ ) ( H ~ O ) ~ .and H~O [Co(oda)(H20)-H20], ], revealed that self-assembly in the synthesis and interconversion of crystalline solids is a thermodynamically controlled process [1351. One-dimensional C0(dien)~(VO~)~-(H~0) was prepared by Lin et al. [I361 and characterized by single crystal XRD and thermal analysis. The structure is composed of infinite one-dimensional chains formed by corner-sharing V04 tetrahedra with co(diengf complex cations and crystallization water molecules

occupying the interchain positions, which are held together in a threedimensional network via extensive hydrogen-bonding interactions. The compound, with a new zig-zag conformation of metavanadate chains, is the first example of vanadium oxides incorporating trivalent transition-metal coordination groups. Nikolova and Popova [137] reported a new cobalt(I1) trihydrogen hexaoxoperiodate tetrahydrate, CoH3106.4H20,synthesized and characterized by quantitative analysis, TG, DTA, DSC and IR spectroscopy. Based on DTA and DSC data, the following thermal decomposition scheme has been proposed:

) ~ Co3O4. the intermediate phase being a mixture of C 0 ( 1 0 ~ and Liu and coworkers [I381 studied Co(NIA)2(H20)4 (NIA = nicotinate), a transition- metal organic-inorganic composite compound with a novel threedimensional supramolecular cavity structure with ladder-type hydrogen-bond chains. The formation of this special structure is attributed to it having two carboxylates and four water molecules, which are strong donor/acceptors in hydrogen bond interactions. Bakhmutova et al. [139] reported the synthesis and both the structural and thermoanalytical characterizations of unusual polymeric cobalt phosphonates containing a clathrated phosphonate anion and a layered bis-phosphonate. The study showed an unusual loss of coordinated water molecules at 90 OC, and a ) ~ residues of P205 proved by the further loss of mass to residue of C O ( P O ~with 1000 OC. The thermal properties of diastereomers of mixed cobalt(II1) complexes with aromatic amino acids and diamine were studied by Miodragovi and coworkers [I401 to obtain information about stereochemical effects on their thermal stability. The thermal decompositions of these complexes were shown to be multi-step degradation processes, some of which can be satisfactorily separated into individual steps, depending on the molecular symmetry. For diastereomers which crystallize with water molecules, preliminary dehydration occurs. The results of a thermoanalytical study of 4(5)-hydroxymethyl-5(4)methylimidazole complexes with divalent cobalt, of general formula CoL4(N03)2, compared with analogous nickel and copper complexes, were reported by Materazzi and coworkers [141]. The thermal stabilities and the decomposition steps were determined and the gaseous products of decomposition were analyzed by TG-FTIR to prove the proposed decomposition steps. The decomposition kinetics were examined using the Flynn-Wall-Ozawa and Kissinger methods. From the former method the E dependencies were presented and connected with the mass losses of the corresponding steps. The

Kissinger method was used to calculate the activation energies and the preexponential factors. 5.5. Copper

The rapid development in the crystal engineering of metal-organic coordination polymers has produced many novel materials with various structural features and properties. One of the most effective and attractive approaches for the assembly of polymeric frameworks is the incorporation of appropriate metal ions and multifbnctional bridging ligands. Xie and coworkers used 5nitroisophthalic acid (nip) as a bridging ligand, phenanthroline (phen) as a second ligand, with copper(I1) ions to assemble a novel coordination polymer [Cu(nip)(phen)], . The constant-volume combustion energy of the complex was determined by rotating bomb calorimetry and the standard combustion enthalpy and the standard formation enthalpy were calculated [142]. Ghosh and Bharadwaj [I431 described discrete hexameric water clusters with puckered-boat conformation acting as pillars between 2-D copper(I1) coordination polymers to form a stable 3-D metal-organic framework. Fujii et al. [I441 studied the complexation of copper(I1) ion with some amide solvents such as N-methylformamide (NMF), formamide (FA), N,Ndimethylacetamide (DMA) and N-methylacetamide (NMA) by titration calorimetry in acetonitrile containing (C2H5)4C104 as an ionic medium at 298 K. The reported results demonstrated that these amides coordinate to the metal ion to form a series of mononuclear complexes, and their formation constants, enthalpies and entropies have been obtained. Hetero-bimetallic coordination polymers of copper(I1) and manganese(I1) bridged by pyridinedicarboxylate ligands were synthesized and characterized by single-crystal XRD and thermal analysis to study the effect of thermal dehydration on the magnetic properties of these systems [145]. Marinho et al. [I461 reported the TG characterization of a 2-D coordination polymer involving monatomic carboxylate bridges, starting from a dinuclear paddle-wheel copper(I1) unit, with the copper ions involved in a CuN20207 chromophore. Zang and coworkers [147] reported the characterization of three homochiral helical coordination polymers of Cu(I1) with 2,2',3,3'-oxydiphthalic dianhydride. TG and powder XRD analyses showed that one of the porous frameworks is stable after the removal of solvent water molecules. In contrast, the others changed their structures to amorphous ones because of the simultaneous loss of solvent and coordination water molecules. Inorganic ion-exchangers with a layered structure, belonging to the class of acid phosphates of tetravalent elements, can exchange transition-metal ions and intercalate organic molecules between the layers to obtain complexes formed in

situ between the layers of the material. Vecchio and coworkers [148] studied the thermal dehydration and decomposition processes of some intercalation compounds by simultaneous TG-DSC and TG-FTIR. y-Zirconium and ytitanium phosphates were intercalated with 1,lO-phenanthroline and subsequently reacted with copper ions to form the complex in situ. Reaction mechanisms for thermal decomposition of all the materials were investigated and proposed according to the mass losses recorded by TG and confirmed by TG-FTIR. The dinuclear metal carboxylates M2(02CR)4represent an important class of transition-metal complexes with respect to the study of structure and metalmetal interactions. Several new chain compounds of dinuclear metal carboxylates have been isolated and some of them show interesting gasadsorption properties, giving a new aspect of chain complexes. Takamizawa and coworkers [149] presented evidence for a thermodynamic correlation between the phase transition and gas adsorption for a copper complex which was constructed by van der Waals interactions among chain-building blocks. Inoue et al. [I501 demonstrated that large amounts of carbon tetrachloride can be absorbed into 1-dimensional tunnels in copper(I1) trans- 1,4-cyclohexane dicarboxylate (Figure 14) under the saturated vapour pressure at room temperature, and the desorption can be performed easily by evacuation above room temperature. It was also confirmed that the absorptionldesorption is reversible. The thermodynamic and structural properties were studied for the empty (non-absorbed) sample and partially-filled (10, 22 and 31% of the full carbon tetrachloride-absorbed) samples, using adiabatic calorimetry between 13 and 300 K and by powder XRD with high-energy synchrotron radiation. The heat- capacity anomaly due to the first-order phase transition observed in the empty sample was not observed in the fully-absorbed sample. However, the partially absorbed samples showed smaller heat-capacity anomalies at lower temperatures than the empty sample. Such phenomena were compared with the previous results for toluene-absorbed samples [15 1- 1531 and the differences were discussed. Ukraintseva et al. 11541 studied the thermal dissociation processes for clathrates [ C U P ~ ~ ( N O ~ )(G= ~ ] . ~THF G and CHC13) by measuring the vapour pressure and mass loss as a function of temperature. The enthalpies of dissociation are 43 and 51 ~ m o l - 'for the THF and CHC13 clathrates, respectively. The activation energy of dissociation of the THF clathrate is 75 kJ mol-' by the ASTM E 698 method.

Figure 14. Structure of copper(I1) trans-1,4-cyclohexane dicarboxylate. Reproduced from reference [15 11 with permission from Elsevier.

A great number of Lewis-base stabilized copper(1) &diketonates have been prepared to date. They are mostly homonuclear hexafluoroacetylacetonate complexes, (hfac)cu(')~.Interest in these compounds was aroused by the fact that they form a pure copper film through the disproportionation: 2(hfac)Cu(I)L + cuO+ ~u(")(hfac)~ + 2L . To take advantage of these compounds for copper deposition, they should be volatile, thermally stable at room temperature and susceptible to thermally induced disproportionation at elevated temperatures. Krysiuk and coworkers reported that the main reason for the volatility and thermal stability of different compounds is due to the different symmetry, and hence low polarity, of their molecules, resulting in a relatively low lattice energy. Higher thermal stability is due to the stronger metal-ligand bonding of dimethylsilyl groups [155].

60

60

20 0

(a)

Temperature ('c)

(h)

100

200 300 Temperaturerc)

400

500

Figure 15. TG and DSC of compounds (a) [ c u 1 ( h f a c ) 1 2 ( ~and ~) (b) [CU' (hfac)12(DVTMSO)heated at 10 OC/min in a dynamic inert atmosphere. Reproduced fiom reference [155] with permission from Elsevier. Complexes with pyrazole-based ligands are a frequent subject of chemical investigations aimed at understanding the relationship between the structure and activity of the active site of metallo-proteins. The metal ion in biological systems is often coordinated to one or more imidazole groups, which are part of histidine fragments of the proteins. A thermoanalytical and structural study of several copper complexes with pyrazole substitutes has been reported [156]. Bajpai and Tiwari [I571 prepared a novel disodium salt of bisdithiocarbamate of urea (UBDT) and its Cu(I1) complex Cu(1I)UBDT. By elemental analysis, NMR and TG, the water-soluble UBDT was shown to possess a good chelating ability for various metal ions, while its copper (11) complex, Cu(II)UBDT, is an amorphous and intractable solid having a polymeric structure. The ligand and the complex were found to possess high thermal stability. UBDT cyclised on heating to yield a heterocyclic compound (probably highly stabilised by resonance) which also showed a coordinating tendency for various metal ions. The compounds, derived fiom L-threonic acid and various biological metal elements, facilitate the combination of the metal ions with amino acids or proteins and improve the efficiencies of the absorption and utilization of these metal ions in the human body. Qing and coworkers investigated the thermodynamic properties of copper L-threonate hydrate by adiabatic calorimetry and combustion calorimetry [ 1581. Reduction of ~ 0to s2~ by~ 2,5-pyridinedicarboxylate under hydrothermal conditions produced a new binuclear copper(I1) coordination polymer [CuS(4,4'-bipy)], (4,4'-bipy = 4,4'-bipyridine). Single-crystal XRD revealed that this compound consisted of sulfur-bridged binuclear copper(I1) units with Cu-Cu bonding which were combined with 4,4'-bipy to generate a 3-D network constructed from mutual interpenetration of two-dimensional (6,3) nets. The TG curve of the compound (shown in Figure 16) indicated three steps of mass loss, the first two of them corresponding to the release of the 4,4'-bipy ligands. The third mass loss was due to the release of part of the sulfur. After the

decomposition of the compound at high temperature, the residue corresponded to Cu2S [159].

0

100

200

300

400

500

600

700

Temperature (OC)

Figure 16. TG curve of the polymeric coordination compound [CuS(4,4'-bipy)],. Reproduced from reference [I591 with permission from Elsevier. Zelenak and coworkers [I601 prepared three copper acetate complex ~H O ~, C O O ) ~ ( and ~~)~ compounds, namely C U ( C H ~ C O O ) ~ ( ~ ~ ) ~C- ~UH( C C ~ ~ ( C H ~ C 0 0 ) ~ ( pwhere y z ) , en = ethylenediamine, tn = 1,3-diaminopropane, pyz = pyrazine. The thermal characterization during heating in argon by TG, DTA and MS showed multistep processes, involving dehydration, deamination and decarboxylation. During the heating of the pyrazine compound, the collapse of the polymeric structure took place rapidly in one step and the processes of deamination and decarboxylation overlapped. However, according to the MS data the deamination takes place first, while the decarboxylation is a slower process. New copper(1) coordination polymers were prepared by the reaction of copper(1) iodide [161] and chloride with 2-ethylpyrazine in water at room temperature or under solvothermal conditions. By thermal analysis (Figure 17) it was demonstrated that the product formation depends on the applied heating rate, which shows that the solid-state kinetics play an important role in such thermal reactions. In poly-[CuCl(k-2-ethylpyrazine-N,N')] (I), "zig-zag"-like CuCl chains are present, which are connected by the 2-ethylpyrazine ligand to a three-dimensional network. By comparison, in catena-[Cu3C13(p2-2ethylpyrazine-N,N')2] (11), six-membered Cu3C13 rings occur, which are connected to chains by the organic ligands. In poly-[Cu2C12(p2-2-ethylpyrazineN,N')] (111), CuCl double chains are found, which are linked by the ligands to form sheets. The thermal behaviour of the different compounds was investigated

using simultaneous TG-DTA-MS, as well as temperature-dependent powder XRD. Two mass-loss steps were found upon heating compound I in a thermobalance at 1 OC/min, where the first loss corresponds to the transformation into compound 111, and the second to the loss of the remaining ligands with the formation of CuC1. If the heating rate is increased to 16 OC/min, compound I1 is formed as an intermediate in a consecutive reaction [162].

4 5 i . I . I ~ I . I . I . I . I . I . I . I

I 75 100 125 150 175 200 225 250 275 300 325 75 100 125 150 175 200 225 250 Temperatur I "C Temperature I'C

50

Figure 17. DTA, TG, DTG and MS curves and heating-rate-dependent TG curves for poly[CuCI(p2-2-ethylpyrazine-N,N')] (I). Reproduced from reference [I621 with permission from Elsevier The factors affecting the general shape of the phase diagram and compound formation in the binary copper(1) halide-alkali-metal halide systems was reported by Wojakowska and Krzyzak [163]. A set of phase diagrams for the systems CuX-AlkX (where Alk = Li, Na, K, Rb or Cs and X = C1, Br or I) is given, basing on authors' investigation and selected literature data. Sizes and valences of ions involved in coulombic and polarization interactions were considered as main factors affecting the phase equilibria. An increasing tendency for compound formation was noted in the series Na+K+Rb+Cs, Na+Li and I-+Br-tCl. The most strongly represented were shown to be compounds of the formulas AlkCu2X3and Alk2CuX3. 5.6. Iron

Because of the relationship between the structure and thermolysis of metal(I1) and metal(II1) complexes with heterocyclic ligands, the study of the influence of

metal and ligand nature on the process of thermal decomposition are of interest. Mojumdar and coworkers [I641 studied the Fe(II1) complexes with nicotinamide, showing that the thermal decompositions of the complexes are multistage processes and produce FeO or Fe203 as the final solid products. Nicotinamide was coordinated to Fe(II1) through the nitrogen atom of its heterocyclic ring. The thermal decomposition of argentojarosite (AgFe3(S04)2(OH)6)has been studied by TG, spectroscopic and infrared emission techniques 11651. Frost and coworkers proved that the dehydroxylation occurs in three stages at 228, 383, 463 "C with the loss of 2, 3 and 1 hydroxyl units. The loss of sulfate occurs at 548 "C and is associated with a loss of oxygen. At 790 "C, the final loss of oxygen leaves only metallic silver and hematite. The characterization of the thermal decomposition of argentojarosite is of interest because it provides understanding of how silver production in ancient and medieval times was carried out. This work suggests that temperatures of around 750 "C are required to produce metallic silver. TG combined with MS has been used to study [166] the thermal decomposition of a synthetic ammonium jarosite ((NH4)Fe3(S04)2(0H)6).Five mass loss steps were observed at 120, 260, 389, 510 and 541 "C. MS confirmed these steps as loss of water, dehydroxylation, loss of ammonia and loss of sulfate in two steps. Changes in the molecular structure of the ammonium jarosite were followed by infrared emission spectroscopy (IES). This technique allows the infrared spectrum at the elevated temperatures to be obtained. IES confirmed the dehydroxylation to have taken place by 300 "C and the ammonia loss by 450 "C. Loss of the sulfate was observed by changes in band position and intensity after 500 "C.

5.7. Lanthanides Lanthanide ions have a high affinity for hard donor atoms, and ligands containing oxygen or oxygen-nitrogen atoms may be employed in the preparation of lanthanide polymeric complexes. Lanthanide ions present only little preference in bond direction, due to the inner positioning of their 4f valence orbitals. The 5-nitro-2-anthranilates of lanthanum(III), samarium(III), terbium(III), erbium(II1) and lutetium(II1) were obtained as hydrates having 2.5 mol of water molecules per 1 mol of compound [167]. The compounds are isostructural. The processes of dehydration and rehydration were investigated. The first step of dehydration does not cause a change of crystal structure. The entire dehydration gives anhydrous compounds with different structures to the structures of hydrates. The dehydration of the La, Sm, Tb and Er compounds was reversible and rehydration gives complexes having the same crystal structures as the initial compounds.

New lanthanide coordination polymer,^ with either salicylic acid and 8hydroxyquinoline [168], or pyridine-3,4-dicarboxylicacid and oxalic acid, were synthesized and characterized by elemental analysis, IR, TG, and single-crystal XRD. Chains were shown to be cross-linked by pyridine-3,4-dicarboxylic ligands into interesting two-dimensional framework structures [169]. Wu and coworkers studied, by a similar approach, three lanthanide(II1) coordination polymers, [{Lnz(bpdc)3(H20)3) ,HzO], (Ln = Sm, Eu, Tb), self-assembled from 2,2'-bipyridine-4,4'-dicarboxylic acid (H2bpdc) and the corresponding lanthanide(II1) salts under hydrothermal conditions. The compounds were found to be isomorphous and isostmctural. TG showed a high thermal stability (decomposition under N2 at T > 470 "C), indicating that the coordination habit of the metal ions with the bpdc ligand has a profound effect on the overall rigidity of the framework and the thermal stability of the compound [170]. Two series of lanthanide benzenedicarboxylates of the general framework formula [M2(2,2'bi~y)~(C~H404)~] (M = Y, Gd, and Dy) have been prepared employing hydrothermal methods, and their structures were determined by single-crystal XRD. TG, carried out in oxygen atmosphere, indicated the loss of the lattice water, followed by the loss of 2,2'-bipy and isophthalate groups. All of the calcined samples were found to be crystalline, and the powder XRD lines match well with the corresponding pure oxides [171]. Rare earth elements complexes with 1,3,5-benzenetricarboxylate ion were prepared as solids of the general formula Ln(C9H306)-nH20,where n=6 for La-Dy and n=4 for Ho-Lu,Y. Hydrated 1,3,5-benzenetricarboxylates lose water molecules during heating in one step (La-Tb), two steps (Y, Ho-Tm) or three steps (Dy, Yb, Lu). The anhydrous complexes are stable up to 573-742 K and decompose to oxides (CeLu) at higher temperatures [172]. Complexes of yttrium(II1) and lanthanides(II1) with 1,2,4,5-benzenetetracarboxylic acid were prepared as crystalline solids, On heating in insoluble in water, of the general formula Ln4(C10H208)3.14H20. air or inert gas atmosphere, all compounds lose water molecules; the remaining anhydrous compounds decompose to oxides. The yttrium and the heavy lanthanide (from Ho to Lu) complexes crystallize in the monoclinic crystal system and dehydration does not change the crystal structure of the compounds [173]. Complexes of heavy lanthanides(II1) (Gd-Lu) and Y(II1) with 4-chlorophthalic acid were prepared and their IR spectra, solubility in water at 295 K, and thermal decomposition were investigated by Kurpiel-Gorgol and Brzyska [174]. When heated, the complexes with general formula Ln2[C1C6H3(C02)2]3-nH20 (where n=6 for Tb, Dy(III), n=4 for Gd, Ho and Er(III), n=2 for Tm-Lu(II1) and n=3 for Y(II1)) are reported to decompose to the oxides Ln203, Tb407 with intermediate formation of oxochlorides LnOC1.

Ferenc and Bocian [175] studied the thermal properties of 5-chloro-2methoxybenzoates of lanthanides(II1) and divalent transition metal ions, in both air and nitrogen atmospheres. The complexes were obtained as mono-, di-, tetraand pentahydrates with a metal-to-ligand ratio of 1:3 (in the case of lanthanides(II1)) and 1:2 (in the case of d-block elements). The compounds showed colours typical for ~ n and ~ M+~ ions + and all of them are polycrystalline compounds. When heated, they dehydrate to form anhydrous salts, which in air are decomposed to the oxides of the respective metals, while in nitrogen to the mixtures of metal oxides, oxychlorides and carbon. Zhang and coworkers [I761 characterized three new lanthanide coordination polymers, prepared by the reaction with 1,2-bis(4-pyridy1)ethane-N,N'-dioxide or with mixed N-oxide ligands in different molar ratios. TG revealed that the ligands are bound differently due to their distinctive coordination modes. New lanthanide(II1) complexes of N,N' bis(2-hydroxyethy1)glycinate were studied by Messimeri et al. [177]. The thermal decompositions of the complexes were determined using TGJDTG and DSC techniques under nitrogen. The complexes release all the water content in two endothermic steps, but stable hydrated intermediates cannot be formed, and clear plateaux are not reached after complete dehydration, because the multi-step decomposition of the anhydrous species starts immediately. A lanthanum citrate trihydrate was synthesized as precursor material for an aqueous solution-gel route to La-containing multi-metal oxides. By means of TG and powder XRD, it was shown that the carboxylate groups of the citrate are coordinated to ~ a in~ monodentate, + bidentate and bridging ways. Also the alkoxide group, which carries the proton, is coordinated to ~ a ~Two + . water molecules complete the coordination sphere, while the third one can be found between the ~a~+-citrate network [178]. Badea and coworkers [I791 reported an investigation on the thermal stability of a series of new complexes with azo- and azomethinic-chromophores of the type [Er(HL)2(H20)2](HO)(where H2L is o,o'-dihydroxy-azobenzene,N-(2-hydroxy1-naphthalidene)aminophenol or N-(2-hydroxy- 1-naphthalidene)anthranilic acid). The thermal behaviour steps of complexes were compared with those of the corresponding ligands. The thermal transformations are complex processes according to TG and DTG curves, including phenol elimination, oxidative condensation and thermolysis processes. The final product of decomposition of all the complexes was Er203. The reactions of LnC13 (Ln = lanthanide cations), 1-(S,S) or d-(R,R)-tartaric acid and molybdate in acidified aqueous solutions gave rise to enantio-pure leftor right-handed double helical coordination metal compounds. TG and powder XRD studies suggested that the backbone collapses with the removal of aqua ligands. However, it reverts easily to the original compound after being

immersed in water, as confirmed by similar XRD patterns. The conductivity studies for these compounds revealed that they are semiconductors. Study of the magnetic susceptibilities revealed that the magnetic behaviours for MoGd, MoDy, MOHOand MoYb obey the Curie-Weiss law [180]. The synthesis, characterization and a TG-DSC study of gadolinium and lutetium methanesulfonate coordination compounds with pyridine-N-oxide and 2-, 3- and 4-picoline-N-oxides, reported by De Moura et al. [181,1821, showed that the thermal stability trend is: 2-picNO < py-NO < 4-picNO < 3-picNO for Gd and 2-picNO < 4-picNO < 3-picNO < py-NO for Lu, while the endothermic release of ligand molecules is followed by the thermal exothermic degradation of the ligands. A complex of holmium perchlorate coordinated with 1-glutamic acid was prepared with a purity of 98.96%. The compound was characterized by chemical, elemental and thermal analysis. Molar heat capacities of the compound were determined by a high-precision automated adiabatic calorimeter fkom 78 to 370 K. The dehydration temperature, dehydration enthalpy, entropy and the standard enthalpy of formation were reported [183]. A Ho(PDC),(o-phen) complex has been obtained and studied by Xie et al. [I841 from the reaction of hydrated holmium chloride, ammonium pyrrolidinedithiocarbamate (APDC) and 1,lO-phenanthroline (0-phen.H20) in absolute ethanol. The enthalpy of complex formation from a solution of the reagents and the molar heat capacity of the complex were determined by using a heat conduction microcalorimeter. The enthalpy of complex formation from the reaction of the reagents in the solid phase, was calculated on the basis of an appropriate thermochemical cycle and other auxiliary thermodynamic data. The thermodynamics of formation of the complex were investigated by the reaction in solution at the temperature range of 292.15 - 30 1.15 K. The constant-volume combustion energy of the complex, A,U, was also determined. Dan and coworkers synthesized and characterized two complexes of general formula [Ln(Ala)2(Im)(H20)](C104)3(Ln = Pr, Gd; Ala = alanine; Im = imidazoles). By using a solution-reaction isoperibol calorimeter, standard enthalpies of reaction at 298.15 K, were determined. Standard enthalpies of formation of the two complexes at 298.15 K were also calculated [185]. Balboul studied the reaction between lanthanum oxide and strontium carbonate non-isothermally between 350 and 1150 "C at different heating rates, and the intermediates and the final solid product were characterized by XRD. The reaction is reported to proceed through formation of lanthanum oxycarbonate La20(C03)2, lanthanum dioxycarbonate La202C03, and non-stoichiometric strontium lanthanum oxide La2Sr0, (x=4+6). La4Sr07 was found to be the final product which begins to form at about 700 OC. ~i+-dopingenhances the formation of the final product as well as the commencement of the reactions at

lower temperatures [186]. ~ a " - and ~b"-based oxynitride perovskites of the AB02N type (A = Ca, Sr, Ba) were synthesized by ammonolysis of complex oxide precursors. These precursors were either crystalline perovskites or amorphous xerogels prepared by solid-solid reaction and by soft chemistry methods, respectively. Their thermal stability was investigated by TG-MS. Oxidation studies revealed an intermediate product that gives rise to a characteristic mass gain in the TG curve. This intermediate was found for all the examined oxynitrides in oxidizing atmosphere. MS confirmed molecular nitrogen evolution indicating that N2 is retained during the oxidation reaction. At higher temperatures (800-1000 OC) the di-nitrogen is released, leading to the formation of the corresponding oxides [187]. The thermal transformation from lanthanum hydroxide (La(OH)3)to lanthanum oxide (La203)was studied by Neumann and Walter [188]. They reported two successive endothermic effects, caused by loss of water. Thermal analysis (DTA-TG, DSC), and high-temperature powder XRD were used to characterize this process. Lanthanum hydroxide oxide (LaOOH) was obtained as a temporary product at about 330 OC. The structure of lanthanum hydroxide oxide was characterized by powder XRD and subsequent Rietveld refinement. The enthalpies of dehydration were calculated by DSC to be 87 kJ mol-' for the transformation La(OH)3 to LaOOH, and 54 kJ mol-' for the transformation LaOOH to La203. The activation energy of the transformation of lanthanum hydroxide to lanthanum hydroxide oxide was estimated by isothermal TG studies. 5.8. Lead

Singh and coworkers synthesized lead(I1) complexes of reduced glutathione (GSH) of general composition [Pb(L)(X)].H20 (where L=GSH; X=Cl, NO3, CH3CO0, NCS). TG and DTA studies indicated the coordination of cysteinyl sulfur with metal ion and the presence of water molecules in the complexes. The thermal behaviour of the complexes shows that the water molecule is removed in the first step, followed by the removal of anions and then by the decomposition of the ligand in the subsequent steps. The thermal decomposition of all the complexes proceeds via first-order kinetics [189]. DSC and TG were used for quality control of the materials and the technological processes during battery manufacture. Matrakova and Pavlov presented the results of an investigation on lead-acid battery pastes and active materials, aimed to estimate the efficiency of the two thermal methods for the analysis and the control of the processes taking place during battery production and operation [190].

5.9. Lithium The recent development and commercialization of high energy-density, rechargeable Li-ion batteries is one of the most important successes in modern electrochemistry. These battery systems are conquering the market, and their range of application continues to expand. A huge amount of scientific work has been devoted to this field by thousands of research groups throughout the world, because of the complexity of these systems, whose operation involves highly complicated Li intercalation reactions, interfacial charge-transfer reactions, passivation and corrosion phenomena, and simultaneous surface and bulk side reactions. At present, all the commercial Li-ion batteries comprise electrolyte solutions based on a LiPF6 salt and alkyl carbonate solvents. In recent years, calorimetric tools such as DSC and accelerating rate calorimetry (ARC) have been widely used in this field to study the thermal stability of Li batteries and their components [191,192]. The crucial role of LiPF6 in the determination of the thermal behaviour of these systems has been clearly demonstrated. The thermal behaviour of LiPF6 itself is thus interesting and important and deserves special attention. The thermal stability of LiPF6 has been studied in recent years using various methods and approaches [193-1971, including calorimetric measurements at constant volume [194, 196, 1971 constant pressure [193, 1941, dynamic or isothermal conditions, DSC with and without gas removal, in closed or open crucibles, and TG under nitrogen flow [193]. LiPF6 is not stable at elevated temperatures and decomposes to LiF and PF5, exhibiting one [I941 or two [I961 endothermic peaks in DSC at constant volume, and more that two peaks [I941 at normal pressure. Because the thermal decomposition of LiPF6 is accompanied by the development of PF5 gas, the onset of these endothermic processes depends on the test conditions. In an open crucible under nitrogen flow at atmospheric pressure, the isothermal TG measurements showed a mass loss related to only one process at temperatures as low as 343 K [193]. Zinigrad and coworkers [I981 presented the results of a comprehensive study of the thermal stability of the salt LiPF6, using both accelerating rate calorimetry (ARC) and DSC. Pressure monitoring during ARC experiments also permitted the study of the endothermic processes. In a confined volume, LiPF6(s) melts reversibly at 467 K. Reversible decomposition to PF5(g) and LiF(s) starts with melting, but the autogenic development of PF5(g) pressure makes the temperature profile of decomposition a function of volume and sample size. The heat of this reaction at constant volume, as determined by a variety of methods, is in the range 60*5 kJmol-1, and is approximately temperature independent in range 490-580 K.

5.10. Magnesium Srinivasan and Sawant investigated, by thermal and spectroscopic methods, the Mg(I1) complexes of nitro-substituted benzoic acids. TG, DTA, DSC and spectroscopic methods were used to investigate the thermal behaviour and structure of the compounds [Mg(H20)6](4-nba)2.2H20, [Mg(H20)6](3nba)2-2H20,[Mg(H20)6](2-nba)2.2H20,with 4-nba = para-nitrobenzoate, 3-nba = meta-nitrobenzoate, and 2-nba = ortho-nitrobenzoate. All the complexes are formulated as consisting of the octahedral [ M ~ ( H ~ o ) ~cation ] ~ + with the carboxylates outside the coordination sphere. The thermal decompositions of these carboxylates are multi-stage processes, and, while heating, the compounds first lose the water molecules, followed by the loss of the nitrobenzoate, resulting in the formation of oxide. The [Mg(H20)6](4-nba)2-2H20 complex can be reversibly hydrated. The final product of thermal decomposition of all of the complexes is MgO 11991. The development of volatile compounds of alkaline-earth metals has attracted attention because of the need for these compounds as precursors for the preparation of thin films by chemical gas phase methods. Hatanpaa and coworkers [200] characterized the effect of ancillary ligands on the structural and thermal properties of several [Mg(thd)2(A)] complexes, in which A is a neutral Lewis-base ligand. They showed that the evaporation processes of diamine adducts contain two overlapping steps, the first step associated with the evaporation of amine and the second with the evaporation of [Mg2(thd)4]dimer. All the complexes containing mines evaporated almost completely, but the complex which contained 1,2-ethanediol, was thermally unstable and decomposed when heated. At temperatures below the dissociation temperature, all adducts of diamines appeared to evaporate intact. Emanation thermal analysis (ETA) was used by Stanimirova and coworkers [201] to characterize the microstructural changes during heating of Mg-A1-C03 layered double hydroxide (LDH) in the temperature range 293-1473 K. ETA confirmed that the formation of an intermediate phase, with grafted ~ 0 anions in the hydroxide layers, took place in the temperature range 508-523 K and the formation of Mg-A1 mixed oxide (MO) occurred in the range 623-773 K (see Figure !8). The small peak of the emanation rate at 603 K indicated the degradation of the layered structure and the broad peak in the range 1073-1273 K characterized the onset of the separation of the decomposition products of MO into MgO and Mg2A1407.The ETA results revealed that dehydration of the ~ occurred at lower temperatures than that of product with grafted ~ 0 3 anions the initial Mg-A1-C03 LDH.

~

~

TcmpernturdK

TernpcraturcK

Figure 18. (left). Comparison of the results of ETA-DTA, TGDTG and MS measurements obtained during heating of the initial Mg-AI-C03 LDH in range 293-1473 K: a - results of ETA (bold line) and DTA (slender line) measurements, b - results of TG (full line) and DTG (dashed line) measurements, c - results of MS - release of H 2 0 (full line) and C02 (dotted line) (&) Comparison of the ETA-DTA, TGDTG and MS results obtained during heating (293-773 K) of the initial sample (slender lines) and the sample obtained after the re-hydration of the LDH heated at 523 K (bold lines); a results of ETA and DTA measurements, b - results of TG (full lines) and DTG (dashed lines) measurements, c - results of MS - release of H 2 0 (full line) and COz (dotted line). Reproduced from reference [201] with permission fiom Springer.

5.11. Manganese Coordination polymers are of special interest for nanotechnology because individual chains of these polymers could be used as molecular wires, with a number of advantages over other materials, such as carbon nanotubes, due to their easier synthesis, or higher reactivity. It is possible to modify the bulk magnetic, electronic, and optical properties of these polymers by tailoring the

molecules. Thermal analysis and other techniques were employed to characterize a [Mn@-0~)(4atr)~], (ox = oxalato and 4atr = 4-amine-1,2,4triazole) complex and the procedures employed to obtain single chains of this coordination polymer opened a route for future nanotechnological applications of these types of materials [202]. Manganese is also one of the trace elements which plays an important role in some biological systems. It is essential for the oxidation of water to O2 in photosynthetic processes and participates in the disproportionation of hydrogen peroxide (catalase activity) in microorganisms and in the formation of various metallo-enzymes such as superoxide dismutase, pyruvate carboxylase, arginase and enolase. These facts have contributed to an increase of interest in the coordination chemistry of manganese, especially complexes with carboxylato ligands. Rzaczynska and coworkers synthesized the manganese(I1) complex [Mn2(C6H604)2(H20)3]. Its repeating dimeric unit consists of Mn(I1) atoms bridged by oxygens from two carboxylate ligands and one water molecule, and the carboxylate ligands occur as tridentate-bridging and monodentate. The thermal characterization was performed by TG, DTA and TG-FTIR, showing that the compound dehydrates at 403 K and then decomposes at 500 K, first to Mn203which transforms into Mn304at 1183 K. The thermal decomposition is connected with release of water (405 K), carbon dioxide (470 K) and hydrocarbons (595 K) [203]. A two-dimensional organic-inorganic polymer Mn2(H20)[02C(CH2)4C02]2was synthesized as single crystals by the hydrothermal reaction of MnC12 with adipic acid in the presence of base [204]. The TG study in nitrogen atmosphere showed two distinct mass-loss regions around 210 and 260 "C. The first loss was due to dehydration of coordinate H 2 0 corresponding to 1.1 equiv of H20 per formula unit. Interestingly, water diffuses irreversibly out of the compound and repeated thermal analyses confirmed that rehydration under ambient conditions does not occur, even in a week. The crystallinity becomes poor after dehydration at 250 OC, which was confirmed by powder XRD. The second mass loss (260 "C) corresponds to the loss of coordinated adipate, giving MnO and carbon. 5.12. Nickel Photodynamic therapy (PDT) is a treatment that is used for the destruction of certain types of tumours, and is based on the administration of a photosensitizer that concentrates in tumour cells and, upon subsequent irradiation with visible light in the presence of oxygen, selectively destroys the cancerous cells. Wei et al. reported the thermal stability, melting temperature, enthalpy of fusion and thermal decomposition kinetics of porphyrins and their nickel complexes [205]. The neutral three-dimensional nickel coordination polymer with 4,4'-bipyridine and 1,2,4,5-benzenetetracarboxylatewas prepared by Wu and coworkers [206]

from hydrothermal reaction of nickel(I1) chloride with mixed ligands in basified aqueous solution and characterized. The TG and powder XRD analyses suggested that the crystallization and the coordination of water molecules play important roles in the formation and stabilization of the polymer and the coordination polymeric compound has potential applications. Single crystals of the tris-(chelated)nickel(II) complex have been prepared and structurally characterized by XRD and thermal analysis. The structure analysis revealed that the complex has an approximate Dg symmetry, while the TG results showed that the water and chloride molecules were removed at 231 "C leading to the compound [Ni(C25H19N5)](s) [207]. The phenomenological, kinetic and mechanistic aspects of the nitrate, chloride, bromide and iodide complexes of nickel(I1) withl,2-(diimino-4'-antipyriny1)ethane (GA) have been studied by TG and DTG techniques. The kinetic parameters like activation energy, preexponential factor and entropy of activation were computed, and the ratecontrolling process in all stages of decomposition was shown to be random nucleation with one nucleus on each particle (Mampel model) [208]. Prediction of the thermal decomposition pathway of metal complexes is very important from the theoretical and experimental point-of-view to determine the properties and structural differences of complexes. In the prediction of the decomposition pathways of complexes, besides the thermal analysis techniques, some ancillary techniques, e.g. MS, have also been used in recent years. From the molecular structures and fragmentation components, it is believed that the thermal decomposition pathway of most molecules is similar to the ionization mechanism occurring during the ionization process in the mass spectrometer. The thermal decomposition pathway of the diaquabis (N,N'-dimethyl-1,2ethanediamine)-Ni(I1)acesulfamate complex has been predicted by Icbudak and coworkers [209] with the help of TG, DTG, DTA and MS fragmentation patterns. The complex decomposed in four stages: a) dehydration between 84132 "C, b) loss of N,N'-dimethylethylenediamine (dmen) ligand, c) decomposition of the remaining dmen and acesulfamato (acs) by releasing SO2, d) burning of the organic residue resulting in NiO. The volatile products observed in the thermal decomposition process were also observed in the mass spectrometer ionization process, except for the molecular peak, and it was concluded that the ionization and the thermal decomposition pathways of the complex resembles each other. various aFourteen chelates of the type [Ni(1I)(Dio~.H)~l,((D~OX.H)~: dioximes) have been studied by means of FTIR, NMR, MS data and various thermoanalytical methods (TG, DTA, DTG, DSC) [210]. In some cases kinetic parameters of the thermal decomposition of the complexes were also calculated using Zsak6's 'nomogram method'.

5.13. Palladium Cyclo-palladated chemistry has undergone important advances mainly due to the potential uses of compounds as technologically relevant materials (e.g. liquid crystals), catalysts, and anti-tumour drugs. De Almeida and coworkers reported a study on the thermal properties of cyclo-metallated compounds of general formula [Pd2(dmba)2X2(bpe)],where X = NO^-, C1-, N ~ ,-NCO-, NCS-, dmba = N,N-dimethylbenzylamine and bpe = trans- l,2-bis(4-pyridy1)ethylene [2 111. The thermal dehydration and decomposition processes of a Pd(I1) coordination compound, [PdL4]C12.3H20 (L = 1-allylimidazole) were studied by simultaneous TG-DSC techniques under constant heating-rate conditions. The gaseous products were analyzed by TG-FTIR, which confirmed that only two ligand molecules were released and that a new 1-allylimidazole Pd(I1) complex, trans-[PdL2C12], was obtained. The same compound was also prepared by heating [PdL4]Cl2-3H20at 413.15 K in an air atmosphere until a constant mass was reached. The thermal decomposition mechanisms for the complexes were proposed according to the three mass loss steps derived by TG [212]. 5.14. Silver The study of the self-assembling process of metal complexes continues to be a theme of considerable current interest in the context of developing new solidstate polymeric materials with specific architectural and functional features. Silver(1) ions are regarded as extremely soft acids favouring coordination to soft bases, such as ligands containing S and unsaturated N. Silver(1) complexes with these soft ligands give rise to an interesting array of stereochemistries and geometric configurations, with coordination numbers of two to six all occurring. Silver(1) is often used, therefore, to construct network structures. Suenaga et al. studied silver(1) complexes of hexakis(tolylsulfany1)benzene (htsb). In these compounds, the silver ion prefers a square-planar coordination geometry, comprised of four S atoms from two different htsb molecules, and producing a zigzag chain structure of Ag-S in the silver coordination polymer. Based on the TG results, two tolylsulfanyl groups were easily eliminated at approximately 21 1 "C. However, the complexes obtained by the reactions in different solvents, showed different colours and thermal degradation behaviours [2 131.

o1

50

I

I

I

200

350

500

J

650

Temperature/ F

Figure 19.Zig-zag chain structure and the TG curve of silver(1) complexes of hexakis(tolylsulfanyl)benzene. Reproduced from reference [2 131 with permission from Elsevier. Paramonov and coworkers synthesized silver(1) carboxylate complexes by the reaction of silver(1) carboxylate with the neutral ligand in absolute ether or ethanol. By thermal analysis and XRD, the relations between volatility and crystal structures of the complexes were discussed [214]. Mercer et al. prepared a series of silver(1) coordination networks based upon non-chelating bidentate thioether ligands. Frameworks using AgOTs as the silver(1) starting material form two-dimensional frameworks and are quite stable as shown by DSC and TG data. The networks are sufficiently robust to maintain the same layered motif when the basic skeleton of the ligand is sequentially derivatized with -OEt, -OBu, and -0Hex groups [2 151. Cagran and coworkers reported an investigation carried out with a fast pulseheating technique on silver and the Ag-28Cu binary alloy in the solid and the molten states. Properties like enthalpy or electrical resistivity of a pulse-heated sample can be obtained for a wide temperature range (from the solid state up into the liquid state) from the directly measured base quantities, namely: current through the sample, voltage drop across the specimen and the pyrometrically determined temperature. As a further result, the enthalpy of fusion is computable from the enthalpy values at the melting transition or the solidus/liquidus transition. These therrnophysical properties (mainly of the melting transition and the subsequent liquid phase) (see Chapter 9 of this Volume) are commonly used as input data for numerical casting simulations. The measurements presented within this work deal with Group-11 elements silver, copper and the binary eutectic 72-28 (mass%) alloy of the two elements, respectively. One of the main goals of this work was to investigate to what extent the thermophysical properties of the two pure materials influence or determine the properties of its

corresponding alloy and whether data for pure materials can be used to predict the thermophysical properties of simple alloys. For this specific copper-silver alloy, the authors showed that there is a certain mutual solid solubility with a quite large miscibility gap: Ag-28Cu is reported to be not a single-phase alloy but an eutectic alloy with two phases [2 161. 5.15. Sodium The thermal decomposition mechanism of a new coordination polymer of sodium trinitrophloroglucinate was studied using DSC, TG/DTG and FTIR techniques. Under a nitrogen atmosphere, with a heating rate of 10 "Clmin, the thermal decomposition of the complex contained one endothermic and five exothermic processes. Two intense exothermic decomposition processes were observed in the range 173-228 OC, suggesting its energetic nature, and the solid decomposition residue at 500 OC was sodium isonitrile. A test of its explosive properties revealed that the compound is sensitive to mechanical stimuli and can act as a component of ecologically-clean initiating compositions [2 171. The influences of the product gases on the kinetics of the thermal decomposition of sodium hydrogen carbonate, NaHC03, were investigated by means of controlled rate evolved gas analysis coupled with TG (CREGA-TG) [218]. From a series of CREGA-TG measurements carried out under controlled gaseous concentrations of C 0 2 and H20 by considering the self-generated C 0 2 and H 2 0 during the course of reaction, anomalous effects of C 0 2 and H 2 0 on the kinetic rate behaviour of the thermal decomposition of NaHC03 were revealed. The reaction is decelerated by atmospheric C 0 2 and accelerated by H20. Heat capacity measurements of sodium alkoxides (methoxide, ethoxide, npropoxide and iso-propoxide) were carried out using DSC in the temperature range 240-550 K by Chandran and coworkers [219]. From the heat capacity values, other thermodynamic functions, such as enthalpy increments, entropies and Gibbs energy functions of these compounds were derived. The Cp,m298 values of sodium methoxide, sodium ethoxide, sodium n-propoxide and sodium iso-propoxide were measured and reported. 5.16. Strontium A novel coordination network of D,L-homocysteic acid with strontium chloride was reported by Liu and coworkers [220]. This compound exhibited an infinite microporous multilayered structure, where chloride anions are intercalated between layers to neutralize the charge. TG and powder XRD showed that the layered structure of the compound, below 326 OC, is sustained in the process of the reversible losslgain of coordinated water, confirming that the network involves a robust coordination-based cationic layer framework but rather

flexible interlayer interactions. This compound and its analogues were suggested to have potential applications in anion exchange and gas storage. Kazak et al. [221] proposed an eight-coordinate strontium(I1) complex (bistriethanolamine-strontium(I1) saccharinate). The anions and cations are linked by the intermolecular hydrogen bonds between the hydroxyl hydrogens of the tea ligands and the amine-N, and carbonyl-0 atoms of the neighboring sac ions, forming one-dimensional chains running parallel to c. The adjacent chains are held together by van der Waals interactions, creating a three-dimensional network. 5.17. Zinc

Nanoporous materials with channels and cavities of molecular dimensions have a number of potential applications, for example, in catalysis, ion-exchange, gas adsorption, etc. Of the mine-templated transition-metal phosphates, the zinc phosphates have the richest structural chemistry and show a variety of materials with different structural dimensionality. Jensen and coworkers [222] prepared amine-templated zinc phosphates that are an interesting example of the correlation between the temperature for the synthesis and the density/structural dimensionality of the synthetic products that may be obtained. Zinc complexes of pyruvic and glyoxylic acid oximes have been investigated [223] as potential precursors for single source chemical vapour deposition (SSCVD) of ZnO thin films. The structural differences, in conjunction with ligand substituent changes give different responses to TG curves for their sublimation. The glyoxylic acid methyloxime complex, Zn(02CCHN(OMe))2 2H20, has volatility and thermolysis properties suitable for SSCVD of ZnO. Zelenak et al. [224] used TG to confirm the removal of the surfactant and to determine the load of two bactericidal zinc(I1) complexes in synthesized mesoporous silica used as a host system. By the reaction of an aqueous solution of Zn(N03)2.6H20 with the tetraethylammonium salt of the compound obtained by the condensation of tris(2-ch1oroethyl)amine with 4-hydroxyethylbenzoate, a porous coordination polymeric structure with infinite interlinked chains of Zn(I1) metallocycles was obtained and characterized by Neogi and Bharadway [225]. TG and XRD showed that an infinite water chain passes through the metallocycles like a thread. Migdal-Mikuli and Szostak [226] detected four solid phases of hexadimethylsulphoxidezinc(II) chlorate(VI1) [Zn(DMS0)6](C104)2 by DSC. Specifically, phase transitions were detected between the metastable phase KII t* the supercooled phase KO, between the stable phase KIb t* the stable phase KIa, between the stable phase KIa t* stable phase KO. At Tm2= 389 K, the

crystals partially melted, while at T,, = 465 K the melting is complete. From values of the entropy changes, it was concluded that the phases KO and KO' are the orientationally dynamically disordered phases, so called ODDIC crystals, and phases KIa, KIb and metastable KII are dynamically ordered but with some degree of positional disorder. Twelve zinc(I1) complexes with thiosemicarbazone and semicarbazone ligands were prepared and characterized by TG-DTA and H and C NMR spectroscopy [227]. Their antimicrobial activities were evaluated by MIC against four bacteria and two yeasts. The authors showed that the properties of the ligands, such as the ability to form hydrogen bonding with a counter anion or hydrated water molecules, or the lesser bulkiness of the N moiety, would be a more important factor for antimicrobial activities than the coordination number of the metal ion for the zinc(I1) complexes. Low-temperature heat capacities of the solid coordination compounds Zn(Leu)S04-1/2H20e, and Zn(His)S04.1/2H20(,, (Leu = Leucine and His = Histidine) were measured by a precision automated adiabatic calorimeter over the temperature range between T = 78 K and T = 371 K. Di and coworkers [228,229] determined the initial dehydration temperature of the coordination compounds by analysis of the heat-capacity curve. The experimental values of molar heat capacities were fitted to a polynomial equation with the reduced temperatures (x), [x = f (T)], by a least-squares method. Enthalpies of dissolution of both the complexes were determined by isoperibolic solutionreaction calorimetry. New complex compounds of general formula Zn(4-C1C6H3-2(OH)C00)2.L2.nH20 (where L=thiourea (tu), nicotinamide (nam), caffeine (caf), n=2,3), were prepared and characterized by Gyoryova and coworkers [230], who studied their thermal properties by TGJDTG and DTA methods. Thermal decomposition of the hydrated compounds starts with the release of water molecules. During the thermal decomposition of the anhydrous compounds, the release of organic ligands take place, followed by the decomposition of the salicylate anion. Zinc oxide was the final solid product of the thermal decomposition performed up to 650 "C. TG, powder XRD, IR spectra and chemical analysis were used for the determination of the products of the thermal decomposition.

6. REFERENCES

S. Vyazovkin, Anal. Chem., 74 (2002) 2749. S. Vyazovkin, Anal. Chem., 76 (2004) 3299. S. Vyazovkin, Anal. Chem., 78 (2006) 3875. S. Materazzi and R. Curini, Appl. Spectr. Rev., 36 (2001) 1. S. Materazzi and R. Curini, Appl. Spectr. Rev., 36 (2001) 169. S. Materazzi, A. Gentili and R. Curini, Talanta, 68 (2006) 489. S. Materazzi, A. Gentili and R. Curini, Talanta, 69 (2006) 78 1. S.L. Simon, Thermochim. Acta, 374 (2001) 55. M.J. Starink, Int. Mater. Rev., 49 (2004) 191. H.J. Seifert, J. Therm. Anal. Cal., 67 (2002) 789. H.J. Seifert, J. Therm. Anal. Cal., 83 (2006) 479. G.M.B. Parkes, G. Bond, P.A. Barnes and E.L. Charsley, Rev. Sci. Instrum., 71 (2001) 168. E.A. Olson, M.Y. Efremov, A.T. Kwan, S. Lai, V. Petrova, F. Skiettakatte, J.T. Warren, M. Zhang and L.H. Allen, Appl. Phys. Lett., 77 (2000) 267 1. K. Tozaki, R. Masuda, S. Matsuda, C. Tokitorno, H. Hayashi, H. Inaba, Y. Yoshimura and T. Kimura, J. Therm. Anal. Calorim., 64 (2001) 331. H. Hayashi, C. Nonaka, K. Tozaki, H. Inaba and C. Uyedab, Thermochim. Acta, 43 1 (2005) 200. M. Bastos, G. Bai, E. Qvarnstrom and I. Wadso, Thermochim. Acta, 405 (2003) 2 1.

J.G.P. Binner, D.M. Price, M. Reading and B. Vaidhyanathan, Rev. Sci. Instrum. 76 (2005) 064902. E.A. Olson, M.Y. Efremov, M. Zhang, Z. Zhang and L.H. Allen, J. Appl. Phys., 97 (2005) 034304. F. Paulik, E. Bessenyey-Paulik and K. Walther-Paulik, Thermochim. Acta, 438 (2005) 76. G. Vanden Poel and V.B.F. Mathot, Thermochim. Acta, 446 (2006) 41. F. Eigenmann, M. Maciejewski and A. Baiker, Thermochim. Acta, 440 (2006) 8 1.

A. Toda, T. Arita and M. Hikosaka, Thermochim. Acta, 43 1 (2005) 98. Z.G. Wang, X.T. Zu and Y. Huo, Thermochim. Acta, 436 (2005) 153. K. Ziewiec, P. Olszewski, R. Gajerski and A. Malecki, J. Alloys Compd., 373 (2004) 115.

M.J. Starink, N. Gao, L. Davin, J. Yan and A. Cerezo, Philos. Mag., 85 (2005) 1395.

C. Tatar and R. Zengin, Thermochim. Acta, 433 (2005) 56.

P. Kost'al, J. Shanelova and J. Malek, J. Non-Cryst. Solids, 35 1 (2005) 3 152. N. Mehta and A. Kumar, J. Them. Anal. Calorim., 83 (2006) 669. P. Vizcaino, R.O. Rios and A.D. Banchik, Themochim. Acta, 429 (2005) 7. C. Tatar and F. Yakuphanoglu, Thermochim. Acta, 430 (2005) 129. I. Katayama, K. Shimazawa, D. Zivkovic, D. Manasijevic, Z. Zivkovic and H. Yamashita, Themochim. Acta, 43 1 (2005) 138. D.S. Kanibolotsky, O.A. Bieloborodova and V.V. Lisnyak, Themochim. Acta, 438 (2005) 5 1. E. Cesari, V.A. Chernenko, J. Font and J. Muntasell, Themochim. Acta, 433 (2005) 153. G. Riontino, G.L. Ferro, D. Lussana and M. Massazza, J. Therm. Anal. Calorim., 82 (2005) 83. G. Riontino, D. Lussana and M. Massazza, J. Therm. Anal. Calorim., 83 (2006) 643. M. Campos, L. Blanco and J.M. Torralba, J. Therm. Anal. Calorim., 84 (2006) 483. G.A. Sharpataya, K.S. Gavrichev, A.V. Khoroshilov, V.N. Plakhotnik and V.M. Gurevich, J. Therm. Anal. Calorim., 82 (2005) 339. Z. Kalva and J. Sestak, J. Therm. Anal. Calorim., 76 (2004) 67. L. Bi, E. Wang, J. Peng, R. Huang, L. Xu and C. Hu, Inorg. Chem., 39 (2000) 671. B.H. Hamilton, K.A. Kelly, T.A. Wagler, M.P. Espe and C.J. Ziegler, Inorg. Chem., 43 (2004) 50. Y. Zhang, J.K. Liang, X.L. Chen, M.He and T. Xu, J. Alloys Comp., 327 (2001) 96. N. Koga, T. Utsuoka and H. Tanaka, J. Therm. Anal. Calorim., 80 (2005) 71. G. Yin, Y. Yao, B. Jiao, S. Chen and S. Gao, Themochim. Acta, 435 (2005) 125. M. Maczka, D. Wosewicz, P.E. Tomaszewski and J. Hanuza, Themochim. Acta, 438 (2005) 112. A.I. Korobov, J. Therm. Anal. Calorim., 74 (2003) 21 1. A.I. Korobov, Russ. J. Phys. Chem., 78 (2004) 584. P.K. Gallagher, J.P. Sanders, P.M. Woodward and I.N. Lokuhewa, J. Therm. Anal. Calorim., 80 (2005) 217. P.K. Gallagherand J.P. Sanders, J. Therm. Anal. Calorim., 82 (2005) 659. N. Koga and H. Tanaka, J. Therm. Anal. Calorim., 82 (2005) 725. Z. Li, X. Shen, X. Feng, P. Wang and Z. Wu, Themochim. Acta, 438 (2005) 102.

R. Strobel, M. Maciejewski, S.E. Pratsinis and A. Baiker, Thermochim. Acta, 445 (2006) 23. B.V. L'vov, Thermochim. Acta, 424 (2004) 183. S. Labus, A. Malecki and R. Gajerski, J. Therm. Anal. Calorim., 74 (2003) 13. J.G.P. Binner, D.M. Price, M. Reading and B. Vaidhyanathan, Thermochim. Acta, 443 (2006) 147. J.G.P. Binner, D.M. Price, M. Reading and B. Vaidhyanathan, Thermochim. Acta, 446 (2006) 156. J.G.P. Binner, G. Dimitrakis, D.M. Price, M. Reading and B. Vaidhyanathan, J. Therm. Anal. Calorim., 84 (2006) 409. R. Turcotte, P.D. Lightfoot, R. Fouchard and D.E.G. Jones, J. Hazard. Mater., 101 (2003) 1. J.C. Oxley, J.L. Smith, E. Rogers and M. Yu, Thermochim. Acta, 384 (2002) 23. G. Singh and S.P. Felix, Combust. Flame, 135 (2003) 145. G. Drake, T. Hawkins, A. Brand, L. Hall, M. Mckay, A. Vij and I. Ismail, Propellants, Explos., Pyrotech., 28 (2003) 174. J.H. Sun, Z.H. Sun, Q.S. Wang, H. Ding, T. Wang and C.S. Jiang, J. Hazard. Mater., 127 (2005) 204. X.R. Li and H. Koseki, Proc. Saf. Environ. Prot., 83 (2005) 3 1. I.M. Tuukkanen, E.L. Charsley, S.J. Goodall, P.G. Laye, J.J. Rooney, T.T. Griffiths and H. Lemmetyinen, Thermochim. Acta, 443 (2006) 116. A. Malecki and B. Malecka, Thermochim. Acta, 446 (2006) 113. V. Berbenni, C. Milanese, G. Bruni and A. Marini, J. Therm. Anal. Calorim., 82 (2005) 401. E. Charrier, E.L. Charsley, P.G. Laye, H.M. Markhama, B. Berger and T.T. Griffiths, Thermochim. Acta, 445 (2006) 36. B.S. Patra, S. Otta and S.D. Bhattamisra, Thermochim. Acta, 441 (2006) 84. Y. Cao, H. Zhang, C. Huang, Q. Yang, Y. Chen, R. Sun, F. Zhang and W. Guo, J. Solid State Chem., 178 (2005) 3563. V. Shvaiah and S.K. Das, Inorg. Chem., 44 (2005) 8846. D. Bayot, B. Tinant and M. Devillers, Inorg. Chim. Acta, 357 (2004) 809. Z. Fu, J. Dai, J. Zhang, S.Hu, R.Fu, W. Du and X. Wu, Inorg. Chem. Commun., 6 (2003) 919. M.A. Mohamed, A.K. Galwey and S.A. Halawy, Thermochim. Acta, 429 (2005) 57. D. Zhana, C. Conga, K. Diakitea, Y. Taoa and K. Zhanga, Thermochim. Acta, 430 (2005) 101.

M. Maciejewski, E. Ingier-Stocka, W. Emmerich and A. Baiker, J. Thermal Anal. Calorim., 60 (2000) 735. E. Ingier-Stocka and M. Maciejewski, Thermochim. Acta, 354 (2000) 45. E. Ingier-Stocka and M. Maciejewski, Thermochim. Acta, 432 (2005) 56. G. Vanhoyland, R. Nouwen, M.K. Van Bael, J. Yperman, J. Mullens and L.C. Van Poucke, Thermochim. Acta, 354 (2000) 145. G. Vanhoyland, A. Le Bail, J. Mullens and L.C. Van Poucke, Inorg. Chem., 43 (2004) 785. A. Ubaldini, C. Artini, G.A. Costa, M.M. Carnasciali and R. Masini, J. Therm. Anal. Calorim., 84 (2006) 207. E. Lamprecht, G.M. Watkins and M.E. Brown, Thermochim. Acta, 446 (2006) 9 1.

M.N. Sokolov, A.L. Gushchin, D.Yu. Naumov, O.A. Gerasko and V.P. Fedin, Inorg. Chem, 44 (2005) 243 1. D. Nelis, D. Mondelaers, G. Vanhoyland, A. Hardy, K. Van Werde, H. Van den Rul, M.K. Van Bael, J. Mullens, L.C. Van Poucke, and J. D'Haen, Thermochim. Acta, 426 (2005) 39. E. Diez, 0 . Monnereau, L. Tortet, G. Vacquier, P. Llewellyn and F. Rouquerol, J. Optoelectron. Adv. Mater., 2 (2000) 552. H. Aono, M. Tsuzaki, A. Kawaura, M. Sakamoto, E. Traversa and Y.J. Sadaoka, Am. Ceram. Soc., 84 (2001) 969. S.P. Tolochko, L.V. Makhnach, I.F. Kononyuk, V.V. Vashuk, V.A. Lomonosov, J. Hauck, H. Altenburg and V.M. Shelekhina, Inorg. Mater., 36 (2000) 1137. 0. Carp, L. Patron, G. Pascu, I. Mindru, N. Stanica, J. Therm. Anal. Calorim., 84 (2006) 391. M.C. Wilding, P.F. McMillan and A. Navrotsky, Physica A, 3 14 (2002) 379.

J.P. Sanders and P.K. Gallagher, Thermochim. Acta, 406 (2003) 241. J.P. Sanders and P.K. Gallagher, J. Therm. Anal. Calorim., 72 (2003) 777. J. Madarasz, M. Okuya, G. Pokol and S. Kaneko, Solid State Ionics, 172 (2004) 5 15.

M. Krunks, I. Oja, K. Tonsuaadu, M. Es-Souni, M. Gruselle and L. Niinisto, J. Therm. Anal. Calorim., 80 (2005) 483. A. Hardy, D. Mondelaers, G. Vanhoyland, M.K. Van Bael, J. Mullens and L.C. Van Poucke, J. Sol-Gel Sci. Technol., 26 (2003) 1103. A. Larraiiaga, J.L. Mesa, J.L. Pizarro, A. Peiia, J.P. Chapman, M.I. Arriortua and T. Rojo, Mat. Res. Bull., 40 (2005) 78 1. C.W. Lu, Q. Wang, Y.X. Chen and J.L. Shi, Thermochim. Acta, 404 (2003) 63.

K.J.D. MacKenzie, T. Kemmitt, R.H. Meinhold, M. Schmiicker and L. Mayer, J. Europ. Ceramic Soc., 20 (2000) 645. P. Sulcova, J. Therm. Anal. Calorim., 82 (2005) 51. P. Sulcova and M. Trojan, J. Therm. Anal. Calorim., 83 (2006) 557. Z. Mesikova, P. Sulcova and M. Trojan, J. Therm. Anal. Calorim., 83 (2006) 561. Z. Mesikova, P. Sulcova and M. Trojan, J. Therm. Anal. Calorim., 84 (2006) 733. P. Sulcova and M. Trojan, J. Therm. Anal. Calorim., 84 (2006) 737. 0.Pozdnyakova, J. Megyeri, E. Kuzmann and L. Szirtes, J. Therm. Anal. Calorim., 82 (2005) 409. A.S. Monayenkova, A.A. Popova, L.A. Tiphlova, Y.V. Gorushkina, A.V. Tsoi, I.P. Romanova and V.A. Alyoshin, Thermochim. Acta, 430 (2005) 83. Y.Z. Guo, R. Weiss and M. Epple, Eur. J. Inorg. Chem., 15 (2005) 3072. S. Yariv and I. Lapides, J. Therm. Anal. Calorim., 80 (2005) 11. Y. Xi, W. Martens, H. He and R.L. Frost, J. Therm. Anal. Calorim., 81 (2005) 91. R. Venkata Krishnan and K. Nagarajan, Thermochim. Acta, 440 (2006) 141. M. Ambrova and J. Jurisova, Thermochim. Acta, 443 (2006) 105. S.M. Pourmortazavi, M. Fathollahi, S.S. Hajimirsadeghi and S.G. Hosseini, Thermochim. Acta, 443 (2006) 129. S. Vyazovkin and T.L. Ferrin, Solid State Commun., 113 (2000) 627. I. Yamashita, T. Tojo, H. Kawaji and T. Atake, Phys. Rev. By 65 (2002) 212104. I. Yamashita, H. Kawaji, T. Atake, Y. Kuroiwa and A. Sawada, Phys. Rev. By67 (2003) 014104. X.Teng, F.Li, P.Ma, Q. Ren and S.Li, Thermochim. Acta, 436 (2005) 30. K. Popa, R.J.M. Konings, P. Boulet, D. Bouexiere and A.F. Popa, Thermochim. Acta, 436 (2005) 5 1. K. Popa and R.J.M. Konings, Thermochim. Acta, 445 (2006) 49. C.V.G. Reddy, S.V. Manorama and V.J.J. Rao, Mater. Sci.-Mater.El., 12 (2001) 137. R. Koferstein, H.P. Abicht, J. Woltersdorf and E. Pippel, Thermochim. Acta, 441 (2006) 176. I. Labidi, G. Kenessey and G. Liptay, J. Therm. Anal. Calorim., 82 (2005) 55. I. LabQdi,L. Horvath, G. Kenessey and G. Liptay, J. Therm. Anal. Calorim., 83 (2006) 247.

119. M. Krunks, J. Madarasz, T. Lekela, A. Mere, L. Niinisto and G. Pokol, J. Therm. Anal. Calorim., 72 (2003) 497. 120. L.A. Silva, J.R. Matos and J.B. de Andrade, Thermochim. Acta, 360 (2000) 17. 121. P. Masset, J.-Y. Poinso, J.-C. Poignet, J. Therm. Anal. Calorim., 83 (2006) 457. 122. E. Tomaszewicz and M. Kotfica, J. Therm. Anal. Calorim., 74 (2003) 783. 123. P. Budmgeac, J.M. Criado, F.J. Gotor, J. Malek, L.A. Perez-Maqueda and E. Segal, Int. J. Chem. Kinet., 36 (2004) 309. 124. P. Thomas and P. Simon, J. Therm. Anal. Calorim., 80 (2005) 77. 125. E. Godocikova, P. Bal, J.M. Criado, C. Real and E. Gock, Themochim. Acta, 440 (2006) 19. 126. H. Hamma, S.B. Rasmussen, J. Rogez, M. Alaoui-Elbelghiti, K.M. Eriksen and R. Fehrmann, Thermochim. Acta, 440 (2006) 200. 127. D. Mansfeld, M. Mehring and M. Schunnann, Inorg. Chim. Acta, 348 (2003) 82. 128. R.R. Jia, C.P. Wu, S. Wu, Y.X. Yang, Y.R. Chen and Y.Q. Jia, Amino Acids, 31 (2006) 85. 129. R.R. Jia, Y.X. Yang, Y.R. Chen and Y.Q. Jia, J. Therm. Anal. Calorim., 76 (2004) 157. 130. F.A. Almeida Paz and J. Klinowski, J. Solid State Chem., 177 (2004) 3423. 131. J.H. Liao, C.Y. Lai, C.D. Ho and C.T. Su, Inorg. Chem. Comm., 7 (2004) 402. 132. Y. Liu, Y. Lu, H. Wu, J. Wang and K. Lu, Inorg. Chem., 4 1 (2002) 2592. 133. V. Parvanova, J. Therm. Anal. Calorim., 84 (2006) 589. 134. A. Eslami, Thermochim. Acta, 409 (2004) 189. 135. C. Jiang and Z.Y. Wang, Polyhedron, 22 (2003) 2953. 136. B.Z. Lin, Z. Li, X.K. Pei and P.D. Liu, J. Mol. Structure, 660 (2003) 181. 137. D. Nikolova and R. Popova, J. Therm. Anal. Calorim., 84 (2006) 401. 138. Y. Liu, B. Ma, X. Zhao, Y. Deng, H. Zhang and Zichen Wang, Themochim. Acta, 433 (2005) 170. 139. E.V. Bakhmutova, X. Ouyang, D.G. Medvedev and A. Clearfield, Inorg. Chem., 42 (2003) 7046. 140. Dj.U. Miodragovi, Z.M. Miodragovi, D. Skala, M.J. Malinar, D.M. Mini and K. Andjelkovic, Themochim. Acta, 436 (2005) 90. 141. S. Materazzi, S. Aquili, S. De Angelis Curtis, S. Vecchio, K. Kurdziel and F. Sagone, Thermochim. Acta, 321 (2004) 19. 142. G. Xie, S. Chen, B. Jiao, S. Gao and Q. Shi, Thermochim. Acta, 443 (2006) 53.

143. S.K. Ghosh and P.K. Bharadwaj, Inorg. Chem., 43 (2004) 5180. 144. K. Fujii, T. Endoh, M. Yokoi, Y. Umebayashi and S.I. Ishiguro, Thermochim. Acta, 43 1 (2005) 29. 145. B.O. Patrick, C.L. Stevens, A. Storr and R.C. Thompson, Polyhedron, 24 (2005) 2242. 146. M.V. Marinho, M.I. Yoshida, K.J. Guedes, K. Krambrock, A.J. Bortoluzzi, M. Holmer, F.C. Machado and W.M. Teles, Inorg. Chem., 43 (2004) 1539. 147. S. Zang, Y. Su, Y. Li, H. Zhu and Q. Meng, Inorg. Chem., 45 (2006) 2972. 148. S. Vecchio, R. Di Rocco, C. Ferragina and S. Materazzi, Thermochim. Acta, 43 5 (2005) 181. 149. S. Takamizawa, E. Nakata and H. Yokoyama, Inorg. Chem. Comm., 6 (2003) 763. 150. M. Inoue, H. Kawaji, T. Tojo and T. Atake, Thermochim. Acta, 446 (2006) 117. 151. M. Inoue, M. Moriwaki, T. Atake, H. Kawaji, T. Tojo and W. Mori, Chem. Phys. Lett., 365 (2002) 509. 152. M. Inoue, T. Atake, H. Kawaji and T. Tojo, Solid State Commun., 134 (2005)303. 153. M. Inoue, H. Kawaji, T. Tojo and T. Atake, Thermochim. Acta, 43 1 (2005)58. 154. E.A. Ukraintseva, V.A. Logvinenko, D.V. Soldatov and T.A. Chingina, J. Therm. Anal. Calorim., 75 (2004) 337. 155. V.V. Krisyuk, A.E. Turgarnbaeva and S.W. Rhee, Polyhedron, 23 (2004) 809. 156. K. Meszaros Szecsenyi, V.M. Leovac, V.I. Cesljevic, A. Kovacs, G. Pokol, G. Argay, A. Kalman, G.A. Bogdanovic, Z.K. Jacimovic and A. Spasojevic-de Bire, Inorg. Chim. Acta, 353 (2003) 253. 157. A. Bajpai and S. Tiwari, Thermochim. Acta, 41 1 (2004) 139. 158. W. Qing, S.P. Chen, S.L. Gao, Z.C. Tan, Y.Y. Di and S. Qi-Zhen, Thermochim. Acta, 44 1 (2006) 132. 159. X. Gua, Z. Shia, J. Penga, Y. Chena, E. Wanga and N. Hu, J. Mol. Structure, 694 (2004) 219. 160. V. Zelenik, Z. Vargova, K. Gyoryoval, E. Vecemikov6 and V. Balek, J. Therm. Anal. Calorim., 82 (2005) 747. 161. C. Nather, I. Jep, N. Lehnert and D. Hinz-Hiibner, Solid State Sci., 5 (2003) 1343. 162. C. Nather and I. Jep, J. Solid State Chem., 169 (2002) 103. 163. A. Wojakowska and E. Krzyzak, J. Therm. Anal. Calorim., 83 (2006) 597.

S.C. Mojumdar, I. Ondrejkovicova, L. Nevidanska and M. Melnyk, J. Anal. Appl. Pyrol., 64 (2002) 59. R.L. Frost, R. Wills, M.L. Weier and W. Martens, Themochim. Acta, 437 (2005) 30. R.L. Frost, R. Wills, J.T. Kloprogge and W. Martens, J. Therm. Anal. Calorim., 84 (2006) 489. M. Sikorska-Iwan, R. Mrozek, A. Ostasz and Z. Rzaczynska, J. Therm. Anal. Calorim., 74 (2003) 529. L. Qiang-Guo, L. Xu, H. Yi, X. Sheng-Xiong, Y. De-Jun, Y. Li-Juan, W. De-Liang and L. Yi, Themochim. Acta, 441 (2006) 195. C. Qin, X. Wang, E. Wang and L. Xu, Inorg. Chim. Acta, 359 (2006) 417. J. Wu, T. Yeh, Y. Wen, J. Twu and K. Lu, Cryst. Growth Des., 6 (2006) 467. A. Thirumurugan and S. Natarajan, (???), 6 (2006) 983. Z. Rzaczynska, A. Ostasz and S. Pikus, J. Therm. Anal. Calorim., 82 (2005) 347. Z. Rzaczynska, A. Ostasz, M. Sikorska-Iwan, H. Gluchowska, E. Olszewska and S. Pikus, J. Therm. Anal. Calorim., 84 (2006) 575. R. Kurpiel-Gorgol and W. Brzyska, J. Therm. Anal. Calorim., 82 (2005) 389. W. Ferenc and B. Bocian, J. Therm. Anal. Calorim., 74 (2003) 521. L.P. Zhang, M. Du, W.J. Lu and T.C.W. Mak, Polyhedron, 23 (2004) 857. A. Messimeri, C.P. Raptopoulou, V. Nastopoulos, A. Terzis, S.P. Perlepes and C. Papadimitriou, Inorg. Chim. Acta, 336 (2002) 8. G. Vanhoyland, J. Pagnaer, J. D'Haen, S. Mullens and J. Mullen, J. Solid State Chem., 178 (2005) 166. M. Badea, R. Olar and A. Emandi, J. Therm. Anal. Calorim., 84 (2006) 60 1. C. Wu, C. Lu, S. Lu, H. Zhuang and J. Huang, Dalton Transactions, 16 (2003) 3 192. M. De Moura, J. Matos, L.P. Mercuri and R.F. de Farias, Themochim. Acta, 4 11 (2004) 1. M. De Moura, J. Matos and R.F. De Farias, J. Serbian Chem. Soc., 70 (2005) 1041. X.M. Wu, Z.C. Tan, X.C. Lvb and S.S. Quc, Themochim. Acta, 441 (2006) 199. G. Xie, S.P. Chen, S.L. Gao, X.X. Meng and Q.Z. Shi, J. Therm. Anal. Calorim., 83 (2006) 693. Y.M. Dan, Y.R. Zhao, Y. Liu and S.S. Qu, J. Therm. Anal. Calorim., 84 (2006) 53 1. B.A.A. Balboul, Themochim. Acta, 445 (2006) 78.

187. A. Rachel, S.G. Ebbinghaus, M. Gungerich, P.J. Klar, J. Hanss, A. Weidenkaff and A. Reller, Thermochim. Acta, 438 (2005) 134. 188. A. Neumann and D. Walter, Thermochim. Acta, 445 (2006) 200. 189. B.K. Singh, R.K. Sharma and B.S. Garg, J. Therm. Anal. Calorim., 84 (2006) 593. 190. M. Matrakova and D. Pavlov, J. Power Sources, 158 (2006) 1004. 191. J. Jian and J.R. Dahn, Electrochem. Solid State Lett., 6 (2004) A180. 192. E.P. Poth and D.H. Goughty, J. Power Sources 128 (2004) 308. 193. X. Zhang, P.N. Ross Jr., R. Kostecki, F. Kong, S. Sloop, J.B. Kerr, S. Striebel, E.J. Cairns and F. McLarnon, J. Electrochem. Soc., 148 (2001) A463. 194. K.S. Gavritchev, G.A. Sharpataya, A.A. Smagin, E.N. Malyi and V.A. Matyukha, J. Therm. Anal. Calorim., 73 (2003) 71. 195. K. Tasaki, K. Kanda, S. Nakarnura and M. Ue, J. Electrochem. Soc., 150 (2003) A1628. 196. G. Gerardine, R. Botte, E. White and Z. Zhang, J. Power Sources, 97 (2001) 570. 197. A. Du Pasquier, F. Disma, T. Bowmer, A.S. Gozdz, G. Amatucci and J.-M. Tarascon, J. Electrochem. Soc., 145 (1998) 472. 198. E. Zinigrad, L. Larush-Asraf, J.S. Gnanaraj, M. Sprecher and D. Aurbach, Thermochim. Acta, 438 (2005) 184. 199. B.R. Srinivasan and S.C. Sawant, Thermochim. Acta, 402 (2003) 45. 200. T. Hatanpaa, J. Kansikas, I. Mutikainen and M. Lesela, Inorg. Chem., 40 (2001) 788. 201. T. Stanimirova, T. Hibino and V. Balek, J. Therm. Anal. Calorim., 84 (2006) 473. 202. U. Garcia-Couceiro, D. Olea, 0. Castillo, A. Luque, P. Roman, P.J. de Pablo, J. Gomez-Herrero and F. Zamora, Inorg. Chem., 44 (2005) 8343. 203. Z. Rzaczynska, A. Bartyzel and T. Glowiak, Polyhedron, 22 (2003) 2595. 204. Y. Kim and D. Jung, Inorg. Chem., 39 (2000) 1470. 205. X. Wei, X. Du, D. Chen and Z. Chen, Thermochim. Acta, 440 (2006) 181. 206. C. Wu, C. Lu, S. Lu, H. Zhuang and J. Huang, Inorg. Chem. Commun., 5 (2002) 171. 207. C. Ruiz-Perez, P. A. Lorenzo Luis, F. Lloret and M. Julve, Inorg. Chim. Acta, 336 (2002) 131. 208. N.T. Madhu, P.K. Radhakrishnan and W. Linert, J. Therm. Anal. Calorim., 84 (2006) 607. 209. H. Icbudak, Z. Heren, A. Uyanik and M. Odabasoglu, J. Therm. Anal. Calorim., 82 (2005) 303.

210. Cs. Vkhelyi Jr., G. Pokol, A. GGmory, A. Ganescu, P. Sohk, G. Liptay and Cs. Virhelyi, J. Therm. Anal. Calorim., 83 (2006) 701. 21 1. E.T. de Almeida, A.M. Santana, A.V.G. Netto, Claudia Torres and A.E. Mauro, J. Therm. Anal. Calorim., 82 (2005) 361. 212. S. Vecchio, S. Materazzi and K. Kurdziel, Int. J. Chem. Kin., 37 (2005) 667. 213. Y. Suenaga, K. Kitamura, T. Kuroda-Sowa, M. Maekawa and M. Munakata, Inorg. Chim. Acta, 328 (2002) 105. 214. S.E. Paramonov, N.P. Kuzmina and S.I. Troyanov, Polyhedron, 22 (2003) 837. 2 15. N.J. Melcer, G.D. Enright, J.A. Ripmeester and G.K.H. Shimizu, Inorg. Chem., 40 (2001) 4641. 216. C. Cagran, B. Wilthan and G. Pottlacher, Thermochim. Acta, 445 (2006) 104. 217. H. Chen, T. Zhang, J. Zhang, X. Qiao and K. Yu, J. Hazard. Mat., 129 (2006) 3 1. 218. S. Yamada and N. Koga, Thermochim. Acta, 431 (2005) 38. 219. K. Chandran, R. Venkata Krishnan, A. Gopalan, K. Nagarajan and V. Ganesan, Thermochim. Acta, 438 (2005) 107. 220. Y. Liu, J. Su, W. Li and J. Wu, Inorg. Chem., 44 (2005) 3890. 221. C. Kazak, S. Hamamci, Y. Topcu and V.T. Yilmaz, J. Mol. Structure, 657 (2003) 35 1. 222. T.R. Jensen, N. Gerentes, J. Jepsen, R.G. Hazel1 and H.J. Jakobsen, Inorg. Chem., 44 (2005) 658. 223. M.R. Hill, A.W. Jones, J.J. Russell, N.K. Roberts and R.N. Lamb, Inorg. Chim. Acta, 358 (2005) 201. 224. V. Zelenak, V. Hornebecq and P. Llewellyn, Microporous Mesoporous Mater., 83 (2005) 125. 225. S. Neogi and P.K. Bharadwaj, Inorg. Chem., 44 (2005) 816. 226. A. Migdal-Mikuli and E. Szostak, Thermochim. Acta, 444 (2006) 195. 227. N. Chikaraishi Kasuga, K. Sekino, M. Ishikawa, A. Honda, M. Yokoyama, S. Nakano, N. Shimada, C. Koumo and K. Nomina, J. Inorg. Biochem., 96 (2003) 298. 228. Y.Y.Di,Z.C. Tan,L.W. Li, S.L. Gao, J. Xing, S.X. WangandL.X. Sun, J. Chem. Eng. Data, 50 (2005) 1641. 229. Y.Y. Di,Z.C. Tan,H.G. Yu, S.L. Gao,Y. LiuandL.X. Sun, Thermochim. Acta, 4 12 (2004) 171. 230. K. Gyoryovi, J. Chomic ,E. Szunyogovi, L. Piknovi, V. Zelenhk and Z. Vargovi, J. Therm. Anal. Calorim., 84 (2006) 727.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors 2008 Elsevier B.V.

Chapter 13

ISOCONVERSIONAL KINETICS Sergey Vyazovkin Department of Chemistry, University of Alabama at Birmingham, 901s 1 4 ~ ~ Street, Birmingham, AL 35294, USA, [email protected] 1. INTRODUCTION Isoconversional kinetics rest upon evaluating a dependence of the effective activation energy on conversion or temperature and using this dependence for making kinetic predictions and for exploring the mechanisms of thermal processes. This chapter presents the evolution of isoconversional methods and discusses their advantages and limitations. The emphasis is put on the development of new techniques and applications that have taken place over the past decade. The applications covered include both physical and chemical processes. Physical processes include crystallization of polymer melts and glasses, the glass transition, and second-order solid-solid transitions. Chemical processes involve reversible decompositions, degradation of polymers, and crosslinking (curing). The chapter also discusses the methods of making isoconversional (model-free) predictions, as well as the techniques for evaluating the pre-exponential factor and the reaction model. Thermal analysis methods are not species specific. By measuring the evolution of overall physical properties of a system, these methods provide information on macroscopic kinetics. The macroscopic kinetics are inherently complex because they include information about multiple steps that occur simultaneously. Unscrambling complex kinetics presents a serious challenge that can only be met by kinetic methods that provide means of detecting and treating multi-step processes. As indicated by the results of the ICTAC Kinetics Project [1,2,3], only the methods that use multiple heating programmes can meet this challenge. Isoconversional methods are definitely the most popular of the methods based on the use of multiple heating programmes. The present chapter provides a brief introduction to isoconversional methods and an overview of their application to the kinetic analysis of various physical and chemical processes.

2. ISOCONVERSIONAL METHODS These methods have their origin in the single-step kinetic equation:

and make use of the isoconversional principle, which states that, at a constant extent of conversion, the reaction rate is a function only of the temperature so that:

[

d ln(da l dt) -

1

la=-%

In equation (1) and (2), A and E are the Arrhenius parameters (the preexponential factor and the activation energy, respectively), A a ) is the reaction model (or conversion function), R is the gas constant, T is the temperature, t is the time, and a is the extent of conversion. Henceforth, the subscript a denotes values related to a constant extent of conversion. A combination of E, A, and A a ) is sometimes called the "kinetic triplet". In order to obtain data on a variation of the rate at a constant extent of conversion (i.e., the left hand side of equation (2)), isoconversional methods employ multiple temperature programmes (e.g., different heating rates and/or temperatures). Although equation (2) is derived from the single-step kinetic equation (I), the fundamental assumption of isoconversional methods is that a single equation (1) is applicable only to a single extent of conversion and to the temperature region (AT) related to this conversion (Figure 1). That is, isoconversional methods describe the kinetics of the process by simultaneously using multiple single step kinetic equations (Figure 1). This attribute of isoconversional methods allows multi-step processes to be detected via a variation of E, with a . Equation (1) is easily rearranged into equation (3):

Figure 1. Each value of E, is associated with a narrow temperature region ATthat changes with a. which is the basis of the differential method of Friedman [4]. The subscript i denotes different heating rates. The application of this method to integral data (e.g., TG data) requires numerical differentiation of experimental a vs T curves that tends to yield quite noisy rate data and, therefore, scattered E, values. Numerical differentiation can be avoided by using integral methods. Integration of equation (1) for isothermal conditions yields:

where g(a) is the integral reaction model. Rearrangement of equation (4) leads to equation (5):

Equation (5) allows for evaluating the E, dependence from a series of runs performed at different temperatures, Ti.

Nonisothermal runs are most commonly performed at a constant heating rate P. For such conditions, integration of equation (1) requires solving the temperature integral I(E, T):

which does not have an analytical solution. Its solution is accomplished by using either approximations or numerical integration. For instance, the use of Doyle's approximation [5] yields equation (7):

which is the foundation of the most popular isoconversional methods by Flynn and Wall [6] and by Ozawa [7]. A more precise approximation by Coats and Redfern [8] gives rise to equation (8):

For even better precision, one can use numerical corrections [9] or numerical integration. The latter, for example, is employed in the method [lo] that makes use of minimizing the following function:

At each particular value of a in equation (9), E, is determined as a value that minimizes @(E,), and the temperature integral, I(E, T ) is solved numerically. The simpler integral methods (e.g., equations (5), (7) and (8)) assume that the value of E, is constant in I(E,T) throughout the whole interval of integration, from 0 to a. This assumption causes a systematic error in the value of E, when E, varies with a. The error can be as large as 20 - 30% [l 11. This error does not appear in the differential method of Friedman. Also, it is easily eliminated in the advanced integral methods of Vyazovkin [10,11,12] because they use numerical integration as a part of E, evaluation. Eliminating the error in equation (9) is accomplished by integration over small temperature segments as follows:

In this case, E, is assumed constant for only a small interval of conversions, Aa. Integration by segments yields E, values that are similar to those obtained by the Friedman method [ l 11. Integration over segments has also been implemented by Simon et al. [13] in an incremental integral isoconversional method. Budrugeac [14] has proposed a nonlinear differential isoconversional method that uses a numerical algorithm similar to that used in the advanced isoconversional method (equation (9)). He has also demonstrated [14] the asymptotic convergence of the advanced integral method with the differential method. Although isothermal and constant heating rate temperature programmes are employed most commonly, they do not exhaust all practical needs. Other important programmes include constant rate cooling and distorted linear heating andlor cooling. The former finds wide use in the melt crystallization studies. The latter is frequently encountered as a result of self-heating or self-cooling when studying processes accompanied by significant thermal effects. Such temperature programmes cannot be handled by simpler integral methods (e.g., equations (7) and (8)) because they have been developed under the assumption that p is constant and positive. Because these assumptions are not made in the differential method of Friedman, it can be applied to handle data obtained under the aforementioned temperature programmes. However, the advanced integral method (equation (9)) can be adjusted to an arbitrary temperature programme T(t) by replacing integration over the temperature with integration over the time. The resulting advanced isoconversional method of Vyazovkin [l 1,121 can be used to handle data obtained under arbitrary temperature programmes, T,(t). In this method, the E, value is found by minimizing the function:

" " J[E,,T.(~,~ J~E,,T, (t, )

@(E,)=CC '=I ,ti

where J[E., T,. (t,

)I

exp[*]dt

I '=-,a

RT, (t)

(12)

The effect of self-heatinglcooling is accounted for in equation (1 1) by using the so-called "sample temperature" whose variation with the time represents the actual temperature programme (i.e., T(t)) experienced by a sample.

Sometimes the popular Kissinger method [15] is erroneously classified as an isoconversional method. The confusion appears to stem from the fact that the basic equation of the method:

looks similar to the isoconversional equation (8). Nevertheless, in equation (13) TPfi is the peak temperature at different heating rates, and the extent of conversion at the peak is known [9] to vary with the heating rate. In addition, the Kissinger method always produces a single value of E for the whole process. This is an important limitation of the method, because the resulting value is sound only if E, does not vary with a throughout the process. Yet, variations are very common, and the Kissinger method cannot detect them. Therefore, the E values determined by this method should be considered with care, unless an isoconversional method has been used to prove that E, is independent of a.

3. CONCEPT OF VARIABLE ACTIVATION ENERGY To our knowledge the first application of an isoconversional method is due to Kujirai and Akahira [16], who studied the decomposition rates of insulating materials. In their work, they used an empirical equation:

where w is the mass decrease in % of the initial value, t is the time to reach the extent of decomposition w at different temperatures, and Q is a "material constant" [16], which was determined as the slope of a plot of logt vs T-'. The meaning of Q and F(w) in equation (14) can be easily established by comparing it with equation (5). Clearly, Q is E/2.303R7and F(w) is -log[g(a)lA]. By using the early data of Kujirai and Akahira one can determine that the activation energy for degradation of the insulating materials studied demonstrates a noticeable variation with w (Figure 2). A similar effect was reported for the thermal degradation of a phenolic plastic by Friedman [4] in the first application of his method and for the thermal degradation of Nylon 6 by Ozawa [7] in the first application of his method. Apparently the phenomenon of a variable activation energy has been with isoconversional methods since their first days of existence. Using simulated data for multi-step reactions, Flynn and Wall [6] demonstrated that experimentally observed variations of the activation energy

--B--silk

--*-.manila paper --O--filter paper - -e- - cotton

Figure 2. Variations in E obtained Figure 3. Arrhenius plot for reaction complicated by difhsion fiom the early data [16] are nothing else but a sign of the process complexity. In spite of this fact, E, dependencies have not been exploited, but rather ignored as a major nuisance. The potential of isoconversional methods had not been fully appreciated until we [17] brought analysis of the Eddependencies to the forefront and demonstrated their utility for predicting kinetics and exploring the mechanisms of processes. We also introduced [18] the concept of a variable activation energy. Although this concept does not sit well with the traditionalist's point of view [19], it does provide a reasonable compromise between the actual complexity of condensed phase reactions and oversimplified methods of describing their kinetics [20]. As we argued [18, 211, the expectation that an experimentally determined activation energy should be a single constant value directly related to an energy barrier, originates from kinetic theories of simple gas phase reactions. For such systems, single steps can be easily isolated and their kinetics can be readily measured. As a result, the experimental activation energy can be directly linked to the energy barrier of a reaction. However, isolation of individual steps is not generally possible when applying TG and DSC to study processes that occur in the solid or liquid media. For these types of processes, two factors need to be considered: the simultaneous occurrence of multiple reaction steps and the presence of diffusion. Diffusion adds an extra step to a chemical reaction so that the overall process rate becomes dependent on the chemical reaction rate as well as on the transport rate of the reactants and products in the reaction medium. In this case, the temperature dependence of the overall rate is described by the effective rate constant, kdas follows:

where kR and kD are the reaction and diffusion rate constants respectively. From equation (15) one can easily derive the effective activation energy for a process that involves both chemical reaction and diffusion:

Because both kRand kDvary with temperature, the effective activation energy in equation (16) is also temperature dependent. An Arrhenius plot of Inkef against T - I for such process is nonlinear so that the activation energy derived from its slope changes with increasing temperature from the activation energy of reaction (ER) to the activation energy of diffusion (ED). The activation energy of diffusion of small molecules in the liquid or solid medium is typically a small value, markedly smaller than that of a chemical reaction. In this situation, the Arrhenius diagram shows a characteristic plot that is bent upwards (Figure 3). Only if one of the two steps is much faster than another, will the overall rate be determined by the slowest step and the experimental value of Eef will become the activation energy of this step, i.e., EefER if kR << kD (so-called kinetic control) or E e ~ EifDkD<< kR(so-called difhsion control). The dependence of the effective activation energy on both temperature and extent of conversion can be exemplified by a decomposition process that occurs via two parallel pathways as follows:

If the pathways follow different reaction models, the overall decomposition rate is given by equation (18):

By taking the logarithmic derivative of the reaction rate at a constant a,we can determine the effective activation energy at each conversion as follows:

A three-dimensional plot of Ea can be visualized by using, for instance, kinetic triplets experimentally determined for parallel channels of decomposition of nickel formate that are as follows:fi(a)=(l-a12", E1=200 kJ mol-', A1=1016minand J;(a)=a(l-a) and E2=100 kJ mol-', A2=107 min-' [18]. Inserting these triplets in equation (19) yields an effective activation energy, which is a function of the temperature as well as of the extent of conversion (Figure 4). The two aforementioned examples show that the experimental activation energy is generally a function of the energy barriers of the individual steps of a process. Note that the examples considered are very simplistic, and real processes tend to involve multiple chemical and diffusion steps. In this situation it may be practically impossible to link the experimental values of the activation energy of a process to the energy barriers of the individual steps. To stress this difference, the experimental values of the activation energy are frequently referred to as an "effective", or "apparent", or "global" activation energy. Because of its composite nature the effective activation energy tends to vary throughout the process. That is why the concept of variable effective activation energy has been proposed [18] as a necessary compromise between the actual complexity of condensed phase reactions and oversimplified methods of describing their kinetics

Figure 4. Simulated surface plot of E, vs a and T (equation (19))

Figure 5. Free energy of nucleation as a function of nucleus radius

4. KINETICS OF PHYSICAL PROCESSES

4.1. Crystallization Most melted substances crystallize on cooling. From the thermodynamic standpoint, crystallization should occur spontaneously, just below the equilibrium melting temperature, T, , because, under these conditions, the Gibbs (free) energy of the solid is lower than that of the liquid phase. The resulting difference in the free energies is a negative value termed the volume free energy, AGV.In reality, no visible crystallization occurs until the melt reaches significant supercooling. The delay in crystallization is caused by a free energy barrier associated with the creation of the solid phase nuclei that is characterized by the surface free energy, AGs. This value represents the difference in the Gibbs energy of the surface and the bulk of the nucleus. Because the free energy of the surface is always larger by the value of the surface energy, o, the value of AGs is positive. The total free energy of the nucleation process is given as: AG = AG,

+ AG,

(20)

The nuclei are usually assumed to have a spherical shape of the radius, r so that AGs is the surface area of the sphere times the surface energy, and AGVis the volume of the sphere times the volume energy per unit volume, AG,:

Equation (21) establishes a dependence of AGs and AGv on the radius of a nucleus (Figure 5). The sum of these two terms (i.e., AG) passes through a maximum that defines the critical size of a stable nucleus. The formation of a nucleus of a larger size would result in its spontaneous growth accompanied by a decrease in AG. The maximum value of AG is the energy barrier for the nucleation process, AG*. The nucleation rate is commonly expressed in the Arrhenius form:

The height of the energy barrier can be found from the condition of a maximum for AG (equation 2 I), which yields:

The value AG, is defined as: AG, =

- AH,AT To

where AT = Tm- T is the supercooling, and Mfis the heat of fusion. The value of AG,, decreases with supercooling (i.e., with decreasing temperature) causing a decrease in the critical nucleus size and the energy barrier (equation (23)). It follows from equations (23) and (24) that:

Because AG* decreases with decreasing temperature (equation (25)), the nucleation rate increases. This situation gives rise to a temperature dependence that is markedly different from a typical Arrhenius dependence. Figure 6 shows an Arrhenius plot for the nucleation rate. The plot has a positive slope that corresponds to the negative temperature coefficient andlor negative effective energy of activation. Also, the slope varies strongly with the temperature, reaching infinity at T=Tm(equation (25)). Not far below the melting point, the nucleation rate quickly increases with decreasing temperature. However, the nucleation rate does not increase monotonously, but passes through a distinct maximum at a certain temperature, T,, (Figure 7). Decreasing the temperature below Tm, results in decreasing the nucleation rate. This occurs because of a significant increase in the melt viscosity that creates an energy barrier, ED associated with diffusion of molecules across the phase boundary. Introduction of the respective energy term into equation (22) gives [22]:

Unlike the AG* term, the ED term represents a typical Arrhenius temperature dependence (Figure 7). Therefore, the product of the two terms (equation (26)) gives rise to a temperature dependence that has a maximum in the nucleation

Figure 6. Arrhenius plot for nucleation in temperature range TgTm( 1 : melt nucleation; 2: glass nucleation)

Figure 7. Temperature dependence of nucleation rate ( 1 e x p ( - ~ ~ * / k T2:) ; exp(-EDlkT);3: product of 1 and 2)

rate. Below the temperature related to the maximum, the process becomes controlled by diffusion that results in a dramatic decrease of the nucleation rate. If the maximum nucleation rate is not very large in a particular substance, its cooling may result in semi-crystalline material that contains a substantial fraction of the amorphous (glassy) phase. This fraction can be increased by increasing the rate of cooling. Ultimately, fast cooling results in the formation of an entirely amorphous substance, i.e., a glass. Glasses can be crystallized by heating above their glass transition temperature. The process is frequently called "cold crystallization". The nucleation rate in the glass is largely limited by diffusion and increases with increasing temperature. The corresponding Arrhenius plot has the normal negative slope that represents the positive temperature coefficient and/or positive effective energy of activation (Figure 6). The slope also decreases with increasing temperature. It follows from the above that crystallization kinetics generally demonstrate non-Arrhenius behaviour (Figure 6). Although for narrow temperature intervals the temperature dependence of the crystallization rate can be approximated by using the Arrhenius equation, for wider intervals one should make an allowance for the temperature variation in E. As seen from Figure 6, the E value should continuously decrease with increasing temperature. It should be positive for the glass crystallization and decrease as the temperature is increased from the glass-, transition temperature, Tg to T,,. For the melt crystallization it should be negative and increase when the temperature is increased from T, to T,.

Note that an allowance for variation in E is not made in the Kissinger method that is very frequently used to determine the so-called "activation energy of crystallization" for melts, as well as for glasses. This method is thus inadequate for crystallization of glasses because it masks the temperature dependence of E. Furthermore, it is inapplicable to the processes that occur on cooling, such as the melt crystallization [23]. The use of an advanced isoconversional method (equation (11)) and the method of Friedman (equation (3)) has been recommended [23] as a viable alternative to the Kissinger method. The application of isoconversional methods to the glass and melt crystallization data results in the general types of the E, dependencies shown in Figure 8. These dependencies reflect the respective temperature dependencies of the slopes of the Arrhenius plots (Figure 6).

Glass: E > 0 T increases

-

Glass crystallization j

----------------....---.--------------

Melt: E < 0 T decreases

T,-T-

Tm

Figure 8. Dependencies of E, on a Figure 9. Variation in E according to the Hoffman-Lauritzen theory for melt and glass crystallization Therefore, one must be careful in assigning a physical meaning to the experimental value of E. It is common practice to call an experimentally determined E value an "activation energy", which is not infrequently interpreted as an "energy barrier". Such interpretations are clearly invalid in the case of crystallization. As seen from Figure 6, by performing crystallization in different temperature regions one can obtain practically any value of E from large positive to large negative numbers and this obviously suggests that E is not an energy barrier of crystallization.

4.2. Melt and glass crystallization of polymers Negative Eff values (Figure 8) have been obtained experimentally by Vyazovkin et al. [24] for the melt crystallization of poly(ethy1ene terephthalate), poly(ethy1ene oxide) [25], and poly(ethy1ene 2,6-naphthalate) [26]. Similar dependencies have also been reported by other workers [27,28,29,30,31,32,33,34,35,36,37,38,39,40,4 1,421 who applied isoconversional methods to crystallization of the melts. More importantly, it has been demonstrated [43] that the resulting Eddependencies can be employed for evaluating the parameters of the Hoffman - Lauritzen theory [44]. According to this theory the linear growth rate of a polymer crystal, G depends on temperature, T as follows:

where Go is the preexponential factor, U* is the activation energy of the segmental jump, AT=Tm-Tis the undercooling, p2Tl(Tm+T) is the correction factor, T, is a hypothetical temperature at which viscous flow ceases (usually taken as Tg - 30K). The parameter Kg is defined as:

where b is the surface nucleus thickness, o is the lateral surface free energy, o, is the fold surface free energy, Tmis the equilibrium melting temperature, Ahf is the heat of fusion per unit volume of crystal, kg is the Boltzmann constant, and n takes the value 2 for crystallization regime I and 111, and 1 for regime 11. The parameter U*is frequently taken as the "universal" value 1.5 kcal mol-' [44]. Vyazovkin and Sbirrazzuoli [43] have used equation (27) to determine the theoretical temperature dependence of the effective activation energy (Figure 9) of the growth rate as:

Numerous experimental data by Toda et al. [45,46] indicate that the logarithmic derivative of the microscopically measured growth rate is equivalent to the logarithmic derivative of the overall crystallization rate obtained from DSC. For this reason, the temperature dependence of the effective activation energy

evaluated from DSC data can be fitted to equation (29) to determine the values of U* and K,. A dependence of E on T can be easily obtained from an isoconversional dependence of E, on a. Each value of E, is linked to a given a , which, in its turn, is related to a narrow temperature region, AT (see Figure 1). Therefore, one can correlate the E, values with the temperature by replacing a with an average temperatures corresponding to this a at different heating rates. Equation (29) suggests that the effective activation energy decreases with increasing the temperature throughout both glass and melt crystallization regions [26]. These two regions are usually separated by the maximum crystallization rate. Therefore, the E value typically changes fkom positive to negative on passing from the glass to melt crystallization region (Figure 9). That is, a change in the temperature region of crystallization would result in changing a value of E. This, as stated earlier, indicates that E is not an "energy barrier" of crystallization, although it is sometimes interpreted as such. In order to stress its difference from the true activation energy, which represents the energy barrier, the E value should be referred to as the effective activation energy. Figure 10 shows the E, dependencies for the glass and melt crystallization of poly(ethy1ene terephthalate). As stated earlier, the glass crystallization data produce positive E, values that decrease with increasing a (i.e., with increasing T). On the other hand, the melt crystallization data yield negative E, values that increase with increasing a (i.e., with decreasing T). 0 Glass crystallization,E > 0 loot

: "%....\

0 Glass crystallization, E > 0 O

10

Melt crystallization, E < 0

E =0

I -

\

0 Melt crystallization,E < 0

Figure 11. E, vs T data converted from E, dependencies (Figure 10) and fitted to equation (29) (dashed line) By converting E, vs. a for the melt data into an E, vs T dependence and fitting the latter to equation (29), one obtains the values Kg= 3.2 x lo5 K~ and U*= 4.3 Figure 10. E, dependencies for the glass and melt crystallization of poly(ethy1ene terephthalate)

kJ mol-' [43]. Vyazovkin et al. have also used this method to determine the Kg and U* values for the melt crystallization of poly(ethy1ene oxide) 1251 and poly(ethy1ene 2,6-naphthalate) [26] and found the values to agree with the literature values derived from microscopic measurements. Similarly good agreement has also been reported by other workers [32,35,40,41] who applied the aforementioned approach. Vyazovkin and Dranca [26] have recently demonstrated that equation (29) is suitable for simultaneously fitting combined melt and glass crystallization data. Figure 11 provides an example of such a fit for poly(ethy1ene terephthalate) that produces the values Kg= 3.6 x lo5 K~ and U*=7.5 kJ mol-'. Compared to the values derived from the melt data alone, the Kg value has not changed much, whereas the U* value has risen closer to the universal value 6.3 kJ mol-l. Similar results has been reported [26] for poly(ethy1ene 2,6-naphthalate) crystallization. The use of the combined data sets [26] improves the precision of the fit as well as the accuracy of the U*value.

4.3. Second-order transitions Second-order transitions do not cause abrupt changes in the volume or enthalpy but in the derivatives of the respective properties. The resulting effects are quite subtle. Figure 12 displays the ferromagnetic to paramagnetic transition in Ni metal. As seen it has the characteristic appearance of the inverted Greek letter 1. That is why the transitions of this type are called 1-transitions. The critical temperature, Tc determined from DSC is 632 K (359 "C). This value is typically found to be somewhat larger than the Curie temperature (354 "C) determined from magnetic measurements [47]. Increasing the heating rate causes a slight shift of the transition to higher temperatures. This allows one to determine the effective activation energy of the transition by using an isoconversional method. It is seen from Figure 13 that the E, value changes dramatically throughout the process. The behaviour of the effective activation energy can be predicted fi-om the Landau theory of phase transitions. According to it, for a second-order transition, the temperature dependence of the relaxation time is given by [48]:

where Tc is the critical temperature at which the two phases would be at equilibrium. From equation (30) one can derive the temperature dependence of the effective activation energy as follows:

As this equation suggests, the effective activation energy should quickly increase to infinity on approaching the critical temperature (see Figure 12). Obviously, the observed E, dependence [49] (Figure 13) tracks the theoretical temperature dependence (Figure 12) and simply reflects the phenomenon of passing through the critical point of the transition. Similar behaviour should be expected for other &transitions.

560

580 600 620 640 660 TIK

Figure 12. DSC trace of Ni around Figure 13. E, dependence for the Athe Curie point. Inset: E vs T depen- transition in Ni obtained from DSC dence by equation (3 1) data (Figure 12) 4.4. Glass transition Many materials can be produced in the glassy state. This is a nonequilibrium state that is thermodynamically driven to relax to the equilibrium liquid state. The relaxation process can be readily followed by various thermal techniques. The rate of relaxation is characterized by the relaxation time, r, whose temperature dependence is described either by the Arrhenius equation:

or, more commonly, by the Williams-Landel-Ferry (WLF) equation:

log-

T

To

=

-C,(T-To) (C,+T-To)

where To is a reference temperature, zo is the relaxation time at To,and CI and C2 are constants. The WLF equation works well for relaxation processes above the glass transition temperature, T,. The Arrhenius equation is typically applied well below T,, but may also serve as an approximation above T,, especially when relaxation is measured over a narrow region of temperatures. DSC is applied routinely for measuring the glass transition. The transition appears as a step in curve of heat capacity against T. An increase in the heating rate causes this step to shift to higher temperatures. This shift can be used for determining the activation energy of the process according to equation (34) 1501:

The value of T, can be defined in several ways. For instance, Moynihan et al. [50] have used T, determined as the extrapolated onset, the inflection point, and the position of a DSC peak obtained on heating. The differently defined values of T, correspond to different stages of the glass transition. It has been reported in several papers [51,52,53] that the E value estimated from equation (34), decreases with increases in the T, value. It is markedly bigger when T, is taken as an onset temperature and smaller when it is taken as the midpoint andlor peak temperature. To explore this phenomenon more closely, Vyazovkin et al. [54] have employed an isoconversional method that allowed them to reveal a variation in E throughout the glass transition. The conversion, a , can be readily determined from DSC data as the normalized heat capacity [55] :

where C, is the observed heat capacity, and C,, and C,,[ are, respectively, the glassy and liquid heat capacities. Because the values of C,, and C , / depend on temperature, they are extrapolated into the glass transition region. Figure 14 shows the normalized heat capacities for the glass transition in maltitol measured at different heating rates. The application of an isoconversional method to these data reveals a decreasing dependence of E, on a (Figure 15).

As noted earlier (section 4.2), each value of E, is related to a given a and, therefore, to a respective temperature region, AT (see Figure 1). This allows one to link the E, values to the average temperature of the AT region. Figure 15 shows the E , vs T dependence obtained by substituting the average T, for a. The dependence shows a decrease that is typical of the glass transition [54]. The

Figure 14. Normalized heat Figure 15. Variation in E with a and capacities at different heating rates T for the glass transition in maltitol (numbers by the line types) trend is predicted by the WLF equation that gives rise to a decreasing temperature dependence of the effective activation energy: C,C,T~ d lnz E = R= 2.303R dT-' (c, +T -T,), The decrease is explained by the co-operative character of the molecular motion. Co-operativity is very strong in the early stages of the glass transition, when the available free volume is too small to allow for independent motion of individual molecules. As a result, the co-operative motion has a large energy barrier. As the temperature rises, the free volume increases, relieving energetic constraints, so that the E value decreases. A correlation between the co-operativity of the molecular motion and the magnitude of E has been revealed by Vyazovkin and Dranca [56], who compared the glass-transition kinetics of poly(styrene) and

poly(styrene)-clay nanocomposite. Because the composite has a brush structure [57], its polymer chains move in a correlated manner that reflects in a markedly larger size of the cooperatively rearranging region and, therefore, in a larger value of the activation energy. A similar correlation has been reported [58] in a comparative study of the glass transition in glucose and maltitol, which is a bulkier derivative of glucose. The variability in E correlates [54,59] with the dynamic fragility of the glass forming systems. The biggest variation has been observed for polymers, having the largest fragility (e.g., poly(viny1 chloride) and poly(ethy1ene terephthalate)), whereas the smallest variation is found in the systems, having the lowest fragility (e.g., poly(n-butyl methacrylate) and boron oxide) [59]. This correlation suggests that kinetic models of the glass transition that assume the constancy of the activation energy are conceptually unsuitable for polymer glasses because they tend to have the largest fragility.

5. KINETICS OF CHEMICAL PROCESSES. 5.1. Reversible decompositions A number of solid-state decompositions occur reversibly in accord with the following general equation:

Because the activities of pure solid phases are taken to be 1, the equilibrium constant of such a process is:

where PC is the partial pressure of the gaseous product C and PO is the standard pressure. The temperature dependence of the equilibrium constant: InK=- -

RT

+ Const

allows one to correlate the pressure and temperature at which all three phases can co-exist in equilibrium. It follows from equation (38) and (39) that the logarithm of the equilibrium pressure has a linear dependence on the reciprocal temperature. A good example is the reversible decomposition of calcium carbonate that occurs as follows:

Figure 16 presents a temperature dependence of the equilibrium pressure of carbon dioxide for the decomposition of calcium carbonate [60]. The straight line:

P[z2(atm) = 4.67 x 10' exp

(41)

correlates the pressure and temperature at which CaC03(,),CaO,,) and COZc,,are at equilibrium with each other. Below this line, calcium carbonate would decompose to calcium oxide and carbon dioxide. While largely simplified, this thermodynamic model introduces an important notion that a reversible decomposition should start when the equilibrium pressure of the

2001

*

& Q ,

Masuda et al Urbanovici, Segal Q Vyazovkin

v

Figure 16. Equilibrium pressure of Figure 17. E, dependencies reported C02 for decomposition of CaC03 by several workers for dehydration of CaC204.H20 gaseous product rises above its partial pressure at a given temperature. For instance, the partial pressure of carbon dioxide in air under normal conditions is ~ which is significantly larger than the equilibrium approximately 3 x 1 0 ~atm, pressure (Figure 16). In order to raise the equilibrium pressure above 3x10-~atm one needs to raise the temperature above 530°C. Then calcium carbonate would start converting to carbon dioxide and calcium oxide. Another obvious way to

initiate decomposition would be to decrease the partial pressure of C 0 2below its equilibrium value at a given temperature. This is accomplished experimentally by performing reversible decompositions in vacuum. When a reversible decomposition takes place under a continuously rising temperature, its initial stage occurs not far from equilibrium. For such conditions, the basic rate equation (I) needs to be expanded to include a pressure dependent term [6 1,621:

where PCand P,'~ are the partial and equilibrium pressure of the gaseous product C in equation (37). According to equation (42), under rising temperature conditions the rate of the initial stage of decomposition will be predominantly determined by a change in the equilibrium pressure with temperature. Once the system is removed far from equilibrium (P,'~>>PC), the pressure dependent term becomes irrelevant so that the temperature dependence of the rate will follow the basic rate equation (1). Therefore, in a wide temperature region, the temperature dependence of rates of reversible reactions tends to demonstrate significant deviations from Arrhenius behaviour. As a result, the effective activation energy tends to vary with temperature as well as with the extent of reaction. Pavlyuchenko and Prodan [63,64] have shown that the effective activation energy of a reversible decomposition can be expressed as:

where E2 is the activation energy of the reverse reaction, A. is the heat of adsorption, rn is a constant (0 < m 5 l), Q is the thermal effect of reaction. Under the rising temperature conditions the value of r;fY continuously increases and causes the last term of the sum in equation (43) to decrease. For the initial stages of the process (i.e., P;~ w PC), the last term in equation (43) is large so large values of the effective activation energy are obtained. In the later stages (i.e., p,'4 >> PC), the last term in equation (43) converges to 1 so that the values of the effective activation energy are close to the activation energy of the forward reaction. The resulting decrease in the effective activation energy is frequently detected when using an isoconversional method for analysis of reversible processes. Figure 17 demonstrates a dependence of the E, value on the extent of reaction for the thermal dehydration of calcium oxalate

monohydrate. The excellent agreement of the E, dependencies reported in three different papers [I 1,65,66] should be noted. A similar type of dependence has also been reported for the thermal dehydration of lithium sulfate monohydrate [67,68].

5.2. Thermal and thermo-oxidative degradation of polymers Degradation of polymers involves scission of the bonds between the individual atoms of a chain. For typical vinyl polymers, that means scission of C-C bonds whose energy is around 350 kJ mol-I. Although this energy is quite large, thermal degradation of vinyl polymers starts rather easily above 200 "C. This is typically explained by the presence of the weak links in the polymer chain. Such links may include head-to-head links, hydroperoxy and peroxy structures that serve as spots where thermal degradation is initiated. The formation of the initial macro radicals is followed by degradation via various radical pathways whose activation energies are markedly smaller than that for the C-C bond scission. Because of the change from the process of radical initiation to propagation, the effective activation energy of thermal degradation tends to change throughout the process. It tends to be lower at earlier stages, which are determined by initiation at the weak links. Once the weak links are exhausted, the effective activation energy rises at the later stages representative of propagation. This variation has been clearly demonstrated by Peterson et al. [69], who applied an advanced isoconversional method (equation (1 1)) to the TG data on the thermal degradation of poly(styrene), poly(ethylene), and poly(propy1ene). The E, dependencies obtained have increasing character [69] as shown in Figure 18 [70]. However, a more complex E, dependence has been observed by Peterson et al. [71] for the thermal degradation of radically polymerized poly(methy1 methacrylate). The dependence demonstrates an increase (60 to 190 kJ mol-I) followed by a quick falling off (190 to 60 kJ mol-') and another increase (60 to 230 kJ mol-'). The complex behaviour is associated with existence various weak links (e.g., the head-to-head linkages, the vinylidene end groups) that give rise to different mechanisms of initiation. In an oxygen containing atmosphere, polymers undergo thermo-oxidative degradation. As a highly reactive radical, oxygen easily initiates low activation energy pathways for thermal degradation so that thermo-oxidative degradation starts at temperatures about 100 "C lower than the respective thermal degradation in an inert atmosphere. Thermo-oxidative degradation of vinyl polymers occurs via the formation of the hydroperoxide radical in the propagation step of the process. For this reason, thermo-oxidative degradation tends to yield activation energies not far from 100 kJ mol-', which is characteristic of the bimolecular decomposition of organic hydroperoxides. The corresponding activation energies are obtained (Figure 18) when applying an

isoconversional method to TG data on the thermo-oxidative degradation of vinyl polymers [69].

Figure 18. E, dependencies for Figure 19. Effect of a nitro-group in thermal (squares) and thenno- amines on the E, values for the oxidative (pentagons) degradation of epoxy-amine curing reaction. poly(styrene) 5.3. Crosslinking Highly crosslinked polymers are commonly produced from epoxy materials such as diglycidyl ether of bisphenol A (DGEBA). DGEBA is easily copolymerized with various substances such as amines or anhydrides. Copolymerization of DGEBA with a monoamine yields linear polymer chains. When DGEBA is copolymerized with a diamine, polymer chains crosslink. The kinetics of crosslinking (curing) are usually complex because of the multitude of steps involved in the process. For example, copolymerization of DGEBA with a diamine involves two steps associated, respectively, with the two hydrogens of a primary amine. Reaction of the first hydrogen with an epoxy group produces a secondary amine as a part of the growing polymer chain. The second hydrogen becomes less accessible and, thus, less reactive with respect to another epoxy group. As a result, the initial stages of copolymerization involve, for the most part, chain extension due to the primary amine reaction. On the other hand, crosslinking associated with the secondary amine reaction tends to take place at later stages. In addition to chemical steps, epoxy-amine curing causes dramatic physical changes of the reaction medium. The changes include an increase in the molecular weight, the viscosity and the glass-transition temperature of the forming polymer. Molecular mobility, however, experiences a dramatic decrease, especially due to crosslinking. The crosslinked chains

cannot move one past another so that the reaction medium turns into a glassy or rubbery solid. The glassy solid is formed when the glass-transition temperature of the medium rises above the current actual temperature of the process. A dramatic decrease in the molecular mobility changes the kinetics of curing to a diffusion mode that is controlled by mass transport of the reactants. The complex kinetic behaviour is revealed by isoconversional methods as a variation of the effective activation energy with the cure progress. A comparison with other kinetic methods reveals [72] the advantage of isoconversional methods in detecting the complexity of the curing kinetics. Yet another benefit is that isoconversional methods yield consistent values of the activation energy for isothermal and nonisothermal conditions [72,73]. Care must, however, be exercised [74] to avoid systematic errors in isoconversional calculations when isothermal curing is carried out below the limiting glass transition temperature, i.e., when the ultimate degree of curing changes with the curing temperature. The application of isoconversional methods to DSC data on epoxy curing frequently demonstrates a characteristic decrease in the E, values in the later stages of the process. For instance, Vyazovkin and Sbirrazzuoli [75] have reported a decrease from 60 to 40 kJ mol-' at a > 0.6. The lower values of E, are characteristic of diffusion of small molecules in a liquid/solid medium, and the effect can be explained by diffusion control associated with vitrification. A correlation of the decrease in E, with a decrease in molecular mobility has been established experimentally [76] by using temperature modulated DSC. A similar correlation has been reported by other workers [77,78] who also combined an isoconversional method with temperature modulated DSC. The correlation of a decrease in the E, values with vitrification can also be demonstrated chemically. For instance, Sbirrazzuoli et al. [73] have applied an isoconversional method to curing DGEBA with 1,3-phenylene diamine (m-PDA). They have analyzed two systems: a stoichiometric one that had two moles of DGEBA per mole of the amine and a nonstoichiometric one that had a fivefold excess of the amine. An excess of amine favors chain extension via the primary amine reaction. As a result, the E, value (-55 kJ mol") does not change with the extent of curing, which suggests that a single step determines the rate of the overall process. In the stoichiometric system, the E, value of the initial stages is similar to that found in the nonstoichiometric system. However, a contribution of crosslinking becomes increasingly important at later stages so that the E, value drops quickly to low values typical of diffusion. The decrease has been reported [73] to correlate with an increase in the shear modulus and a decrease in the complex heat capacity. The earliest stages of curing (a+O) frequently demonstrate large E, values that quickly drop to the regular values of the activation energy around 50 - 60 mol-I. This phenomenon has been discussed by Vyazovkin and Sbirrazzuoli [79], who

attributed it to the temperature dependence of viscosity in the systems of high viscosity. The hypothesis has been experimentally supported by the fact that at a+O the E, values determined from DSC data are similar to the activation energy of viscous flow estimated from rheological data. Alternatively, the rapid decrease in E, in the early stages is frequently interpreted as being due to a change from a non-catalyzed to an autocatalytic mode. Although it may be a plausible explanation, one needs to make sure that the values of the activation energy of the non-catalyzed reaction (i.e., E, at a+O) are not unreasonably large. A reasonable reference value for the uncatalyzed reaction can be found in paper by Swier et al. [80], who have demonstrated that the cure initiated by the reaction of the primary amine with epoxy-aniline complex has an activation energy of 80 kJ mol-'. The E, dependencies can also provide valuable insights into the chemical mechanisms. For example, by applying an isoconversional method to an epoxyanhydride curing reaction catalyzed by tertiary amine, Vyazovkin and Sbirrazzuoli [81] have obtained an E, dependence that increases from 20 to 70 kJ mol-' with increasing a. For the same reaction without anhydride they have found a practically constant value E, 20 kJ mol" that provides an estimate of the activation energy of initiation. Consequently, 70 kJ mol-' gives an estimate of the activation energy for propagation. Also, the use of an isoconversional method for kinetic analysis of curing of diglycidyl ether of 4,4'-bisphenol (DGEBP) with nitro-substituted phenylenediamine has allowed Zhang and Vyazovkin [82,83] to discover a strong effect on the amine reactivity of the position of the nitro-group. This effect manifests itself as a change in the activation energy from 50 to over 100 kJ mol-I. Figure 19 demonstrates that the presence of a nitro-group next to an amino-group in 2,4-dinitroaniline (2,4DNA) causes E, to reach values above 100 kJ mol", whereas in the 4-nitro-1,2phenylenediamine (4-NPDA)/DGEBP system the curing process has E, of about 50 kJ mol-I, which is a typical value for epoxy-amine reactions. This obviously suggests that a nitro-group separated from an amino-group by more than two carbons has no effect on the curing process. In its turn, a 3-nitro-1,2phenylenediamine (3-NPDA)/DGEBP system presents an intermediate case, where one of the amino-groups is located next to the nitro-group and another is separated by more than two carbons. As a result, the system demonstrates an increase in E, from 50 to 100 kJ mol-I, which reflects progress of the reaction from the amine in position 1 to the amine in position 2.

-

6. ISOCONVERSIONAL METHODS AND THE KINETIC TRIPLET 6.1. Is it really needed? It is a commonly-held view that a kinetic description is incomplete until the whole kinetic triplet has been determined. For this reason, isoconversional methods are commonly criticized as being incapable of directly determining the pre-exponential factor and the reaction model. These two components of the triplet typically appear in isoconversional equations in a conjoint form (cf., equation (5)) so that their separation ultimately requires one to choose the reaction model. Although the pre-exponential factor is usually determined by choosing the reaction model, it should be stressed that a method of its evaluation in a model-independent way also exists [17]. Once the pre-exponential factor is determined, one can also numerically reconstruct the reaction model. The respective methods are illustrated below (section 6.3). Nevertheless, it is our opinion [84] that, compared to the activation energy, the pre-exponential factor and reaction model contribute little extra to understanding the process kinetics. The real need in determining the whole kinetic triplet is that in general it is believed to be necessary for making kinetic predictions. However, this belief is false as can be concluded from the following section that demonstrates that successful kinetic predictions can be accomplished without the pre-exponential factor and reaction model. 6.2. Isoconversional kinetic predictions Kinetic analysis allows one to solve an important practical task of predicting thermal stability of materials outside the temperature region of experimental measurements. Thermal stability can be evaluated as the time to reach a certain extent of conversion at a given temperature. Rearrangement of equation (4) gives

where t, is the time to reach the extent of conversion a at a given temperature, To, in an isothermal run. The major problem of equation (44) is that it is typically used in conjunction with a single heating rate method for evaluating kinetic triplets. Since such methods tend to yield significantly differing kinetic triplets for the same dataset, substitution in equation (44) of the triplets obtained results in highly erratic predictions [85]. The problem is partially addressed in the ASTM E698 method [86] that utilizes the following predictive equation:

where the value of E is determined by the Kissinger method (equation (13)). The method uses multiple heating rates and yields a model-independent value of E. However, as mentioned earlier, the application of this method is problematic in the case of complex (multi-step processes) that cannot be described by a single value of E. Also, equation (45) assumes explicitly that the process kinetics obey the first-order model, g(a)=-ln(1-a). The same assumption is made [86] to determine the pre-exponential factor as

On the other hand, reliable kinetic predictions can be accomplished in entirely model-free way by using the dependence of E, on a determined by an isoconversional method. The relevant predictive equation [17,87] was originally obtained in the following form

and later modified to employ data from arbitrary heating programmes, as follows

The respective predictions can be called "model-free predictions", because they do not require the reaction model to be used explicitly in the numerator of equation (47) and (48). Note, that the pre-exponential factor is also unnecessary for making predictions by equation (47) and (48).

It has been experimentally demonstrated [17,75,85] that the model-free equations give rise to reliable predictions, whereas substitution of the kinetic triplets, obtained from a single heating rate run, into equation (44) yields fundamentally erroneous predictions. It has also been shown that the model-free predictions are superior to the predictions based on the ASTM method (equation

t / min

a

Figure 20. Isoconversional (2) and Figure 21. Evaluating log& by equation ASTM (3) predictions against actual (50) from E, data on decomposition of data (1). Inset: E, dependence for the ammonium nitrate curing process. (45)). Figure 20 provides an example of using equations (45) and (47) for predicting the curing progress of an epoxy material at To=lOO "C. It is seen that the model-free prediction (equation (47)) compares very well with the actual measurement, whereas the ASTM prediction deviates markedly from the experimental data. This example emphasizes the importance of accounting for reaction complexity when making kinetic predictions. In equation (47), this complexity is accounted for by using an isoconversional dependence of E, on a determined for the curing process (Figure 20). Note that according to equation (47), each value oft, is predicted by using the respective value of E, that varies with a from -90 to -40 kJ mol-'. On the other hand, ASTM predictions rely on the Kissinger method (equation (13)) that treats any process as a simple singlestep reaction, which can be represented by a single value of the activation energy. For the process considered this value is 63 kJ mol-'. This value is

consistent with the E, value related to a = 0.45, which is the extent of conversion accomplished at the DSC peak maximum for this process. 6.3. Evaluating the pre-exponential factor and the reaction model. The fundamental flaw of single heating rate methods is that they produce significantly differing kinetic triplets, most of which provide quite satisfactory description of the same dataset [85]. This occurs because of the mutually compensating correlation of E and A. Known as a compensation effect, this correlation takes the following form log A, = aE, - b

(49)

where a and b are constants, and Ai and Ei are Arrhenius parameters associated with a particular reaction model gi(a) orJ(a). Table 1 provides an example of a compensation effect for the thermal decomposition of ammonium nitrate [88]. It is readily noticeable that larger values of E correspond to larger values of logA and the other way around. Note that an increase in A is equivalent to an increase in the rate, whereas an increase in E causes the rate to decrease (see equation (1)). Therefore, almost the same rate can be maintained by simultaneously increasing or decreasing both parameters. Table 1. Kinetic triplets for the decomtlosition of ammonium nitrate r881 i Reaction Model dal EJkJ mol-' l o g ~ ~ ~ l m i n - ' ~ 114 1 power law a 11.5 -0.2 1I3 2 power law a 17.7 0.6 112 3 power law a 30.1 2.0 3 I2 4 power law a 104.5 10.2 5 one-dimensional diffusion a2 141.6 14.2 -ln(l - a ) 8 1.5 8.2 6 Mampel (first order) 114 7 Avrami-Erofeev [-ln(1 - a)] 15.1 0.4 8 Avrami-Erofeev [-ln(1 - a)]" 22.5 1.3 112 9 Avrami-Erofeev [-ln(1 - a)] 37.2 3.1 10 three-dimensional diffusion [ l - (1 156.7 15.3 1 1 contracting sphere 1 - (1 - a)Ln 74.8 6.8 1 - (1 - ' 72.4 6.6 12 contracting cylinder One of the useful properties of the compensation effect is that it includes the correct values of the Arrhenius parameters [89]. Therefore, if the value of E is known, one can substitute it into equation (49) to determine logA. For

isoconversional methods, the use of the compensation effect allows one to estimate the pre-exponential factor in a model-independent way, i.e., without choosing a suitable reaction model [17]. For instance, the application of an isoconversional method to the thermal decomposition of ammonium nitrate results in the dependence of E, [88] shown in Figure 21. On the other hand, the logAi and Ei values (Table 1) demonstrate a compensation effect of the following functional form:

By substituting the values of E, (Figure 21) into equation (50), one obtains estimates for the preexponential factor, log& (see Figure 21). Once both E, and logA, dependencies are known, it is possible [17] to numerically reconstruct the reaction model in either integral or differential form. The integral form is reconstructed by substituting the values E, and A, into equation (5 1):

where T, is the experimental value of the temperature corresponding to the conversion a at given heating rate, P. The differential form can be reconstructed by substituting the values E, and A , in rearranged equation (1):

The models are reconstructed in the numerical form, i.e., as a set of g(a) orAa) values corresponding to different values of a. The analytical form (i.e., equation) can be established by plotting the g(a) or A a ) values against the theoretical dependencies obtained from the model equations (e.g., Table 1). Figure 22 provides an example of reconstructing the integral reaction model for the thermal decomposition of ammonium nitrate. It is seen that no single model from Table 1 fits perfectly the numerical values of g(a) obtained by substituting the values E, and A, (Figure 21) into equation (5 1). The first-order model seems to fit data well at a < 0.6, whereas the Avrami-Erofeev model appears to be the best fit at a > 0.6. This situation reflects the general problem of model-fitting procedures. No matter how large the pool of models (Table 1) is, there is no assurance that the correct model is included. Nevertheless, the use of this

procedure has allowed Zhou et al. [90] to determine that the first step of dehydration of the drug, nedocromil sodium trihydrate, follows the single zeroorder kinetic model, g(a)=a. This procedure has also been helpful in establishing that crystallization of nifedipine from the glassy state obeys the Avrami-Erofeev model [9 11.

Figure 22. Experimental values of g(a) obtained by equation (5 1) (squares) and theoretical dependencies (lines). Numbers represent models fiom Table 1

In conclusion, a word of caution must be voiced about reconstructing the reaction model. The procedures described (equations (5 1) and (52)) are based on the assumption that a single-step kinetic equation (1) or (6) holds for the process under study. This assumption can be valid only if the experimental values of E, demonstrate no significant systematic variation with a (cf., Figure 21). If E, varies significantly with a , the process involves multiple steps so that establishing a single reaction model for it has little sense. 7. CONCLUSIONS Isoconversional kinetics is an efficient compromise between the common single-step Arrhenius treatment and the predominantly encountered processes whose kinetics are multi-step and/or non-Arrhenius. Isoconversional methods are capable of detecting and handling such processes in the form of a

dependence of the effective activation energy on the extent of conversion and/or temperature. Although the resulting activation energies usually are effective or composite values, they can be employed successfully for making reliable kinetic predictions, for learning about complex mechanisms, and, ultimately, for accessing intrinsic kinetic parameters. 8. REFERENCES 1. M.E. Brown, M. Maciejewski, S. Vyazovkin, R. Nomen, J. Sempere, A. Burnham, J. Opfermann, R. Strey, H.L. Anderson, A. Kemmler, R. Keuleers, J. Janssens, H.O. Desseyn, C.R. Li, T.B. Tang, B. Roduit, J. Malek and T. Mitsuhashi, Thermochim. Acta, 355 (2000) 125. M. Maciejewski, Thermochim. Acta, 355 (2000) 145. 2. A.K. Burnham, Thermochim. Acta, 355 (2000) 165. 3. H. Friedman, J. Polym. Sci. C, 6 (1964-65) 183. 4. 5. C.D. Doyle, J. Appl. Polym. Sci. 6 (1962) 639. J.H. Flynn and L.A. Wall, J. Res. Nat. Bur. Standards A, 70 (1966) 487. 6. 7. T. Ozawa, Bull. Chem. Soc. Japan, 38 (1965) 1881. A.W. Coats and J.P. Redfern, Nature, 201 (1964) 68. 8. N. Sbirrazzuoli, Y. Girault and L. ElCgant, Thermochim. Acta, 9. 293 (1997) 25. 10. S. Vyazovkin and D. Dollimore, J. Chem. Inf. Comp. Sci., 36 (1996) 42. 11. S. Vyazovkin, J. Comput. Chem., 22 (2001) 178. 12. S. Vyazovkin, J. Comput. Chem., 18 (1997) 393. 13. P. Simon, P.S. Thomas, J. Okuliar and A.S. Ray, J. Therm. Anal. Calorim., 72 (2003) 867. 14. P. Budrugeac, J. Therm. Anal. Calorim., 68 (2002) 131. 15. H.E. Kissinger, Anal. Chem., 29 (1957) 1702. 16. T. Kujirai and T. Akahira, Sci. Papers Inst. Phys. Chem. Res. Tokyo, 2 (1925) 223. 17. S. Vyazovkin, Int. J. Chem. Kinet. 28 (1996) 95. 18. S. Vyazovkin, Int. Rev. Phys. Chem., 19 (2000) 45. 19. A.K. Galwey, Thermochim. Acta, 397 (2003) 249. 20. S. Vyazovkin, Thermochim. Acta 397 (2003) 269. 21. S. Vyazovkin, New J. Chem. 24 (2000) 913. 22. J.C. Fisher and D. Turnbull, J. Chem. Phys. 17 (1949) 71. 23. S. Vyazovkin, Macromol. Rapid Commun. 23 (2002) 771. 24. S. Vyazovkin and N. Sbirrazzuoli, J. Phys. Chem. B, 107 (2003) 882. 25. S. Vyazovkin, J. Stone and N. Sbirrazzuoli, J. Therm. Anal. Calorim., 80 (2005) 177. 26. S. Vyazovkin and I. Dranca, Macromol. Chem. Phys., 207 (2006) 20.

P. Supaphol, N. Dangseeyun, P. Srimoaon and M. Nithitanakul, Thermochim. Acta, 406 (2003) 207. D.S. Achilias, G.Z. Papageorgiou and G.P. Karayannidis, J. Polym. Sci. B, 42 (2004) 3775. N. Somrang, M. Nithitanakul, B.P. Grady and P. Supaphol, Eur. Polym. J., 40 ( 2004) 829. M.Q.Ren, J.B. Song,Q.X.Zhao,Y.S.Li,Q.Y.Chen,H.F.Zhangand Z.S. Mo, Polym. Int., 53 (2004) 1658. P. Supaphol, N. Dangseeyun and P. Srimoaon, Polymer Test., 23 (2004) 175. D.S. Achilias, G.Z. Papageorgiou and G.R. Karayannidis, Macromol. Chem. Phys. 206 (2005) 1511. J.L. Cai, Y. Han, Z.R Yuan, R.S Cheng, Z.L. Wang, L.S Jiang, J.F. Bi and G.X. Li, Polym. Int., 53 ( 2004) 1127. M.Q. Ren, Q.Y. Chen, J.B. Song, H.L. Zhang, X.H. Sun, Z.S. Mo, H.F. Zhang, X.Q. Zhang and L.S. Jiang, J. Polym. Sci. B, 43 (2005) 553. J.L. Cai, T. Li, Y. Han, Y.Q. Zhuang and X.Q. Zhang, J. Appl. Polym. Sci., 100 (2006) 1479. M.T. Run, J.G. Gao and Z.T. Li, Thermochim. Acta 429 (2005) 171. G.Z. Papageorgiou, D.S. Achilias, D. N. Bikiaris and G.P. Karayannidis, Thermochim. Acta, 427 (2005) 117. X.P. Xia, C.S. Xie and S.Z. Cai, Thermochim. Acta, 427 (2005) 129. M.Q. Ren, J.B. Song, C.L. Song, H.L. Zhang, X.H. Sun, Q.Y. Chen, H.F. Zhang and Z.S. Mo, J. Polym. Sci. B, 43 (2005) 3231. E. Botines and J. Puiggali, Eur. Pol. J., 42 (2006) 1595. M.T. Run, Y .J. Wang, C.G Yao and J.G. Gao, Thermochim. Acta, 447 (2006) 13. K.W. Smith, F.W. Cain and G. Talbot, J. Agric. Food Chem., 53 (2005) 303 1. S. Vyazovkin and N. Sbirrazzuoli, Macromol. Rapid Commun. 25 (2004) 733. J.D. Hoffman, G.T. Davis and J.I. Lauritzen Jr., Treatise on Solid State Chemistry. Vol. 3. (Ed. N. B. Hannay) Plenum Press, New York, 1976. p. 497. A. Toda, T. Oda, M. Hikosaka and Y. Saruyama, Polymer, 38 (1997) 233. A. Toda, T. Arita, C. Tomita and H. Masamichi, Polym. J., 3 1 (1999) 790. A.L. Greer, Thermochim. Acta, 42 (1980) 193. E.M. Lifshitz and L.P. Pitaevskii, Physical Kinetics, Pergamon, Oxford, 1981. I. Dranca and S. Vyazovkin, unpublished data.

C.T. Moynihan, A.J. Eastel, J. Wilder and J. Tucker J. Phys. Chem., 78 (1974) 2673. C.A. Angell, R.C. Stell and W. Sichina, J. Phys. Chem., 86 (1982) 1540. D. Lacey, G. Nestor and M.J. Richardson, Thermochim. Acta, 238 (1994) 99. B.C. Hancock, C.R. Dalton, M.J. Pika1 and S.L. Shamblin, Pharm. Res., 15 (1998) 762. S. Vyazovkin, N. Sbirrazzuoli and I. Dranca, Macromol. Rapid Commun., 25 (2004) 1708. I.M. Hodge, J. Non-Cryst. Solids, 169 (1994) 2 11. S. Vyazovkin and I. Dranca, J. Phys. Chem. B, 108 ( 2004) 11981. K. Chen, M.A. Susner and S. Vyazovkin, Macromol. Rapid Commun., 26 (2005) 690. S. Vyazovkin and I. Dranca, Pharm. Res., 23 (2006) 2158. S. Vyazovkin, N. Sbirrazzuoli and I. Dranca, Macromol. Chem. Phys., 207 (2006) 1126. Handbook of Chemistry and Physics, 53rded, (Ed. R.C. Weast), Chemical Rubber Pub. Co., Cleveland, OH, 197211973, F-66. A.W. Searcy and D. Beruto, J. Phys Chem. 82 (1978) 163 J.M. Criado, M. Gonzalez, J. Malek and A. Ortega, Thermochim. Acta, 254 (1995) 121 M.M. Pavlyuchenko and E.A. Prodan, Doklady Akad. Nauk SSSR, 136 (1961) 651. M.M. Pavlyuchenko and E.A. Prodan, in 5th Int. Symp. Reactivity Solids, Munich, 1964, Elsevier, Amsterdam, 1965, p. 409. E. Urbanovici and E. Segal, Thermochim. Acta, 94, (1985) 399. Y. Masuda, Y. Ito, R. Ito and K. Iwata, Thermochim. Acta., 99 (1986) 205. J. Huang and P.K. Gallagher, Thermochim. Acta., 192 (1991) 35. H. Tanaka, N. Koga and J. Sestak, Thermochim. Acta, 203 (1992) 203. J.D. Peterson, S. Vyazovkin and C.A. Wight, Macromol. Chem. Phys., 202 (2001) 775. K. Chen and S. Vyazovkin, unpublished data. J.D. Peterson, S. Vyazovkin and C.A. Wight, J. Phys. Chem. B, 103 (1999) 8087. S. Vyazovkin and N. Sbirrazzuoli, Macromol. Chem. Phys., 200 (1999) 2294. N. Sbirrazzuoli, S. Vyazovkin, A. Mititelu, C. Sladic and L. Vincent, Macromol. Chem. Phys., 204 (2003) 1815. S. Vyazovkin and N. Sbirrazzuoli, Macromol. Rapid Commun., 2 1 (2000) 85.

S. Vyazovkin and N. Sbirrazzuoli, Macromolecules, 29 (1996) 1867. N. Sbirrazzuoli and S. Vyazovkin, Thermochim. Acta, 388 (2002) 289. E. Leroy, J. Dupuy and A. Maazouz, Macromol. Chem. Phys., 202 (2001) 465. L.A. Mercado, G. Ribera, M. Galia and V. Cadiz, J. Polym. Sci. A, 44 (2006) 1676. S. Vyazovkin and N. Sbirrazzuoli, Macromol. Chem. Phys., 201 (2000) 199. S. Swier, G. van Assche, B. van Mele, J. Appl. Polym. Sci., 91 (2004) 2798. S. Vyazovkin, N. Sbirrazzuoli, Macromol. Rapid Commun., 20 (1999) 387. Y.X. Zhang and S. Vyazovkin, Macromol. Chem. Phys., 206 (2005) 1084. Y.X. Zhang and S. Vyazovkin, Macromol. Chem. Phys., 206 (2005) 1840. S. Vyazovkin, J. Them. Anal. Calorim., 83 (2006) 45. S. Vyazovkin and C.A. Wight, Thermochim. Acta, 3401341 (1999) 53. Standard Test Method for Arrhenius Kinetic Constants for Thermally Unstable Materials (ANSIIASTM E698 - 79), ASTM, Philadelphia, 1979. S.V. Vyazovkin and A.I. Lesnikovich, Russ. J. Phys. Chem., 62 (1988) 1535. S. Vyazovkin, J.S. Clawson and C.A. Wight, Chem. Mater., 13 (2001) 960. S. Vyazovkin and W. Linert, Chem. Phys., 193 (1995) 109. D. Zhou, E.A. Schmitt, G.G.Z. Zhang, D. Law, C.A. Wight, S. Vyazovkin and D.J.W. Grant, J. Pharm. Sci., 92 (2003) 1367. D. Zhou, E.A. Schmitt, G.G. Zhang, D. Law, S. Vyazovkin, C.A. Wight and D. J.W. Grant, J. Pharm. Sci., 92 (2003) 1779.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

THERMOCHEMISTRY M.V. Roux and M. Temprado Institute of Physical-Chemistry "Rocasolano", C.S.I.C. Serrano 119,28006-Madrid, Spain 1.

INTRODUCTION

1.1. The objectives of thermochemistry Two basic properties of chemical compounds are the structures and the energies of their molecules. These are intimately related because the energy associated with a particular structure depends on the atoms, types of bonds and angles that form the structure. The thermochemist is interested in enthalpy changes accompanying reaction, but even more in the enthalpies of formation of compounds from their elements. This valuable and fundamental thermodynamic property of a material is defined as the enthalpy change that occurs upon the formation of a compound from its component elements in their standard states, at a determined temperature of reference, usually 298.15 K, and a standard pressure, currently taken to be 101.325 kPa (1 atmosphere). Values of enthalpies of formation provide a measure of the relative thermochemical stabilities of molecules, intimately related to their structures. It is, however, necessary to eliminate the intermolecular and network energies and to refer the enthalpies of formation to the gaseous state. From values of the enthalpies of formation of the molecules in gaseous state, the steric, electronic and electrostatic effects of different substituents can be evaluated. Comparison of the enthalpies of formation of isomeric compounds is particularly useful because it shows their relative stabilities and provides evidence on the interactions that are responsible for the enthalpy of formation of each isomer. For example, Figure 1 depicts graphically the enthalpies of formation of the three isomers of formula C5H12, pentane [I], 2-methylbutane [I] and 2,2-dimethylpropane [I]. All of them are more stable (have lower enthalpy) than five carbon atoms and six hydrogen molecules in their standard states. Nevertheless, the enthalpies of formation of the pentane isomers reveal that 2-methylbutane is more stable than pentane by 6.9 kJ mol-' and the most branched hydrocarbon, 22-dimethylpropane, is 21.1

kJ mol-' more stable, which shows that the stability increases with branching of the hydrocarbon.

Figure 1. The enthalpies of formation of pentane, 2-methylbutane and 2,2dimethylpropane, illustrating the use of A,H: values to determine the relative stabilities of isomeric compounds [2]. (With permission from Macmillan.) By combining values of the enthalpies of formation of reactants and products, the enthalpy changes for every possible chemical reaction can be calculated. When these thermochemical data are combined with entropy data, Gibbs energies and equilibrium constants can be also determined. This chapter is mainly focused on experimental thermochemistry using calorimetric methods for the study of organic compounds. The main differences between organic and inorganic thermochemistry are related to the states of the compounds and to the experimental techniques. Organic thermochemistry is concerned with molecules in all the states of matter, but more especially with the gas state, and high precision combustion calorimetry is considered to be the best method to determine the enthalpy of formation. Inorganic thermochemistry is concerned mainly with materials in the solid state and reaction calorimetry, especially in solution, is used. Organic thermochemistry frequently needs data obtained from inorganic thermochemistry, because of the use of inorganic reactants, whose enthalpies of formation must be known.

1.2. Short historical introduction Considering that thermochemical research is concerned with the study of the thermal effects of chemical reactions and of associated physical processes involving compounds of well defined composition [3,4], it is important to consider when the concept of quantity of heat was introduced in science. The concept of heat as a property that can be measured was developed by Joseph Black (1728-1799). In 1890 Berthelot [5] considered that Lavoisier and Laplace were the ones who in their publication Me'moire sur la chaleur [6] had established the fundamentals of thermochemistry. This is supported by Partington in his History of Chemistry [7]. The word "thermochemistry" initially written as thermo-chemistry is due to Germain Hess and is found for the first time in the title of a report (Recherches thermo-chimiques), written in French, that was presented at the Imperial Academy of Sciences of St. Petersburg in 1840 [8] and was subsequently published in their bulletin [9]. In that report, he also established the so-called Hess's Law [9] that enables thermochemical values that are difficult to obtain directly, to be calculated. In 1890, Ostwald [lo] considered Hess to be the founder of thermochemistry, although experiments on thermochemistry had been initiated long before then. In the 1920s the developing chemical technology required reliable physicochemical data. The accuracy of the thermochemical data available was hardly adequate for the calculations of enthalpies of chemical reactions and completely insufficient for the calculation of equilibrium constants. If reliable values of equilibrium constants were to be calculated, the accuracy of the thermochemical data would have to be increased by about one order of magnitude. The classical period of thermochemistry may be considered to have ended with this understanding. Modern thermochemistry was inaugurated when the National Bureau of Standards (NBS), Washington D.C., published a series of articles with new standards of accuracy. Scientists with extraordinary experience in thermometry and calorimetry were working in the Heat division of the NBS. Dickinson had designed a calorimeter afterwards used by Jessup in high precision combustion calorimetry. Mueller built a new model of the Wheatstone resistance bridge (named the Mueller bridge for the measurement of resistances) and Meyers manufactured high precision platinum thermometers. The Electricity division was responsible of the maintenance of the resistances and batteries of patrons. Washburn gave the first detailed analysis of the processes occurring in combustion experiments in a bomb-calorimeter [l 11, Rossini published in 1931 his accurate value of the enthalpy of formation of water that, later on, was corrected to standard states [12], Jessup and Green determined an accurate value for the energy of combustion of the primary reference compound, benzoic

acid, [13] and the enthalpy of formation of carbon dioxide was determined by Dewey and Harper [14], Jessup [15], Prosen and Rossini [16,17] setting up the fundamentals of modern organic thermochemistry. A thorough study of the history of thermochemistry from 1775 until approximately 1950 has been made by MCdard and Tachoire 141. 2. EXPERIMENTAL DETERMINATION OF THE ENTHALPIES OF FORMATION OF ORGANIC COMPOUNDS

2.1. Introduction One of the aims of the calorimetry is the development of experimental methods for the direct determination of the quantities of heat produced in chemical reactions and in associated physical processes [IS, 191. In the thermochemistry of organic compounds, the experimental determination of the enthalpies of formation can be carried out both by reaction calorimetry and by combustion calorimetry. The differences between these classes of calorimetric experiments are related to the changes produced in the carbon skeleton of the molecules. In reaction calorimetry, the energy or enthalpy of any chemical reaction is determined and, in these reactions, the carbon skeletons of the molecules are generally maintained. In combustion calorimetry, the energy of combustion in an oxygen atmosphere at high pressure is measured and there is a total breakdown of the carbon skeleton.

2.2. Combustion calorimetry 2.2.1. Introduction In a calorimetric experiment, a chemical reaction is carried out under welldefined conditions and the change in energy accompanying the reaction is determined. Some calculations must be done to obtain the energy change under standard-state conditions with each pure reactant and product and in its stable solid, liquid, or gaseous state at the temperature of reference, usually 298.15 K and at a pressure of 101325 Pa (1 atm). A single calorimetric experiment consists of three different parts: a) a calorimetric part where the energy developed in the combustion of the compound under the experimental conditions of the combustion bomb is determined; b) a chemical part, where the initial and final states of the reaction of combustion are characterized with high precision, and c) a third part in which the energy of combustion in the standard state at T = 298.15 K is determined from the results obtained in the other parts. From this value, using Hess's law, the standard enthalpy of formation in the crystalline state can be calculated.

2.2.2. Basic thermodynamicprinciples of combustion calorimetry The processes in combustion calorimetry are based on the first law of thermodynamics 1201:

where Uf and U, are the internal energies of the system in the final and initial states respectively; Q is the quantity of heat evolved between the initial and final states; and W is the work done by the system against the external pressure. In a combustion experiment the heat is determined under constant volume conditions, hence W = 0. Under these conditions:

Therefore, the quantity of heat absorbed or released by the system to the surroundings at constant volume is equal to the change in the internal energy of the system. To adjust the experimental energy change determined in a combustion calorimetric experiment for a given reaction, Qv, to the enthalpy change at constant pressure, we have to use the relation between Qvand Q,:

where An is the change in number of moles of gases during the reaction, R the gas constant (8.3 14472 J K-'mol-') and Tthe absolute temperature:

2.2.3. Combustion calorimeters A combustion calorimeter is an instrument used to determine the energy of combustion of substances in a gas, which is usually oxygen for organic compounds. Samples are burnt in a closed reaction vessel, usually called the bomb, in an oxygen atmosphere at a pressure great enough, usually of 3 MPa, to obtain a complete reaction. The bomb is inside the calorimeter that is usually a can filled with water. There are other types of calorimeters [18,21], such as aneroid calorimeters, where the calorimeter is a metal block and does not contain any liquid. In heat-conduction calorimeters, the heat released by a combustion reaction is quantitatively transferred from the measuring vessel to the surrounding sink (metal block) through a thermopile wall [22,23]. Most combustion calorimeters are of the isoperibol type, which means that the calorimeter is surrounded by a constant-temperature jacket, the thermostat, which is maintained at, or close to, 298.15 K. The calorimeter system is the

combination of the bomb, the water-filled can, and auxiliary equipment that includes the thermometer, the stirrer and the heater. In an isoperibol calorimeter the heat, q, released in the combustion reaction produces an increase of the temperature of the water inside it, which is measured by means of a suitable thermometer (usually a platinum resistance thermometer, a quartz thermometer or a thermistor). The stirrer ensures that the energy released in the combustion reaction is transferred to the calorimeter so that the combustion bomb, the water and the can are quickly brought to the same temperature. The heater is used to take the temperature of the calorimeter to the starting point of the experiment. It is desirable to have an experimental temperature increment of at least 1 K. Figure 2 shows a typical temperature-time curve obtained in an isoperibol combustion calorimeter.

ti

tf

Time

Figure 2. Temperature-time curve obtained in an isoperibol combustion calorimeter. This curve can be divided in three different parts that correspond to three distinct parts of the experiment. These are usually named the fore-period (between t, and ti), the main-period, (between ti and tf), and the after-period (between tf and tb). In the fore-period the temperature increases slowly and uniformly because the temperature of the surrounding is higher that the temperature of the water in the can and there is also an increase of the temperature produced by the heat of stirring and the heat dissipated by the thermometer. When regular increment of the temperature has been recorded, the sample is ignited at the time ti, by discharging of a capacitor through a platinum wire, which ignites a fuse, usually a cotton thread, and the temperature rises rapidly (main period) as a result of the exothermic combustion reaction. The main period ends at tf when the drift rate is once again uniform. The temperature

of the calorimeter is then recorded during an after-period of approximately 20 min and at the time tb the measurements are over. The aim of the measurement of temperature in the fore- and after-periods is to calculate the correction that must be applied to the observed temperature rise taking into account the external energy supplied by the stirring, the thermometer, and the heat exchanged between the calorimeter and its environment. This correction is based on Newton's law of cooling and the procedure has been described in detail by Hubbard et al. [24] and Sunner [25]. If an adiabatic calorimeter is used the correction to the temperature rise is negligible [26]. The greater usage of isoperibol calorimeters is due to the fact that they are easier to construct and operate than adiabatic calorimeters. Between 1881 and 1905, Berthelot and co-workers developed the first combustion calorimeter, precursor of the modem calorimeters with static bomb [27-301. 2.2.4. Calibration of the calorimetric system The energy equivalent of the calorimeter, 4calor) is defined as the amount of energy required to increase the temperature of the calorimeter by 1 K. The most precise determination of 4calor) is based on the transfer of a determined quantity of electrical energy through a heater placed at the same location as the combustion crucible. Because most of the calorimeters used are of the isoperibol type and are not equipped for electrical calibration, a standard reference material, benzoic acid, is used. Its certified energy of combustion in 0 2 must have been measured in an electrically calibrated calorimeter. Because the conditions under which the specific energy of combustion reported on the certificate was determined usually differ from those ones used in combustion calorimeters, certain corrections must be applied [31]. Details of these corrections are given in the certificate.

2.2.5. Standard state corrections When a substance is combusted, the energy released may differ significantly from that expected for the combustion reaction under standard conditions. This problem was studied by Washbum [ l l ] , who analyzed the corrections that must be applied to the calorimetric data in order to obtain values of the standard change in the internal energy. In high-precision combustion calorimetry, it has become standard practice to apply these corrections. The physical effects [32] for which corrections must be applied are: (a) Compression of condensed phases. The substance, auxiliary materials and the water, initially present in the bomb, are under a final pressure greater than 1 MPa and a correction for the energy stored in these species is required. (b) Compression of the gaseous phase. Since neither the initial nor the final gaseous states in the bomb are at very low

pressures, a thermal correction for the non-ideality of the gases is required. (c) Solution of the gases in the aqueous phase. In the initial state some oxygen will be dissolved in the water, or water solution, placed in the bomb, and in the final state some of the gas will be dissolved in the aqueous phases. A thermal correction for dissolution of the gases is required. (d) Vaporization of water. When water or water solution initially is placed in the bomb, the gas-space becomes saturated with water vapour and remains so throughout the experiment. The concentration of water vapour may change between the initial and final conditions. The necessary thermal corrections require a knowledge of the enthalpy of vaporization of water at 25 OC and the number of moles of water present in the initial and final states. (e) Dilution of the aqueous phase. The energy of formation of the final aqueous phase will depend on its concentration. It is necessary to know the energy of dilution from the calculated concentration of the final solution to an arbitrarily selected reference concentration, e.g. H2S04.115H20 in the case of combustion of compounds containing sulfur. (0 Non-isothermal reaction. It is necessary to correct the energy of combustion obtained at the temperature Ti, bU;, ,to 298.15 K by equation (5). Au:n = -Cp,initial (Tfina~ - Tinitia~)+ ACv(298.15 - Tfinal)

(5)

The calculation of the term AC, requires knowledge of the constant-volume heat capacities of the substance(s) to be burned, of the water placed in the bomb, of the oxygen, of the gas mixture produced, and of the final aqueous solution. Washburn's first calculations were made for compounds containing only carbon, oxygen and hydrogen, CaHbOc,however, later these corrections were applied by Prosen [33] to compounds of the general formula C,HbOcNd,and by Hubbard et al. [34] to compounds of the general formula CaHbOcSd. Washburn's corrections for halogen compounds of the general formula CaHbOcCld, CaHbOcBrd,and CaHbOcIdhave been calculated and are given in references [24] and [35]. For compounds containing other heteroatoms that produce solutions on combustion for which the key auxiliary quantities required in Washburn's correction are not available, comparison experiments, for the determination of the energetic equivalent of the calorimeter, have to be done. Comparison experiments [36] are calibration experiments designed so that their final states are close duplicates of those in the main experiment. This is achieved by burning, in the presence of a suitable liquid reagent, a mixture of benzoic acid and another substance of well-known energy of combustion. The combustion of both substances should result in the increment of the temperature, the amount of carbon dioxide and the composition of the final liquid solution formed, being approximately the same as that obtained in the combustion experiment with the compound under investigation. By using this "effective energy equivalent",

determined in this way, to calculate the energy liberated in combustion experiments, any error in the calculation of the correction to standard states is largely eliminated.

2.2.6. Evaluation of errors and accuracy in combustion calorimetry High-precision combustion calorimetry is considered to be one of the most difficult experimental procedures [37]. The precision required in combustion experiments would have to be k0.01-0.02% in order to have an uncertainty in the enthalpy of formation of approximately f1 kJ mol-', which is the precision necessary to obtain reliable thermochemical data. This includes errors in all three different parts of a combustion experiment. In the calorimetric part there are errors in weighing the water in the calorimetric jacket and also in the temperature measurements. In the chemical part there are errors in weighing the sample and in the data for auxiliary materials (benzoic acid, cotton, Vaseline, polythene, etc.), errors in the combustion process caused by production of either carbon monoxide or soot and, in the case of compounds with S or N, errors arising fiom the production of SO and NO instead of SO2 and NOz. Important errors may arise fiom sample impurities, water being one of the most important and difficult, because many compounds are hygroscopic. In the third part there are errors in the corrections to the standard state. Thus errors in any part of experiment should be kept under f0.01%. In a typical experiment, if 500 mg of the compound is used, it must be weighed with an uncertainty of less than 50 pg. Usually the combustion process increases the temperature of the calorimeter by about 2 "C, which must be determined with an uncertainty of less than 0.2 millidegree. To ensure complete combustion, the contents of the bomb are slowly released through a carbon dioxide absorbent, which is then weighed. The ratio of the amount of carbon dioxide found to that expected should be between 1.0001 and 0.9999. All the other experimental variables must be determined equally well. Finally, the data must be reduced to standard conditions (25' and 1 MPa) making use of the Washburn corrections [ l l ] . It is important to emphasize that at least five experiments without detectable errors must be done. For compounds containing sulfur or halogens, which lead to sulfuric acid or hydrogen halides as reaction products, the contents of the bomb must be completely defined and mixed. This requires the use of a rotating oxygen bomb calorimeter [38], which is considerably more complex than a standard oxygen bomb calorimeter. Conventional stainless-steel bombs are not adequate for use with sulfur and halogen organic compounds, and platinum-lined bombs with crucible, electrodes and other fittings made of platinum are not attacked by the gases produced.

In compounds containing sulfur, 10 cm3 of distilled water is added to the bomb, and this, which initially contains air at atmospheric pressure, is assembled and charged with oxygen without previous flushing. The mass of the sample to be burned has to be calculated to have about 8 mmol of sulfur [39]. Under these conditions the nitrogen acts as a catalyst and hence H2SO4 and HN03 are produced. The absence of ~ 0and NO; ~ ~ can be - checked [40]. The generalized combustion reaction for compounds with molecular formula CaHbOcSdis given by the equation (6) [35]: CaHbO,Sd(cd) + (a + bl4 - c12 + 3d12) O2(g) + (nd + d - bl2) Hz0 (1) + a C 0 2 (g) + d [H2S04 n H20] (1)

(6)

The calculation of the enthalpies of formation of organic sulfur compounds from the experimental results for their energies of combustion, requires an accurate knowledge of the enthalpy of formation of aqueous sulfuric acid. Enthalpies of formation of [H2SO4n H20] where n = 1 to co are available from the NBS Tables of Chemical Thermodynamic Properties [41]. The value usually taken for n in combustion calorimetry is n = 115 being A, H ; ([H2S04. 115 H20](1)) = 3887.81 1 f 0.042) kJ mol-'. The combustion in oxygen of organic halogen compounds in a bomb that contains water is represented by the general reactions: a) compounds containing fluorine. C,HbOcNdF,(cd) + (a + bl4 - c12 - e14) 0 2 (g) + [(n + 112) e (1-x) - b/2)1 H20 (1) -,(a - ex14) C02 (g) + ex14 CF4 (g) + e (1-x) [HF - n H201 (1) + dl2 N2 (g) (7) x being the fraction of the total fluorine which produces carbon tetrafluoride. This fraction depends on the ratio e/b and, generally, only fluorine is found if e/b < 1 [35,42]. Values for A,H;(HF . nH20) where n = 1.3 to 400 have been published [43]. b) compounds containing chlorine. CaHbOcCld (cd) + [a+(b-d)/4-cl2) O2 (g) + (nd-(b-d)/2) H20 (1) + a C 0 2(g) + d [HCl - n Hz03 (1)

(8)

A significant proportion of the chlorine contained in the compound gives, as product of the reaction, free chlorine C12 and a solution of arsenious oxide must be used to reduce chlorine to chloride ion [35]. Enthalpies of formation of HC1.nH20 for values of n in the range 3.6 1 to 3300 have been published [43].

c) compounds containing bromine. C,HbO,Brd (cd) + [a+(b-d)/4-c12) 0 2 (g) + (nd-(b-d)/2) H20 (1) + a CO2 (g) + d [HBr . n Hz01 (1) (9) For organic bromine compounds, large quantities of free bromine Br2 are produced and larger quantities of the reducing agent, arsenious oxide, must be used [35]. Values for A,H;(HBr.nH20) where n = 700 - 2100 are listed in reference [43]. d) compounds containing iodine. CaHbO,Id(cd) + [a + (b-d)l4 - c12) 0 2 (g) + a CO2 (g) + bl2 H2O (1) + dl2 I2 (cd)

(10)

In the combustion of organic iodine compounds in oxygen, there are no side reactions that interfere in the main reaction, and only iodine, 12,is produced [35]. For these families of compounds and similarly for organic compounds containing boron, silicon, phosphorous, arsenic or selenium, special care must be taken in the analysis of the chemistry of the combustion reaction and, in order to obtain well-defined final solutions, auxiliary substances must be used [35]. The introduction of chemical reagents that react with some of the products of combustion means that the measured energy of reaction has to be corrected for the energy of side reactions before the energy of combustion of the compound can be calculated. A study of the auxiliary substances and the reactions where they are included, so as to get accurate results in combustion experiments with organic compounds containing heteroatoms, has been done by Head and Good [35] and Minas da Piedade [44]. As a result, fewer data are available for these compounds. It must be remembered that the important quantity is the accuracy, not the precision. The reported precision of a set of measurements refers only to random errors and, in accordance with normal thermochemical practice, the uncertainty assigned is twice the overall standard deviation of the mean and includes the uncertainties in calibration and in the values of the auxiliary quantities [45]. When high precision is achieved, systematic errors may be considerably larger than random errors, leading to an inaccuracy that it is much larger than the uncertainty interval. To check the accuracy of the results obtained, reference materials with the same elements and states that the compound has must be used. Once the enthalpy of combustion of a compound has been obtained, the more useful quantity is the enthalpy of formation of that compound fi-om its elements.

For example, for 2-furancarboxylic acid, using Hess's law, the following thermodynamic cycle may be written [46].

The enthalpy of combustion of graphite (the standard form for carbon) is -393.51 *0.13 kJ mol-' and the enthalpy of combustion of hydrogen gas (the standard form for hydrogen) is -285.830 h 0.040 kJ mol-' [47]. The expected enthalpy of combustion of 5 gram-atoms of graphite and 2 mols of hydrogen would then be -2539.21 kJ mol-'. The difference between this and the observed enthalpy of combustion of 2-hrancarboxylic acid, - 2040.67 kJ mol-', is the enthalpy of formation of 2-furancarboxylic acid from its elements at 298.15 K (25OC), - 498.54 kJ mol-'.

2.3. Reaction calorimetry Reaction calorimetry (other than combustion) [37,48,49] is concerned with the measurement of the enthalpy changes accompanying chemical reactions. Enthalpies of formation of compounds cannot generally be determined directly, but the results obtained from reaction calorimetry may frequently be combined with other data to provide enthalpies of formation. The requirements of precision in the results obtained from reaction calorimetry are considerably less than the ones obtained from combustion calorimetry, because one measures a small quantity directly, instead of obtaining it as the small difference between two large quantities. Enthalpies of reaction in solution are generally measured in an isothermal jacketed calorimeter. This consists of a calorimetric vessel that contains a certain amount of one of the reactants that is either a liquid or, if a solid is involved, it has been dissolved in a suitable solvent. The other reactant is sealed in a glass ampoule that is placed in a holder. The vessel is enclosed in a container, which is placed in a thermostatted bath with the temperature controlled to k 0.001 "C. When the system has reached thermal equilibrium, the ampoule is broken and the reaction is initiated. Throughout the experiments the temperature is measured as a function of the time and a temperature-time curve with approximately the same shape as the ones obtained in combustion calorimetry, with fore-period, reaction-period and after-period is obtained. The observed temperature rise is due to several sources: the heat transferred from the thermostatted bath, the energy of the reaction and the stirring energy. To correct

the temperature increment observed in the experiment to that due to the reaction only, Newton's law of cooling is used. It is necessary to know the heat capacity of the system to calculate the enthalpy of the reaction. This can be determined by measurement of the temperature increment produced by the reaction of a reference material, or by electrical calibration supplying a determined quantity of electrical power from a heater during a known time. There are other types of reaction calorimeters [50], such as adiabatic calorimeters, where the jacket is maintained at the same temperature as the reaction vessel during the whole experiment, and no corrections need to be applied to the observed temperature rise [5 11. In titration calorimeters [52] for a reaction such as:

a determined amount of the compound A in solution is titrated with a solutionof B. During the titration heat is released while there is enough A to react, but when A has been consumed, there is no further reaction and no heat is liberated (and the temperature does not increase). Titration calorimetry has been used to analyze acid-base, oxidation-reduction, precipitation and other types of reactions [53-601.

2.4. Thermochemistry of phase changes 2.4.1.Introduction Organic thermochemistry usually deals with molecules in the gaseous state in order to study their intrinsic stabilities in the absence of a crystal lattice, intermolecular bindings in the liquid state, or solvation forces. Therefore the determination of the enthalpy of vaporization or the enthalpy of sublimation is an essential step in obtaining the enthalpy of formation in the gas phase. The enthalpy of sublimation can be obtained by combination of the enthalpy of vaporization and the enthalpy of fusion (equation (12)). The enthalpy of fusion is easily and reliably obtained by DSC.

Enthalpies of sublimation are also useful for studies of molecular packing in the solid phase and of polymorphism and enthalpies of vaporization provide information on the intermolecular interactions present in the liquid phase.

2.4.2. Experimental techniques Despite the large number of different techniques and procedures used to measure enthalpies of vaporization and of sublimation [18,6 1-99], all measurements can be classified as either: a) direct isothermal calorimetric measurements, or b) indirect procedures. Among direct calorimetric methods there are two main modifications: adiabatic [65-721 and conduction [73-781 calorimeters. Adiabatic calorimeters have been widely employed to determine enthalpies of phase changes for volatile compounds, whereas conduction calorimeters are among the most accurate methods for compounds with low vapour pressures. Modern adiabatic calorimeters employ a technique whereby the enthalpy of vaporization is measured under conditions in which a measured amount of electrical energy is supplied to a heater immersed in the sample to compensate for the heat absorbed by the substance during the evaporation and hence the temperature is kept constant. The main differences among adiabatic calorimeters are that the vapour flows out of the calorimeter at atmospheric pressure (those of Mathews and Fehlandt [65]), into a vacuum, [67,69-711 into a gas stream [68], or into a closed recirculation system with continuous fluid flow [66]. The most used conduction instruments are Calvet calorimeters [73,74], in which the calorimeter has two Knudsen cells, one for the sample and the other as reference. The whole system is thermo-regulated and the heat flow to the reference cell is measured with two thermopiles that surround both cells. A number of sublimation enthalpies have been measured using drop calorimeters [79]. In such experiments, a sample and a reference vessel are dropped into a heated conduction microcalorimeter, evacuated and the output measured. This method is useful with substances that decompose on heating under reduced pressure before a detectable amount of vapour has been formed, and in the study of substances available in small quantities (it requires approximately 10 mg per experiment). Sublimation enthalpies have also been measured using various modifications of commercial differential scanning calorimeters (DSC) and thermogravimetric analysis (TGA) instruments [80], but the accuracy of these methods is lower than the other well-established techniques. The enthalpies of phase changes of low-volatility compounds are not generally determined directly, but are derived from the measured relation between vapour pressure and temperature with the help of the Clausius-Clapeyron equation (13).

where p is the vapour pressure at temperature T, R the gas constant and A a constant. Vapour pressures can be measured directly using a differential ebulliometric method [81,82], or by static measurements with different kinds of manometers [83-851. Among the most reliable methods for measuring the vapour pressures of solids as a function of temperature are the mass loss [86] or torsion [87] Knudsen effusion techniques. The torsion-effusion method is a complementary method to mass-loss effusion and both have often been measured simultaneously [88,89]. This combination provides additional information useful in assessing the presence of association in the vapour because vapour pressure information is provided that is dependent (mass loss) and independent (torsion) of the molar mass of the effusing vapour. Transpiration or gas saturation techniques have been widely used for the measurements of vapour pressures [90,91]. An inert gas is passed over the sample and the amount of material transported as a function of temperature is determined. Different analytical methods have been used to quantify the mass transferred. The main advantages of this method are the large temperature range accessible and the small amount of sample needed (-30 mg) for the experiments. Head-space analysis has also been used by various investigators. The vapour in equilibrium with a solid is either measured directly by an absorption technique [92] or indirectly [93]. Vapour pressures can be also measured indirectly by using a quartz resonator where the frequency of the quartz crystal changes as a function of the thickness of the material deposited on its surface [94,95]. A simple and frequently used technique to determine the vaporization enthalpies of volatile liquids is the isoteniscope method [96,97]. The isoteniscope was devised to trap a small amount of the liquid and its vapour in a part of a vessel that is totally immersed in a thermostatted bath, separated from the measuring device by an open-end manometer which is also immersed in the bath. The liquid in this inner manometer is usually the liquid being studied. The pressure in the outside space is measured by conventional methods. When the inner manometer shows that the outside pressure is the same as that inside (the vapour pressure of the liquid at the bath temperature) the outside pressure is registered. This method has been widely employed to obtain vapour pressures of mixtures [98]. Correlation gas chromatography is an indirect method to determine the enthalpies of vaporization of both solids and liquids [99]. The quantity directly measured is the enthalpy of transference from the condensed state in the column to the gas state. The enthalpy of vaporization is obtained by using the equation obtained by regression analysis between the enthalpies of transference and the

literature data for the reference compounds. The main advantage of this method is that the purity of the sample does not affect the measurements. Where there is a scarcity or complete lack of experimental data, several methods have been evolved to predict enthalpies of phase changes [loo-1031 based mainly on a group contribution approach. 2.4.3. Errors in enthalpy measurements ofphase changes The uncertainties associated with measurements of phase changes of solid and liquid with low vapour pressures (compounds that have large sublimationlvaporization enthalpies) are large and the accuracy is sometimes less than desirable. Furthermore, these measurements have to be performed at elevated temperatures and corrections to T = 298.15 K introduce additional errors in the values. In addition, a lack of standards in this vapour pressure range produces less reliable values. For solids, an important source of discrepancy among literature data is the occurrence of polymorphism and solid-solid phase transitions [104]. In measuring enthalpies of vaporization of liquids, one of the main sources of error is the presence of air dissolved in the liquid. Thus, it is very important to de-gas the sample prior to the experiment. One of the most important reasons to the lack of accuracy is due to systematic errors associated with the measurement. Every technique has some systematic error inherent in the method. The enthalpies of phase changes of many organometallic compounds are not very accurate. The uncertainties associated with such data are usually high and discrepancies among different values obtained by different laboratories and techniques are observed. This can be mainly explained due to the low volatility and thermal instability of some of the organometallic compounds, as well as the lack of appropriate standards.

2.5. Additional techniques Themochemistry is not only concerned with the enthalpies of formation of compounds. Some energies of reaction types, such as ionization potentials, electron affinities, oxidation or reduction potentials, bond dissociation energies and energies associated with acid-base processes are also studied by thermochemical means. The adiabatic ionization energy and the adiabatic electron affinity of a molecule AB are the minimum energies required to remove or attach an electron from the isolated molecule at zero Kelvin (reactions (14) and (15)) respectively. That means that the transition is between the ground states of all molecules present, assumed to be perfect gases with no interaction among them.

The appearance energy of a cation A+ can be defined as the standard enthalpy of reaction (16), where a molecule AB in the gas phase at T = 0 K is ionized and excited to a state AB+* by means of an electron or a photon, followed by decomposition into the fragments A+ and B, provided that A+and B are in their ground states and all the species, including the electron, have zero translational energies.

The bond dissociation enthalpy (BDH or DH298.1S(A-B)) for a species A-B is the enthalpy required for homolytic bond cleavage at T = 298.15 K, reaction (17), and depends exclusively on the relative enthalpies of formation of the reactant and product states.

The bond dissociation energy (BDE or Do(A-B)) is the energy required for the same reaction at T = 0 K and can be related to the depth of the potential well (D,(A-B)) by the zero point energy (ZPE), equation (18) [105]:

The bond dissociation enthalpy is a valuable quantity for deriving enthalpies of formation of organic radicals, provided that all the enthalpies of formation of the neutral species present in the homolytic cleavage reaction are known. For ionic and radical species, using all of the previous reactions following a thermochemical cycle, see Figure 3, various quantities can be derived if the rest of the energetic data present in the cycle are known [105-1091.

A-B

Electron transfer Figure 3. Relationship between heterolysis, homolysis and electron transfer processes. All of the techniques cited in the preceding sections may be used to study stable long-life species, mainly neutral molecules. Thermochemical data for short-lived and unstable molecules, ions and radicals are measured by other means. Some standard analytical techniques, such as NMR, gas chromatography or mass spectrometry, have been used to derive quantitative thermochemical data [110]. Reliable enthalpies of formation of radicals and anions can be derived from ionization potentials, ion intensities and ionic-appearance energies obtained by mass spectrometry [ I l l ] . The main techniques used, based on measurements of ion intensities are: electron photodetachment spectroscopy [112], flowing afterglow [113], guided ion-beam mass spectrometry [114], highpressure mass spectrometry [115], ion cyclotron resonance mass spectrometry [116], kinetic energy release distributions [117], Knudsen cell-mass spectrometry [118], a mass spectrometry-kinetic method [119], photoionization mass spectrometry [120] and pulsed high-pressure mass spectrometry [121]. A large number of bond dissociation enthalpies in solution for the N-H and 0H bonds have been determined by Bordwell and coworkers by the combination of PKH* values with the oxidation potentials of the conjugate bases [122]. Reversible redox potentials for species in solution can be obtained by electrochemical measurements by using techniques such as cyclic voltammetry. From these potentials it is possible to derive Gibbs energies and bond dissociation enthalpies. Kinetic measurements of the Gibbs energies of reactions permit the study of transition states with very short lifetimes by the use of the Eyring equation [123]. Comparing the rates of two compounds undergoing a similar reaction under similar conditions, the free energy difference between the two transition states can be determined by combining the difference between the two ground states of the reactants and the difference between both free energies of activation [124].

Similarly, differences in conformational and isomeric energies can be determined from equilibrium measurements by NMR and gas chromatography [125,126]. Photoacoustic calorimetry is another widely used technique to study the thermochemistry of organic systems in solution [127-1291. Light is used to initiate a reaction of interest, and the heat liberated from the reaction is detected acoustically. From the amplitude of the photoacoustic signal and the solution transmittance, the enthalpies of reaction and bond dissociation enthalpies can be derived. 3. REFERENCE MATERIALS

How can the accuracy of enthalpies of formation be determined? One way is to cany out the measurement in several different laboratories. We would be able to find systematic errors that were the same from laboratory to laboratory. Therefore, if values agree within their uncertainty intervals, the accuracy and the uncertainty interval may be adequate. Unfortunately, such replicate experiments are not frequently available. Thus, it is important to check the results obtained using reference materials. The result obtained with the reference material will confirm that the results are not subject to systematic errors. A reference material should satisfy the following requirements: (a) it has to be easily obtained in a pure state, (b) be stable, (c) be non-hygroscopic, (d) be easily handled for the measurements, (e) be physiologically harmless, ( f ) it must not react with the instrumental material, (g) nor react with the surrounding atmosphere, and (h) photoreactions must not occur. Reference materials for combustion calorimetry have been recommended by the IUPAC Commission "Physicochemical Measurements and Standards" [1301321 and by the ICTAC "Thermochemistry Working Group" [133]. They are classified as primary, secondary and tertiary reference materials and when their properties are certified by a national or international organization, agency or laboratory authorized they are called "certified reference materials". For combustion calorimetry, the recommended reference materials must be selected according to the atoms in the molecule and its physical state [133]: benzoic acid (C,H,O, cr, primary), succinic acid (C,H,O, cr, secondary), hippuric acid (C,H,O,N, cr, tertiary), acetanilide (C,H,O,N, cr, secondary), nicotinic acid (C,H,O,N, cr, tertiary), 1,2,4-triazole (C,H,O,N, cr, secondary)[l34], urea (C,H,O,N, cr, tertiary), thiantrene (C,H,O,S, cr, secodary), 4-fluorobenzoic acid (C,H,O,F, cr, secondary), pentafluorobenzoic acid (C,H,O,Fl, cr, tertiary), 4chorobenzoic acid (C,H,O,Cl, cr, secondary), 4-bromobenzoic acid (C,H,O,Br, cr, tertiary), 4-iodobenzoic acid (C,H,O,I, cr, tertiary), triphenylphosphine oxide

(C,H,O,P, cr, secondary), 2,2,4-trimethylpentane (C,H,O, 1, secondary), a , a , a trifluorotoluene (C,H,F, 1, tertiary). Reference materials [133] for processes described as reaction, dissolution and dilution, except for reactions carried out in combustion bombs, are also included in this section. Potassium chloride is currently used as a reference material for the enthalpy of solid-solution calorimetric determinations, and it is available as a certified material. Other recommended reference materials are: tris(hydroxylmethy1)-aminomethane (secondary), 4-aminopyridine (secondary), sulfuric acid solution + sodium hydroxide solution (primary). 1-propanol, hexafluorobenzene, benzene, n-alkanes and water have been recommended as primary standards for determinations of enthalpies of vaporization [133]. For sublimation enthalpies, benzoic acid, naphthalene, ferrocene, and anthracene have been recommended as primary, pyrene and iodine as secondary, and benzophenone, biphenyl, phenanthrene, trans-stilbene, 1,3,5-triphenylbenzene as tertiary reference materials [I331. 4. THERMOCHEMICAL DATA BASES FOR ORGANIC COMPOUNDS The critical evaluation of published data [135], constitutes the viable end product of experimental thermodynamics for practical and industrial purposes. Before 1939 the available data were the International Critical Tables [I361 and an extensive and uncritical summary of enthalpies of combustion published by Kharasch [137]. Bichowsky and Rossini published a comprehensive, critical, survey of the thermochemistry of C1 and C2 organic compounds published up to 1933 [138]. This work was followed by the American Petroleum Institute [139,140], the manufacturing Chemists Association [I411 and the LandboltBornstein uncritical compilation of thermodynamic data for inorganic and organic compounds [142]. Circular 500 from the National Bureau of Standards lists internally consistent, selected, values of the thermodynamics properties of inorganic and C1 and C2 organic compounds [143]. Stull, Westrum and Sinke published in 1969 a book [144] where thermodynamic data for approximately 4400 pure organic compounds, from 298 to 1000 K are tabulated. One year later Cox and Pilcher published a compilation of thermochemical data, critically examined from original papers for organic and organometallic compounds 131. In 1978 the available thermochemical data for benzene derivatives were analyzed by Cox [145] and in 1985 Pedley published a book with selected values for the standards enthalpies of approximately 3000 organic compounds of the elements C, H, 0, N, S, F, C1, Br, and I [146]. This book was updated in 1994 [I471 with the results re-organized into approximately 350 sets of compounds containing specific functional groups and/or ring systems.

In 1989 the Committee on Data for Science and Technology (CODATA) of the International Council for Science conducted a project to establish internationally agreed values for the thermodynamic properties of key chemical substances [47]. In 1959 the Dow Thermal Laboratory was commissioned to prepare the Joint Army-Navy-Air Force Thermodynamic Tables (JANAF Thermodynamic Tables) that are issued from time to time [148]. TRC Thermodynamics Tables is 14-volume set containing critically evaluated data covering physical and thermodynamic properties of all classes of hydrocarbons and certain classes of sulfur derivatives of hydrocarbons present in petroleum and coal. The hydrocarbons collection contains information on more than 3700 compounds arranged in more than 200 groups [149]. The NIST Chemistry WebBook contains thermochemical data, reaction thermochemistry data, spectra, spectroscopic data, ion energetics data, thermophysical property data [150]. Several compilations of enthalpies of vaporization and sublimation measured by different means can be found in the literature [3,151- 1541. A compilation of bond dissociation energies of organic compounds has recently been published by Luo [155].

5. RECENT DEVELOPMENTS IN EXPERIMENTAL TECHNIQUES 5.1. Combustion calorimetry The major developments in the experimental technique carried out in the last few years have been concerned with the development of micro-scale combustion calorimeters [156- 1641. One of the main limitations of the macro-scale combustion calorimeter is the quantity of sample needed to determine the enthalpy of combustion of a compound of purity degree around 99.9 %. In each experiment samples of mass between 0.5 g to 1 g are used and at least five experiments are needed. The development of synthetic organic chemistry has made available an enormous range of new chemical compounds with fascinating structural features. Knowledge of their enthalpies of formation is important in assessing the relationship between molecular structure and energy. Syntheses of many of these compounds produces only very small amounts, or are time consuming and expensive. In addition, the combustion of samples often becomes a cleaner reaction when small quantities of samples are used. For these reasons, research groups have been directing their efforts toward miniaturization of combustion bombs and calorimeters. The aim is the development of methods that work with samples of a few milligrams and lead to results of approximately the same accuracy as those obtained with macro-bombs. For a long time, an enormous effort has gone into the development of microcalorimeters [165-1761. Recently, several systems working in the range of

microcombustion calorimetry, using highly specialized instrumentation, have been reported [156- 1641. The main problem associated with these microcalorimeters is the difficulty in the analysis of their combustion products. Recently, due mainly to the development of analytical techniques, microcombustion calorimeters with moving bombs have been used to study compounds containing halogens [177-1811, sulfur [74,182-1851 and ferrocene [186]. Because high-precision combustion calorimeters equipped with rotating macrobombs are not commercially available, a new rotating-bomb combustion calorimeter has recently been re-designed [187] and has yielded good results in combustion experiments with the reference compound thianthrene. A new state-of-the-art reference gas calorimeter has been developed recently for the more accurate determination of the calorific values of pure hydrocarbons, as well as for fuel-gas mixtures [188].

5.2. Enthalpies of sublimation and vaporization As cited previously, one of the most successful methods for the indirect determination of enthalpies of sublimation is based on the Knudsen-cell massloss effusion methods. Recently equipment has been described with three [I891 and nine [190] Knudsen cells used for simultaneous measurements. The torsion effusion method has been used recently to determine enthalpies of sublimation of uracil derivatives [19 11 and sublimation enthalpies have also been determined in drop calorimeters [192,193]. Several modifications of differential scanning calorimeters, DSC, have been developed during the last few years for the measurement of enthalpies of phase changes [194-1971 and Knudsen cells have been used as calorimetric cells. Modulated temperature thermogravimetry has also been proposed for the measurement of enthalpies of sublimation [198]. During the last decade, a number of sublimation enthalpies have been measured by transpiration or transference techniques [9 1,199-2041. The data obtained with this method are in agreement with those obtained by the Knudsen cell mass-loss effusion method, correlation gas chromatography [205] and thermogravimetry [206]. For compounds with low vapour pressures, several quartz crystal resonators have been developed to determine enthalpies of sublimation [207,208]. Two new modifications of a static apparatus for vapour pressure measurement of metal organic precursors have recently been constructed by Sime~ekand coworkers [209].

6. COMPUTATIONAL THERMOCHEMISTRY

Even though this chapter is mainly focussed on experimental thermochemistry using calorimetric methods, it is important to mention computational thermochemistry due to the growing relevance and development of this field [2 101. The number of reports pertaining to computational calculations related to thermochemical problems has increased enormously during the last years, corroborating the continuing interest and importance of these methods. Improvements in computational hardware and software make possible the application of computational thermochemistry to molecules of increasing size. Nowadays theoretical calculations with high accuracy are possible for a large number of molecules and, in some cases, the values obtained by high-level calculations question the experimental results, for example the enthalpy of formation of the silicon atom and thus the value for Si2derived from it [211], the value for F2C=0 [2 12,2131, or for diazomethane [2 141. A detailed explanation of these methods is beyond the scope of this chapter and can be found elsewhere in the literature [210,215-2171. Among the general theoretical models that can be applied for a wide range of molecules, the socalled Gaussian-n theories, the complete basis set models (CBS) and DFT methods have been the most succesful. Despite the failure of the first version of the Gaussian-n family (Gl), [218] the successive versions G2 [219] and G3 [220] generally agree well with the experimental values. These models are complex protocols that aproximate QCISD(T)I6-311+G(3df;2p) calculations by combining a series of smaller calculations. Several modifications of these schemes have been developed to reduce the computational cost and increase the speed, without an important reduction of the accuracy [22 1-2231. The CBS family employs the asymptotic convergence of pair natural orbital expansions to extrapolate to the MP2 limit and the higher order contributions for the various components of the electronic energy are evaluated with a smaller basis set selected to balance the errors and CPU times [224]. DFT methods are based on the electron density being dependent exclusively on three spatial variables, instead of being based on a wave function that is dependent on a large number of variables. They have the advantage of being applicable to bigger molecules due to their computational speed and low computational cost, when compared to the previously cited families. The most accurate and used method to date among these DFT models is the B3LYP functional, employing Becke's gradient corrected exchange functional [225], the Lee-Yang-Parr correlation h c t i o n a l [226], and three parameters fitted to the original G2 test set.

The accuracy of the calculated enthalpies of formation in the gas phase for some of these models named as G2, G2(MP2), G2(MP2,SVP), CBS-Q, CBS-q, CBS-4 and B3LYPl6-31 l+G(3dJ;2p) ranged from 2.3 to 4.9 kJ mol-' taken as the mean absolute deviations from experiment for a set of 61 compounds containing C, H, 0, N, S, F and C1 (extended G2 data set) [227]. The maximum deviations from experiment among this data set for these kinds of compounds range from 11.3 to 17.6 kJ mol-' depending on the method used, however larger deviations are obtained when the enthalpies of formation for molecules containing atoms like Si are calculated [227]. In computational thermochemistry, the enthalpy of formation is usually obtained through atomization or isodesmic reactions (reactions, real or hypothetical, in which the types of bonds made informing the products are the same as those that are broken in the reactants). Petersson et al. [227] demonstrated the improvement in the results when the enthalpies of formation obtained through atomization reactions are corrected with the spin-orbit lowerings for the atoms and empirical bond additivity corrections (BAC). The main problem with isodesmic reactions is different reactions can be used for the same molecule. The selection of the isodesmic reaction can drastically affect the result obtained. So reference molecules with a well-known experimental enthalpy of formation and an experimental uncertainty not exceeding 4 kJ mol-' should be selected, because errors increase with molecular size. Raghavachari et al. [228] have proposed to use a standard set of isodesmic reactions, the "bond separation reactions", [215] where all formal bonds between non-hydrogen atoms are separated into the simplest parent molecules containing these same kinds of linkages. The main problem for these kinds of isodesmic reactions is the large number of molecules involved in the reaction for relatively big molecules and therefore the possible amplification of errors associated with any molecule present in the reaction [227].

*

7. THERMOCHEMISTRY AS A POWERFUL TOOL TO SOLVE ACTUAL CHEMICAL PROBLEMS Some examples in which thermochemistry data are used to solve actual chemical problems are described below. 7.1. Thermochemistry of cyclobutadiene: Enthalpy of formation, ring strain, and anti-aromaticity Recent studies on cyclobutadiene have been carried out [229,230] to determine the enthalpy of formation of this molecule. Conventional calorimetric studies are not useful in this determination due to the extraordinary reactivity of this molecule. Several values have been obtained using more stable derivatives as

models and following a thermodynamic cycle with the help of computational calculations. In spite of this, the values for the enthalpy of formation of cyclobutadiene existing in the literature are very scattered, and a range spanning from 377 to 477 kJ mol-' can be found. However, sophisticated computational methodologies reduce the range to from 414 to 435 kJ mol". Deniz et al. [229], using photoacoustic calorimetry and molecular mechanics and semi-empirical computational calculations, obtained a value of A,H; (g, 298.15 K) = 477 f 46 kJ mol-' and derived from it the ring strain (134 kJ mol-') and anti-aromatic destabilization energies (200 kJ mol-I). The new experiments of Kass and coworkers [230] on the acidity of the cyclobuten-3-yl cation and the ionization potential of the cyclobuten-3-yl radical, combined with the values of the allylic C-H bond dissociation energy of cyclobutene, the ionization potential of the hydrogen radical, the bond dissociation energy of the hydrogen molecule and the enthalpy of formation for cyclobutene, yielded a value of the enthalpy of formation for the desired product, cyclobutadiene, AfH;(g, 298.15 K) = 428 f 16 kJ mol-'. From these results, ring strain and anti-aromatic destabilization energies of 149 f 12 and 131- 136 kJ mol-I respectively, were suggested. 7.2. Thermochemistry of cubane and cuneane Through combined combustion calorimetry, correlation-gas chromatography and crystallographic and computational effort, the first definitive determinations of the structure and enthalpies of formation of a pair of cubane and cuneane isomers were made [23 11. Dimethyl cuneane-2,6-dicarboxylatewas found to be some 190 kJ mol-' more thermodynamically stable than its synthetic precursor, the more symmetric, dimethyl cubane-1,4-dicarboxylate.From the values of the enthalpies of formation of these compounds, accompanied by quantum chemical calculations, values of A,H; (g, 298.15 K) = 613.0 9.5 kJ mol-' for cubane and 436.4 rt 8.8 kJ mol" for cuneane were obtained. From these values, strain enthalpies of 68 1.0 f 9.8 and 504.4 f 9.1 kJ mol-' were calculated for cubane and cuneane, respectively, by means of isodesmic reactions.

+

7.3. Enthalpy of formation of buckminsterfullerene, Cso Since the discovery by Kroto et al. [232] of the third molecular form of carbon, C60, named buckminsterfullerene, and especially after the development of effective ways of production of these type of molecules, there has been great interest in the thermochemistry of these molecules. The enthalpy of formation of C60is a key value in establishing its thermodynamic stability. Several micro- and macro-combustion calorimeters have been used for the experimental determination of the enthalpies of formation of C60 in the crystalline state at 298.15 K [233]. A graphical representation of the available results collected in

reference [233] is given in Figure 4. The results obtained are spread from 2422.3 to 2273.4 kJ mol-I, showing the difficulty of the measurements. C60 Fullerene

Figure 4. A@,

(cr, 298.15 K) values existing in the literature for C60,

The weighted mean value, p, of the enthalpy of formation in the condensed state obtained from all of these results, except the highest value, is ~,H;(cr, 298.15 K) = 23 18.4 k 22.2 kJ mol-I. For the enthalpy of sublimation of C60, the experimental available data are given in reference [152]. The measured enthalpies of sublimation at 298.15 K are spread between 165.5 and 184.1 kJ mol-', with most of them being around 182 kJ mol-'. The weighted mean value obtained for the enthalpy of sublimation from the eight available data at 298.15 K, except the lowest, is A,,H0(298.15 K) = 182.0 1.8 kJ mol-'. From the values of the enthalpy of formation in the condensed state and the enthalpy of sublimation, the experimental value for the enthalpy of formation at 298.15 K in the gas phase for C60is calculated to be A,H;(~,298.15 K) = 2500 22 kJ mol-I.

+

+

7.4. Steric, stereo-electronic and electrostatic interactions in oxanes, thianes and sulfone and sulfoxide derivatives Thermochemical data, and in particular the enthalpies of formation of oxygenand sulfur-containing six-membered heterocycles, provide essential information on the factors responsible for the contrasting behaviour (structural, conformational and reactivity) between these types of compounds. Understanding of the experimental thennochemical observations has required theoretical modeling to confirm the relative importance of the steric, electronic, electrostatic and stereo-electronic interactions that are responsible for the enthalpies of formation for 1,3- and 1,4- dioxanes, 1,3,5-trioxane and their

corresponding sulfur analogues, thiane sulfoxide, thiane sulfone, 1,3-dithiane sulfoxide and 1,3-dithiane sulfone [234]. The most characteristic feature of these kinds of compounds is the presence or absence of stabilization, through nx + ( J * ~ hyperconjugation, - ~ for molecules that present the fragment X-C-Y (X = 0 , S; Y = 0, S, SO, SO2). This stabilization is found for 1,3-dioxane and 1,3,5trioxane but is not effective in the analogous sulfides. In the case of 1,3-dithiane sulfone and 1,3-dithiane sulfoxide, computational calculations of the molecular structure pointed to the existence of hyperconjugation but the additional stabilization is overridden by electrostatic repulsion between the positive charges located at sulfur atoms. 7.5. Keto-en01 tautomerism and the enthalpy of mixing between tautomers of acetylacetone 2,4-pentanedione (acetylacetone) is the prototype of keto-en01 tautomerism and of a low-barrier hydrogen bond and forms complexes with a variety of metals. Correlation gas chromatography has proven useful in the study of species that coexist in equilibrium [235] and was used to evaluate the vaporization enthalpies of the pure tautomers of acetylacetone [236]. Values of 5 1.2 f 2.2 and 50.8 f 0.6 k~.mol-',respectively, were measured for pure 2,4-pentanedione and Z 4-hydroxy-3-penten-2-one. The value of 50.8 f 0.6 kJ mol-' can be contrasted with a value of 43.2 f 0.2 kJ.mol-' calculated for pure Z 4-hydroxy-3-penten-2one, using the calorimetric value of the enthalpy of vaporization previously accepted. The difference was attributed to an endothermic enthalpy of mixing which destabilizes the mixture relative to the pure components. Table 1 collects the new enthalpies of formation recalculated according to these findings. Table 1. Summary of standard molar enthalpies of formation at T = 298.15 K of the two acetylacetone tautomers. All values in kJ mol-'. Values in brackets are the previously accepted ones. Compound 2,4-pentanedione

Afflrn(l)

-410.1 f 1.2 [-416.3 f 1.11 Z 4-hydroxy-3-penten-2-one -429.0 f 1.0 [-427.6 f 1.11 Mixture [-425.5 k 1.01 (8 1.4% en01 1 18.6% diketo)

AlgHrn

51.2 f 2.2 [41.9 f 0.11

50.8 f 0.6 [43.2 f 0.11

Afflrn(g) -358.9 [-374.4 -378.2 [-384.4

f 2.5 f 1.31 f 1.2 f 1.31

A comparison of the new gas-phase enthalpies of formation of the two tautomers results in a difference of -19.3k2.8 kJmol-', in excellent agreement with the NMR value reported by Folkendt et al. of -19.5 0.77 kJ rnol-' [237].

+

7.6. Radical generation by using organometallic complexes of Group 6 metals Combination of thermochemistry, kinetics, spectroscopy and computational calculations can help predict under what conditions a reaction pathway will be followed. Hoff and coworkers use this approach to investigate under what conditions a free PhE (E = S, Se, Te) radical might be generated when a PhEEPh interacts with a specific metal. The knowledge of the PhE-EPh bond strength (an organic problem) [238] and also the LnM-EPh (L,M = metal complex) bond strength (an inorganic problem) [239] are important in understanding the oxidative addition reactions of transition-metals and the dichalcogenides. The thiyl radical (.SR), free or bound to a metal (LnM-(.SR)), plays an important role in biological systems, however thermochemical studies on these molecules are scarce and even more in the case of the heavier chalcogenides, Se and Te. In the studies of Hoff and coworkers the interactions w[(P(~P~)~)~(co)~], and Mo[N('~u)Ar]3and PhEbetween MO[(P(~P~)~)~(CO)~], EPh have also been investigated. Oxidative addition to W is more exothermic complex and is thermodynamically more than for Mo in the M[(P(~P~)~)~(co)~] favoured for MO[N(~BU)A~]~. In spite of that, the addition is faster for MO[(P(~P~)~)~(CO)~], compared to MO[N('BU)A~]~. 7.7. Application to biochemical systems Several applications of isothermal titration calorimetry to biochemical systems have been published recently. Franghanel et al. [240] proposed a new calorimetric method to investigate slow conformational changes in proteins and peptides, Tsamaloukas et al. [241] studied the interactions of cholesterol with unsaturated phospholipids and D'Amico et al. [242] determined the binding enthalpies of a-amylases to different substrates.

7.8. Thermochemistry of reactions in gas phase for compounds with important implications as catalysts. The important catalytic applications of transient metal sulfides have motivated the study of the gas-phase thermochemistry of the early cationic transition-metal sulfides of the second row: YS', Z~S'and N ~ S '[243] by guided ion-beam and ion-cyclotron resonance mass spectrometry. Lago and Baer have studied the photo-ionization of bromine species that play an important role in the catalytic depletion of the Earth's protecting ozone layer by the threshold photoelectron photo-ion coincidence (TPEPICO) technique

12441 and derived the enthalpies of formation of several neutral and cationic bromine species. 8. CONCLUSIONS

As can be seen from the examples cited, themochemistry is a suitable and powerful tool for solving chemical problems in fields as different as organic, inorganic, organometallic and physical chemistry, biochemistry, catalysis, materials, atmospheric studies, and can deal with neutral, radical and ionic species. 9. REFERENCES 1. W.D. Good, J. Chem. Thermodyn., 2 (1970) 237. 2. A. Streitwieser, C. H. Heathcock, E.M. Kosover, Introduction to Organic Chemistry, 4thEdition, Macmillan, New York, 1992, Ch. 5. 3. J.D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds, Academic Press, London and New York, 1970. 4. L. MCdard and H. Tachoire, Histoire de La Thermochimie, PrClude B la Thermodynamique Chimie Publications de L'UniversitC de Provence, France, 1994. 5. M. Berthelot, La RCvolution Chimique, Lavoisier, Alcan, Paris, 1890, p. 99. 6. A.L. Lavoisier and P.S. de la Place, MCm. Acad. R. Sci. Pour 1'AnnCe 1780, (1784) 355. 7. J.R. Partington, A History of Chemistry, Vol. 3, Macmillan, London, 1962, p. 426. 8. G.H. Hess, Bull. Sci. Acad. Imp. Sci. St-PCtersbourg, 7 (1840) 257. 9. G.H. Hess, Bull. Sci. Acad. Imp. Sci. St-Petersbourg, 7 (1 840) 263; Ann. Chim. Phys. 75 (1840) 89. 10. G.H. Hess, in Themochemische Untersuchungen, (Ed. W. Ostwald), Vo19 of Klassiker der Exakten Naturwissenschaften, Engelmann, Leipzig, 1890. 11. E.W. Washburn, J. Res. Natl. Bur. Stand., 10 (1933) 525. 12. F.D. Rossini, J. Res. Natl. Bur. Stand., 22 (1939) 407. 13. R.S. Jessup and C.B. Green, J. Res. Natl. Bur. Stand., 13 (1934) 469. 14. P.H. Dewey and D.R. Harper, J. Res. Natl. Bur. Stand., 21 (1938) 457. 15. R.S. Jessup, J. Res. Natl. Bur. Stand., 2 1 (1938) 475. 16. E.J. Prossen and F.D. Rossini, J. Res. Natl. Bur. Stand., 33 (1944) 439. 17. E.J. Prossen, R.S. Jessup and F.D. Rossini, J. Res. Natl. Bur. Stand., 33 (1944) 447.

18. W. Zielenkiewicz, in Calorimetry (Ed. P. Gierycz) ,Institute of Physical Chemistry of the Polish Academy of Sciences, Warszawa, 2005. 19. W. Hemminger and S.M. Sarge, in Handbook of Thermal Analysis and Calorimetry, Vol 1, Principles and Practice, (Ed. M.E. Brown), Elsevier, Amsterdam, 1998, Ch. 1. 20. P.J. van Ekeren, in Handbook of Thermal Analysis and Calorimetry, Vol 1, Principles and Practice, (Ed. M.E. Brown), Elsevier, Amsterdam, 1998, Ch. 2. 21. R.B. Kemp, in Handbook of Thermal Analysis and Calorimetry, Vol. 1, Principles and Practice, (Ed. M.E. Brown ), Elsevier, Amsterdam, 1998, Ch. 14. 22. E. Calvet and H. Prat, Recent Progress in Microcalorimetry (Ed. and transl. by H.A. Skinner), Pergamon Press, Oxford, 1963. 23. M. Laffite, in Combustion Calorimetry, (Eds S. Sunner and M. Mansson), Pergamon Press, Oxford, 1979, Ch. 17. 24. W.N. Hubbard, D.W. Scott and G. Waddington, in Experimental Thermochemistry, (Ed. F.D. Rossini),Interscience, New York, 1956, Ch. 5. 25. S. Sunner, in Combustion Calorimetry, (Eds S. Sunner and M. Miinsson), Pergamon Press, New York, 1979, Ch. 2. 26. D.R. Kirklin and E.S. Domalski, J. Chem. Thermodyn., 15 (1983) 941. 27. F.D. Rossini (Ed.), Experimental Thermochemistry, Interscience, New York, 1956. 28. H.A. Skinner (Ed.), Experimental Thermochemistry, Interscience, London, 1962. 29. S. Sunner and M. Miinsson (Eds), Combustion Calorimetry, Pergamon, Oxford, 1979. 30. M. Frenkel (Ed.), Thermochemistry and Equilibria of Organic Compounds, VCH Publishers, New York, 1993. 3 1. J.D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds, Academic Press, 1970, p. 68. 32. J.D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds, Academic Press, 1970, p. 72. 33. E.J. Prosen, National Bureau of Standards, Report No 1119. 34. W.N. Hubbard, D.W. Scott and G. Waddington, J. Phys. Chem., 56 (1954) 152. 35. A.J. Head and W.D. Good, in Combustion Calorimetry, (Eds S. Sunner and M. Miinsson), Pergamon, Oxford, 1979, Ch. 9. 36. M. Miinsson and W.N. Hubbard, in Combustion Calorimetry, (Eds S. Sunner, and M. Miinsson), Pergamon, Oxford, 1979, Ch. 5.

37. K.B. Wiberg, in Molecular Structure and Energetics, Vol. 2. Physical measurements, (Eds J.F. Liebman and A. Greenberg), VCH Publishers, New York, 1987, Ch. 4. 38. W.D. Good, D.W. Scottt and G. Waddington, in Experimental Thermochemistry, Vol. 2, (Ed. H.A. Skinner), Wiley-Interscience, New York, 1962, p. 15. 39. S. Sunner and B. Lundin, Acta Chem. Scand., 7 (1953) 122. 40. W.N. Hubbard, C. Katz and G. Waddington, J. Phys. Chem., 58 (1954) 142. 41. D.D. Wagman, W.H. Evans, V.B. Parker, R.H. Schumm, I. Halow, S.M. Bailey, K.L. Churney and R.L. Nuttall, J. Phys. Chem. Ref. Data, 11 (1982) 58. 42. V.P. Kolesov, in Combustion Calorimetry, (Eds S. Sunner and M. Mhsson), Pergamon, Oxford, 1979, Ch. 13. 43. V.B. Parker, D.D. Wagman and D. Garvin, Selected Thermochemical Data Compatible with the CODATA Recommendations, Natl. Bur. Stand. (US), Interim Report NBSIR 75, January 1976. 44. M.E. Minas da Piedade, in Energetics of Stable Molecules and Reactive Intermediates, (Ed. M.E. Minas da Piedade), NATO-AS1 Series No. C535, Kluwer, Dordrecht, 1999, Ch. 2. 45. G. Olofsson, in Combustion Calorimetry, (Eds S. Sunner and M. Minsson), Pergamon, Oxford, 1979, Ch. 6. 46. M.V. Roux, M. Temprado, P.JimCnez, J. PCrez Paraj6n and R. Notario, J. Phys. Chem., A 107 (2003) 11460. 47. J.D. Cox, D.D. Wagman and V.A. Medvedev, CODATA Key Values for Thermodynamics, Hemisphere Publishing Corp., New York, 1989. 48. H.A. Skinner, in Experimental Thermochemistry, (Ed. F.D. Rossini), Interscience, New York, 1956, Ch. 11. 49. H.A. Skinner, in Experimental Thermochemistry, (Ed. H.A. Skinner), Interscience, London, 1962, Ch. 8. 50. K.N. Marsh and P.A.G. O'Hare (Eds), Solution Calorimetry, Experimental Thermodynamics, Blackwell Scientific Publications, Oxford, 1994. 5 1. H.A. Skinner, J.M. Sturtevant and S. Sunner, in Experimental Thermochemistry, (Ed. H.A. Skinner), Interscience, London, 1962, Ch. 9. 52. J.J. Christensen, in Thermochemistry and its Applications to Chemical and Biochemical Systems, (Ed. M.A.V. Ribeiro da Silva), NATO AS1 Series, No. C119, Reidel, Dordrecht, 1982, p. 1. 53. H.J.V. Tyrrell and A.E. Beezer, Thermometric Titrimetry, Chapman and Hall, London, 1968. 54. J. Barthel, Thermometric Titrations, John Wiley & Sons, New York, 1975. 55. G.A. Vaughn, Thermometric and Enthalpimetric Titrimetry, Van Nostrand Reinhold Co., London, 1973.

56. J.L. Oscarson, R.M. Izatt, J.O. Hill and P.R. Brown, in Solution Calorimetry, (Eds K.N. Marsh and P.A.G. O'Hare), Blackwell Scientific Publications, Oxford, 1994, Ch. 10. 57. H. Hemminger and G. Hohne, Calorimetry: Fundamentals and Practice, Verlag Chemie, Weinheim, 1984. 58. G. Olah, G.K.S. Prakash and J. Sommer, Superacids, Wiley, New York, 1985. 59. D.J. Eatough, J.J. Christensen and R.M. Izatt, Experiments in Thermometric Titrimetry and Titration Calorimetry, Brigham Young University Press, Provo, Utah, 1974. 60. R.J. Angelici, in Energetics of Stable Molecules and Reactive Intermediates, (Ed. M.E. Minas da Piedade), NATO-AS1 Series No. C535, Kluwer, Dordrecht, 1999, Ch. 4. 61. G.W. Thomson, in Techniques of Organic Chemistry, (Ed. A. Weissberger), 3rdEd., Part 1, Interscience Publishers, New York, 1965, p. 401. 62. R. Cooper and D.R. Stranks, in Techniques of Inorganic Chemistry, (Ed. A. Weissberger), Interscience Publishers, New York, 1966, p. 1. 63. V. Majer, V. Svoboda and J. Pick, Studies in Modem Thermodynamics, 9: Heats of vaporization of fluids, 1989, Elsevier, Amsterdam. 64. J.S. Chickos, Netsu Sokutei, 30 (2003) 116. 65. J.H. Mathews, P.R. Fehland, J. Am. Chem. Soc., 53 (193 1) 3212. 66. K.S. Pitzer, J. Am. Chem. Soc., 63 (1941) 2413. 67. E. Morawetz and S. Sunner, Acta Chem. Scand., 17 (1963) 13. 68. I. Wadso, Acta Chem. Scand., 20 (1966) 536. 69. E. Morawetz. Acta Chem. Scand., 22 (1968) 1509. 70. E. Morawetz, Chemica Scripta, 1 (1971) 103. 71. S. Sunner and C. Svensson, J. Chem. Soc. Faraday Trans., 1 (1979) 2359. 72. R. Jochems, H. Dekker, C. Mosselman and G. Somsen, J. Chem. Thermodyn., 14 (1982) 395. 73. E. Calvet and H. Prat, Microcalorimetrie. Applications Physico Chimiques et Biologiques, Mason, Paris, 1956. 74. R. Sabbah and C. Minadakis, Thermochim. Acta, 43 (198 1) 269. 75. G. Della Gatta and G. Varone, Proc. Second. CODATA Symposium on Critical Evaluation and Prediction of Phase Equilibria in Multicomponent Systems, Paris, 1985, p. 576. 76. K. Kusano, Thermochim. Acta, 88 (1985) 109. 77. S. Murata, M. Sakiyama and S. Seki, Thermochim. Acta, 88 (1985) 121. 78. P. Backman, M. Bastos, D. HallCn, P. Lonnbro and I. Wadso, J. Biochem. Biophys. Methods, 28 (1994) 85. 79. F.A. Adedeji, L.S. Brown, J.A. Connor, M.L. Leung, M. Paz-Andrade and H.A. Skinner, J. Organometal. Chem., 97 (1975) 221.

80. L.A. Torres-Gomez, G. Barreiro-Rodriguez and A. Galarza-Mondragon, Thermochim. Acta, 124 (1988) 229. 8 1. E.F.G. Herington and J.F. Martin, Trans. Faraday Soc., 49 (1953) 154. 82. D. Ambrose, J. Phys., 1E (1968) 41. 83. W.H. Mears, R.F. Stahl, S.R. Orfeo, R.C. Shair, L.F. Kells, W. Thompson and H. McCann, Ind. Eng. Chem., 47 (1955) 1449. 84. D.R. Douslin and A.E. Osborn, J. Sci. Instr., 42 (1965) 369. 85. C.G. de Kruif and H.A.J. Oonk, Chemie Ing. Techn., 45 (1973) 455. 86. M. Knudsen. Ann. Phys., 28 (1909) 75. 87. D. Ferro, V. Piacente, R. Gigli and G. D'Ascenzo, J. Chem. Thermodyn., 8 (1976) 1137. 88. C.G. De Kruif and C.H.D. Van Ginkel, J. Chem. Thermodyn. 9 (1977) 725. 89. C.G. de Kruif, in Thermochemistry and Its Applications to Chemical and Biochemical Systems, (Ed. M.A.V. Ribeiro da Silva), NATO AS1 Series, No C119, Reidel, Dordrecht, 1982, p. 143. 90. J.L. Hales, R.C. Cogman and W.J. Frith, J. Chem. Thermodyn., 13 (1981) 591. 91. S.P. Verevkin, J. Chem. Eng. Data, 45 (2000) 953. 92. R. S. Dygdala, K. Stefanski and J. Wolnikowski, Bulletin De L'academie Polonaise Des Sciences, 15 (1977) 439. 93. T.P. Melia and R. Merrifield, J. Inorg. Nucl. Chem., 32 (1970) 1489. 94. A.B. Teplitskii, I.K. Yanson, O.T. Glukhova, A. Zielenkiewicz, W. Zielenkiewicz and K.L. Wierzchowski, Biophys. Chem., 11 (1980) 17. 95. P.M. Burkinshaw and C.T. Mortimer, J. Chem. Soc. Dalton Trans., (1984) 75. 96. S.A. Greene, Anal. Chem., 28 (1956) 428. 97. V.G. Yurkin and A.M. Rozen, Z. Fizicheskoi Khimii, 47 (1973) 492. 98. F.B. Belaribi, A. Aitkaci and J. Jose, J. Them. Anal., 46 (1996) 245. 99. J.S. Chickos, S. Hosseini and D.G. Hesse, Thermochim. Acta, 249 (1995) 41. 100. I. Wadso, Acta Chem. Scand., 20 (1966) 544. 101. S.W. Benson, F.R. Cruickshank, D.M. Golden, G.R. Haugen, H.E. O'Neal, A.S. Rodgers, R. Shaw and R. Walsh, Chem. Rev., 69 (1969) 279. 102. A. Bondi, J. Chem. Eng. Data, 8 (1963) 371. 103. J.S. Chickos, W. Acree, Jr, J.F. Liebman, in Computational Thermochemistry. Prediction and Estimation of Molecular Thermodynamics, (Eds D. Fmrip and K. Irikura),ACS Symposium Series 677, Washington DC, 1988, p. 76. 104. M.V. Roux, M. Temprado and J.S. Chickos, J. Chem. Therrmodyn., 37 (2005) 941. 105. S.J. Blanksby and G.B. Ellison, Acc. Chem. Res., 36 (2003) 255.

106. E.M. Arnett, Acc. Chem. Res., 6 (1973) 404. 107. E.M. Arnett, K. Amarnath, N.G. Harvey and J.P. Cheng, Science, 247 (1990) 423. 108. F.G. Bordwell and X.M. Zhang, Acc. Chem. Res., 26 (1993) 510. 109. F.G. Bordwell, A.V. Satish, S. Zhang and X.M. Zhang, Pure Appl. Chem., 67 (1995) 735. 110. J.A.M. Sim6es and M.A.V. Ribeiro da Silva, in Energetics of Stable Molecules and Reactive Intermediates, (Ed. M.E. Minas da Piedade), NATO-AS1 Series No. C535, Kluwer, Dordrecht, 1999, Ch. 1. 111. A. Macoll, Org. Mass. Spectrom., 17 (1982) 1. 112. D.M. Wetzel and J. I. Brauman, Chem. Rev., 87 (1987) 607. 113. R.R. Squires, Acc. Chem. Res., 25 (1992) 46 1. 114. P.B. Armentrout, Acc. Chem. Res., 28 (1995) 430. 115. M.T. Bowers, Acc. Chem. Res., 27 (1994) 324. 116. A.G. Marshall, Int. J. Mass. Spectrom., 200 (2000) 33 1. 117. J.L. Beauchamp, P.A.M. Van Koppen, in Energetics of Organometallic Species, (Ed. J.A. Martinho SimGes), NATO-AS1 Series No. C367, Kluwer, Dordrecht, 1992, Ch. 17. 118. G. De Maria, in Thermochemistry and its Applications to Chemical and Biological Systems, (Ed. M.A.V. Ribeiro da Silva), NATO-AS1 Series No. C119, Reidel, Dordrecht, 1984, p. 157. 119. R.G. Cooks, J.S. Patrick, T. Kotiaho and S.A. McLuckey, Mass Spectrom. Rev., 13 (1994) 287. 120. J.E. Bartmess, J. Phys. Chem., 98 (1994) 6420. 121. P. Kebarle and S. Chowdhury, Chem. Rev., 87 (1987) 5 13. 122. Y.Y. Zhao, F.G. Bordwell, J.P. Cheng and D.F. Wang, J. Am. Chem. Soc., 119 (1997) 9125; F.G. Bordwell, S.Z. Zhang, I. Eventova and Z. Rappoport, J. Org. Chem., 62 (1997) 5371; and references cited therein. 123. S. Glasstone, K.J. Laidler and H. Eyring, The Theory of Rate Processes, McGraw Hill, New York, 1941. 124. E.M. Arnett and D.R. McKelvey, Record of Chemical Progress, 26 (1965) 185. 125. E.L. Eliel, N.L. Allinger, S.J. Anguila and G.A. Morrison, Conformational Analysis, Wiley-Interscience, New York, 1965. 126. K. Pihlaja and E. Taskinen, in Physical Methods in Heterocyclic Chemistry, Vol. 6, (Ed. A.R. Katritzky), Academic Press, New York, 1974, Ch. 5. 127. J.M. Kanabus-Kaminska, B.C. Gilbert and D. Griller, J. Am. Chem. Soc., 111 (1989) 3311. 128. S.E. Braslavsky and G.E. Heibel, Chem. Rev., 92 (1992) 1381. 129. K.S. Peters, Angew. Chem. Int. Ed. Engl., 33 (1994) 294.

130. J.D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds, Academic Press, London, 1970, Ch. 3. 131. J.D. Cox, Pure Appl. Chem., 40 (1974) 399. 132. A.J. Head and R. Sabbah, in IUPAC Recommended Reference Materials for the Realization of Physicochemical Properties, (Ed. K.N. Marsh), Blackwell, Oxford, 1987, Ch. 9. 133. R. Sabbah, A. Xu-wu, J.S. Chickos, M.L. Planas, M.V. Roux and L.A. Torres, Thermochim. Acta, 331 (1999) 93. 134. M.V. Roux, L.A. Torres and J.Z. DAvalos, J. Chem. Thermodyn., 33 (2001) 949. 135. S. Abramowitz, D.D. Wagman, V.B. Parker and D. Garvin, in Thermochemistry and its Applications to Chemical and Biochemical Systems, (Ed. M.A.V. Ribeiro da Silva), NATO AS1 Series, No. C119, Reidel, Dordrecht, 1982, p. 803. 136. E.W. Washburn (Ed.), International Critical Tables of Numerical Data, Physics, Chemistry and Technology, McGraw-Hill, New York and London, in 8 parts, 1926-1933. 137. M.S. Kharasch, J. Res. Natl. Bur. Stand., 2 (1929) 359. 138. F.R. Bichowsky and F.D. Rossini, Thermochemistry of the Chemical Substances, Reinhold, New York, 1936. 139. F.D. Rossini, K.S. Pitzer, W.J. Taylor, J.P. Ebert, J.E. Kilpatrick, C.W. Beckett, M.G. Williams and H.G. Werner, Selected Values of Properties of Hydrocarbons, American Petroleum Institute Research Project 44, Natl. Bur. Stand. Circular 461, U.S. Government Printing Office, Washington D.C, 1947. 140. Selected Values of Properties of Hydrocarbons, American Petroleum Institute Research Project 44, Carnegie Institute of Technology, Pittsburgh, Pa., 1953. 141. Selected Values of Properties of Chemical Compounds, Manufacturing Chemists Association Research Project, College station, Texas, 1961. 142. Landbolt-Bornstein, Zahlenwerte und Funktiones, 6thedition, band 11, Teil4, Springer-Verlag, Berlin, Gottingen, Heidelberg, 1961. 143. F.D. Rossini, D.D. Wagman, W.H. Evans, S. Levine and I. Jaffe, Selected Values of Chemical Thermodynamic Properties. Circular of the National Bureau of Standards 500, US, Washington DC, 1952. 144. D.R. Stull, E.F.W. Westrum, Jr, and G.C. Sinke, The Chemical Thermodynamics of Organic Compounds, John Wiley & Sons, New York, 1969. 145. J.D. Cox, A Method for Estimating the Enthalpies of Formation of Benzene Derivatives, in the Gas State, NPL Report CHEM 83, 1978.

146. J.B. Pedley, R.N. Naylor and S.P. Kirby, Thermochemical Data of Organic Compounds, 2ndEdn, Chapman and Hall, London, 1986. 147. J.B. Pedley, Thermochemical Data and Structures of Organic Compounds, Vol.1, TRC Data Series, Thermodynamic Research Center, Texas, 1994. 148. M.W. Chase, Jr., NIST-JANAF Thermochemical Tables, 4thEdn, J. Phys. Chem. Ref. Data., Monograph 9, 1998. 149. M. Frenkel (Ed.), TRC Thermodynamic Tables. National Institute of Standards and Technology, Boulder, Colorado, 2003. 150. H.Y. Afeefy, J.F. Liebman and S.E. Stein, in NIST Chemistry Webbook, NIST Standard Reference Database, (Eds P.J. Linstrom and W.G. Mallard), Number 69, June 2005, National Institute of Standards and Technology, Gaithersburg MD, 20899. http://webbook.nist.gov/chemistry/ 151. J.S. Chickos and W.E. Acree, Jr., J. Phys. Chem. Ref. Data, 32 (2003) 5 19. 152. J.S. Chickos and W.E. Acree, Jr., J. Phys. Chem. Ref. Data, 31 (2002) 537. 153. V. Majer and V. Svoboda, Enthalpies of Vaporization of Organic Compounds: A Critical Review and Data Compilation, Blackwell Scientific Publications, Oxford, 1985. 154. A. Tamir, E. Tamir and K. Stephan, Heats of Phase Change of Pure Components and Mixtures, a Literature Guide, Elsevier, 1983, Amsterdam. 155. Y .R. Luo, Handbook of Bond Dissociation Energies in Organic Compounds, CRC Press, Boca Raton, 2003, p. 380. 156. T. Kiyobayashi, Ph. D. Thesis, Faculty of Sciences, Osaka University, 1995. 157. H.P. Diogo and M.E. Minas da Piedade, J. Chem. Thermodyn., 27 (1995) 197. 158. X. An, B. Chen and J. He, Science in China, Series B: Chemistry, 41 (1998) 548. 159. Y. Nagano, J. Therm. Anal. Cal., 57 (1999) 867. 160. J.Z. Dhvalos and M.V. Roux, Meas. Sci. Technol., 11 (2000) 1421. 161. M. Sakiyama and T. Kiyobayashi, J. Chem. Thermodyn., 32 (2000) 269. 162. A. Rojas-Aguilar, J. Chem. Thermodyn., 34 (2002) 1729. 163. A. Rojas-Aguilar and M. Martinez-Herrera, J. Chem. Thermodyn., 36 (2004) 619. 164. E.A. Camarillo and H. Flores, J. Chem. Thermodyn., in press. 165. M. Padoa and B. Foresti, Gazz. Chim. Ital., 53 (1923) 494. 166. W.A. Roth, H. Ginsberg and R. Lassw, Z. Elektrochem., 30 (1924) 4 17. 167. W. Mac Ewan and C. Anderson, Rev. Sci. Instrum., 28 (1955) 280. 168. E. Calvet, H. Moureu, P. Chovin and H. Tachoire, J. Chim. Phys., 57 (1960) 593. 169. V.V. Ponomarev and T.A. Alekseeva, Zhur. Fiz. Kim., 35 (1961) 1629. 170. H. Mackle and P.A.G. O'Hare, Trans. Faraday Soc., 59 (1963) 2693. 171. M. MAnsson, J. Chem. Thermodyn., 5 (1973) 72 1.

172. M. Mksson, J. Chem. Thermodyn., 6 (1974) 1018. 173. R. Sabbah and M. Coten ,Thermochim. Acta, 49 (198 1) 307. 174. M. Mhsson, Pure Appl. Chem., 5 1 (1983) 417. 175. R.M. Metzger, C.S. Kuo and E.S. Arafat, J. Chem. Thermodyn., 15 (1983) 841. 176. H.-D. Beckhaus, C. Ruchardt and M. Smisek, Thermochim. Acta, 79 (1984) 149. 177. R.C. Santos, H.P. Diogo and M.E. Minas da Piedade, J. Chem. Thermodyn., 31 (1999) 1417. 178. R. Sabbah, and A.R. Aguilar, J. Chem. Thermodyn., 27 (1995) 685. 179. R. Sabbah and A.R. Aguilar, J. Chem. Thermodyn., 28 (1996) 221. 180. R. Sabbah and A.R. Aguilar, J. Chem. Thermodyn., 28 (1996) 1071. 181. R. Sabbah, and A.R. Aguilar, Struct. Chem., 7 (1996) 383. 182. R. Sabbah, Bull. Soc. Chim. Fr., 1 (1979) 434. 183. R. Sabbah and L.A.T. G6mez7Thermochim. Acta, 52 (1982) 285. 184. L.A.T. G6mez and R. Sabbah, Thermochim. Acta ,57 (1982) 67. 185. R. Sabbah and I. Antipine, Bull. Soc. Chim. Fr., 3 (1987) 392. 186. R. Sabbah and J.A.G. PCrez, Thermochim. Acta, 297 (1997) 17. 187. H. Flores, J. Mentado, P. Amador, L.A. Torres, M. Campos and A. Rojas, J. Chem. Thermodyn., 38 (2006) 756. 188. A. Pramann, A. Benito, P.L. Cremonesi, J.-R. Filtz, R. Foster, F. Haloua,

B. Hay, M. Jaeschke, C. Ryszfeld, S.M. Sarge and P. Ulbig, Abstracts of papers of ICTAC 13 Conference, Chia Laguna, 2004. 189. M.A.V. Ribeiro da Silva and M.J.S. Monte, Thermochim. Acta, 171 (1990) 168. 190. M.A.V. Ribeiro da Silva, M.J.S. Monte and L.M.N.B.F. Santos, J. Chem. Thermodyn., 38 (2006) 778. 191. B. Brunetti, V. Piacente and G. Portalone, J. Chem. Eng. Data, 45 (2000) 242. 192. T. Kiyobayashi and M.E. Minas da Piedade, J. Chem. Thermodyn., 33 (2001) 11. 193. M.A.V. Ribeiro da Silva, L.L.M. Spencer and L.M.P.F. Amaral, J. Chem. Thermodyn., 35 (2003) 1343. 194. T.M.R. Maria, F.S. Costa, M.L. Planas-LeitBo and J.S. Redinha, Thermochim. Acta, 269 (1995) 405. 195. M. Casserino, D.R. Blevins and R.N. Sanders, Thermochim. Acta, 284 (1996) 45. 196. A. Rojas-Aguilar, E. Orozco-Guareiio and M. Martinez-Herrera, J. Chem. Thermodyn., 33 (2001) 1405. 197. A. Rojas-Aguilar, F. GimCnez-Carbajal, F.E. Orozco-Guareiio and H. Flores-Segura, J. Them. Anal. Cal., 79 (2005) 95.

198. A. Schiraldi and D. Fessas, J. Therm. Anal. Cal., 71 (2003) 225. 199. J.W. Hastie and D.W. Bonnell, in Themochemistry and Its Applications to Chemical and Biochemical Systems, (Ed. M.A.V. Ribeiro da Silva), NATO AS1 Series, No. C119, Reidel, Dordrecht, 1982, p. 183. 200. R.L. Blaine and B.K. Hahn, J. Therm. Anal. Cal., 54 (1998) 695. 201. K. Bayerenther, G. Braner, M. Fanker and K.-H, Schmidt, Int. Laboratory News, May 1994. 202. W. Zielenkiewicz, G.L. Perlovich and M. Wszelaka Tylik. J. Therm. Anal., 57 (1999) 225. 203. G.L. Perlovich, Ph. D. Thesis, Ivanovo, Russia, 2001. 204. D. Kulikov and S.P. Verevkin, Fluid Phase Equilibr., 192 (2001) 187. 205. J.S. Chickos, D.M. Hillesheim, S.P. Verevkin, M.V. Roux, M. Temprado, M. Segura, R. Notario, D.E. Demasters and J.F. Liebman, Mol. Phys., 101 (2003) 1311. 206. D. Hinks, M.I. Rafiq, D.M. Price, G.A. Montero and B. Smith, Color Technol., 119 (2003) 84. 207. L.A. Torres, I. Hernindez-Contreras, J.A. Guardado and M.G. GonzSlez, Meas. Sci. Technol., 5 (1994) 5 1. 208. M.A.V. Ribeiro da Silva, M.L.C.C.H. Ferriio, M.J.S. Monte, J.M. Gonqalves and F. Jiye, J. Chem. Thermodyn., 31 (1999) 1067. 209. M. Fulem, K. RfiiiEka, V. RfiiiEka, T. SimeEek, E. Hulicius and J. Pangrac, J. Chem. Themodyn., 38 (2006) 312. 210. K.K. Irikura and D.J. Frurip (Eds), Computational Themochemistry, ACS Symposium Series, No. 677, Washington DC, 1998. 21 1. J.W. Ochterski, G.A. Petersson and K.B. Wiberg, J. Am. Chem. Soc., 117 (1995) 11299. 212. J.A. Montgomery, Jr., H.H. Michels and J.S. Francisco, Chem. Phys. Lett., 220 (1994) 391. 213. R.S. Grev and H.F. Schaefer 111, J. Chem. Phys., 97 (1992) 8389. 2 14. D.A. Dixon, W.A. de Jong, K.A. Peterson and T.B. McMahon, J. Phys. Chem., A 109 (2005) 4073. 215. W.J. Hehre, L. Radom, P. v. R. Schleyer and J.A. Pople, Ab Initio Molecular Orbital Theory, Wiley-Interscience, New York, 1986. 216. F. Jensen, Introduction to Computational Chemistry, John Wiley & Sons, Chichester, 1999. 217. C.J. Cramer, Essentials of Computational Chemistry, Theories and Models, 2"d Edn, John Wiley & Sons, Chichester, 2004. 21 8. J.A. Pople, M. Head-Gordon, D.J. Fox, K. Raghavachari and L.A. Curtiss, J. Chem. Phys., 90 (1989) 5622. 219. L.A. Curtiss, K. Raghavachari, G.W. Trucks, and J.A. Pople, J. Chem. Phys., 94 (1991) 7221.

220. L.A. Curtiss, K. Raghavachari, P.C. Redfern, V. Rassolov and J.A. Pople, J. Chem. Phys., 109 (1998) 7764. 221. L.A. Curtiss, K. Raghavachari and J.A. Pople, J. Chem. Phys., 98 (1993) 1293. 222. L.A. Curtiss, P.C. Redfern, K. Raghavachari, V. Rassolov and J.A. Pople, J. Chem. Phys., 110 (1999) 4703. 223. A.G. Baboul, L.A. Curtiss, P.C. Redfern and K. Raghavachari, J. Chem. Phys.,llO (1999) 7650. 224. J.W. Ochterski, G.A. Petersson and J.A. Montgomery, Jr, J. Chem. Phys., 104 (1996) 2598. 225. A.D. Becke, J. Chem. Phys., 98 (1993) 5648. 226. C.T. Lee, W.T. Yang and R.G. Parr, Phys. Rev., B 37 (1988) 785. 227. G.A. Petersson, D.K. Malick, W.G. Wilson, J.W. Ochterski, J.A. Montgomery, Jr. and M.J. Frisch, J. Chem. Phys., 109 (1998) 10570. 228. K. Raghavachari, B.B. Stefanov and L.A. Curtiss, J. Chem. Phys., 106 (1997) 6764. 229. A. Fattahi, L. Lis, Z. Tian and S.R. Kass, Angew. Chem. Int. Ed., 45 (2006) 4984. 230. A.A. Deniz, K.S. Peters and G.J. Snyder, Science, 286 (1999) 1119. 231. M.V. Roux, J.Z. Davalos, P. JimCnez, R. Notario, 0. Castaiio, J.S. Chickos, W. Hanshaw, H. Zhao, N. Rath, J.F. Liebman, B.S. Farivar and A. BashirHashemi, J. Org. Chem. 70, (2005) 5461. 232. H.W. Kroto, J.R. Heath, S.C.O'Brien, R.F. Curl and R.E. Smalley, Nature 318 (1985) 162. 233. A. Rojas-Aguilar and M. Martinez-Herrera, Thermochim. Acta, 437 (2005) 126, and references cited therein. 234. E. Juaristi, R. Notario and M.V. Roux, Chem. Soc. Rev., 34 (2005) 347. 235. M. Temprado and J.S. Chickos, Thermochim. Acta, 435 (2005) 49. 236. M. Temprado, M.V. Roux, P. Humnahanant, H. Zhao and J.S. Chickos, J. Phys. Chem., B 109 (2005) 12590. 237. M.M. Folkendt, B.E. Weiss-Lopez, J.P. Chauvel Jr. and N.S. True, J. Phys. Chem., 89 (1985) 3347. 238. J.E. MacDonough, M.J. Carlson, J.J. Weir, C.D. Hoff, O.P. Kryatov, E.V. Rybak-Akimova and C.C. Cummins, Inorg. Chem., 44 (2005) 3 127. 239. J.E. MacDonough, J.J. Weir, K. Sukcharoenphon, C.D. Hoff, O.P. Kryatov, E.V. Rybak Akimova, B. Scott, G.J. Kubas, F.H. Stephens, A. Mendiratta and C.C. Cummins, J. Am. Chem. Soc., 128 (2006) 10295. 240. J. Franghanel, S. Wawra, C. Luke, D. Wildemann and G. Fischer, Anal. Chem., 78 (2006) 45 17. 241. A. Tsamaloukas, H. Szadkowska and H. Heerklotz, Biophys. J., 90 (2006) 4479.

242. S. D'Amico, J.S. Sohier and G. Feller, J. Mol. Biol., 358 (2006) 1296. 243. I. Kretzschrnar, D. Schroder, H. Schwarz and P.B. Armentrout, Int. J. Mass. Spectrom., 249-250 (2006) 263. 244. A.F. Lago and T. Baer, J. Phys. Chem., A 110 (2006) 3036.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 15 THERMAL ANALYSIS AND RHEOLOGY

Mustafa Versan Kok Department of Petroleum and Natural Gas Engineering, Middle East Technical University, 0653 1, Ankara-Turkey 1. INTRODUCTION In recent years the application of thermal analysis techniques to the rheological behaviour of fossil fuels has gained a wide acceptance among research workers, which is of outstanding significance for industry and for the economy. Fossil fuel research has become one of the modem fields of application of thermal analysis and, over the last two or three decades, thermal analysis has passed through phases of full recognition and consolidation to rapid expansion owing to advances in several directions. New methods and techniques have been evolved and several techniques have been employed together for the determination of several parameters under identical conditions. The purpose of this chapter is to provide information on how thermal analysis techniques can play a key role in examining the rheological behaviour of crude oils. Crude oil is a mixture of paraffin, naphthenes and aromatic hydrocarbons, and small quantities of oxygen-, sulfur- and nitrogen-containing compounds and ashing solids. The majority of crude oils and crude oil products contain substantial amounts of petroleum wax. Generally two types of wax are encountered depending on the kind of crude oil. These are macro-crystalline waxes, composed of mainly n-alkanes with varying chain-length, and microcrystalline waxes which contain high proportions of iso-paraffins and naphthenes. Waxes are usehl byproducts in the petroleum industry, but at the same time they can cause a lot of trouble. Because waxy crude oils have high pour-points, their transportation, production and storage may raise a lot of problems due to wax separation and precipitation. Wax precipitation, which generally results in decreased carrying capacity of the fluid solvent, occurs when the crude oil is cooled below its cloud point. The wax constituents tend to crystallize and separate from the liquid phase causing difficulties during production, transportation and storage. The crystallization behaviour is strongly dependent on the wax distribution and composition, the asphaltene and resin

contents, and on some physical parameters, such as cooling rate, shearing and thermal cycling. Reservoir crude oils are very close to thermodynamic equilibrium, but during production, changes of temperature and pressure causes a disturbance of the equilibrium and phase separations, such as liberations of gas and precipitation of asphaltene and wax, may result. Suitable wax crystal modifiers are partly able to solve these low temperature troubles by modifying the rheological properties of waxy crude oils and by decreasing wax deposition, resulting in better pumpability. This preventive method is economical and effective. The goals of treatment of high-pour crude oils with a polymeric pour-point depressant chemical are to reduce the temperature of congealing, to inhibit waxes from precipitating out of solution, to reduce the yield strength of congealed crude oil and to decrease the viscosity. 2. PARAFFIN WAXES Paraffinic hydrocarbons or paraffins are straight-chain or branched saturated organic compounds with the composition CnHZn+>The term paraffin waxes is used for mixtures of various hydrocarbon groups, especially paraffin and cycloalkanes, that are solid at ambient temperature. Paraffins are present in large amounts in nature, but are also produced synthetically and are formed as byproducts in processing certain natural substances. Paraffin waxes, solid at ambient temperature, are obtained fiom lubricating oil fractions having various average boiling points, from distillation residues resulting from the vacuum distillation of hydrocarbon crude oils, and from the so-called tank waxes and pipeline waxes separated during the storage and transport of such crude oils. Crude oils and their products contain a large number of individual paraffins. Even the lower boiling-point fractions of petroleum contain, depending on the source of the crude oil, in addition to alkanes, varying amounts of other hydrocarbons, namely cycloalkanes and aromatic compounds. Paraffin waxes form essentially two groups known as macro-crystalline and micro-crystalline. Macro-crystalline paraffin waxes are mixtures which consist of saturated normal C18-C30hydrocarbons and smaller amounts of isoalkanes and cycloalkanes. The molecular weights of the components vary between 250 and 400 and melting points between 40-60 OC. Crystals are plate- or needle-shaped. Micro-crystalline paraffin waxes contain, in addition to normal hydrocarbons, large amounts of isoalkanes and naphthenes with long alkyl side-chains. The isoalkanes form compounds. micro crystals and the major part of these waxes consists of C40-C55 Paraffin waxes with macro-crystalline structure can be classified with respect to their melting point, or to the extent of refining. Depending on the degree of refining, one can classify paraffin waxes as technical, semi-refined and refined

grade waxes. Technical grade paraffin waxes usually contain less than 6 wt% oil, these are products obtained by de-waxing process from the slacks. Semirefined paraffin waxes may contain a maximum of 3 wt% oil and their color is light-yellow to white. Finally, refined paraffin waxes contain 0.4-0.8 wt% oil, they are completely colorless, odorless and do not contain substances detrimental to health. All paraffin waxes obtained from petroleum are crystalline below their setting point. The size of the crystals decreases with the increasing boiling point of the paraffin wax. Micro-crystalline paraffin waxes have higher molecular weights, densities and refractive indices than macrocrystalline paraffin waxes. From the point-of-view of both processing and applications, it is an important property of micro-crystalline paraffin waxes that they are capable of retaining more oil than macro-crystalline waxes.

3. EXPERIMENTAL TECHNIQUES

3.1. Introduction Determining the wax content of crude oil is of great importance for petroleum industry, especially for production, storage and transportation of waxy crude oils. There are several methods for measurement of wax content. The standard acetone method (UOP method 46-64) and its modified versions appear to be the industrial practice. In addition, gas chromatography, pulsed nuclear magnetic resonance (NMR), thermomicroscopy, and density measurements have been used for measuring the wax content. The procedure of the standard acetone method is very complex and some toxic solvents (such as toluene, benzene, etc.) have to be used. The gas chromatographic method and pulsed NMR method have poor accuracy and low repeatability. The density measurement technique requires specialized equipment. In recent years, researchers have tried to find more convenient and reliable methods to determine the wax content of crude oils. DSC is well documented as a powerful technique for investigating the characteristics of crude oils [I] and thermomicroscopy allows the observation of changes in the sample to be monitored at selected magnifications using an optical microscope. Thermomicroscopic investigations are useful for correlating morphological or structural changes with the thermal effects and can provide information regarding the internal structure of materials, including such phenomena as crystallization or phase separation. Polarized light microscopy is most commonly employed to observe anisotropic behaviour within substances which exhibit more than one refractive index. These are termed birefringent materials. Small regions within the sample may appear white or colored under polarized light. A particular molecular lattice structure is usually apparent. Other materials, such as polymers and glasses, are isotropic and will not affect the light passing through them. Phase contrast microscopy may be used to increase

the contrast within unstained specimens when the refractive index within a region of interest is very close to that of the surrounding matrix. This technique transforms the differences in refractive indices into differences in intensity of transmitted light. By combining these two methods, one can observe both wellcrystallized domains (polarized light) and amorphous or slightly crystalline fractions contained in a glassy matrix. The pour points of the samples can be determined following ASMT procedure (D97-87) and rheometry can be used to determine the rheological behaviour of the samples. 3.2. Differential scanning calorimetry (DSC) When crude oils are examined by DSC several thermal effects are observed [2]. These are i) an increase in the heat capacity around the glass-transition temperature, T,, (the glass-transition temperature of the hydrocarbon matrix of each crude oil was assumed to be the temperature at the maximum of the first derivative of the calorimetric signal, recorded during heating). Values of Tg were in the range of -120 OC to -80 OC depending on the properties of the crude oil. ii) A small exothermic effect that: occurs after the T, , due to the crystallization of species which cannot crystallize on cooling. iii) Finally a broad endothermic effect shows the dissolution of n-paraffins in the hydrocarbon matrix. Determination of the paraffin content requires the estimation of the baseline in order to determine the total effect, Q, of the dissolution of paraffins, and use of an equation AHdiss = f(T) which allows the computation of the paraffin content. The baseline for crude oils is assumed to be the line between the end of the exothermic effect after Tg, and the end of the dissolution of the paraffins. If dS is the area between the calorimetric signal and the baseline in the temperature interval T and T+dT, a thermal effect dQ can be computed as: dQ = k.dS, where dQ is the heat absorbed by the sample for the dissolution of the mass of paraffins and k is the calibration factor. Using the relationship AHdiss= f(T), a corresponding mass of paraffins is computed. Integration from the beginning to the end of the thermal effects gives the mass of paraffins. The equation AHdiss = f(T) has been established by successive approximations. The dissolution function AHdiss,fitted with the correct values of the mass of paraffin is shown in Figure 1. The crystallization temperature (T,) commonly called the wax appearance temperature (WAT) is determined from the intersection of the baseline and the extrapolation of the peak in cooling experiments. Routine use of DSC apparatus shows a thermal noise of 7 pw. Therefore, the crystallization temperature is defined as the temperature at which the thermal power developed by the heat of crystallization of paraffins is 15 pw. In order to have a reasonable sensitivity, a smaller cooling rate (-1-2 OCImin) must be used. This rate of cooling allows a better approach to the solid-liquid equilibrium temperature. Determination of

the amount of precipitated waxes thus requires the computation of the baseline and a knowledge of the experimental relation AHdiss= f(T). The mass of the precipitated waxes is directly calculated from the enthalpy measurements. The calorimetric signal above the wax appearance temperature is always a straight line corresponding to a linear variation of the heat capacity of the hydrocarbon matrix. Therefore the baseline is assumed to be a straight line that is computed by least squares fitting with the values of the calorimetric signal in the temperature range of 10 OC before the crystallization temperature. For diesel fuels, ideal behaviour of the solution of paraffins was found and a linear relationship between enthalpies and temperature was determined [3]. For crude oils, the comparison between the enthalpies of melting and precipitation shows a lowering of 20 J/g of the value of the enthalpy of precipitation. This difference can be attributed to the non-ideal behaviour of the solution of nparaffin in the crude-oil matrix. Another relation between enthalpy and temperature must thus be chosen. Using the complex mixture of paraffins, it was found that the values of enthalpies of precipitation are close to 200 J/g. This constant value is used in the computer programs for the determination of the amount of precipitated waxes. For non-doped diesel fuels, the pour point is reached when 1 wt% of paraffins contained in the he1 has precipitated. Concerning the crude oils, the knowledge of the enthalpy of precipitation allows the amount of precipitated waxes at different temperatures and the pour point to be calculated.

Figure 1. Computation of the paraffin content [4,5].

3.3. Thermomicroscopy and rheology To correlate morphological or structure changes with the thermal effects observed by DSC, thermomicroscopy experiments must be performed under the same experimental conditions. Micrographs of the crude oil samples must be taken before and after the crystallization temperature and at the pour point corresponding to 2 wt% of precipitated waxes. Waxy and high pour-point crude oils have low viscosities and behave as Newtonian fluids at high temperatures, but exhibit non-Newtonian behaviour as the crude oil is cooled, owing to the precipitation of waxes. At the pour-point temperature a solid mass is formed. For this reason transportation of these types of crude oils creates several technical and economic problems. The magnitude of this problem depends on the pour points of the samples and the ambient temperature. Therefore the rheological behaviour of the samples must be determined with rotational viscometers. With rotational viscometers, the fluids whose viscosities are to be measured are located between two rigid boundary surfaces, which are symmetrical from the point-of-view of rotation and arranged coaxially. One of theses surfaces rotates at a constant angular velocity. Moments of torque act on each boundary surface, of the same amounts but in opposite directions. From the relation between the moment of torque and the angular velocity, the viscosity is determined.

4. APPLICATIONS The majority of crude oils contain waxes which can precipitate during cold weather and cause problems such as deposition in pipelines and production equipment. Transportation of waxy crude oils creates technical and economic problems, the magnitude of which depends on several parameters such as pour point, transportation method and the ambient temperature. Therefore the wax appearance temperature (WAT), at which visible crystallization occurs, is an important parameter. For example, crude oil with a pour point of 32.2 "C will present congealing problems under most ambient conditions. It may congeal on the walls of tankers resulting in stock loss, or if it congeals in a pipeline, the required pressure to restart may exceed the burst pressure of the line. By contrast, a crude oil with a pour point of 10 "C presents few congealing difficulties, but wax may precipitate out of solution and stick to pipe walls or form a sludge at the bottom of a storage tank even in warm climates. This wax deposition can block flow lines and inhibit the performance of metering devices that measure transferred crude oil [ 6 ] . Eight crude oils from several sources, covering a wide range of fluid composition and properties, were studied using DSC, thermomicroscopy and viscometry. The paraffin contents of the crude oils were determined by DSC

analysis and the pour points of the samples were determined following the ASTM (D97-87) procedure (Table 1). The DSC was calibrated for temperature and heat flow measurements using the melting point and heat of melting of high purity compounds. The apparatus was flushed with argon. 20-25 mg samples of crude oils were contained in sealed aluminum crucibles. Experiments were performed on cooling in the range of +80 "C and -20 "C at 2 "Clmin. cooling rate. Table 1. Prouerties of crude oil sam~les

[ Crude Oil

Crude Oil-1 Crude Oil-2 Crude Oil-3 Crude Oil-4 Crude Oil-5 Crude Oil-6 Crude Oil-7 Crude Oil-8

I OAPI Gravity 39.0 36.3 31.1 40.0 38.3 33.6 28.9 35.2

I % Paraffin (DSC) 24.3 12.9 8.7 16.9 17.6 22.3 7.8 11.8

I Pour Point PC 24.0 18.0 -27.0 3.0 15.0 28.0 -27.0 6.0

In DSC experiments, wax appearance temperatures (WAT) were determined manually by intersection of the baseline and extrapolation of the peak. Repeatability of the WAT determinations depends on the shape of the DSC profile of the crude oil and the accuracy can be estimated to be in the order of 2 "C. Typical DSC curves of different crude oils are given in Figure 2. Pedersen et al. [7] measured the wax precipitation temperatures on sixteen crude oils by DSC and concluded that the amount of wax precipitated as a hnction of temperature at atmospheric pressure can be predicted with reasonable accuracy. Hansen et al. [8], used DSC to study wax precipitation from a series of North Sea crude oils by measuring glass-transition temperatures, wax precipitation and dissolution temperatures and wax precipitation and dissolution enthalpies in the temperature range from +70 to -140 OC. Measured precipitation and dissolution enthalpies based on total sample amounts ranged from 2.3 to 41.0 Jlg and from 1.3 to 44.5 Jlg, respectively. A similar study was performed by Jayalakshmi et al. [9], where paraffin waxes from different crude oil sources were characterized using DSC and high-temperature capillary GC techniques. DSC curves of the waxes were correlated with their n-paraffin contents. This correlation was further validated by obtaining n-paraffin contents using a high- temperature GC technique. Elsharkawy et al. [lo] measured the wax contents of crude oils by acetone precipitation techniques, as well as wax appearance temperatures (WAT) by viscosity measurements and DSC, of eight different stock-tank crude oils. Comparison of WATs measured by DSC and viscometry indicates that the viscosity method overestimates the WAT.

Comparison between predicted and measured results showed that the WAT measured by DSC compares very well with that predicted from the model for most crude oils. Srivastava et al. [I 11 used DSC to study the temperatures and enthalpies of wax crystallization and of melting in the middle distillate obtained from crude oil. The variations in WAT with wax concentration, as measured by DSC, were identical with those measured by optical microscopy and the ASTM cloud-point method. DSC values were, however, lower than microscopic values and higher than ASTM cloud-point measurements. Jiang et al. [12] used temperature modulated differential scanning calorimetry (TMDSC) to measure the wax appearance temperature (WAT) of crude oil samples. Changes in the TMDSC signals exhibit excellent correlations with WATs measured using conventional DSC. It was observed that, for oil samples having low wax contents, TMDSC is a more sensitive technique for identifying the onset of wax crystallization. Another new DSC method to measure the wax content of crude oils was developed by Chen et al. [13]. Fourteen crude oils were studied and the wax contents determined were in good agreement with those determined by the standard acetone method. This method has an advantage over reported DSC methods in which the exact dissolution or precipitation enthalpy of wax is required. In addition and more significantly, the new method can be applied to improve the accuracy in determining the amount of precipitated wax in a waxy crude oil at different temperatures.

f-'

A

--

6 c-7;

CRUDE 4

I

I

+40

+30

+20

+10

TEMPERATURE ("C)

Figure 2. DSC curves of different crude oils [4,5]

0

Jimenezmateos et al. [14] used temperature modulated differential scanning calorimetry (TMDSC) for the first time to measure the wax appearance temperature (WAT) of crude oil samples. Changes in the TMDSC signals exhibit excellent correlations with WATs measured using conventional DSC. For oil samples having low wax contents, TMDSC is a more sensitive technique for identifying the onset of wax crystallization. The phase angle between the heating rate and the temperature modulations is extremely sensitive to this process. Thermomicroscopy experiments were performed with a microscope which was equipped with polarized light and phase contrast devices, and a hot stage unit which can operate in the range of 300 OC to -70 OC. By this method, well-crystallized areas, corresponding to the precipitated waxes in the hydrocarbon matrix, can be observed. The crude oil sample (-7-8 mg.) was put in to a glass crucible and cooled from +80 OC to -20 OC at 2 OClmin. Measurements of the intensity of transmitted light through the crude oils used are given in Figure 3. During the experiments, vaporization of light components of the crude oil can occur, leading to a decrease on the solubility of waxes and a value of WAT higher than the real value. Determination of the WATs on the same crude oil samples were repeated several times and no significance deviation in the values was observed. Repeatability of the WAT determinations by this technique can be estimated to be of the order of 2 OC. As expected, the size and number of crystals increases because continuous precipitation of paraffins occurs in the sample during cooling. In all experiments, very small crystals are observed. These observations suggest that each crude oil contains nuclei in the liquid matrix. When paraffins precipitate, heterogeneous germination occurs and small crystals are always obtained. In a similar study, Letoffe et al. [15] characterized the mixtures of 2 and 4 wt% of pure paraffins in a crude oil matrix and fourteen crude oils by DSC and thermomicroscopy at a cooling rate of 2 "Clmin in the range from +80 to -20 OC. By thermomicroscopy, it was observed for mixtures of pure paraffins and crude oil matrices that the size of the crystals is small and depends on the length of paraffinic chains at the pour point. For crude oils, at the pour point the same final state is obtained (unlike diesel fuels). Li et al. 1161 pointed out that the amount of precipitated wax is one of the key factors that govern the flow properties of waxy crude oils. Experimental results for twenty-four crude oils have shown that approximately 2 wt% precipitated wax is sufficient to cause gelling of a virgin waxy crude. Accordingly, a correlation was sought between the pour point and the temperature at which 2 wt% of wax has precipitated out from a crude oil. The development of the gel-point correlation and further verification of the pourpoint correlation indicate that there is a relationship between the gelling of virgin waxy crude oils and the amount of precipitated wax. Experimental results

showed that approximately two or three times the amount of precipitated wax was present at the gelling temperature when the oils were in their beneficiated conditions.

CRUDE 1

A CRUDE 3

35°C

-----------

30°C

I

I

I

I

I

+45

+40

+35

+30

+25

TEMPERATURE ("C)

Figure 3. Intensity of transmitted light vs temperature for different crude oils [4,5]

Rheology experiments also give information in the determination of wax appearance temperatures of crude oils. In this research, WATs of crude oils were determined by viscometry from the point where the experimental curve deviates from the extrapolated Arrhenius curve (Figure 4). It was observed that all crude oils, except highly asphaltenic samples, are Newtonian fluids above their wax appearance temperatures. The flow behaviour of crude oils is considerably modified by the crystallization of paraffins corresponding to the variation of the apparent viscosity with temperature. Below the WAT, flow becomes nonNewtonian and approaches that of the Bingham and Casson plastic model [17,18]. Visintin et al. [19] investigated a waxy crude oil which gels below a threshold temperature under static and dynamic conditions, using a combination of rheological methods, optical microscopy, and DSC. The gels displayed a strong dependence of the yield stress and modulus on the shear history, cooling rate, and stress loading rate. Gelation of the waxy crude oil studied was suggested to be the result of the association between wax crystals, which produces an extended network structure, and the system displayed features common to attractive colloidal gels. Fasano et al. [20] reviewed a series of mathematical models formulated for the flow of waxy crude oils, that is, of mineral oils with a high content of paraffinic hydrocarbons which may be dissolved or segregated as solid crystals at sufficiently low temperatures. The crystals have a tendency to form aggregates, producing a gel-like structure. The resulting product can be modeled as a Bingham fluid, but its rheological parameters depend on the amount and state of the segregated phase, whose evolution is in turn influenced by the flow. Wax can form a solid deposit at the pipe wall, reducing the pipe radius, and this phenomenon is also taken into account. The models presented have different degrees of complexity, depending on which phenomena they include. In presenting each of them, their expected ranges of validity are discussed.

I

I

I

I

I

I

+50

c45

+40

+35

+30

+25

TEMPERATURE ("C)

Figure 4. Apparent viscosity vs temperature for different crude oils [4,5]

Kane et al. [21] studied the relation between the rheological properties and the structures of waxy crude oils cooled in quiescent conditions and under flow. In the dynamic regime, measurements were performed on gels cooled slowly in order to allow paraffin crystallization to reach equilibrium. The amounts of crystalline material at various temperatures were measured independently by DSC or NMR and the morphology was observed by transmission electron microscopy. The relation between the shear modulus and the crystal content was established and analyzed in the framework of colloidal gels with a fractal structure. A series of experiments is presented, with various shearing protocols, leading the oils into different states. Steady viscosities were measured under various shear rates and at constant temperatures. Cooling was performed under constant shear stresses. Li and Zhang [22] developed a model based on shear-rate dependent viscosity by applying the theory of suspension rheology. This model is characterized by its capability to predict viscosities of crude oils with various thermal and shear histories and beneficiated with pour-point-depressants (PPD). Once viscosities are known at only two temperatures above the wax appearance temperature, and apparent viscosities at one temperature in the non-Newtonian regime, viscosities, or apparent viscosities, at any temperatures above the gel point can be predicted by using the model, together with the concentration of precipitated wax at that specified temperature. Furthermore, parameters of rheological models, such as the consistency coefficient, K, and the flow behaviour index, n, of the power law model may be obtained by regression of the predicted viscosity data and corresponding shear-rates. Wax appearance temperatures of crude oils determined by DSC, thermomicroscopy and viscometry (Figure 5) are given in Table 2. An examination of the experimental curves obtained by the three different techniques shows a slow precipitation of waxes which cannot be detected by thermomicroscopy and viscometry. For thermomicroscopy and viscometry, good correlation is seen for all the crude oil samples studied. In some cases, WATs from thermomicroscopy and viscometry are higher than DSC values. This fact can be explained by the appearance of non-crystalline particles (asphaltene) in the crude oils which cannot be detected by DSC. Thermomicroscopy will, however, detect a change in the intensity of transmitted light. Both techniques, DSC and thermomicroscopy, should be used together to provide a better understanding of the thermal behaviour of crude oils on cooling. Ronningsen et al. [23] compared three different methods for determination of the wax precipitation temperature (WPT), namely polarization microscopy, DSC and viscometry. Microscopy invariably gave the highest WPTs and probably the most relevant values for predicting the onset of wax deposition on cold surfaces.

TEMPERATURE ("C)

Figure 5. Determination of wax appearance temperatures of Crude oil 5 using DSC, viscometry and thermomicroscopy [4,5]

The WPTs from microscopy were found to depend on factors such as thickness of the sample film and cooling rate. Thermomicrosco

Crude oils contain substantial amounts of waxy materials. When a crude oil containing wax is cooled below its cloud point, wax constituents tends to separate from the liquid phase of the crude oil and start to crystallize and this causes several problems during production, storage and transportation. The crystallization behaviour of crude oils is mainly dependent on the wax composition and distribution. This type of crude oil has low viscosity and behaves as a Newtonian fluid at high temperatures, but, during cooling, owing to the precipitation of waxes, the oils show non-Newtonian behaviour. To decrease the viscosity of the crude oil and to lower the congealing temperature and yield strength, chemical additives and pour-point depressants should be used. Their structure is mostly polymeric and improves the flow properties of crude oils and decreases the surface deposition of crystals. Among the polymers and copolymers claimed to improve the low temperature behaviour of waxy crude oils, only two types seem to be generally used. These polymers are polyalkylacrylates, or copolymers, containing esterified derivatives of maleic anhydride [24]. Efficient additives for modifying the wax crystallization are also able to reduce the pour point and viscosity at low temperatures. Three different additives (Additive 1, 2 and 3) were used in order to see their effects on Crude oil-5. The viscosity of the crude oil is decreased considerably by treatment with the additives. Measurements of the shear stress-shear rate relationships were performed at constant temperature and experimental data were fitted to the Bingham plastic model using a linear regression program. When the amount of additive was increased from 500 ppm to 1000 ppm, the pour point of the crude oil was decreased significantly (Table 3). Denis and Durand [25] determined the low temperature properties of petroleum products that had been improved either by refinery processing or by adding wax

crystallization modifiers. They concluded that the crystallization of paraffin in waxy crude oils and petroleum products are governed by three phenomena: nucleation, growth and agglomeration or gelling.

Coutinho and Daridon [26] discussed and compared several cloud-point measurement techniques. Some of these techniques, such as viscosity, filter plugging, and DSC can only be used under very favorable circumstances, but because every technique requires some finite, often large, amount of solid to detect the presence of a new phase, the cloud point, defined as the temperature for which the first solid appears in the oil, is not accessible experimentally. So, unless a very detailed compositional analysis is available, it is impossible to predict the cloud point accurately with a thermodynamic model. The effect of the paraffin distribution in the oils on the cloud point detection is discussed, and it was shown that the compositional information can be used to assess the uncertainty of the measured cloud points. Karacan et al. [27] suggested the use of X-ray computerized tomography (CT) as an alternative tool for cloud-point determinations of crude oils and dark fuel oils and presented the results for artificially prepared transparent oils. The technique is fully computerized and data gathering and analysis are achieved by taking the advantage of the processing of the CT images. The cloud points of diesel oil samples containing 5%, 10% and 15% additional wax were determined with this new technique. Results showed that the cloud points determined with this technique and the standardized ASTM D-3 117 method were very close. This encourages the use of the proposed technique to determine the cloud points of transparent distillate fuels and dark fuel oils and crude oils, whose cloud points cannot be determined easily and accurately. Kruka et al. 1281 proposed that the conventional ASTM procedures for cloud-point determination are not applicable to dark crude oils and also do not account for potential sub-cooling of the wax. A review of possible methods and testing with several crude oils indicate that a reliable method consists of determining the temperature at which wax deposits begin to form on a cooled surface exposed to warm, flowing oil. A concurrent thermal

analysis of the waxy hydrocarbon can indicate the presence of possible multiple wax-precipitation temperature regions in the solution.

5. CONCLUSIONS The results indicate that, for crude oil samples with different contents of wax, differential scanning calorimetry (DSC), thermomicroscopy and rheometry provide excellent methods of measuring the wax appearance temperatures (WATs). Because of the diversity of the results, it is not possible to affirm which technique is most suitable for determining the onset of wax appearance. It is suggested that DSC and thermomicroscopy should be used together for a better understanding of the determination of WATs of crude oils. Viscometry should be used to study the flow properties of crude oils below the WAT, in particular when flow improvers are added. 6. REFERENCES 1. M.V. Kok, J.M. Letoffe, P.Claudy, D. Martin, M. Garcin and J.L. Volle, J. Therm. Anal. Calorim., 49 (1997) 727. 2. P. Claudy, J.M. Letoffe, B. Chague and J. Orrit, Fuel, 67 (1988) 58. 3. P. Claudy, J.M. Letoffe, B. Neff and B. Damin, Fuel, 65 (1986) 861. 4. M.V. Kok, J.M. Letoffe and P.Claudy, D. Martin, M. Garcin and J.L. Volle, Fuel, 75 (1996) 787. 5. M.V. Kok, J.M. Letoffe and P.Claudy, J. Therm. Anal. Calorim., 56 (1999) 959. 6. D.S. Schuster and J.H. Magill, Polymers as Rheology Modifiers, (1991) 301. 7. K.S. Pedersen, P. Skovborg and H.P. Ronningsen, Energy and Fuels, 5 (1991) 924. 8. A.B. Hansen, E. Larsen, W.B. Pedersen, A.B. Nielsen and H.P. Ronningsen, Energy and Fuels, 5 (1991) 914. 9. V. Jayalakshmi, V. Selvavathi, M.S. Sekar and B. Sairam, Petroleum Sci. Technol., 17 (1999) 843. 10. A.M. Elsharkawy, T.A. Al-Sahhaf and M.A. Fahim, Fuel, 79 (2000) 1047. 11. S.P. Srivastava, T. Butz, P.S. Verma, R.C. Purohit and I. Rahimian, Petroleum Sci. Technol., 20 (2002) 83 1. 12. Z. Jiang, J.M. Hutchinson and C.T. Imrie, Fuel, 80 (2001) 367. 13. J. Chen, J.J. Zhang and H.Y. Li, Thermochim. Acta, 410 (2004) 23. 14. J.M. Jimenezmateos, L.C. Quintero and C. Rial, Fuel, 75 (1996) 1691. 15. J.M. Letoffe, P.Claudy, M.V. Kok, M. Garcin and J.L. Volle, Fuel, 7 (1995) 810. 16. H.Y. Li, J.J. Zhang and D.F. Yan, Petroleum Sci.Technol., 23 (2005) 1313.

17. E.C. Bingham, Fluidity and Plasticity, ,McGraw-Hill, New York, 1962. 18. N. Casson, Rheology of Disperse Systems, Pergamon Press, New York, 1959. 19. R.F.G. Visintin, R. Lapasin, E. Vignati, P. D'Antona and T.P. Lockhart, Langmuir, 21 (2005) 6240. 20. A.Fasano, L. Fusi and S. Correra, Meccanica 39 (2004) 441. 2 1. M. Kane, M. Djabourov and J.L. Volle, Fuel, 83 (2004) 1591. 22. H.Y. Li and J.J. Zhang, Fuel, 82 (2003) 1387. 23. H.P. Ronningsen, B. Bjorndal, A.B. Hansen, E. Larsen and W.B. Pedersen, Energy and Fuels, 5 (1991) 895. 24. B.J. Musser and P.K. Kilpatrick, Energy and Fuels, 12 (1998) 715. 25. J.Denis and J.P.Durand, Revue de L'Institut Du Petrole, 46 (1991) 637. 26. J.A.P. Coutinho and J.L. Daridon, Petroleum Sci. Technol., 23 (2005) 1113. 27. C.O. Karacan, M.R.B. Demiral and M.V. Kok, Petroleum Sci. Technol., 18 (2000) 835. 28. V.R. Kruka, E.R. Cadena and T.E. Long, J. Petroleum Technol., 47 (1995) 681.

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter I6 POLYMORPHISM Mino R. Caira

Department of Chemistry, University of Cape Town, Rondebosch 7701, South Africa 1. INTRODUCTION

The ubiquity of crystal polymorphism, coupled with novel methodology for generating polymorphs, high-throughput polymorph isolation, and increasing regulatory demands on the need for control of crystallized forms in many areas of solid-state chemistry, presents on-going challenges in the identification of new polymorphs and their comprehensive structural and thermal characterization. To achieve such 'comprehensive' characterization, polymorphic systems ideally require investigation by a battery of techniques that includes diffraction and spectroscopic methods for structural elucidation, thermal analysis methods to identify phase transformations and establish thermodynamic relationships among the various crystalline forms, and dissolution techniques to determine solubility trends. To these can be added theoretical techniques, developed more recently, and aimed at the prediction of the crystal structures of new polymorphs, as well as their morphologies. To a large extent, this holistic approach to the study of crystal polymorphism has been driven by regulatory bodies responsible for ensuring robust manufacture, safety and efficacy of pharmaceutical products [I]. In this chapter, describing the use of thermal methods of analysis to investigate polymorphism, information gleaned from complementary techniques will therefore also feature. In particular, since X-ray diffraction still retains its status as the defmitive method for polymorph identification, studies describing the complementary use of thermal and X-ray methods will be cited frequently. Furthermore, while a diversity of compounds is addressed here, systems of pharmaceutical relevance will be encountered much more frequently since this is an area of particular innovation as far as polymorphic research is concerned. In a treatment of polymorphism and its investigation by thermal methods of analysis, a number of closely related solid-state phenomena cannot be neglected. Thus, systems displaying solvatomorphism (the ability of a substance to exist in two or more crystalline phases arising from differences in their solvation states) molecular inclusion and isostructurality (the inverse of polymorphism), are also given due attention in this chapter.

Several reviews on the applications of thermal analysis to the elucidation of polymorphism and polymorphic systems have appeared during the last five years. A selection of these and a summary of their contents follow. Giron [2,3] has emphasized the frequency and relevance of polymorphism and solvatomorphism for pharmaceuticals, indicating the importance of the manufacture of the various solid phases of a drug, as well their characterization for gaining an understanding of the thermodynamic and kinetic factors involved in the process parameters of both the API (active pharmaceutical ingredient) and the dosage form. The theoretical distinction between enantiotropy and monotropy, as well as its practical determination from DSC traces, are described [2]. (This continues to be a crucial aspect that guides the selection of appropriate solid phases for formulation and is discussed in the present chapter with illustrative case studies). Combined use of DSC, microcalorimetry and TG with microscopy, XRD and spectroscopic techniques in the investigation of pharmaceutical polymorphism was described as 'state of the art', allowing reliable interpretation and analytical quantification. More recently, Urakarni reviewed the use of isothermal calorimetry for estimating enthalpies of solution and transition of pharmaceutical polymorphs [4]. Ampoulebased isothermal microcalorimetry sheds light on the kinetics and thermodynamics of polymorphic transitions, allowing an assessment of the feasibility of employing metastable drug polymorphs in formulations. Phipps has also treated applications of isothermal microcalorimetry techniques in the pharmaceutical setting [5] emphasizing their use in studying drug polymorphic transformations, as well as in the detection of low levels (
The next section of this chapter describes significant recent developments in research on crystal polymorphism, with an emphasis on new methods of producing polymorphs because this has important implications for their rapid characterization by analytical methods. In the subsequent section, the role of thermal analysis in elucidating the nature of polymorphs, solvates, and their transformations will be outlined. The major section of this chapter will describe recent case studies illustrating the use of thermal methods and complementary techniques to identify polymorphs and clarify thermodynamic relationships in polymorphic systems. 2. RECENT DEVELOPMENTS IN POLYMORPHIC RESEARCH 2.1. Introduction At least three areas of research on crystal polymorphism merit discussion and updating here. These include recently discovered procedures for generating polymorphs, high-throughput screening for polymorphism and the prediction of polymorphic structures from single molecules based on computational methods. All of these developments are contributing to a proliferation of crystal forms, especially of drug candidates, necessitating more rapid identification and characterization using thermal analysis and allied techniques. 2.1. I . Polyrnorph preparation Guillory has provided a usefkl summary of procedures for generating polymorphs and solvates of organic compounds [lo]. The methods described for obtaining polymorphs include sublimation, crystallization from a single solvent, evaporation from a binary mixture of solvents, vapour difhsion, thermal treatment, crystallization from the melt, rapidly changing solution pH to precipitate acidic or basic substances, thermal desolvation of crystalline solvates, growth in the presence of additives, and grinding. Solvated forms may arise from the recrystallization experiments referred to above and in general have the potential to form whenever solvent molecules are available in the liquid or vapour states. Solvates can also be obtained via solvent exchange e.g. the formation of a methanol solvate by recrystallization of a hydrate form dry methanol, or the formation of an ethanolate by exposing a methanolate to a saturated atmosphere of ethanol. Related to the preparation of polymorphs and solvates is the generation of amorphous forms, where the most common procedures are solidification of the melt, reduction of particle size, spray-drying, lyophilisation, removal of solvent fkom a solvate, and precipitation of acids and bases by pH alteration. Braga and Grepioni have recently described the generation of novel solid

forms, including polymorphs, via solid-vapour reactions and crystal-seeding, especially in the context of organometallic chemistry [ l 11. While many literature reports and patents refer to highly specific recipes for obtaining unique crystalline forms of parent molecules, several more general procedures for producing polyrnorphs have been published in recent years. Some of these focus on the isolation of hitherto unknown metastable polymorphs of pharmaceuticals, because their higher solubilities may lead to improved bioavailability. Methods using supercritical fluids (SFs) to control crystal form are gaining in popularity [12-141 owing to the possibility of reproducible preparation of specific polymorphs by appropriate adjustment of the operating conditions such as temperature, pressure, substrate concentration, flow ratio of SF to substrate solution, and the nature of the solvent. A recent example involves the preparation of a new, metastable polymorph of deoxycholic acid (DCA, Figure 1, 1) using supercritical C02 [14]. The new form was detected by the appearance of an exothermic peak at 428 K in DSC curves followed by the fusion of bulk DCA at 448 K. Solvent-drop grinding is a novel, efficient and 'green' procedure for polymorphic control [15,16]. It involves grinding a known polymorph of a compound with a very small amount of solvent added in order to effect a polymorphic transformation. Variation of the solvent may lead to different polymorphic outcomes. For example, polymorphic Form I of anthranilic acid (Figure 1, 2), ground in the presence of heptane, transforms to Form 11, which in turn converts to Form I11 when ground in the presence of chloroform [15].

Figure 1. Chemical structures of DCA 1, anthranilic acid 2, tolbutamide 3, glycine 4.

Isolation of intermediate, metastable polymorphs can generally be achieved by inhibiting their transformation to stable forms through the mediation of additives. A recent report on the isolation of a metastable form (Form IV) of tolbutamide (Figure 1, 3) is a case in point [17]. This species crystallized exclusively from an aqueous solution containing dimethylated P-cyclodextrin (DMB), whereas the stable form (I) crystallized in the absence of DMB. The proposed mechanism involves inhibition of the solution-mediated transformation of Form IV to Form I by complexation of tolbutamide with DMB. While not yet established as a general method, it is likely that this approach will be applied to the isolation of metastable polyrnorphs of other organic substrates. A method based on exposing a solution of a substrate to a very high, static electric field of 500 000 Vlm to induce crystallization of a specific polymorph has been described [IS] and shown to effect nucleation of the y-polymorph of glycine (Figure 1, 4). The same polymorph has been shown to crystallize selectively when a supersaturated solution of glycine is irradiated with laser light [19]. Though currently limited in their application, these methods may likewise find more widespread use in the future.

-

2.1.2. High-throughput screening In contrast to traditional manual methods of generating polymorphs, modern highthroughput (HT) screening is performed via multiple parallel crystallizations which use relatively small amounts of material and explore a wide variety of physical conditions (variations in solvent, temperature, substrate concentration), typically using commercially designed, multiple-well apparatus. This approach is having a dramatic impact on the diversity of crystalline forms that may be isolated from a single substrate. In a recent assessment of the effectiveness of HT-screening [20], it was shown that while manual screening procedures involving -100 crystallization experiments yielded nine distinct crystalline forms of a drug candidate, the HT approach based on 1500 crystallization trials yielded twice as many distinct solid forms. Significant reductions in both the timescale as well the amounts of material used in HT-screening are important advantages. Novel crystal forms were identified using Raman spectroscopy, thermal analysis and powder X-ray diffraction (PXRD). 2.1.3. Polymorph prediction The prediction of the crystal structures that a given molecule may assume remains a challenging problem in the research on polymorphs, especially for molecules having a high degree of conformational flexibility [21]. More particularly, while the use of computational methods based on minimisation of lattice energy may lead to a series

of potential crystal structures (polymorphs), predicting which of these will crystallize under given experimental conditions is generally not possible due to the fact that both thermodynamic and kinetic factors determine the outcome. However, the finding, based on theoretical calculations, that a particular molecule can assume several hypothetical polymorphic structures with energies spanning a narrow range does give an indication of its potential for polymorphism, and may encourage fixther attempts to isolate new polymorphs experimentally. In this sense, polymorph prediction may also contribute to the proliferation of new crystal forms. A striking example of this relates to aspirin (Figure 2, S), which has been intensively studied in an effort to discover novel crystalline forms that might have superior delivery properties to those of the original Form I. Despite systematic studies performed in the 1960s and 1970s aimed at isolating such forms, only an indication of the possible existence of a metastable polymorph was forthcoming. In 2004, a computational study led to the prediction of a low energy polymorph of aspirin having a small shear constant that suggested a low barrier to transformation [22]. Very recently, the 'predictably elusive7Form I1 of aspirin was finally isolated [23]. The X-ray analysis of the new form confirmed that its crystal structure is consistent with that predicted. In this case, isolation of the new polymorph was serendipitous, this form appearing on attempted co-crystallization of aspirin and either acetamide or the drug levetiracetam. Under ambient conditions, aspirin Form I1 crystals convert to Form I. These forms are distinguishable from their DSC traces which reveal endotherms at 135.5 and 143.9 "C respectively. In concluding this section, the issue of co-crystallization merits further mention because this has recently emerged as a h i t f u l means of generating novel crystalline forms of organic compounds, especially those with potential for pharmaceutical development. Currently, pharmaceutical co-crystals are regarded as entities formed between a molecular or ionic active pharmaceutical ingredient and a co-crystal former that is a solid under ambient conditions [24]. Examples of such species are the 1:l piracetam-gentisic acid, the 2: 1 carbamazepine-benzoquinone, and the 1:1 sulfadimidine-anthranilic acid co-crystals (Figure 2, 6), in all of which, hydrogen bonding is responsible for binding the partner molecules. Given the large number and structural variety of co-crystal formers that have GRAS ('generally regarded as safe') status, as well as the possibility of a range of stoichiometries for pharmaceutical co-crystals, it is clear that the potential exists for generating very large numbers of novel candidates in this class. Such co-crystals themselves may display polymorphism andlor solvate formation, yielding yet more possible solid forms with potential practical application. As for any solid intended for medicinal use, comprehensive physicochemical characterization of each such form

would be mandatory. Thus, in addition to the developments in polymorphic research referred to earlier, the increasing level of active discovery of pharmaceutical cocrystals places yet more pressure on the analyst to identify and characterize new compounds rapidly and reliably using thermal analysis and allied techniques.

Figure 2. Chemical structures of aspirin 5 and the sulfadimidine-anthranilic acid cocrystal 6 . 3. THERMAL ANALYSIS IN STUDIES OF CRYSTAL POLYMORPHISM

3.1. Introduction Some of the examples quoted earlier alluded to the use of thermal methods of analysis to distinguish crystal polymorphs. However, the wealth of information on polymorphic, solvatomorphic and related systems that is accessible by thermal analysis goes far beyond that of simple identification. In this section, a brief survey of the multiple uses of hot stage microscopy (HSM), thermogravimetric analysis (TG) and differential scanning calorimetry (DSC) in the investigation of these systems follows. An introduction to the thermodynamic principles underlying polymorphism is also provided because this is required for an understanding of the role of the DSC method, especially in connection with the derivation of energy-temperature diagrams [25]. Detailed treatments of the theory of polymorphism and the characterization of polymorphs and solvates using thermal analysis have been published [26,27].

3.1.1. Analysis ofpolymorphs HSM permits visual ob'servation of morphology, colour changes, phase transitions, recrystallization phenomena, fusion and decomposition, together with the recorded temperatures of these events, during programmed heating of crystals of a polymorph. These observations are frequently used to support the interpretation of thermal events

observed in TG and DSC curves. Major advantages of HSM include the small sample required and the rapidity of measurement. TG on polymorphs, in principle, yields zero mass loss prior to compound decomposition but may reveal traces of surface solvent. The major utility of TG is in the investigation of solvates (see below). The DSC technique remains central to polymorphic investigation, allowing accurate measurement of onset and peak temperatures, as well as enthalpy changes for polymorphic transitions and fusion. For a pharmaceutical displaying polymorphism, it is vital to establish the relative thermodynamic stabilities of the various polymorphs identified in order to guide the selection of the most appropriate one for dosage formulation. This is usually the most stable polymorph, because it has no tendency to transform, thereby ensuring constancy of the product performance. However, metastable forms (with higher solubility and bioavailability) may be preferred and are used if they can be stabilised. Determining the stability relationships among the various polymorphs is therefore a fimdamental requirement and is usually achieved using data drawn from DSC analyses and any other supporting information (e.g. relative stabilities determined from observed solvent-mediated polymorphic transformations). An outline of the thermodynamic considerations follows. Relative stabilities are usually discussed in terms of the Gibbs free energy G (= H - TS), the most stable polymorph possessing the lowest value of G (and the lowest solubility). In a dimorphic system at constant pressure, the plot of the Gibbs free energy of each polymorph as a function of temperature has a slope (dGi/dT), = -Si which is negative. At the point of intersection of these G-T curves, the two phases they represent are in thermodynamic equilibrium at their transition temperature T,. Each polymorph thus has a temperature range within which it is stable. Here, the distinction between enantiotropically and monotropically related polymorphs is crucial. Their depiction in schematic Gibbs free energy-temperature diagrams is shown in Figure 3. Enantiotropy refers to the case where two polymorphs can interconvert at a specific transition temperature (T,) below the melting points of the individual polymorphs (Figure 3 (a)). Specifically, at temperatures below T,, the lower-melting polymorph is the stable form while above T,, the higher-melting polymorph is stable. If T, happens to be near to ambient temperature, local temperature fluctuations may result in undesirable transformation of one polymorph into another. If the compound in question is a drug, its delivery properties could be seriously compromised in this situation. In contrast, for polymorphs that are monotropically related, there is no observable transition temperature; one polymorph (the higher-melting form) is stable up to the melting point while the other (lower-melting) is always metastable with respect to it. In this

case, the only spontaneous transition that can occur is that fiom the lower-melting form to the higher-melting form. Giron has summarised the experimental DSC profiles that can be expected for both enantiotropy and monotropy [2], in each case considering experiments in which the DSC trace is recorded starting with either the stable or the metastable polymorph. These DSC traces provide the necessary experimental data, including transition temperatures and enthalpies, to enable elaboration of the basic G-T curves shown in Figure 3(a), (b) by addition of the corresponding H-T curves for each phase present, including that for the common liquid phase. The resulting 'energy-temperature' diagram, of which Figure 3(c) is an example, is an extremely valuable record of the relative stabilities of the various polymorphs present under given conditions. In the hypothetical case shown below, polymorphs I and I1 are enantiotropically related, while polymorph 111 is metastable with respect to both I and I1 over the temperature range for which data for I11 were recorded. The enantiotropic relationship between I and I1 is typically deduced by direct observation of an endotherm (AHt = HII - HI > 0 at TJ upon heating the stable form I, corresponding to its transition to Form I1 [ 2 ] .On cooling 11, the corresponding exotherm for its transformation to I should in principle be observed, reversibility being the hallmark of enantiotropy. The value of T, might have been alternatively deduced fiom the intersection of the van't Hoff plots of ln(solubi1ity) versus 1/T for the two polymorphs, and AHt from the difference between the enthalpies of dissolution.

(b)

GI % stable

t I

I,

TI, 11 TI, I

T

Figure 3. Schematic energy-temperature relationships for (a) an enantiotropic system, (b) a monotropic system, (c) a hypothetical trimorphic system.

In Figure 3(c), the arrows denote the relative sizes of the enthalpies of fusion of Forms I and 11. Here, the enthalpy of fusion of the lower melting form exceeds that of the higher melting form, a requirement for enantiotropy, as embodied in the 'heat of fusion rule', one of four thermodynamic rules established by Burger and Ramberger [28]. These rules relating to polymorphic transitions have been surnrnarised and discussed in more detail elsewhere [29]. Their utility in constructing energytemperature diagrams should be evident from the above discussion. The application of these rules was extended by Yu a decade ago [30]. This work led to formulas for calculating AG, the Gibbs energy difference between polymorphs, as well as its temperature dependence, AG(T). This allows estimation of the relative stabilities of two polymorphs by extrapolation of the value of AG(T) to any temperature of interest. The transition temperature (T,) is thus easily obtained by setting AG(T) = 0. This approach has been used frequently, especially to estimate the virtual temperature of intersection of the G-T curves for a monotropic system, a point that is required to define the corresponding E-T diagram completely. An alternative method to using the van't Hoff plot for estimating the transition temperature for polymorphic pairs is that using enthalpies of solution and solubility data. An account of this procedure appeared recently 1311. However, this was found to be equivalent to that described previously by other researchers [32]. In summary, derivation of the energy-temperature diagram for polymorphic systems, firmly based on accurate DSC data, other supporting information and the theoretical considerations outlined above, is a primary goal in many investigations of drug polymorphism. Studies reporting such diagrams and their relevance are cited later in this chapter. 3.1.2. Analysis of solvates As noted in the Introduction to this chapter, a treatment of the thermal analysis of polymorphism inevitably embraces related phenomena such as solvatomorphism, which occurs frequently for compounds in all classes. (This phenomenon has also been called 'pseudopolymorphism' but there is much controversy in the current literature concerning the use of this term [33]). Desolvation of solvated compounds is one of the procedures for deriving polymorphs and indeed may produce either a polymorph that cannot be obtained by other means, or a known polymorph whose texture is different from that of the same form prepared by recrystallization from solution. An interesting and commercially important example of the latter relates to a 1,Cdioxane solvate of the analgesic paracetamol, whose desolvation yielded a polymorph with altered texture and consequently superior technological properties to those of the same (monoclinic) polymorph obtained in the commercial preparation of

this drug [34]. Pharmaceutical hydrates represent an extremely important class of solvates owing to their frequent incorporation in drug formulations, their possible inadvertent dehydration to unstable polymorphic forms in response to changes in local conditions or during processing, and the possible difficulty of maintaining a specific hydrate isolated in the laboratory during scale-up operations [35]. Such hydrates may themselves display polymorphism; nitrofurantoin, for example, has been isolated as two distinct monohydrates crystallizing in the monoclinic and orthorhombic systems [36]. With the increasing appearance of solvated forms of both organic and inorganic compounds, instances of isostructurality are becoming more frequent [37]. This term applies to a series of solvates of a given parent or 'host' compound whose members crystallize with the host molecules adopting essentially the same three-dimensional structural arrangement, and in which the various guest molecules (typically solvent molecules) are accommodated in crystallographically equivalent voids. Figure 4 is a schematic diagram representing two isostructural crystals. Well-known examples, studied by several groups, include series of solvates of the bile acids deoxycholic acid (DCA, Figure 1) and cholic acid (CA). In the solid-state, these host molecules assemble in bilayers with the guest molecules enclosed within common cavities or helical channels. CAYfor example, forms an isostructural pair with the guests ethyl acetate and ethylpropionate, the two solvates crystallizing in the same space group and with very similar unit cell dimensions. This tendency for isostructurality extends to larger host molecules such as calixarenes and cyclodextrins, which in addition may contain fairly large guest molecules, alone or in a mixture with smaller molecules of solvent. The larger components may be volatile oils, and species such as camphor or menthol, which are released upon heating the inclusion compound.

Figure 4. Schematic representation of isostructural crystals, with different guest molecules encapsulated in equivalent voids of the common host framework.

DSC and TG are employed to investigate both the thermodynamics and the kinetics of desolvationlguest-loss processes for compounds in all of the above categories and a discussion of both the structural aspects of solvates and the methodology for investigation of their thermal properties follows. The well-known HSM technique of identifying solvated forms by observing bubble formation on heating crystals under an inert medium (e.g. silicone oil) continues to be widely used as a screening method owing to its rapidity and the small sample requirement. It is common practice to record the HSM events on a video-camera and furnish a series of micrographs in papers and reports. The HSM technique is typically employed to distinguish polymorphs and solvates emanating hom the pharmaceutical high-throughput screening referred to earlier. The TG method plays a major role in quantifying mass loss during heating of solvates, allowing determination of the host-guest stoichiometry. It also reveals the nature of the event (single- or multistep), and the temperature range over which the guest is lost, these parameters being dependent on (inter alia) the particular mode of inclusion of the solvent molecules within the solvate crystal. Enthalpies of desolvation are accessible using the DSC technique, which also permits identification of the desolvated forms (one or more polymorphs) from their temperatures and enthalpies of transition or fusion. The results of all three of the above techniques are usually combined to obtain a reliable interpretation of the complete series of events occurring upon heating the material in question. Figure 5 illustrates combined TG, DSC and HSM data for the dehydration of a cyclodextrin inclusion complex containing a non-volatile pharmaceutical excipient as guest [38]. The water content was determined from the mass loss occurring over the temperature range 30-100 "C, the broad peak A in the DSC trace corresponding to the dehydration event. Loss of water from the complex crystals is evident from the slow release of bubbles into the silicone oil in which the crystals are immersed. Endotherm B in the DSC trace represents melting of excess excipient in the sample, while C corresponds to the fusion of the anhydrous inclusion complex. Dynamic techniques such as variable-temperature PXRD provide valuable supporting information in analyses of the type shown in Figure 5, revealing for example the sequential appearance of significantly new diffraction patterns at temperatures close to those observed by thermal analysis for desolvation steps and polymorphic transitions. Because solvation/desolvation phenomena have an important bearing on the physical properties of a material, especially in the context of pharmaceutical processing, manufacture and storage, many studies are directed towards understanding these processes at the molecular level. The X-ray crystal structures of solvates are frequently determined in order to identify the precise mode

of inclusion of solvent molecules and their interactions with each other and with the parent molecules. Ultimately, it should be possible to reconcile these structural features with the thermal behaviour of the solvate.

% 80 -

.40

100

TCA I

90

wt.

70-

I

35 Heat flow

C A

I

I

Figure 5. TG, DSC and HSM analyses of a hydrated cyclodextrin inclusion complex [381. Pharmaceutical hydrates have been the subject of particularly intensive study and Morris has classified these on the basis of the types of inclusion of water molecules observed in their crystals [35]. In the class of hydrate described as 'isolated lattice site', the water molecules occupy voids that are shielded from one another by intervening host molecules. Cephadrine dihydrate belongs to this category. In this case, the water molecules occur in isolated pairs, arranged at regular intervals in the crystal. Loss of water molecules from the surface of such a hydrate crystal thus leaves voids that are not accessible by other water molecules in the crystal, resulting in a relatively stable hydrate. Cephadrine dihydrate accordingly yields two sharp, closely spaced DSC endotherms at -100 O C for dehydration, the TG trace showing mass loss over the same temperature range. Sharp peaks for u(0-H) are also observed in DRIFT spectra of this hydrate. The second class of hydrates ('lattice channels') includes those in which water molecules form a continuum within channels in the crystal (e.g. ampicillin trihydrate). Because these channels span the entire length of the macroscopic crystal, dehydration can commence at the two surfaces where the channels terminate. Increasing the temperature of the crystal accelerates the

dehydration. This may leave the host framework intact or result in collapse of the structure. Early onset of dehydration and its continuous nature, revealed in DSC and TG traces, are typical of channel hydrates (cf. Figure 5). Morrison linther subdivides lattice channel hydrates into 'expanded channels' (non-stoichiometric species whose unit cell dimensions vary to accommodate the degree of hydration or dehydration), and 'planar' hydrates. In planar hydrates, the water molecules are located in planes or layers that alternate with layers of parent molecules, and dehydration typically occurs along the planar axes. Finally, 'ion associated hydrates' are those containing metalcoordinated water molecules. Many pharmaceutical salts containing ~ a +K',, ca2+ and M ~ belong ~ + to this category. In addition to metal-coordinated water molecules, such crystals may contain un-coordinated water molecules. TG and DSC curves may then be quite complex, revealing dehydration at a series of temperatures, depending on the relative strengths of binding of the various water molecules. The appearance of TG and DSC curves for crystalline solvates, as well as the parameters that characterize their thermal desolvation (temperature range for guest loss, desolvation onset and peak temperatures, enthalpy of desolvation), depend on essentially three factors. One of these was mentioned above, namely the mode of inclusion of the solvent molecules (location in isolated sites, channels or layers). Another is the three-dimensional structure of the host framework and the strengths of the intermolecular interactions that determine it. The weaker these are, the lower is the energy input required to disrupt the host assembly and facilitate guest escape. A third factor is the strength of interaction between host and guest molecules. These range from weak van der Waals interactions to very strong, directional hydrogen bonds, so that in the first case, solvent molecules are often disordered within the cavities they occupy whereas in the second, their atoms are usually ordered and the extent of their thermal motion is relatively low (as reflected in relatively small crystallographic thermal displacement parameters). Given this variety of factors influencing thermal behaviour, it is often meaningless to compare parameters such as onset temperatures and enthalpies of desolvation derived from TGIDSC curves for solvates of different chemical composition with respect to both host and guest. Isostructural solvates, on the other hand, present a unique opportunity for investigating the role of host-guest interactions in determining the thermal properties of solvates because they have common host frameworks and contain different guest molecules in voids of essentially identical topology (see schematic in Figure 4). Thus, two of the above factors are eliminated and the parameters for thermal desolvation should depend primarily on the name and strengths of the host-guest interactions. Such solvates would therefore appear to be of considerable interest, but examples of studies that focus on this aspect of their desolvation are, however, uncommon [37].

Another interesting aspect of isostructural solvates, alluded to earlier, is their ability to permit solvent exchange within a limited range of solvent molecular sizes. A void occupied by a methanol molecule could alternatively be filled by e.g. two water molecules, or an ethanol molecule. Given a hostlparent compound that is able to form isostructural solvates with e.g. both water and methanol, an experiment to investigate guest exchange would involve exposing one of the two crystalline solvates to the vapour of the second solvent. Experiments could be conducted at a single temperature or at a series of constant temperatures. Samples of the solid phase would be removed periodically and subjected to TG and DSC measurements to determine the relative amounts of the two solvents present in the crystals. In this way, the thermodynamic and kinetic course of guest exchange could be followed quantitatively. A reference to studies of this type is cited in the material that follows. In concluding this section, we emphasise the growing trend towards the use of coupled techniques in the investigation of polymorphs and solvates [2].Application of the most widely used thermal methods, DSC and TG, to the study of these systems is rendered more powerful by their coupling with other techniques. The combined use of HSM, TG and DSC has already been mentioned. For studying solvates, obvious advantages are offered by TG-IR or TG-MS techniques, namely identification and quantification of single or mixed solvents present in the crystals (useful also for monitoring guest exchange processes), as well as evolved components associated with the decomposition of the ensuing polymorph. Other coupled techniques and a discussion of their merits for examining polymorphic and solvatomorphic systems have been reviewed by Le Parlouer [ 6 ] . TG-Raman spectroscopy, for example, is particularly useful for the study of hydrates. The utility of gravimetric sorption methods (for studying hydrate and solvate formation), isothermal microcalorimetry, and the thermally stimulated current (TSC) method in the context of polymorphism and solvatomorphismhas also been discussed in the above review.

4. RECENT STUDIES 4.1. Characterization of polymorphs and polymorphic transformations A selection of recent studies in which thermal analysis features in the characterization of polymorphs is discussed here with the aim of illustrating some of the conceptual issues, as well as the use of coupled techniques, described in the previous sections. As stated in the Introduction, crystal polymorphism is ubiquitous, affecting all classes of solid compounds ranging from elemental substances and simple inorganic compounds to complex molecules, including proteins. This survey commences with one example of thermal analysis used to investigate the

polymorphism of a simple inorganic compound and proceeds to summarise more complex instances of polymorphism occurring in several synthetic polymers. The remaining case studies in this section refer to compounds of pharmaceutical interest. Both drugs and excipients are important, but space permits discussion of the former only. The polymorphism of calcium carbonate is still not completely understood despite its very long history and the appearance of many published studies. In addition to calcite, aragonite and vaterite, non-crystalline forms of CaC03 of biological origin exist. An account of the relationships between these solid phases has appeared recently [39] together with a summary of thermodynamic and kinetic data for the transformations of the metastable polyrnorphs aragonite and vaterite to the stable calcite. These authors describe the preparation of non-crystalline calcium carbonate and report preliminary values of the transition temperature and enthalpy change for its crystallization to calcite. The DSC method, supported by TG and PXRD, was used in this study. Polymorphism is encountered frequently in the food industry, cocoa butter and milk fat being well known materials that display this phenomenon. The materials investigated here by thermal analysis are generally highly complex mixtures, with variable composition, that yield broad endothems in DSC. One such example is natural lard, whose important constituents are long-chain (CI6-Cl8)fatty acids and triacylglycerols (TAGs). Examination of a typical sample of lard using DSC reveals a large number of transitions on heating and cooling. These are attributed to polymorphic transformations of the TAGs, each of which may exist in a number of polymorphic forms. This is an area of polymorphic investigation that depends critically on the use of coupled techniques to unravel the complexities typical of multi-component systems. In a recent extensive study of a commercially available sample of lard, Kalnin et al. employed coupled DSC and XRDT (variable temperature XRD), with synchrotron radiation in an attempt to infer structural information and correlate these data with observed thermal events [40]. The use of both small-angle XRD (SAXS) and wide-angle XRD (WAXS) employing synchrotron radiation yielded structural information relating to the longitudinal packing and the lateral organization of the chains of the TAG molecules respectively. Several polymorphs were identified in this study and their principal crystallographic parameters estimated. These findings have practical implications for the macroscopic behaviour of lard on fast cooling. Synthetic polymers such as polypropylene also exhibit polymorphism and their properties are highly dependent on their mode of preparation. Using a combination of the WAXS technique to identify polymorphs, and HSM for crystallization studies

from the melt, van der Burgt et al. investigated the influence of thermal treatments on the polymorphism in stereo-irregular isotactic polypropylenes (iPP) derived from two sources [41]. The known polymorphs of iPP (a, P, y) have crystal structures that are based on 31-helical chains and the frequency of their occurrence is a function of the isotacticity of the sample. This study focused on the crystallization behaviour of samples of iPP prepared using both Ziegler-Natta and metallocene catalysts, which produce different types of stereo-defect distributions. The former generally yields stereo-block type distributions whereas metallocene-derived catalysts produce iPP containing more randomly distributed stereo- andlor regio-defects along the polymer chain. This study revealed a strong dependence of the phase behaviour of iPP on tacticity and cooling rate (0.5-40 K min-' range investigated) in proceeding from the melt to the solid state, lower cooling rates generally favouring the formation of the ypolymorph relative to the a-polymorph. This effect is more pronounced for samples of P P containing random defect distributions. A further study of the crystal polymorphism of poly(buty1ene-2,6-naphthalate) (PBN) was recently reported by Ju et al. [42]. This material and related polymers such as poly(ethy1ene terephthalate) (PET) and poly(ethy1ene-2,6-naphthalate) (PEN) have been the subjects of numerous studies since the 1950s. For PBN, two polymorphs, a and p, are known, both crystallizing in the triclinic system with one repeat unit in the crystal unit cell. At the molecular level, the main difference between these forms was attributed to a conformational difference in the respective glycol residues [43]. Ju et al. examined the effects of various thermal treatments on the polymorphism of PBN using a combination of techniques including DSC, WAXD (wide-angle X-ray diffraction), HSM with polarized-light, and FTIR spectroscopy. Crystallization and annealing steps were performed on samples of PBN placed directly in the DSC cell. Crystallization of PBN pellets from both the melt and solid states were conducted under isothermal and non-isothermal conditions. The a-form was identified in samples of annealed PBN following quenching, or by crystallizing from the static melt at a lower temperature. Instead, non-isothermal crystallization from the melt at 0.1 K min-' yielded a polyrnorph with a WAXD profile similar to that reported for the known p-form, but with significant differences in d-spacings for the (0-1 1) and (0 10) planes. This evidently represents a new polymorph of PBN, accordingly designated p'. Interestingly, annealing of this form at 220 "C for 12 h in the solid state leads only to a sharpening of the WAXD peaks, attributed to improvement in the quality of the PBN crystals, without polymorphic change. Previous indications that no solid-state polymorphic transitions are observed for PBN were confirmed in this study. As to the structural nature of the new thermally-derived p'-form, a detailed comparison of the

FTIR data led the authors to conclude that it differs from the a-form in having more tightly packed chains, consistent with the higher density of the p'-form. Furthermore, FTIR data indicated that the glycol residue evidently adopts essentially the same conformation in the a- and p'-forms, so that the existence of three distinct polymorphs of PBN seems conclusive. In the case of PBN above, no solid-state polymorphic transitions were observed. However, for some polymeric materials, a recurring difficulty in the DSC analysis of polymeric materials is that of peak overlap due to polymorphism. This hinders quantification of the polymorphs present, preventing reliable analysis of the kinetics of polymorphic transformation. The problem is usually addressed by employing methods that deconvolute the total heat effect into its individual components. Alfonso et al. carried out a systematic investigation of four procedures for achieving this in a study of the polymorphic transformation in isotactic polybutene-1 (PBu-1), from the tetragonal modification (Form 11) to the hexagonal modification (Form I) [44]. Five isotactic PBu-1 homopolymers in the form of flat disks punched from quenched compression-moulded films were examined by DSC. A common set of DSC curves, recorded after various ageing times of melt-crystallized polymer, was examined. Melt-crystallization results in the formation of Forms I and I1 only (the latter being kinetically favoured), though other polymorphs of PBu-1 have been identified. Ageing at room temperature results in Form I1 transforming into Form I, whose melting point is only -10 K higher than that of Form 11. In this case, therefore, the problem addressed was the deconvolution of the two overlapping peaks representing the fusion of these two polymorphs. On the assumption that the sum of the crystallinities, (xo), of the two polymorphs upon ageing is constant (supported by the experimental procedures), the time-dependent degree of polymorphic transformation, 4t), was simply expressed in terms of the rate of change of the crystallinity of one of them i.e. as xl(t)lxo. The four analytical methods employed to deconvolute the DSC peaks included that of partial areas, a simulation approach (based on the linear combination of DSC signals obtained by heating the samples immediately after crystallization or after extended ageing), use of the PEAK-FIT deconvolution program, and total melting enthalpy method (which relies on direct measurement of the entire bimodal endotherm to follow the polymorphic transformation). Each of these approaches is described in some detail in the above reference. The authors found that, except for the method based on partial areas, all methods of data analysis yielded comparable values of 4t). On increasing the ageing time, the method of partial areas underestimates the amount of Form I present. The total enthalpy method, being less subjective, was recommended as the most reliable.

Proceeding now to studies relating to pharmaceutically relevant materials, the first concerns a similar problem of DSC peak overlap to that described above and was encountered in a recent investigation of polymorphs of the antihistaminic terfenadine (Figure 6, 7). The conformational flexibility of the terfenadine molecule and the presence of donor and acceptor functional groups are conducive to polymorphism, borne out by previous identification of several polymorphic forms.

Figure 6 . Chemical structures of terfenadine 7 and tulobuterol8. In a study by Leitiio et al., overlapping DSC melting curves recorded for terfenadine samples prepared by recrystallization from different solvents were analysed in order to enumerate the distinct crystal forms present in each case [45]. Recrystallization of terfenadine fiom methanol, ethanol and ethanoywater mixtures was carried out under a variety of conditions and the resulting products were identified. These included both crystalline and amorphous phases, and mixtures comprising solvates, amorphous and crystalline products. Heating these products at 120 OC for 1 h yielded crystalline solids in all cases and these were subsequently analysed by the DSC method. In most cases, the traces showed two overlapping endothermic peaks attributed to fusion. However, terfenadine samples recrystallized from an ethanoywater mixture yielded a single DSC endotherm and were considered as representing a single polymorph (Form I). One of these was selected as a standard and the properties of its endotherm (height of peak maximum, peak position and asymmetry parameters) used to develop an empirical, multi-parametric equation for the heat flux and the onset temperature. Subsequent deconvolution of the DSC curves for the remaining terfenadine samples was simulated using both flexible and fixed peak profiles, the latter turning out to be more appropriate for the system under study. For some terfenadine samples, deconvolution of the experimental double endotherm yielded two peaks (corresponding to two polymorphs) while for others, three peaks (three polymorphs) were indicated. The complete set of predicted peak temperatures for a total of four identified polymorphs based on the peak-fitting analyses spanned the fairly narrow

range of 144.6-151.8 O C , the highest temperature corresponding to the melting of Form I. This implies that these polymorphs have lattice energies that likewise span a narrow range, consistent with the concomitant polymorphism typically observed for this drug.Thus, of the twelve terfenadine samples analysed, only two corresponded to a single polymorphic form (viz. Form I), all others being mixtures of two or three polymorphs. The authors reported the relative amounts of the various polymorphs present in each sample estimated from the peak areas of the deconvoluted peaks. This study of terfenadine polymorphism is particularly significant owing to the lack of Xray data characterizing its polymorphs and it is also of general interest in the context of analogous polymorphic systems displaying closely melting temperatures. The dimorphism of the bronchodilator tulobuterol (Figure 6, 8) as a racemate was recently reported [46],Form I crystallizing in the monoclinic space group P2Jn with Z =12 and Form II in the triclinic space group P(-1)with Z = 6 molecules per unit cell. These unusual values of Z suggested a common motif comprising three molecules as the asymmetric unit and this was confirmed by single crystal X-ray analyses (Figure 7).

Figure 7. The trimer of the bronchodilator tulobuterol occurring in two polymorphs. The representative trimer shown contains two molecules with configuration R- and one molecule with configuration S- at the chiial centre, the centric space groups requiring the presence in the racemic crystals of the corresponding trimer of chirality (S-, S-, R-). As indicated in Figure 7, three homodromic O-H.-N hydrogen bonds are responsible for the stability of the trimer, leaving no capacity for inter-trimer hydrogen bonding in the crystals, which are maintained by weak van der Waals, CHa--0 and C-H.-n interactions only. As mentioned earlier, it is important to

ascertain whether drug polymorphs are enantiotropically or monotropically related, and to construct their energy-temperature diagrams whenever possible. Examination of the crystals by HSM and DSC revealed no transformations. Accurate measurement of the melting temperatures and enthalpies of fusion (AHf) yielded values of 90.9 0.2 "C, 27.1 f 0.3 kJ mol-' for Form I, and 80.0 0.1 "C, 25.4 f 0.1 kJ mol-' for Form 11. The low values for these parameters are consistent with the weak intermolecular interactions observed in the crystals. As the higher melting form has the higher enthalpy of fusion, these polymorphs are monotropically related (see section 3.1.I.), with an energy-temperature diagram of the type shown in Figure 3(b). The method of Yu [30] was applied to the above data to yield AGO= -773 J mol-' for the transformation of Form I1 to Form I at an extrapolated virtual temperature (T,) exceeding 2000 K. This should be considered a nominal value only as it is subject to a large error, even with the relatively small error estimates for the fusion data. Foppoli et al. investigated the polymorphism of the aspirin derivative NCX4016 (Figure 8, 9) which belongs to a new family of non-steroidal anti-inflammatory drugs (NSAIDs) incorporating nitric oxide [47]. A pre-formulation study using SEM, PXRD, FTIR, DSC, HSM, TG and intrinsic dissolution rate (IDR) measurements revealed two polymorphs of NCX4016, Forms I and 11. As for the tulobuterol case above, these were also found to be monotropically related. The treatment of Yu [30] applied to the melting data of NCX4016 yielded a T, value of 94 "C. In this case, solubility data for Forms I and I1 were also available, enabling an independent determination of T, from van't Hoff plots of ln(1DR) versus 1IT. The extrapolated value was 95 OC, in remarkably good agreement with the value based on melting data.

+

Figure 8. Chemical structures ofNCX4016 9 and lifibrol10.

+

A comprehensive pre-formulation study of the cholesterol-lowering drug, lifibrol (Figure 8, 10) using HSM, DSC, FTIR, PXRD, density and solubility measurements, was carried out to determine the best form for development [48]. Three polymorphs were identified, two of them (Forms I and 11) related enantiotropically, and the third (Form 111) bearing a monotropic relationship to the first two. Using analogous data to those used for tulobuterol above, an energytemperature diagram, closely resembling Figure 3(c), was established for these polymorphs. As stable forms are generally preferred, no further consideration was given to the metastable polymorph 111. The transition temperature range for the enantiotropic pair was established thermomicroscopically as 70-85 "C, leading to Form I1 (the more stable form at temperatures below 70°C, and therefore at roomand body-temperature) as the polymorph recommended for development. Furthermore, Form I1 had better filtration and tabletting properties than Form I. Nonetheless, Form I (metastable with respect to Form I1 at room-temperature) was considered worthy of further consideration because its higher solubility could outweigh the advantages offered by Form 11. Pharmacokinetic data could settle the final choice. This study clearly indicates the significant practical role of thermal analysis, and of the derived energy-temperature diagram in particular, in guiding decision-making in the highly competitive pharmaceutical industrial environment. More recent examples of the construction of energy-temperature diagrams for drug polyrnorphs include those published by Schmidt and co-workers for two local anaesthetics, hydroxyprocaine hydrochloride (HPCHC, Figure 9, 11) and salicaine hydrochloride (SLCHC, Figure 9, 12) [49,50]. Comprehensive solid-state analysis of these compounds included HSM, DSC, TG, FTIR and Raman spectroscopy, SSNMR, PXRD, single crystal XRD and vapour pressure sorption analysis.

Figure 9. Chemical structures of local anaesthetics: hydroxyprocaine hydrochloride 11 and salicaine hydrochloride 12.

Only the thermal analysis results are highlighted here. HSM revealed two polymorphs of HPCHC, a form that is thermodynamically stable at 20 OC (designated Mod. 11°), and a second form (Mod. I) that is stable at higher temperatures. A DSC trace recorded at 5 K mine' yielded a single endothermic peak at 160.5 OC for Mod. I corresponding to its fusion. A DSC scan for Mod. 11° displayed an endotherm at 154.5 OC corresponding to fusion, followed immediately by an exotherm representing recrystallization, and a final endothermic peak at 160.5 OC. This behaviour is consistent with an enantiotropic system, Mod. 11° being the more stable form below T, and Mod. I being the stable form above T,. Accurate measurement of the enthalpies of fusion showed that the difference AHf (11") - AHf(I) was -10 kJ mol-', which confirms the enantiotropic relationship through the heat of fusion rule (see section 3.1.1.). The derived energy-temperature diagram for HPCHC thus resembles the simple one shown in Figure 3(a). This study was complicated by the simultaneous occurrence of a hydrate of HPCHC, whose thermal properties were also reported. The energy-temperature diagram for SLCHC polymorphs is a slightly modified version of that shown in Figure 3(c), a total of three polymorphs having been identified in this system. Here, Mod. I11 is the least stable form and is monotropically related to Mod. 11°, as confirmed by the exothermic DSC peak for transformation of the first to the second. Mod. 11" is the stable form below 140 OC (confirmed by the enthalpies of fusion, crystal densities and observed DSC transformations), while Mod. I is the high temperature form, enantiotropically related to the other forms. Issues of polymorphic purity frequently arise in the fine chemicals and pharmaceutical industries. For example, if a mixture of drug polymorphs is employed in a dosage form, constant monitoring of the composition of this mixture during manufacturing is mandatory. If the crystallization conditions are not carefully controlled, isolation of pure polymorphic materials may be compromised by 'concomitant polymorphism', the simultaneous precipitation of more than one form [I]. Contamination of a stable polymorph by traces of a highly metastable form is of particular concern, since this could induce further transformation of the stable form. Such contamination of the stable form can also arise during its mechanical treatment e.g. by high-energy milling. Polymorphic impurity at a level of -5% is generally difficult to quantify by the usual methods of polymorphic characterization. Tong et al. recently described a method based on DSC for quantifying trace materials at significantly lower levels than this for Form I samples of salmeterol xinafoate containing trace impurities of Form I1 [51]. This method involves measuring the recrystallization rate of Form I1 from the Form I melt by DSC performed at various scanning speeds. Data from both the tested material and reference samples comprising mixtures of Forms I and I1 of known

composition were converted to a-t curves which were then fitted by a non-linear iterative method to the Avrami-Erofe'ev (AE) equation. The AE rate constants determined for the reference mixtures were then related to polymorphic composition using an equation based on an instantaneous nucleation model. The latter equation was finally used to estimate the amount of Form I1 in each sample from the computed AE rate constant. This method, based on estimating the rate of formation of Form I1 from the Form I melt, has been shown to be capable of quantifying Form I1 at as low a level as 0.1% in mixtures. In a more recent report by the same group, a slightly modified procedure was used to estimate impurity content (a metastable polymorph) at as low a level as 0.004% in samples of the antiviral drug ribavirin [52]. In studies of polymorphic transformations, it is necessary to ensure that the starting material is polymorphically pure. Of the three known polymorphs of glycine (H2NCH2-COOH), a-, P- and y-glycine, the least stable is the p-form, which is known to convert to the a-form under certain conditions. The study of this transformation has been hampered by the difficulty of isolating the p-form in high purity and sufficient amount. Drebushchak et al. recently described a novel synthesis that resulted in pure p-glycine [53], and proceeded to investigate the p-a transformation using DSC analysis. In addition to determining the heat capacities of these phases, they established that the transformation is exothermic, the very small value of the enthalpy change (-200 J mol-') being used to rationalise the persistence of the p-form for long periods under normal conditions. Reports on the kinetics of polymorphic transformation appear less frequently than those relating to the kinetics of desolvation, the latter including the dehydration of pharmaceutical hydrates, a process having practical implications for drug stability and storage. The kinetics of polymorphic transformation are commonly studied by XRD methods. For example, the kinetics of the transition of modification I1 of barbital to modification I were recently investigated using the PXRD technique for quantitation of the relative amounts of the two forms present at different times during milling [54]. In this type of study, the kinetic course is followed by integration of selected, nonoverlapping diffraction peaks of either the reactant or the product. A novel procedure using FTIR microspectroscopy coupled with a hot stage has been used to monitor the transformations among acetaminophen polymorphs 1551. This method is useful when different polymorphs display significant differences in their FTIR spectra, allowing kinetic data to be extracted from the variation of the FTIR peak intensity during polymorphic transformation. The in situ kinetics of transformation of carbamazepine Form I11 to Form I were recently studied by FT-Raman spectroscopy under isothermal conditions at a series of temperatures in the range 130-150 "C [56]. With the sample housed in a novel

environmental chamber, the course of the transformation was followed by measuring the relative intensities of spectral peaks arising from two C-H bending modes. Various solid-state kinetic models were found to be consistent with the data and E, values in the range 344-386 kJ morl were established for the transformation. Some aspects of the kinetics of solid-state reactions in the context of polymorphism are encountered again in the next section, which deals with the characterization of organic solvates and their desolvation.

4.2. Characterization of solvates and desolvation processes Thusfar, this chapter has focused on the use of thermal analysis to investigate polymorphism. Some aspects of the related phenomenon of solvatomorphism are treated in this section, though in a necessarily limited way since the subject is very broad. A thorough treatment would include aspects of crystal engineering and supramolecular chemistry, among whose goals is the design of host compounds with the ability to encapsulate solvent molecules selectively to form their crystalline solvates [57]. The general relevance of solvated compounds, the significance of their crystal structures in determining their desolvation tendencies, and the thermal methods of analysis used to characterize solvates were outlined earlier. Here, a number of practical issues and case studies are described illustrating the use of thermal analysis to investigate thermodynamic and kinetic aspects of their desolvation. For compounds of pharmaceutical relevance, solvates comprise an extremely important class and many drugs are prolific solvate formers, an extraordinary case being that of the drug sulfathiazole, for which over 100 solvates have been reported [58]. Determination of the stoichiometry of a solvate by TG is a basic analytical measurement that complements data from other techniques, such as elemental analysis. A solvate may generally be formulated as H.nl(Gl).n2(G2).... where H is the host (or parent) compound, and G1, G2, ... are chemically different (volatile or nonvolatile) guest molecules encapsulated within the crystalline framework or matrix of H. In general the coefficients n,, n2,... need not be integers. For example, in a ternary system, such as a hydrated cyclodextrin (CD) inclusion complex of an organic guest with formula CDTI~(G~).~~(H~O)...., (e.g. that described earlier in Figure 5), n2 is usually non-integral and the n2 water molecules are frequently disordered in the crystal i.e. they occupy more than n2sites. The value of n2in such complex hydrates is often in the range 15-30, and attempting to quantify these by single crystal X-ray analysis is fraught with difficulty because of the correlation between the atom siteoccupancy factors and their thermal displacement parameters. In such cases, it is essential to obtain a reliable estimate of n2 and its standard error by TG and to ensure

that the refined X-ray crystallographic model is consistent with that estimate. The example of a CD inclusion complex was specifically cited above because the characterization of such species often presents experimental challenges. Giordano et al. have reviewed the applications of thermal analysis to CDs and their inclusion complexes, with special reference to their dehydration [59]. Bettinetti et al. recently described the thermal behaviour of several solvates of novel CD derivatives [60]. These compounds display a variety of solid-state phenomena including polymorphism, solvatomorphism and amorphism. Commonly quoted parameters derived from DSC that describe desolvation include onset and peak temperatures (To,, Tpeakrespectively) and the enthalpy change, AH. The value of To, is one measure of stability of a solvate. For some solvates (e.g. those of the 'lattice channel' type), desolvation may be visible upon removal of crystals from their mother liquor under ambient conditions; crystals begin to turn opaque and immediate mass loss is recorded in TG, reflected in an endotherm in the DSC. In other cases (e.g. with solvent molecules located in 'isolated sites'), very high To, values are observed. Thus, for water loss from hydrates of organic host compounds, To, values spanning a range from 20 to 200 OC have been reported. The parameters ToJTb or To,-Tb (where Tb is the normal b.p. of the solvent) have been proposed as alternative measures of solvate stability [61]. Measured values of AH for desolvation derived from DSC are dependent on sample particle size and heating rate. These factors have been discussed elsewhere [61]. The enthalpy change is a composite quantity taking into account all of the steps involved in the desolvation mechanism. The latter has been thoroughly reviewed by Byrn et al. [62] who considered the role of atmosphere, crystal packing and crystal defects in the desolvation process, as well as the nature of the final product. Desolvation of a crystalline solvate may involve a major or a minor change in crystal structure, or produce an amorphous form. This is assessed using PXRD. A rational approach to the study of solvates should aim at reconciling their crystal structures with their thermal stability and kinetics of desolvation. Nassimbeni has reviewed this subject, focusing on the correlation between solvate lattice energies, calculated using atom-atom potentials, with both the thermal data for their desolvation, as well as the selectivity that a given host compound displays for a specific guest [63]. One example is cited here to illustrate this approach [64]. This study refers to the properties of the solvates formed between the host binaphthol and the isomers of lutidine (Figure 10). These solvates were prepared by dissolving binaphthol in each of the liquids and allowing the respective crystalline solvates to precipitate by slow solvent evaporation. Mass loss from TG occurred in a single step in all cases and indicated the compositions 13.2(2,6-L), 13.2(2,4-L) and

13.3,5-L. The DSC traces were also unremarkable, all displaying an endotherm for desolvation, followed by an endotherm corresponding to fusion of the desolvated binaphthol. Single crystal X-ray analyses were performed to establish the precise mode of inclusion of the solvent molecules in the crystals as well as the molecular interactions between the host and guest molecules. All of the solvent molecules are located within channels in the respective solvates and the primary interactions are hydrogen bonds of the type 0-H.-N with 0.-N in the range 2.794(2)-2.826(3) A.

Figure 10. Chemical structures of the host molecule binaphthol 13 and solvent molecules 2,6-lutidiie 2,6-L, 2,4-lutidine 2,4-L and 3,5-lutidiie 3,5-L. Two-way and three-way solution competition experiments, carried out to establish the enclathration selectivity of binaphthol for the three lutidine isomers, yielded the order of preference as 2,6-L 2,4-L > 3,5-L, suggesting that the solvates containing the first two solvents should have similar stabilities but that both should be more stable than the solvate containing 3,5-L. Calculation of the lattice energies of the solvates, including contributions from the hydrogen bonds, yielded values of -494.2, -495.1 and -353.2 kJ mol" for 13.2(2,6-L), 13.2(2,4-L) and 13.3,5-L respectively. The stabilities based on lattice energies are thus consistent with those predicted by the competition experiments. The kinetics of desolvation of these solvates were determined by isothermal TG performed at selected temperatures. For 13.2(2,6-L) and 13.2(2,4-L), the a-t curves were deceleratory and fitted the kinetic model R2 (contracting area) [65], whereas 13.3,5-L yielded sigmoidal a-t curves with a best fit to the Avrami-Erofe'ev model A3. The calculated E, values were in the range 80(4)-101(7) kJ mol-'. For solvates of binaphthol containing other solvents in 'channel-type' inclusion, previous estimates of E, values for their desolvation were in the range 60-150 kJ m o ~ ' .

-

In the above system, although all solvent inclusion could be described as 'channel-type', the topologies of the channels and hence solvent inclusion differ in the three crystals. Comparison of quantities such as Tonand AH for desolvation in such cases may therefore not be very meaningful. However, as indicated earlier, such comparison becomes appropriate for a series of isostructural solvates, in which different solvent molecules occupy common voids or channels. This is the case for a series of solvates of the drug tetroxoprim (W,Figure 11, 14) containing water, ethanol and methanol [66]. Figure 11 shows the crystal packing in the methanol solvate as representative of the series.

Figure 11. Structure of tetroxoprim (TXP)14 and crystal packing in the solvate W. CH30H, a representative member of a series of isostructural solvates. The TXP molecules are shown in stick mode while the CH30H molecules are shown in spacefilling mode. The host TXP framework is extensively hydrogen-bonded and is practically superimposable in this series, water molecules and ethanol molecules therefore occupying identical cavities ('isolated sites') in their respective solvates as those occupied by methanol molecules in Figure 11. Table 1 lists solvate compositions and relevant thermal data. An intriguing, counter-intuitive feature of the data is the increasing trend in Tonwith the decreasing trend in b . ~in. proceeding ~ ~ from water to methanol. In this situation, the dominant factor in determining solvate stability is the strength of the host-solvent interactions. A detailed examination of the relevant hydrogen bonds that bind the host to the solvent molecules showed that their strengths are in the order water < ethanol < methanol, thus providing an

explanation for the trend in Tonvalues. The AHBSexceed the corresponding AHPS values in all cases, with MBS/AHPS 1.3, 1.6 and 1.35 for the respective solvents. This trend was explained on the basis of the increasing steric bulk of the solvent molecules, which hinders their diffusion through the solid phase and their eventual release.

-

Table 1. Thermal data relating to three isostructural solvates of tetroxoprim (TXP) Parameter

TXP. 0.67H20

Ton (OC) 99.7(8) b.p.ps a ("C) 100.0 AHPS(kJ mol-')a 40.7 AHBS(kJ m01-I)~ 53(2) a PS = pure solvent BS = bound solvent

TXP.0.5C2H50H

TXP.0.5CH30H

107.6(4) 78.5 40.5 64(4)

136.3(1) 64.7 39.2 53(1)

Another feature of isostructural solvates is their tendency for guest exchange. A number of isostructural series of solvates have been reported to undergo solvent exchange (e.g. H.Gl(s) transforming to H.Gz(s) by exposure of the former to G2(vapour)) [37]. This process should be more facile for 'channel-type' solvent inclusion. In cases such as the above, with solvent molecules trapped in isolated sites, the concentration gradient set up by incoming vapour evidently results in small conformational flexure of the host arrangement, allowing passage of solvent molecules through the intact crystal. The process can be monitored using DSC to observe the migration of component endothermic peaks as a function of time. Many other examples could have been selected in the area of solvate investigation. Numerous reports of dehydration kinetic studies on drug hydrates are found in the pharmaceutical literature. An instructive recent case concerns piroxicam monohydrate (PM) [67]. Model-free kinetics showed that the dehydration kinetics of PM under isothermal and non-isothermal conditions were different. Complementary structural studies revealed a complex hydrogen-bonding pattern that was reconciled with the observed dehydration behaviour of PM.

5. CONCLUSIONS

Polymorphic and solvatomorphic systems are enjoying increasing interest for both academic and commercial reasons, and the number and variety of such systems is set to increase significantly given the recent discovery of novel techniques for their production and the growing popularity of high-throughput preparative methodology. While the application of thermal methods of analysis has been emphasised, the aim of many of the examples cited above has been to highlight the importance of using complementary analytical techniques for gaining a better understanding of the relationship between the structures of polymorphs and solvates and their thermodynamic and kinetic behaviour. 6. ACKNOWLEDGMENT

The author is grateful to the University of Cape Town and the NRF (Pretoria) for research support. Thanks are due to Kate Davies for assistance with references.

7. REFERENCES J. Bernstein, Polymorphism in Molecular Crystals, Clarendon, Oxford, 2002. D. Giron, J. Therm. Anal. Cal., 64 (2001) 37. D. Giron, J. Therm. Anal. Cal., 68 (2002) 335. K. Urakami, Curr. Pharm. Biotech., 6 (2005) 193. M. Phipps, Abstracts, 36thMiddle Atlantic Regional Meeting of the American Chemical Society, Princeton, NJ, USA, June 8-1 1 (2003) 258. 6. P. Le Parlouer, STP Pharma Pratiques, 13 (2003) 236. 7. W.J. Sichina, PerkinElmer Instruments, Proceedings of the NATAS Annual Conference on Thermal Analysis and Applications, 29th, 2001, pp. 225-234. 8. M. Rajic and M. Suceska, pp. 290-302 in New Trends in Research of Energetic Materials 200 1, Edited by S Zeman, Proceedings of the Seminar, 4", Pardubice, Czech Republic, Apr. 11-12,200 1. 9. D. Braga and F. Grepioni, pp.325-373 in Perspectives in Supramolecular Chemistry 2003, 7 (Crystal Design), John Wiley & Sons Ltd., 2003. 10. J.K. Guillory, in Polymorphism in Pharmaceutical Solids 1999, (Ed. H.G. Brittain), Marcel Dekker, New York, 1999 p. 183-226. 11. D. Braga and F. Grepioni, Chem. Commun., (2005), 3635. 12. A.H.L. Chow, H.H.Y. Tong and B.Y. Shekunov, Drugs Pham. Sci.,

1. 2. 3. 4. 5.

13. I. Pasquali, R. Bettini and F. Giordano, Eur. J. Pharm. Sci., 27 (2006) 299. 14. Y. Tozuka, D. Kawada, T. Oguchi and K. Yamamoto, Int. J. Pharm., 263 (2003), 45. 15. A.V. Trask, N. Shan, W.D.S. Mothenvell, W. Jones, S. Feng, R.B.H. Tan and K.J. Carpenter, Chem. Commun., 7 (2005), 880. 16. A.V. Trask, W.D.S. Mothenvell and W. Jones, Chem. Commun., 7 (2004) 890. 17. Y. Sonoda, F. Hirayama, H. Arima, Y. Yamaguchi, W. Saenger and K. Uekama, Cryst. Growth Des., 6 (2006) 1181. 18. B.A. Garetz, A.S. Myerson, S. Arnold and J.E. Aber, U.S. Pat. Appl. Publ., (2005). 19. A.S. Myerson, B.A. Garetz and J. Matic, Abstracts of Papers, 223rdACS National Meeting, Orlando, FL, USA, April 7-1 1,2002 (2002). 20. 0. Almarsson, M.B. Hickey, M.L. Peterson, S.L. Morissette, S. Soukasene, C. McNulty, M. Tawa, J.M. MacPhee and J.F. Remenar, Cryst. Growth Des., 3 (2003) 927. 21. J. Bernstein, ACA Transactions, 39 (2004), 14. 22. C. Ouvrard and S.L. Price, Cryst. Growth Des., 4 (2004) 1119. 23. P. Vishweshwar, J.A. McMahon, M. Oliveira, M.L. Peterson and M.J. Zaworotko, J. Am. Chem. Soc., 127 (2005) 16802. 24. P. Vishweshwar, J.A. McMahon, J.A. Bis, M.J. Zaworotko, J. Pharm. Sci., 95 (2006) 499. 25. A. Grunenberg, J.-0. Henck and H.W. Siesler, Int. J. Pharm., 129 (1996), 147. 26. S. Lohani and D.J.W. Grant, in Polymorphism in the Pharmaceutical Industry 2006, (Ed. R. Hilfiker), Wiley-VCH, Weinheim, 2006, p. 21-41. 27. D. Craig, in Polymorphism in the Pharmaceutical Industry 2006, (Ed. R. Hilfiker), Wiley-VCH, Weinheim, 2006, p.43-77. 28. A. Burger and R. Ramberger, Mikrochim. Acta [Wien], I1 (1979) 259-271. 29. D.J.W. Grant, in Polymorphism in Pharmaceutical Solids 1999, (Ed. H.G. Brittain), Marcel Dekker, New York, 1999, p. 1-33. 30. L. Yu, J. Pharm. Sci,. 84 (1995) 966. 31. K. Urakami, Y. Shono, A. Higashi, K. Umemoto and M. Godo, Chem. Pharm. Bull., 50 (2002) 263. 32. K. Urakami, Y. Shono, A. Higashi, K. Umemoto and M. Godo, Chem. Pharm. Bull., 52 (2004) 783. 33. K.R. Seddon, Cryst. Growth Des., 4 (2004) 1087. 34. J.M. Fachaux, A.-M. Guyot-Hermann, J.C. Guyot, P. Conflant, M. Drache, J.P. Huvenne, R. Bouche, (1992) Congr. Int. Technol. Pharm., 6th5:213. 35. K.R. Morris, in Polymorphism in Pharmaceutical Solids 1999,

(Ed. H.G. Brittain), Marcel Dekker, New York, 1999, p. 125-181. 36. M.R. Caira, in Design of Organic Solids 1998, (Ed. E. Weber), Springer, Berlin, 1998, p.164-208. 37. M.R. Caira, in Encyclopedia of Supramolecular Chemistry, Marcel Dekker, Inc., New York, 2004, p. 767-775. 38. M.R. Caira and E.J.C. de Vries, in preparation (2006). 39. G. Wolf and C. Giinther, J. Therm. Anal. Cal., 65 (2001) 687. 40. D. Kalnin, P. Lesieur, F. Artzner, G. Keller and M. Ollivon, Eur. J. Lipid Sci. Technol., 107 (2005) 594. 41. F.P.T.J. van der Burgt, S. Rastogi, J.C. Chadwick and B. Rieger, J. Macromol. Sci., B41 (2002) 1091. 42. M-Y. Ju, J-M. Huang, F-C. Chang, Polymer, 43 (2002) 2065. 43. H. Koyano, Y. Yamamoto, Y. Saito, T. Yamanobe and T. Komoto, Polymer, 39 (1998) 4385. 44. G.C. Alfonso, F. Azzurri and M. Castellano, J. Therm. Anal. Cal., 66 (2001) 197. 45. M.L.P. Leitgo, J. Canotilho, M.S.C. C m , J.C. Pereira, A.T. Sousa and J.S. Redinha, J. Them. Anal. Cal., 68 (2002) 397. 46. M.R. Caira, S.A. Bourne and C.L. Oliver, J. Therm. Anal. Cal., 77 (2004) 597. 47. A. Foppoli, M.E. Sangalli, A. Maroni, A. Gazzaniga, M.R. Caira and F. Giordano, J. Pharm. Sci., 93 (2004) 521. 48. A. Burger and A. Lettenbichler, Eur. J. Pharm. Biopharm., 49 (2000) 65. 49. A.C. Schmidt and I. Schwarz, J. Mol. Stmct., 748 (2005) 153. 50. A.C. Schmidt, I. Schwarz and K. Mereiter, J. Pharm. Sci., 95 (2006) 1097. 5 1. H.H.Y. Tong, B.Y. Shekunov, P. York and A.H.L. Chow, Pham. Res., 20 (2003) 1423. 52. H.H.Y. Tong, B.Y. Shekunov, J.P. Chan, C.K.F. Mok, H.C.M. Hung and A.H.L. Chow, Int. J. Pharm., 295 (2005) 191. 53. V.A. Drebushchak, E.V. Boldyreva, T.N. Drebushchak and E.S. Shutova, J. Cryst. Growth, 24 1 (2002) 266. 54. S. Nurono, Moegihardjo, R. Mauludin and Subarno, Majalah Farmasi Indonesia, 11 (2000) 150. 55. S-L. Wang, S-Y. Lin and Y-S. Wei, Chem. Pharm. Bull., 50 (2002) 153. 56. L.E. O'Brien, P. Timmins, A.C. Williams and P. York. J. Pharm. Biomed. Anal., 36 (2002) 335. 57. J.W. Steed and J.L. Atwood, Supramolecular Chemistry, John Wiley & Sons, Ltd., New York, 2000. 58. A.L. Bingham, D.S. Hughes, M.B. Hursthouse, R.W. Lancaster, S. Tavener

and T.L. Threlfall, Chem. Commun., (2001) 603. 59. F. Giordano, C. Novak and J.R. Moyano, Thermochim. Acta, 380 (2001) 123. 60. G. Bettinetti, M. Sorrenti, L. Catenacci, F. Ferrari and S. Rossi, J. Pharm. Biomed. Anal., 41 (2006) 1205. 61. M.R. Caira and L.R. Nassimbeni, pp.825-850, in Comprehensive Supramolecular Chemistry 1996, Vo1.6, edited by J.L. Atwood, J.E.D. Davies, D.D. Macnicol and F.Vogtle, Elsevier Science, New York, 1996. 62. S.R. Byrn, R.R. Pfeiffer and J.G. Stowell, Solid-State Chemistry of Drugs, SSCI, Inc., Indiana, USA, 1999. 63. L.R. Nassimbeni, Acc. Chem. Res., 36 (2003) 631. 64. E. de Vries, L.R. Nassimbeni and H. Su, Eur. J. Org. Chem., (2001) 1887. 65. M.E. Brown, Introduction to Thermal Analysis -Techniques and Applications, 2nd Edn, Kluwer, Dordrecht, 2001. 66. M.R. Caira, G. Bettinetti and M. Sorrenti, J. Pharm. Sci., 91 (2002) 467. 67. A.R. Sheth, D. Zhou, F.X. Muller and D.J.W. Grant, J. Pharrn. Sci., 93 (2004) 3013.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 17

APPLICATIONS OF THERMAL ANALYSIS TO THE STUDY OF DENTAL MATERIALS William A. Brantley Section of Restorative and Prosthetic Dentistry & Director, Graduate Program in Dental Materials Science, College of Dentistry, The Ohio State University, Columbus OH, USA 1. INTRODUCTION A wide variety of dental materials is used in the oral environment for restorative, prosthetic and implant applications, and these materials are described at length in textbooks [I-31. While their physical and mechanical properties have been studied extensively, there has been relatively little use of thermal analysis techniques to gain insight into dental materials. Our group has performed extensive research on several metallic and polymeric dental materials, principally utilizing conventional DSC and temperature-modulated DSC (TMDSC). These studies and thermal analysis studies of dental materials by other research groups are reviewed in this chapter. Although much novel information has been provided, numerous matters are discussed that require additional research. 2. NICKEL-TITANIUM ALLOYS IN DENTISTRY 2.1. Metallurgy background Nickel-titanium alloys, based upon the equi-atomic intermetallic compound NiTi, have very low values of elastic modulus (approximately 35 GPa), compared to stainless-steel alloys (approximately 160-180 GPa) [4]. As a consequence, nickel-titanium alloys have considerable clinical importance for endodontic instruments, permitting negotiation of curved root canals with much greater facility than traditional stainless-steel instruments [5], and for orthodontic wires that have a much more favourable light-force delivery for tooth movement than traditional stainless-steel orthodontic wires [6,7]. NiTi can exist in three phases [8-101: (1) a high-temperature, low-stress austenite form, (2) a low-temperature, high-stress martensite form, and (3) the

R-phase, which serves as an intermediate phase to facilitate the transformation between martensite and austenite. Formation of the R-phase is reported to arise from the presence of dislocations and precipitates [I 11. A substantial dislocation density is expected in the wrought nickel-titanium endodontic instruments and orthodontic wires, which are subjected to extensive mechanical deformation during manufacturing processes [12]. Microstructural precipitates are a consequence of the inevitable deviation of the nickel-titanium alloy composition from the equi-atomic NiTi composition [13,14]. The structural transformation between austenite and martensite occurs when the mechanical stress attains a certain level, or with an appropriate temperature change. A reversible twinning process takes place at the atomic level, which can result in superelastic behaviour and shape memory [a]. The properties of the nickel-titanium endodontic instruments and orthodontic wires depend critically upon the nature and proportions of the NiTi phases in their microstructures, as discussed in the following sections. While X-ray diffraction has been used to study the phases in nickel-titanium endodontic instruments [ 15,161 and orthodontic wires [7,17,18], this analytical technique is limited to a near-surface region less than 50 ym in depth for metallic materials [19], and study of the phase transformations with temperature is not generally convenient. In contrast, DSC can provide information about the phases present in bulk nickel-titanium endodontic instruments and orthodontic wires with facility, and the effect of temperature changes on the NiTi phase transformations is easily studied.

2.2. Nickel-titanium endodontic instruments Endodontics is concerned with the removal of damaged, diseased or necrotic pulp tissue from root canals of teeth, followed by placement of filling and sealing materials. In current practice, rotary drills in the dental handpiece are commonly used to rapidly enlarge root canals to facilitate subsequent removal of pulp tissue. The manufacturing of nickel-titanium rotary endodontic instruments, which involves machining of a wire blank into a variety of cross-sectional shapes that depend upon the particular product, is described in a recent review article [20]. In that article it is stated that the nickel-titanium alloy for these instruments is in the superelastic condition, for which the alloy has the austenitic structure. This statement is highly plausible, because extensive reversible elastic strain (up to approximately 10% for uniaxial tension) could then occur in the instrument when the stress in the root canal reaches the level that causes transformation from austenite to martensite [21]. The first published verification [22] of this superelastic condition was obtained by our research group from DSC experiments on nickel-titanium rotary instruments in the as-received condition. A subsequent study evaluated the rotary instruments after clinical use [23].

Two clinically popular instruments (ProFile 0.04 taper, Dentsply Tulsa Dental, Tulsa, OK, USA; LightSpeed, Lightspeed Technology, San Antonio, TX, USA) were analysed in the as-received condition [22]. The original 1.0 mm wire blanks used for machining the ProFile instruments were also obtained from the manufacturer for analysis. Five randomly selected specimens were analysed for the 25 mm long, I S 0 size ProFile instruments. Four specimens were randomly selected for analysis from a package of 25 mm long LightSpeed instruments that contained I S 0 sizes 40,42.5,45 and 47.5. Two specimens from the starting wire blanks for the ProFile instruments were also analysed. The DSC test specimens consisted of two to four segments, approximately 4 - 5 mm in length, that were carefully cut from the instruments, or starting wire blanks, using a slow-speed water-cooled diamond saw to avoid the introduction of stresses that would alter the NiTi phases. The DSC analyses (Model 2910 DSC, TA Instruments, Wilmington, DE, USA) for the as-received instruments and wire blanks [22] were performed from -125 to 100 OC, using a liquid-nitrogen cooling accessory (TA Instruments) to achieve subambient temperatures. For each analysis, the test specimen was cooled from room temperature to -125 "C, heated to 100 OC, and then cooled back to -125 "C. The DSC plots were obtained for both the heating and cooling cycles, since the phase transformation behaviour for the nickel-titanium alloy differed in the forward and reverse directions, and the linear heating and cooling rate was 10 "C per minute. The DSC cell was purged with nitrogen gas during the analyses, and n-pentane, deionized water, and indium were used for temperature calibration. Because the imposition of sufficiently high stress on the nickel-titanium alloy segments could alter the proportions of the NiTi phases, the test segments for each specimen were placed in an open pan; an empty aluminium pan served as the inert control specimen. Interpretation of the DSC analyses of the nickel-titanium rotary instruments was based upon articles that discussed the DSC plots for nickel-titanium engineering alloys [lo] and orthodontic wires [7,24,25]. To investigate the effects of clinical use, single instruments (25 mm long, IS0 size 25) from another package of the same two nickel-titanium rotary drills were subjected to one, three, or six uses of simulated clinical instrumentation in extracted teeth [23]. A fourth instrument, in the as-received condition from the same package, served as a control for DSC analyses of the used instruments. Each test specimen was a single segment of 4 - 5 mm length that had been carefully cut from the tip region, or the adjacent two regions on the same instrument shaft, using the slow-speed, water-cooled diamond saw. Figures 1 and 2 present DSC plots for test specimens from as-received ProFile and LightSpeed instruments, respectively [22]. The lower plot (solid curve) is for the initial heating cycle, and the upper plot (dashed curve) is for the

subsequent cooling cycle. The two endothermic peaks during the heating cycle correspond to the initial transformation from martensite to R-phase, followed by transformation at higher temperature from R-phase to austenite, which has a greater enthalpy change. The enthalpy change for the first transformation is less than that for the second transformation in Figure 1, whereas the relative magnitudes of the enthalpy changes for these two transformations are reversed in Figure 2. The broad, asymmetric shape of the single exothermic peak on the cooling curves in Figures 1 and 2 suggests that two unresolved peaks are present, in which the higher-temperature peak, corresponding to transformation from austenite to R-phase, has a greater enthalpy change than the lowertemperature peak for transformation from R-phase to martensite. This interpretation is consistent with recent temperature-modulated DSC (TMDSC) studies of nickel-titanium orthodontic wires, which are discussed in Section 2.3. These studies have shown that R-phase generally appears during the forward and reverse transformations between martensite and austenite. Future TMDSC studies should be performed on the rotary nickel-titanium instruments to confirm that these same two-step phase transformations take place. The DSC plots for test specimens from the starting wire blanks for the ProFile instruments were very similar to Figure 1.

-0.10

0.075 -75

Exo Up

-25

25

Temperature ("C)

75 Universal V2.5H TA Instruments

Figure 1. DSC curves for a specimen of several segments from an as-received ProFile rotary NiTi instrument [22]. For Figs. 1 to 6, the solid line is the heating curve, and the dashed line is the cooling curve. Reproduced with permission fiom Elsevier Science.

4

-0.15 -75

Exo Up

1 0.00 -25

25

Temperature ("C)

75 UniversalV2.5H TA Instruments

Figure 2. DSC curves for a specimen of several segments from an as-received LightSpeed rotary NiTi instrument [22]. Reproduced with permission from Elsevier Science. Table 1 summarizes the DSC results for all as-received ProFile and LightSpeed test specimens, including the wire blanks [22]. This table presents approximate values for: (a) the range of AH values for the overall forward transformation from martensite to austenite on heating; (b) the range of AH values for the overall reverse transformation from austenite to martensite on cooling; (c) the range of temperatures for the peak associated with the transformation from martensite to R-phase on heating; (d) the range of temperatures for the peak associated with transformation from R-phase to austenite on heating; (e) the range of temperatures for the peak associated with the transformation from austenite to R-phase on cooling; and ( f ) the approximate austenite-finish (Af) temperature at which the transformation to austenite during heating was completed. The DSC experiments showed that the nickel-titanium alloys for both the ProFile and LightSpeed instruments were in the expected [20] superelastic condition under clinical conditions, since transformation to austenite is completed at approximately 25 "C. The AH values in Table 1 lie within the range of 1.7 to 19.2 Jlg reported for nickel-titanium orthodontic wires [7]. From DSC results for shape memory orthodontic wires [7,25], these nickel-titanium rotary instruments might possess shape memory at room temperature (and in the oral environment) if manufacturers used the appropriate processing steps [26]. However, future

experimental measurements of the loading and unloading behaviour of these instruments are required to establish the existence of shape memory. It is can be seen in Table 1 that the A H values for the overall forward and reverse transformations between martensite and austenite are different, particularly for the LightSpeed instruments. When comparing different brands (such as ProFile and LightSpeed) of nickel-titanium rotary instruments, it is tempting to assume that a greater AH for the transformation from martensite to austenite corresponds to a superior alloy containing less non-transforming, work-hardened martensite . However, further research is required to test this correlation between DSC results and clinical performance. Table 1 Approximate values from DSC plots for two brands of as-received nickeltitanium endodontic instruments [22]. Reproduced with permission from Elsevier Science. Property

AH (martensite to austenite) Heating, Jlg AH (austenite to martensite) Cooling, Jlg Peak temperature, martensite to R-phase Heating, "C Peak temperature, R-phase to austenite Heating, "C Peak temperature, austenite to R-phase Cooling, "C Af temperature Heating, "C

ProFile*

Lightspeed

3.3 to 4.9

7.4 to 8.4

2.8 to 3.6

1.8 to 2.6

-25 to -14t

-18 to -16

+1 to +10

+2 to +3

-6 to +5

-3 to 0

+25

+25

"Includes data for instruments and blanks. ?Not clearly defined for samples from three instruments, i.e., based upon samples from two instruments and two wire blanks. Figure 3 presents the DSC plots for the tip segment from another ProFile instrument in the as-received condition [23]. The single endothermic peak on the heating curve is again assumed to consist of two unresolved overlapping peaks for the transformation from martensite to R-phase at lower temperatures, followed by transformation from R-phase to austenite. For the cooling curve, the higher-temperature exothermic peak (AH = - 1.1 Jlg) is assumed to correspond to initial transformation from austenite to R-phase, followed by

transformation from R-phase to martensite at lower temperatures, rather than to direct transformation from austenite to martensite. The broad, low-temperature exothermic peak complex below -60 OC on the cooling curve is assumed to arise from twinning within the martensite. The reasons for these interpretations are discussed in Section 2.3.

-0.4 -1 -100

Exo Up

-75

-50

-25

Temperature ("C)

0

10.1 25

Universal V2.5H TA Instruments

Figure 3. DSC curves for a specimen consisting of the tip segment from another as-received ProFile rotary instrument [23]. Reproduced with permission from Elsevier Science. Figure 3 shows that the nickel-titanium alloy for the single tip segment from this as-received ProFile instrument is also in the superelastic condition under clinical conditions, because the Af temperature is less than 0 OC [23]. However, the DSC curves in Figure 3 are considerably different from those in Figure 1 for a test specimen consisting of several segments from another as-received ProFile instrument. This difference presumably arises from the greater amount of work hardening experienced by the nickel-titanium alloy in the tip region for the instrument in Figure 3 during its fabrication. When two adjacent segments from the same as-received ProFile instrument were included with the tip region for DSC analysis, the resulting curves were very similar to those in Figure 1 [23]. Thus, the DSC curves in Figure 1 represent a range of work-hardening conditions, because the test specimen consisted of several adjacent segments from the same instrument. Figure 4 presents the DSC curves for the tip segment from another ProFile instrument after one simulated clinical use [23]. There was little difference for

the DSC curves from the tip segments of other ProFile instruments subjected to three and six clinical uses. The likely interpretation for the single endothermic peak on the heating curve is that there are two unresolved overlapping peaks due to the initial transformation from martensite to R-phase, followed by the transformation from R-phase to austenite. The transformations on the cooling curve are shifted to much lower temperatures, where the onset for the broad peak complex is nearly -70 "C. The shape of this complex peak suggests that two or three incompletely resolved overlapping peaks are present, tentatively interpreted as a continuum of transformations from austenite to R-phase to martensite, with perhaps subsequent twinning of martensite.

4

-0.12 -100 Exo Up

0.18 -75

-50

-25

Temperature ("C)

0 Universal V2.5H TA Instruments

Figure 4. DSC curves for a specimen consisting of the tip segment from a ProFile rotary instrument subjected to one simulated clinical use [23]. Very similar DSC curves were obtained for tip segments hom other ProFile rotary instruments that had been subjected to three and six simulated clinical uses. Reproduced with permission from Elsevier Science. Figures 5 and 6 show DSC curves for the tip segment from other LightSpeed instruments in the as-received condition and after one clinical use, respectively [23]. Comparing the DSC curves in Figure 2 for several segments from a different as-received LightSpeed instrument, it is evident that the tip segment for Figure 5 is highly work hardened. It can also be seen that the DSC curves for the tip segments from the as-received LightSpeed instrument in Figure 5 and from the instrument that had been subjected to one simulated clinical use in Figure 6 are very similar. Very similar DSC curves were also obtained for the tip segments from the two other LightSpeed instruments that were subjected to three and six clinical uses.

, ,

-

//'<.

/'-

-

-0.15 - -89.82"C

2

.,X/+'-,+

2 -

-0.25 -

--------

/-//

4 ' '

-81.77"C 1.555Jlg

P

- o.6

r , / ' -

-

E

-6.44"C 0.4315Jlg

a,

I

-ch 7.93%

a,

r3 T

- 0.4 1

I -0.35

L

-

-0.45 7 -100

t

-75

-50

-25

Temperature ("C)

Exo Up

0

0.2

25

Universal V2.5H TA Instruments

Figure 5. DSC curves for a test specimen consisting of the tip segment from another as-received LightSpeed rotary instrument [23]. Reproduced with permission from Elsevier Science.

-0.2 -100

Exo Up

0.15 -75

-50

-25

Temperature ("C)

0

25

Universal V2.5H TA Instruments

Figure 6. DSC curves for a test specimen consisting of the tip segment from the LightSpeed rotary instrument that was subjected to one simulated clinical use [23]. Reproduced with permission from Elsevier Science.

There was also little difference in the DSC curves for the tip segment fiom a LightSpeed instrument subjected to six simulated clinical uses. The two endothermic peaks on the heating curves in Figures 5 and 6 are interpreted as due to transformation from martensite to R-phase, followed by transformation from R-phase to austenite. The low-temperature exothermic peaks on the cooling curves in Figures 5 and 6 are attributed to the same continuum of transformations proposed for Figure 4, but the enthalpy change in the former figures has been decreased due to greater work hardening in these tip segments from the LightSpeed instruments after more extended use. Detailed consideration of the transformation temperatures and enthalpy changes for the tip segment, middle segment, and third segment hrthest from the tip, for the control and used ProFile instruments, revealed no consistent and noteworthy effect of the simulated clinical use, as shown in Table 2 [23]. Table 2 Properties determined fiom DSC plots for ProFile instruments after simulated clinical use [23]. Reproduced with permission from Elsevier Science. Specimen Heating Cooling R,

A,

AH

R, or M,

AH

Mpeak

AH(J/g)

Control: tip segment Control: middle segment Control: third segment Control: all three segments One use: tip segment One use: middle segment One use: third segment Three uses: tip segment Three uses: middle segment Three uses: third segment Six uses: tip segment Six uses: middle segment Six uses: third segment

Notes: R,, A, and M, are starting temperatures for transformations to R-phase, austenite and martensite, respectively. AH is for the overall transformation between martensite and austenite. The lack of definite trends for AH values in Table 2 suggests that there is variability in work hardening of the NiTi microstructural phases along the instrument shafts and also with nominally equivalent instruments for the same ProFile product. Recent Vickers hardness measurements of clinically used

nickel-titanium rotary instruments have shown [27] that the tip regions did not experience significant work hardening beyond the values of Vickers hardness reported [7] for several nickel-titanium orthodontic wires that possess in vivo shape memory. Measurements of Vickers hardness at other positions along the instrument shafts revealed [27] substantial variability, which is consistent with the variability of the DSC results in Table 2 [23]. Since our pioneering study on as-received [22] nickel-titanium endodontic instruments, other DSC studies have confirmed that as-received instruments are in the superelastic condition, which persists after numerous sterilization cycles [28,29]. As would be expected, mechanical properties of these instruments are related to the phase transformation behaviour of the nickel-titanium alloy [30]. Accordingly, suitable elevated-temperature heat treatment may favourably alter the mechanical properties of these instruments [31,321, as was previously found for nickel-titanium orthodontic wires [2 1,331. Future basic scientific research is needed to elucidate the fundamental mechanisms for the notable differences in DSC plots for as-received rotary instruments, as well as for the profound effects of clinical use on NiTi structural transformations without loss of the superelastic character of the instruments. Evidently, these variations are due to complex effects of work hardening on the NiTi microstructural phases.

2.3. Nickel-titanium orthodontic wires Orthodontics is concerned with tooth movement to optimal positions, using metallic archwires ligated to brackets bonded to enamel or dental restorations by adhesive resins, as well as using suitable other metallic appliances, to provide appropriate forces and bending moments in vivo. The force generated by a bent orthodontic wire is proportional to its elastic modulus, and relatively light and continuous forces are considered to be optimum. There is considerable interest in nickel-titanium orthodontic wires, which have the lowest elastic modulus of the major wire alloys [7]. The nickel-titanium orthodontic wire (Nitinol, 3M Unitek, Monrovia, CA, USA) that was first marketed [6] had a heavily cold-worked stable martensite structure [25], but superelastic wires [21] were subsequently introduced and followed [34] by wires having shape memory in the oral environment [7]. For the most efficient treatment, the archwire that is bent by the clinician for treating malpositioned teeth should return completely to the initial undeformed state during the process of tooth movement. Complete recovery will not occur for superelastic wires if there is permanent deformation 1331, whereas full recovery will take place for nickel-titanium wires with in vivo shape memory. Our research group performed a DSC study [25] to investigate the differences in phase transformation behaviour for two superelastic, one non-superelastic and

two shape-memory nickel-titanium orthodontic wires in clinically popular sizes. Four wire products were round (0.016 inch diameter), and the fifth product was rectangular (0.016 inch x 0.022 inch). Specimens for each wire consisted of five straight 5 mm segments, which were carefully cut from the curved arch blanks received from the manufacturer, using a water-cooled, diamond saw. Open aluminium pans were used for the DSC analyses (TA 9 10, TA Instruments), and the analyses were performed from approximately -125 to 100 "C, using a scanning rate of 10 "C per minute with dry nitrogen as a purge gas. The DSC plots showed that the two superelastic wires were characterized by two peaks on the heating curve that were similar to those in Figures 1 and 2, corresponding to the initial transformation from martensite to R-phase and the subsequent transformation fi-om R-phase to austenite [25]. In contrast to Figures 1 and 2, the two peaks on the cooling curve for these two superelastic wires, corresponding to transformation fi-om austenite to R-phase followed by transformation from R-phase to martensite, were much more widely separated (approximately 60 OC and 80 "C). The DSC plots for the two shape-memory wires were both characterized by a single peak on the heating curve, indicative of direct transformation from martensite to austenite, whereas two peaks were observed on the cooling curves, corresponding to transformation fi-om austenite to R-phase followed by transformation from R-phase to martensite [25]. In contrast to Figures 1 and 2, these two peaks on the cooling curves for the shape-memory wires were much more widely separated (by approximately 50 "C). The DSC plot for the non-superelastic wire was characterized by very shallow peaks: two widely separated peaks (approximately 70 OC) on the cooling curve and a single broad peak on the heating curve [25]. The A H value for the transformation on heating from martensite to austenite for this wire was an order of magnitude lower than that for the two shape-memory wires and one superelastic wire, and 20% of that for the other superelastic wire, showing that the non-superelastic wire largely consisted of stable, work-hardened martensite. In pioneering studies [35-371, our research group employed TMDSC to obtain new insight into the structural transformations in nickel-titanium orthodontic wires. These studies (Model 2910 DSC, TA Instruments, Wilmington, DE) were performed over the temperature range from approximately -125 to 100 "C, using a linear heating and cooling rate of 2°C per minute, with a superimposed sinusoidal oscillation having an amplitude of 0.3 18 "C and period of 60 seconds to provide heating-only conditions. Dry helium was used as the purge gas at rates of 25 cm3/minute through the purge port and 100 cm3/minute through the vacuum port. Temperature calibration of the TMDSC apparatus was performed with the use of n-pentane, deionized water and indium, whose melting points spanned the range of temperatures used for analyses of these orthodontic wires.

Three nickel-titanium wires studied by TMDSC had been previously investigated by conventional DSC [25]: superelastic Nitinol SE (3M Unitek); Neo Sentalloy, which has in vivo shape memory (GAC International, Bohemia, NY, USA); and nonsuperelastic Nitinol. The Nitinol SE and Neo Sentalloy archwires had 0.016 inch x 0.022 inch cross-sections, and the Nitinol archwires were 0.019 x 0.025 inch, also a clinically popular size. In addition, the effect of permanent deformation on transformations in Nitinol SE and Neo Sentalloy was studied after bending the archwires to 135' with orthodontic pliers. The archwires were carefully cut into approximately 3-4 mm length segments with a water-cooled diamond saw, and each test specimen consisted of five or six such segments. For the three wire products where bent specimens were analysed, each of the wire segments for the test specimen was the portion of an archwire that had been bent. The test specimens were placed in open aluminium pans, and an empty aluminium pan again served as the inert control specimen. The TMDSC plots for the heating cycle of superelastic Nitinol SE are shown in Figure 7 [37], where two peaks involving the R-phase are evident.

Exo Up

Temperature ("C)

Universal V2.5H TA Instruments

Figure 7. TMDSC reversing and nonreversing heat flow curves for the heating cycle of as-received Nitinol SE [37]. Reproduced with permission from Elsevier Science. Figure 8 shows that the transformation from R-phase to martensite on cooling appears as an unresolved left shoulder on the reversing heat flow peak for the

transformation fi-om austenite to R-phase. This transformation also appears as part of an unresolved, broad lower-temperature peak on the nonreversing heat flow curve. The Af temperature for Nitinol SE in Figure 7 for the heating cycle is near 70 OC, in good agreement with 63 OC found by conventional DSC [25].

Exo Up

Temperature ("C)

Universal V2.5H TA Instruments

Figure 8. TMDSC reversing and nonreversing heat flow curves for the cooling cycle of Nitinol SE [37]. Reproduced with permission from Elsevier Science. Broad, low-temperature exothermic peaks appear on the nonreversing heat flow curves in Figures 7 and 8, indicating transformation within martensite to another structure (M'). Low-temperature peaks had previously been reported from DSC [lo] and electrical resistivity measurements 1381 of engineering and NiTi orthodontic alloys, respectively. However, our group was unable to confirm these peaks on DSC plots for five representative orthodontic wires[25]. Our recent low-temperature transmission electron microscopy study [39] has shown that these peaks are due to twinning within martensite. This process releases energy during heating or cooling, thereby yielding exothermic peaks. Similar results were obtained for TMDSC analyses of the in vivo shapememory wire, Neo Sentalloy, shown in Figures 9 and 10 [37]. While our previous DSC study [25] suggested that direct transformation of martensite to austenite occurs during heating, the nonreversing heat flow curve in Figure 9 shows that a two-step transformation involving R-phase actually takes place.

1

-0.08 -150

I-

-100

-50

0

Temperature ("C)

Exo Up

50

-0.02 100

Universal V2.5H TA Instruments

Figure 9. TMDSC reversing and nonreversing heat flow curves for the heating cycle of Neo Sentalloy [37]. Reproduced with permission from Elsevier Science.

4

0.00 -150 EXO

Up

-100

-50

0

Temperature ("C)

50

! -0.02 100

Universal V2.5H TA Instruments

Figure 10. TMDSC reversing and nonreversing heat flow curves for the cooling cycle of Neo Sentalloy [37]. Reproduced with permission from Elsevier Science.

The TMDSC plots in Figures 11 and 12 for Nitinol are complex with shallow peaks [36]. Because of the high-sensitivity setting on the vertical axis, the reversing heat flow curve is not a straight line. -0.014

,

~

- 0.010

I

I

'

I

-0.018 -150

-1 00

-50

0

Temperature ("C)

Exo Up

-0.010 100

50

Universal V2.5H TA Instruments

Figure 11. TMDSC reversing and nonreversing heat flow curves for the heating cycle of Nitinol [36]. Reproduced with permission from Elsevier Science.

4

0.0145 -130

Exo Up

. -80

-30

20

Temperature ("C)

-0.03

70 Universal V2.5H TA Instruments

Figure 12. TMDSC reversing and nonreversing heat flow curves for the cooling cycle of Nitinol [36]. Reproduced with permission from Elsevier Science.

Interpretations of the TMDSC plots for Nitinol are more tentative. For the heating cycle in Figure 11, the broad peak on the reversing heat flow curve is assumed to correspond to the two-step transformation process of M+R followed by R-+A, while the broad lower-temperature peak complex on the nonreversing heat flow curve is assumed to correspond to twinning within martensite (Mf-+M)followed by transformation M+R. For the cooling cycle in Figure 12, the broad peak on the reversing heat flow curve may correspond to A+R, followed by R+M and then three successive twinning episodes within martensite. X-ray diffraction analyses of Nitinol over the temperature range for Figures 11 and 12 would be required to verify these tentative interpretations. The placement of permanent bends of 135" in the nickel-titanium archwires failed to yield definitive increases (AH values) for the low-temperature martensite peaks on the TMDSC plots, compared to those for the as-received wires [35,37]. It was assumed that bends with acute angles would be needed to increase the amount of work-hardened martensite sufficiently to cause significant increases in the enthalpy changes associated with these lowtemperature peaks.

3. DENTAL POLYMER MATERIALS 3.1. Silicone maxillofacial materials Maxillofacial materials are used in dentistry for prosthodontic reconstruction, for example with patients who have experienced accidents or cancer surgery that involved the orofacial complex. Currently, silicone and polyurethane elastomers are employed for these applications [1,2]. Essential requirements for maxillofacial materials are biocompatibility, the capability of shade matching to the surrounding soft tissues, and adequate mechanical properties to provide acceptable clinical longevity. It is desirable that the viscoelastic response and creep compliance of these materials mimic the in vivo force response of these soft tissues under functional loading without undergoing excessive dimensional changes over time. Since the viscoelastic and creep properties of polymeric materials are often correlated with their glass-transition temperatures (T,) [40], we considered that DSC would be a useful technique to investigate commercial maxillofacial materials. In our pioneering study [41], the effects of pigments on the DSC plots were studied for a heat-polymerized maxillofacial silicone (MDX 4-45 15, Dow Coming, Midland, MI, USA), which served as a model material. Four different shades of the silicone, within the colour space of human subjects, were obtained, using 7 different pigmenting species in minute concentrations not exceeding 0.2 mass % (and often in much lower concentrations) with master batches that were blended into the elastomer [41].

Processing (polymerizing) for these shades of the silicone was performed at 100 "C for 1 hr, and clear unpigmented silicone specimens were also prepared with processing times of 1, 4 and 20 hours at 100 "C. A starting, partially polymerized, unpigrnented silicone specimen that was not heat-processed was included for comparison. The DSC analyses were performed from -150 to 200 OC (TA 910, TA Instruments) at a scanning rate of 10 OC per minute, using an empty aluminium pan as the inert reference material and nitrogen as the purge gas. Five specimens were prepared for each pigmented condition to permit statistical comparisons. Figure 13 shows the DSC heating curve for the lightest shade of the silicone, which contains the melting peak of the crystalline polymer [41]. The specimens for all pigmented conditions had very similar DSC plots, and mean values differed by only small amounts for peak temperature (2.8 OC range) and AH (3.0 J/g range). No significant differences in the mean peak temperatures were found for the unpigmented silicones. For the four pigmented silicones, the mean AH was significantly lower for the lightest shade, and the mean peak temperature was significantly higher for the darkest shade.

Temperature YC)

Figure 13. Heating DSC plot for lightest shade of MDX 4-4515 silicone maxillofacial material [41]. Reproduced with permission from Quintessence Publishing Company, Inc. Careful examination of the lowest-temperature portions of the DSC plots revealed glass transitions between approximately -125 "C and -130 "C, in good agreement with the reported T, for polydimethylsiloxane (PDMS) [42].

Figure 14 shows this weak glass-transition for the maxillofacial silicone with a dark shade.

-0.320 -140

-135 Temperature ("C)

-130

Figure 14. Glass transition for a dark shade of MDX 4-4515 silicone [41]. Reproduced with permission from Quintessence Publishing Company, Inc. A TMDSC analysis was performed (TA 2100) from -160 OC to -100 OC on a polymerized unpigmented specimen, using a heating rate of 5 OC per minute, a superimposed sinusoidal waveform of h0.7 "C amplitude and a 60 second period. Helium was the purge gas. Figure 15 shows a weak glass-transition at approximately -125 OC on the reversing heat-flow curve and that the polymer melting peak has both reversing and nonreversing character. The general explanation [43] of this behaviour is that the reversing character occurs during the early stages because numerous crystallites are undergoing melting and crystallization during temperature modulation, whereas the nonreversing character is associated with the final stages after nearly all crystallites have melted and substantial recrystallization is no longer possible. The reported range of 30 to 36 Jlg for the enthalpy of fusion for crystalline PDMS [42] is substantially greater than the AH range of 13.5 to 16.5 J/g for the melting peak of the MDX 4-45 15 maxillofacial silicone [41], which contains a small percentage of vinyl side-chains and a silica filler [2]. The small differences in mean values of the melting peak temperatures and the M s for the different pigmented specimens suggests that there may be differences in their viscoelastic behaviour and creep. However, the MDX 4-4515 model silicone selected for our study is no longer marketed, and h r e DSC studies are needed to evaluate the effects of pigmentation on current maxillofacial silicone and

polyurethane elastomers and to assess the correlation between such studies, the clinically relevant mechanical properties of the materials, and their in vivo performance in patients.

-3

0 - -0.1 = .-

a (I)

c 0, c --

.-

t? (I)

- -0.2 g>

Temperature' (OC)

Figure 15. TMDSC curves showing reversing, nonreversing and total heat flow for the unpigmented MDX 4-4515 silicone after 1 hour processing [41]. Reproduced with permission from Quintessence Publishing Company, Inc.

3.2 Elastomeric impression materials The preceding research on the model maxillofacial material was followed by TMDSC study of several representative elastomeric impression materials, which are extensively used in dentistry for the accurate fabrication of inlays and crowns from dental alloys, metal-ceramic restorations, and fixed and removable partial dentures [I-31. There have been numerous studies reporting the clinically relevant properties of these impression materials (viscosity before setting by polymerization, strain in compression after setting, permanent deformation for simulated in vivo removal of the impressions, and tear strength of the thin impressions). However, only minimal research has been reported [44] on some thermal properties of impression materials obtained by conventional DSC. Our pioneering TMDSC study [45] was designed to obtain fundamental information about impression materials and seek correlations with their relevant properties. Two vinyl polysiloxane (silicone) impression materials [Examix Light-Body, GC America, Alsip, IL, USA; Reprosil Light-Body, LD Caulk, Milford, DE,

USA], one polyether impression material [Impregum-Medium Body, ESPE America, Norristown, PA, USA], and one polysulfide impression material [CoeFlex Injection (Light-Body), GC America] were selected. The terms Light-Body and Medium-Body refer to viscosity levels provided by the manufacturers [I-31. Appropriate procedures were employed for mixing each material, and samples of uniform thickness (approximately 0.5 mm) were obtained by compressing the mixed material between glass microscope slides [45]. The TMDSC analyses were conducted (Model 2910, TA Instruments) from -150 "C to 200 "C, using a linear heating rate of 2 "C per minute and helium as purge gas. Because of concern with the modulation condition in our preliminary investigation [46], two different heating-only modulation conditions were employed [45]: (a) a period of 60 seconds with an amplitude of 0.3 18 "C (from TA manual); (b) a period of 100 seconds with an amplitude of 0.50 "C. The smaller amplitude in condition (a) was used to minimize inducing crystallization peaks on the nonreversing heat flow curve for a crystalline polymer melting peak. The larger amplitude in condition (b) was used to increase the capability of detecting the very weak glass-transitions in the vinyl polysiloxane impression materials. Baseline calibrations were carried out for the two modulation conditions, and temperature calibrations were performed using n-pentane, acetone, water and indium. 0.00

-0.30 -150

Exo Up

0.14

-0.28 -50

50

Temperature ("C)

150 Universal V2.3C TA Instruments

Figure 16. TMDSC plot for Examix-Light Body and modulation condition (b) [45]. The total heat flow is shown in the top curve. Reproduced with permission from Elsevier Science. The TMDSC results for Examix silicone impression material and modulation condition (b) in Figure 16 are similar to those for the maxillofacial silicone in

Figure 15. The very weak glass-transition near -125 "C is barely evident in Figure 16. As expected, for modulation condition (a) the relative nonreversing character of the crystalline polymer melting peak near -50 "C was substantially less than its reversing character (relative AH values). Similar TMDSC results were found for the second silicone impression material Reprosil [45]. The TMDSC results are shown for the Impregum polyether impression material and the two modulation conditions in Figures 17 and 18, respectively.

-0.75-1 -150

Exo Up

-

, -100

.

, -50

.

, 0

.

, 50

Temperature ("C)

.

.

100

.

.

150

.

C-0.40 200

Universal V2.3C TA Instruments

Figure 17. TMDSC plot for Impregum Medium-Body and modulation condition (a) [45]. Top curve: total heat flow. Reproduced with permission from Elsevier Science. This polyether impression material undergoes a glass transition near -80°C and has melting peaks near -20 "C and 50 "C. While there are exothermic peaks indicative of crystallization on the nonreversing heat-flow curve for modulation condition (b) in Figure 18, only endothermic peaks that would not correspond to crystallization are found on the nonreversing heat-flow curve for modulation condition (a) in Figure 17. The TMDSC results are shown for the Coe-Flex polysulfide impression material and modulation condition (a) in Figure 19. There is a glass transition near -55 "C and an apparent weak crystalline polymer melting peak near 70 "C. No evidence of an exothermic crystallization peak can be seen on the bottom nonreversing heat-flow curve. The peak near 190 "C requires further study, but is not clinically relevant for the properties of the impression material.

-

3 0.06

-

TT---------

-82,400c

--+

--dm-

0.09

-2O.8S0C 29.21Jlg

47.70°C 25.02Jlg

LL

a,

I

1

-7.78%

6 0.03

-,

C

52.47%

Z 0.00

-0.60 I -150

-

,

-

I

-100

.

I

-50

'

r

0

50

T

'

100

Temperature ("C)

Exo Up

-

150

-0.40

200

Universal V2.3C TA Instruments

Figure 18. TMDSC plot for Impregum Medium-Body and modulation condition (b) [45]. Top curve: total heat flow. Reproduced with permission fi-om Elsevier Science.

-0.34 -150

Exo Up

-0.10 -100

-50

0

50

Temperature ("C)

100

150

200

Universal V2.3C TA Instruments

Figure 19. TMDSC plot for Coe-Flex Injection Light-Body and modulation condition (a) [45]. Top curve: total heat flow. Reproduced with permission from Elsevier Science.

Additional TMDSC study of other vinyl polysiloxane, polyether and polysulfide impression materials is important to verify if the polymer transitions shown in Figures 16 to 19 generally exist in different products and to investigate the effects of other temperature modulation conditions. Complementary research on correlations with clinically relevant mechanical properties of the elastomeric impression materials is needed to verify if these thermal analyses have useful predictive power. Interestingly, when compared at apparently similar viscosities, the reported values of the elastic modulus [3] are highest for the vinyl polysiloxane silicone impression materials, intermediate for the polyether impression materials, and lowest for the polysulfide impression materials, in reverse order to the relative values of Tgfound in our thermal analyses [45]. Our X-ray diffraction and scanning electron microscopic study [47] of these impression materials has shown that they contain substantial amounts of crystalline filler particles in the micron size range, which are incorporated by manufacturers to achieve the clinically desired viscosity levels. These filler particles should have considerable influence on the mechanical properties of the impression materials. 3.3. Orthodontic elastomeric modules Elastomeric modules manufactured from polyurethanes, typically in spool chain form with links corresponding to individual teeth, are extensively used for space closure following tooth extraction and to provide correct tooth rotation during orthodontic treatment. While there have been many studies of mechanical properties and clinical performance of these viscoelastic materials [48], the use of thermal analysis to gain insight into commercially available products had been neglected until our pioneering DSC investigation [49]. The objective of that study was to compare products from three manufacturers having three popular colours (gray, red and purple) used in clinical treatment. The approach was to determine values of Tg for the as-received products and for modules retrieved after clinical use in non-smoking orthodontic patients for 4 months. Following pilot experiments to determine the appropriate sample size for statistical comparisons, seven specimens were analysed for the as-received condition of each product/colour combination. Each test specimen consisted of two links with a connector, cut from the same spool to avoid any effect of different batches. For the clinical testing, seven specimens of each product/colour combination were randomly placed in the mouths of the patients. The DSC analyses were performed from approximately - 6 0 "C to 150 "C, using a heating rate of 10 "C per minute. Dry ice was employed to cool test specimens to the starting temperature, and nitrogen was the purge gas. Figure 20 shows a representative DSC plot for an as-received test specimen, and the same general plots were obtained for all product/colour combinations,

differing only in the mean values for T,. Two products (Ormco, Glendora, CA, USA; G&H Wire Company, Greenwood, IN, USA) had similar mean values of T,, ranging from -39 "C to -46 "C. The third product [Rocky Mountain Orthodontics (RMO) Denver, CO, USA] had substantially higher mean values of T,, ranging from -21 "C to -24 "C. These results suggest [40] that there should be significant differences in the clinical force degradation for the Ormco and G&H products, compared to the RMO products, and that there should be little effect on clinical performance for the three pigments of each product, although there was greater variability in T, for different pigments of the G&H product.

-

0 -60

.

0 -40

9

! -20

.

~ 0

.

~ 20

. 40

~

. 60

Temperature ("C)

x

. 80

a

.

,

.

~

.

,

.

,

.

100 120 140 DSC V4.08 Dupont 2100

Figure 20. Representative DSC plot for an as-received red-pigmented Ormco test specimen, showing T, of -41 "C [49]. Reproduced with permission from Elsevier Science. After four weeks of clinical exposure to the oral environment, a second highertemperature glass-transition was observed for all product/colour combinations (Figure 21). A new endothermic peak was also present on the DSC plots. The values of the characteristic lower-temperature T, for the as-received modules lie within the range reported for biomedical polyurethanes [50]. For each product, the four-week exposure to the oral environment caused a decrease in this characteristic T, for two products: approximately 5 "C for the three colours of Ormco modules and approximately 4 to 9 "C for the three colours of RMO modules. For the G&H product, there was minimal change in T, for the red modules, a 4 "C decrease in T, for the gray modules, and a surprising 15 "C increase in T, for the purple modules. Because a previous study [51] reported

~

~

some variability in force delivery for purple Ormco modules, compared to other pigments, further DSC study of as-received batches of this product is needed.

-011

!

-60

.

.

-40

.

.

-20

.

, 0

.

, 20

.

, 40

.

,

.

60

Temperature ("C)

, 80

. ,

.

,

.

,

100

120 140 DSC V4.08 Dupont 2100

Figure 21. Representative DSC plot for a red Ormco test specimen after four weeks in an oral environment, showing the new higher-temperature glasstransition with T, of -26 OC [49]. Reproduced with permission from Elsevier Science. At present the origin of the new higher-temperature glass-transition that occurred in the polyurethane modules after four weeks in vivo is unknown. Calcium phosphate precipitates and biofilms with high protein density have been observed on orthodontic modules after three weeks in vivo [52], and some diffusion of ion species from salivary fluid into the polymer matrix of the polyurethane modules would be expected. Further DSC study of clinically used orthodontic modules is needed to determine the origins of the new highertemperature glass-transition and the associated endothermic peak.

3.4. Resin composites and other dental polymers Numerous thermal analysis studies have been performed by other groups on resin composites, which are used for small anterior restorations that do not experience substantial stress and with caution for posterior teeth because of concern about wear [I-31. The polymer matrix contains the oligomer bis-GMA (bisphenol A-glycidyl methacrylate) or urethane dimethacrylate, and triethylene glycol dimethacrylate is a diluent. Filler particles are silane-coated for chemical bonding with the matrix. Free-radical polymerization is carried out by chemical

means using an organic peroxide initiator and tertiary amine accelerator, or by visible light using the photoinitiator camphoroquinone. The cross-linked polymer contains a substantial percentage of residual C=C bonds, and the degree of conversion to single bonds is important for properties of the composite. Resin composites can be classified according to filler particles as fine-particle, hybrid, microhybrid and microfilled; other classifications such as flowable or packable are related to their manipulation [I-31. Quartz and glass (several types) fillers in fine-particle composites have sizes of about 0.5 to 3 pm. Microfilled and hybrid composites contain colloidal silica particles of 0.01 to 0.02 pm diameter incorporated in the polymer matrix. The microfilled composites also contain these submicron particles in ground 10 to 20 pm filler particles of the polymerized oligomers. The filler volume fraction for composite products varies widely fi-om about 20% to 70%. Clinical selection of composites depends upon strength, wear resistance and esthetics needed for the particular tooth restoration. Thermal analysis has been used extensively to characterize resin composites, which have been under development for several decades. Conventional DSC is an excellent method for investigating polymerization of the composites [53-571. The glass-transition temperature, which is highly relevant for the mechanical properties of the composites, can be readily found by dynamic mechanical analysis (DMA) [58-611, as shown in Figure 22, and by thermomechanical analysis (TMA) [62,63], where there is a discontinuity in the slopes of the plot of length change as a function of temperature below and above T,.

.

-

-50

0

50

100

Temperature ( O C ) Figure 22. DMA curves for a model cured bis-GMAITEGDMA specimen for resin composites [59]. Reproduced with permission from Elsevier Science.

Thermal analyses of other dental polymers have also been reported. The compositions of waxes have been studied by differential thermal analysis (DTA) [64-661 and DMA [67]. Conventional DSC and DMA recently showed the superiority of microwave-processed denture base resins to those processed by the slower traditional dental laboratory heating regimen [68]. In contrast, comparisons of T, and other relevant properties for a soft denture liner that was processed by the dental laboratory procedure and a more convenient chairside procedure indicated that both procedures yielded equivalent results [69]. 4. ACKNOWLEDGMENTS The author acknowledges extensive collaborative contributions to the original research summarized in this chapter from numerous colleagues: Satish Alapati, Thomas Bardin, Gerard Bradley, Bill Culbertson, Patrick Gallagher, Thomas Grentzer, Masahiro Iijima, John Powers, Timothy Svec, Michele Renick, Jeannie Vickery and Blaine Weddle.

5. REFERENCES

K.J. Anusavice (Ed.), Phillips' Science of Dental Materials, 11" Edition, Elsevier/Saunders, St. Louis, MO, 2003. J.M. Powers and R.L. Sakaguchi (Eds.), Craig's Restorative Dental Materials, 12" Edition, ElsevierMosby, St. Louis, MO, 2006. W.J. 07Brien(Ed.), Dental Materials and Their Selection, 3rdEdition, Quintessence, Chicago, 2002. M.K. Asgharnia and W.A. Brantley, Am. J. Orthod., 89 (1986) 228. H. Walia, W.A. Brantley and H. Gerstein. J. Endod., 14 (1988) 346. G.F. Andreasen and R.E. Morrow, Am. J. Orthod., 73 (1978) 142 W.A. Brantley in Orthodontic Materials: Scientific and Clinical Aspects, (Eds W.A. Brantley and T. Eliades), Thieme, Stuttgart, 2001, Chap. 4. C.M. Wayman and T.W. Duerig in Engineering Aspects of Shape Memory Alloys, (Eds T.W. Duerig, K.N. Melton, D. Stockel and C.M. Wayman), Butterworth-Heinemann, London, 1990, p. 3. K. Otsuka in Engineering Aspects of Shape Memory Alloys, (Eds T.W. Duerig, K.N. Melton, D. Stockel and C.M. Wayman), ButterworthHeinemann, London, 1990, p. 36. T. Todoroki and H. Tamura. Trans. Japan. Inst. Metals, 28 (1987), 83. S. Miyazaki and K. Otsuka, Iron and Steel Inst. Japan Int. 29 (1989) 353. G.E. Dieter, Mechanical Metallurgy, 3rdEdition, Mc-Graw-Hill, New York, 1986, Chap. 5 , 6 and 19. D. Goldstein, L. Kabacoff and J. Tydings. J. Metals, 39 (1987) 19.

K.N. Melton in Engineering Aspects of Shape Memory Alloys, (Eds T.W. Duerig, K.N. Melton, D. Stockel and C.M. Wayman), ButterworthHeinemann, London, 1990, p. 2 1. G. Kuhn, B. Tavernier and L. Jordan, J. Endod., 27 (2001) 516. M.G. Bahia, R.C. Martins, B.M. Gonzalez and V.T. Buono, Int. Endod. J., 38 (2005) 795. T.A. Thayer, M.D. Bagby, R.N. Moore and R.J. DeAngelis, Am. J. Orthod. Dentofacial Orthop., 107 (1995) 604. M. Iijima, W.A. Brantley, I. Kawashima, H. Ohno, W. Guo, Y. Yonekura and I. Mizoguchi, Biomaterials, 25 (2004) 171. B.D. Cullity, Elements of X-Ray Diffiaction, 2ndEdition, AddisonWesley, Reading, MA, USA, 1978, p. 292. S.A. Thompson. Int. Endod. J., 33 (2000) 297. F. Miura, M. Mogi, Y. Ohura and H. Hamanaka, Am. J. Orthod. Dentofacial Orthop., 90 (1986) 1. W.A. Brantley, T.A. Svec, M. Iijima, J.M. Powers and T.H. Grentzer, J. Endod., 28 (2002) 567. W.A. Brantley, T.A. Svec, M. Iijima, J.M. Powers and T.H. Grentzer, J. Endod., 28 (2002) 774. T. Yoneyama, H. Doi, H. Hamanaka, Y. Okamoto, M. Mogi and F. Miura, Dent. Mater. J., 11 (1992) 1. T.G. Bradley, W.A. Brantley and B.M. Culbertson, Am. J. Orthod. Dentofacial Orthop., 109 (1996) 589. S. Civjan, E.F. Huget, L.B. DeSimon, J. Dent. Res., 54 (1975) 89. S.B. Alapati, W.A. Brantley, J.M. Nusstein, G.S. Daehn, T.A. Svec, J.M. Powers, W.M. Johnston, W. Guo. J. Endod., 32 (2006) 1191. G.B. Alexandrou, K. Chrissafis, L.P. Vasiliadis, E. Pavlidou and E.K. Polychroniadis, J. Endod., 32 (2006) 675. G. Alexandrou, K. Chrissafis, L. Vasiliadis, E. Pavlidou and E.K. Polychroniadis, Int. Endod. J., 39 (2006) 770. K. Miyai, A. Ebihara, Y. Hayashi, H. Doi, H. Suda and T. Yoneyama, Int. Endod. J., 39 (2006). G. Kuhn and L. Jordan, J. Endod., 28 (2002) 716. S.B. Alapati, Ph.D. Dissertation, The Ohio State University, Columbus, OH, USA, 2006. S.E. Khier, W.A. Brantley and R.A. Fournelle, Am. J. Orthod. Dentofacial Orthop., 99 (1991) 3 10. M.L. Fletcher, S. Miyake, W.A. Brantley and B.M. Culbertson, J. Dent. Res., 71, Special Issue, (1992) Abstract No. 505. W.A. Brantley, M. Iijima and T.H. Grentzer, NATAS Notes, 34 (1) (2002) 5.

W.A. Brantley, M. Iijima and T.H. Grentzer, Thermochim. Acta, 392-393 (2002) 329. W.A. Brantley, M. Iijima and T.H. Grentzer, Am. J. Orthod. Dentofacial Orthop., 124 (2003) 387. R. Chen, Y.F. Zhi, M.G. Arvystas, Angle Orthod., 62 (1992) 59. W.A. Brantley, W.H. Guo, W.A.T. Clark and M. Iijima. J. Dent. Res., 82, Special Issue, (2003) Abstract No. 1535. S.L. Rosen, Fundamental Principles of Polymeric Materials, 2ndEdition, Wiley-Interscience, New York, 1993, Chap. 8 and 18. J.M. Vickery, M.W. Paulus, W.A. Brantley, B.M. Culbertson and W.M. Johnston, Int. J. Prosthodont., 9 (1995) 221. S.J. Clarson and J.A. Semlyen, Siloxane Polymers, Prentice-Hall, Englewood Cliffs, NJ, USA, 1993, p. 2 16. S. Sauerbrunn and P. Gill. Amer. Lab., 25 (9) (1993) 54. M. Pamenius and N.G. Ohlson, Dent. Mater., 8 (1992) 140. J.M. Vickery, W.A. Brantley and T.A. Bardin, Thermochim. Acta, 367-368 (2001) 177. J.M. Vickery, W.A. Brantley, B.J. Weddle and P.K. Gallagher, Proc. 26th NATAS Conference, September 1998, p. 527. J.M. Vickery, J.C. Mitchell and W.A. Brantley, J. Dent. Res., 78 (1999) 550 (Abstract No. 3559). T. Eliades, G. Eliades, D.C. Watts and W.A. Brantley in Orthodontic Materials: Scientific and Clinical Aspects, (Eds W.A. Brantley and T. Eliades), Thieme, Stuttgart, 2001, Chap. 8. M.R. Renick, W.A. Brantley, F.M. Beck, K.W.L. Vig and C.S. Webb. Am. J. Orthod., 126 (2004) 337. N.M.K. Lamba, K.A. Woodhouse and S.L. Cooper, Polyurethanes in Biomedical Applications, CRC Press, Boca Raton, FL, USA, p. 5 and 68. D.L. Baty, J.E. Volz and J.A. von Fraunhofer, Am. J. Orthod. Dentofacial Orthop., 106 (1994) 40. T. Eliades, G. Eliades and D.C. Watts, Eur. J. Orthod., 21 (1999) 649. A. Maffezzoli, A. Della Pietra, S. Rengo, L. Nicolais and G. Valletta, Biomaterials, 15 (1994) 1221. T. Hayakawa, K. Takahashi, K. Kikutake, I. Yokota and K. Nemoto, J. Oral Sci., 41 (1999) 9. Y. Nomura, W. Teshima, N. Tanaka, Y. Yoshida, Y. Nahara and M. Okazaki, J. Biomed. Mater. Res., 63 (2002) 209. S.G. Pereira, T.G. Nunes and S. Kalachandra, Biomaterials, 23 (2002) 3799. N. Emami and K.J. Soderholm, Mater. Sci. Mater. Med., 16 (2005) 47. T.W. Wilson and D.T. Turner, J. Dent. Res., 66 (1987) 1032.

L.G. Lovell, H. Lu, J.E. Elliott, J.W. Stansbury and C.N. Bowman, Dent. Mater., 17 (2001) 504. J. Nie and C.N Bowman, Biomaterials, 23 (2002) 1221. D. Truffier-Boutry, S. Demoustier-Champagne, J. Devaux, J.J. Biebuyck, M. Mestdagh, P. Larbanois and G. Leloup, Dent. Mater., 22 (2006) 405. R.E. Kerby, A. Tiba, L.A. Knobloch, S.R. Schricker and 0 . Tiba, J. Oral Rehabil., 30 (2003) 780. I. Sideridou, D.S. Achilias, E. Kyrikou, Biomaterials, 25 (2004) 3087. J.M. Powers and R.G. Craig, J. Dent. Res., 46 (1967) 1090. J.M. Powers and R.G. Craig, J. Dent. Res., 57 (1978) 37. E. Kotsiomiti and J.F. McCabe, J. Oral Rehabil., 23 (1996) 114. E. Kotsiomiti and J.F. McCabe, J. Oral Rehabil., 24 (1997) 5 17. R.D. Phoenix, M.A. Mansueto, N.A. Ackerman and R.E. Jones. J. Prosthodont., 13 (2004) 17. G.R. Parr and F.A. Rueggeberg, J. Prosthodont., 8 (1999) 92.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors 2008 Elsevier B.V.

Chapter 18

MEDICAL APPLICATIONS OF THERMAL METHODS Beverley D. Glass School of Pharmacy and Molecular Sciences, James Cook University, Townsville, QLD, 48 15 Australia.

1. INTRODUCTION The medical applications of thermal methods have evolved and increased substantially in recent years. This may be attributed to the improved sensitivity and usability of the instrumentation allowing for detailed investigations of thermal properties and predictions of thermal stability, whether they are of the skin, drug penetration and delivery, implants or prosthetics. Differential scanning calorimetry (DSC) is a sensitive method which has enabled changes in biological systems to be examined providing medical researchers with distinct opportunities. Other thermoanalytical techniques such as thermogravimetry (TG) and isothermal titration calorimetry (ITC) are useful in the evaluation of thermal stability and to measure the heat generated in physical and chemical reactions. Thermoanalytical techniques have had a major application in the understanding of transitions in the skin and of drug penetration of the skin. The application of thermoanalytical techniques to prosthetics and implants is also discussed, as are recent DSC investigations of the oesophagus that provided the information on thermal stability required for successful stent implantation. The use of thermoanalytical techniques such as modulated temperature differential scanning calorimetry (MTDSC) has been be used to characterise polymeric material in order to determine whether there are interactions with drug substances to control and predict drug delivery. The ability of thermoanalytical techniques to provide information on the effects of alteration of biological systems is not only important in medical research but also in advanced patient-care.

2. APPLICATION TO PENETRATION OF DRUGS INTO THE SKIN

2.1. Introduction Because of its excellent barrier properties, the skin has until recently provided an insignificant entry point for the systemic absorption of drugs. Systemic absorption of a drug implies that it reaches the circulation and diffuses into all tissues, including the site of therapeutic action. Much research has been undertaken to improve the permeability of the skin to drugs (for transdermal administration either in patches or various semi-solid bases such as creams, ointments and gels ) for the following reasons: drugs administered by this route are not destroyed by the acidic environment of the stomach, avoid the 'firstpass' metabolism, aid patient compliance, maintain effective drug levels in the blood, are easily identifiable, may be removed immediately and result in less gastrointestinal (GI) side effects for those acidic drugs such as the antiinflammatories which are prone to cause GI side effects by primary insult. Drugs may be applied topically for both their local and systemic effects. However until recently most topical applications were designed to deliver drugs with the skin being the target organ, which is not the case for transdermals where the skin is not the target organ and systemic absorption is required. The ability of a drug to penetrate and cause systemic effects is determined not only by the drug itself (penetrant) and its formulation (including penetration enhancers) but also their effect on skin permeability [I]. Figure 1 is a schematic diagram of the structure of human skin, showing the SC.

Epidermis

{

*e:!px\~~cOrneum

Sebaceous gland

Dermis Hair follicle Sweat gland Subcutaneous and fatty tissue

Blood vessels and capillaries

Figure 1. The structure of human skin. (Adapted from [I].)

Because the activity of the drug in various formulations, especially those in semi-solid bases such as creams, ointments and gels, can be improved by penetration enhancers, it is important to understand thermally induced changes in the skin, in order to understand its molecular structure and the mechanism of action of penetration enhancers. DSC has been used to investigate thermal transitions in the SC, including thermotropic lipid transitions in the SC [2]. Drugs which are suitable for administration via the transdermal route include those of low molecular weight (400-800). Other factors, such as increasing the temperature and pH, decreasing skin thickness and increasing hydration by using occlusive bases such as ointments and the use of patches, can facilitate penetration of drugs into the skin. Transdermal delivery of drugs for systemic absorption is often more convenient than the oral route, since drugs avoid the the liver metabolism thus requiring lower doses and less potential for side effects. Thermoanalytical measurements of 'the state of hydration and thermal transitions in the skin can thus provide us with insight into the mechanisms of drug penetration [2]. 2.2. Thermoanalytical techniques and the skin Thermoanalytical experiments have been conducted, not only on intact skin but also on fractionated human stratum corneum and material that has been extracted from the skin using lipids. DSC, sensitive to the temperature dependence of the skin's heat capacity, shows transitions between 30-120 "C due to lipids and proteins in the human SC. Such studies can provide understanding of the changes caused by heat on the structure of the skin and also insight into its functioning. Lipid transitions in the skin have been studied by Guia et al. [2] using DSC experiments on fractionated skin samples and solvent extracts. Three major transitions were observed in all samples at 65, 80 and 95 "C as shown in Figure 2, while a small peak (not observed in all samples) is seen at 35 "C. When these samples were cooled and reheated, the transitions at 35 and 65 "C remained unchanged and the peak at 95 "C disappeared, while the peak at 80 "C was decreased in size and shifted to a lower temperature. To explain these observations, fractionated samples and samples extracted using solvents were studied. On extraction, all transitions below 90°C disappeared, while the transition at 95 "C remained in the extracted sample. On concentration of the extracts, a peak remained at 65 "C [2]. When horny cell membranes are prepared from the stratum corneum, the thermal profile shows 2 peaks at 65 and 75 "C (as shown in Figure 3), while a reheat of this sample shows a peak at 65 "C and a minor peak at 70 "C.

Comparing Figures 2 and 3, shows that some similarity exists in the profiles between 60 and 80 "C for the two samples. These results are confirmed by IR data, which show two areas of rapid thermal change from 55-65 "C and 70-85 "C respectively, corresponding to the thermal transitions seen in Figure 2. It can thus be concluded from this data that the transitions at 65 and 85 "C resulted due to acyl chain motion of the lipids in the SC.

I

l

I

I

25

I

I

I

45

l

I

I

65

l

1

1

85

1

105

Temperature ("C)

Figure 2. Typical DSC trace of the human SC. (Adapted from [2]).

L

I

25

I

I

45

I

I

I

I

I

65

I

85

I

(

I

I

105

Temperature ("C)

Figure 3. DSC trace of the horny cell membrane. (Adapted from [2]).

Since the hydration of the SC facilitates percutaneous absorption, it is an important factor to consider. The effect of hydration on the thermal profiles may be investigated by examining changes in the thermal transition midpoint (T,) and its thermal sharpness measured from the ratio of the DSC peak height to its peak width. Results show that T, decreases with increased hydration for those transitions near 65 and 80 "C, with a corresponding increase in thermal sharpness for the DSC peaks. DSC studies also allow for the detection of structural transitions which are thermally induced. The transition occurring at 95 OC (highest temperature transition) occurs in the intact SC and in the lipid extracted SC, but not in the lipid extract. The lack of this transition in the reheated samples or the profiles of the horny layer (which lacks intracellular keratin), suggests that this is a transition which involves keratin. The transition at 35 "C, although small and variable, is thermally reversible, confirming that it is lipid in nature, however its variability indicates that the lipid is loosely associated with the SC, possibly from the sebaceous glands. The two major transitions at 65 and 80 "C have been proven by DSC experiments to be related to lipids, because extraction of the stratum corneum lipids with a mixture of organic solvents resulted in the disappearance of these peaks from the DSC curves. A decrease in T, and an increase in thermal sharpness, indicative of increasing hydration and of a water-lipid system, is further evidence of the lipid origin of these transitions. However the fact that the lipid extract from the stratum corneum contains only the transition at 65 "C, which is unchanged on heating, suggests that this transition is due only to lipid, and thus represents the extended lamellae area in the intercellular space. Because the transition at 80°C, only in the intact SC, is also seen in the horny cell membrane (keratin-free), it may be assumed that this transition is due to a protein-lipid complex in the horny cell membrane because of the changes associated with T, on reheating. Because hydration plays a role in the SC lipid structure and in drug penetration into the SC, DSC experiments can be valuable in assessing the water content and thus allowing the prediction of drug penetration into the SC. The role of water may be explained as follows: water becomes inserted amongst the polar heads and loosens the packing, thus lowering the attraction of the hydrocarbon chains for each other and decreasing T, [2]. Since the skin presents a barrier to the penetration of drugs, penetration enhancers may be used which have the capacity to alter the barrier properties of the skin. Although most work has been done on the skin transitions at temperature ranges above 0 "C, some penetration enhancers, such as propylene

glycol have low melting points( propylene glycol at -100 i 1 OC), and therefore exhibit sub-zero transitions. Thermal analyses at lower temperatures, ranging from -130 to 120 "C, may indeed be needed to understand the penetration enhancing effects of these substances [3]. Endothermic transitions between -20 and 0 OC for the SC are related to bound water. After isolation and dehydration, a peak was isolated at -9 OC which was thought to represent water, proteins or lipids. To ascertain whether the peak was due to water, further hydration and dehydration experiments were undertaken. On heating, the peak was still evident, while hydration resulted in the appearance of a new peak at -18 "C, whilst maintaining the existing peak. Based on these results, the possibility that this transition is due to water was refuted. When heating to 120 "C caused the peak at -9 OC to persist, it was confirmed that this peak was not due to protein, because proteins denature on heating with a resulting transition at 100 OC. By a process of elimination this endothermic transition was attributed to lipids. To verify this, the authors [3] undertook lipid extraction and pre-treatment of the SC with propylene glycol and a mixture of propylene glycol and oleic acid. After lipid extraction, the transitions at 40, 70 and -9 "C disappeared, providing evidence of the lipid origins of this peak. When skin is pre-treated with propylene glycol, disappearance of the transitions at 100 OC was attributed to the ability of propylene glycol to extract water from protein. While hydration resulted in the maintenance of this peak, dehydration resulted in its removal and this was explained in terms of the lack of affinity of propylene glycol for lipids and the resulting depression of their phase transition temperatures. Pre-treatment of the SC with oleic acid and propylene glycol resulted in the appearance of two new transitions at -5 and -12 "C, with no evidence of the peak at -9 OC. Because the profiles appeared similar for both the hydrated and dehydrated samples, this peak cannot be related to water. The peak at -12 OC, was attributed to the formation of an eutectic mixture between the oleic acid and the stratum corneum and the peak at -5 "C was due to excess oleic acid. The peak at -9 OC is thus assigned to lipids and because of its occurrence at below 0 "C is referred to as the sub-zero lipid peak [3]. 2.3. Thermoanalytical techniques and drug penetration into the skin The ability of a drug to penetrate the skin and exert a therapeutic effect depends initially on its ability to difhse from the formulation to the skin surface, and then the drug must be able to penetrate the skin [I]. This is a passive process and is rate limiting. While the formulation may be altered to facilitate the difhsion of the drug to the skin surface, skin penetration may also be accelerated by the addition of a penetration enhancer.

There is considerable interest in penetration enhancers, which may exert their action by either chemical or physical means. Chemical enhancers may increase penetration of the SC by altering its physicochemical properties to facilitate the diffusion of the drug through the skin [4]. Penetration enhancers may thus either increase drug solubility by altering the partitioning of the drug between the SC and the formulation, or by enhancing the fluidity of the lipid bilayers. Thermal analysis has been used to evaluate the mode of action of penetration enhancers Substances that have proved to be effective penetration enhancers include: acetone, azone, dimethyl acetamide, dimethyl formamide, dimethyl sulfoxide, ethanol, oleic acid, polyethylene glycol, and sodium lauryl sulphate [5, 61. In choosing a penetration enhancer, penetration however is not the only criterion to consider. There has been much controversy due to their absorption into the skin and resulting toxicity and skin irritation [7]. Aioi et al. [7] have reported on the use of vitamin E and squalene to alleviate the skin irritation caused by these enhancers. Both vitamin E (30%) and squalene (10%) also caused improved penetration with a 1.5 and 2.5 fold increase in penetration observed. Note must also be taken of the physicochemical and biological compatibility of these substances with the systems used [S]. Differential thermal analysis (DTA) was used in a study of the effects of fatty acid (linoleic acid shown in Figure 4) penetration enhancers on permeation of piroxicam (1% in a poloxamer 407 gel) through rat skins (shown in Figure 5). The abilities of the fatty acids to disorder the lipid bilayers were correlated with their penetration enhancing effects. Endotherms occurred at 65, 75 and 105 "C due to transitions in the human and porcine SC. In this study, a broad endotherm at 57.5 "C was observed in the SC that had not been treated with the enhancer. Linoleic acid showed the greatest enhancing effect in this study with an enhancement factor (EF) of 1.76 at 57.5 "C. A broad endotherm at 44.6 "C occurred in the skin treated with linoleic acid, with decreased sharpness compared to the endotherm for the intact SC. The thermal profiles of the skin treated with these fatty acids showed that their incorporation disordered the lipids in the skin. Comparison of these results with those obtained from histological examination of the skin showed the treated skin with poloxamer 407 gels to be loosely layered with wide intercellular spaces compared to the control [9]. These results correlated with previous findings, which suggest that this transition is due only to lipid and thus represents, the extended lamellae area in the intercellular space [2].

Figure 4. Molecular structure of linoleic acid.

Figure 5. Molecular structure of piroxicam. Laurocapram (or azone, shown in Figure 6) is reported to be an effective penetration enhancer for both hydrophilic and lipophilic drugs because it solubilizes the intercellular lipids and / or disrupts the lipid lamellae [lo]. Monitoring of the heat generated, using isothermal calorimetry (ITC), allowed the reaction parameters to be determined. The lipid composition of the skin, which differs from that of other biological membranes, is comprised of ceramides and cholesterol (shown in Figure 6) and free fatty acids in the ratio of 50:25:10. ITC was used to investigate the interactions and binding of azone and these lipids to determine the way in which azone works to enhance penetration of drugs into the skin. Binding between azone and cerarnide is enthalpy-driven, whereas with cholesterol it is entropydriven, and there is no evidence of binding with the free fatty acids. These results suggest that azone extracts cholesterol from the skin, binds to the ceramides and has no interaction with the free fatty acids in the skin. Their molecular structures are shown in Figure 6. This binding between cholesterol and azone results in a more disordered state, which suggests that the cholesterol molecules are extracted from the lamellae of the SC. The addition of azone to propylene glycol increases the solubility of cholesterol. Thus, the complex formed between azone and cholesterol dissolves into propylene glycol more easily than that of ceramide-3 and azone, with this dissolution attributed to inter- and intra-molecular hydrogen bonding. While there has been some controversy about the use of chemical penetration enhancers, this information as to their mechanism of action is positive due to the fact that only part of the SC lipids is disturbed and the corneocytes are unaffected. This allows the skin to continue to fulfil its very important barrier

function. Some physical enhancers, on the other hand, may be more harmful due to their ability to facilitate multiple passages through the skin.

Azone

Cholesterol Figure 6. Molecular structures of azone, ceramide-3 and cholesterol. Tenjarla et al. [ l 11 used DSC to investigate a series of N-acetylproline esters as penetration enhancers with benazepril and hydrocortisone as model drugs applied to the skin in propylene glycol, using azone as a control. The compounds (shown in Figure 7) studied included the 11-carbon ester (UNAP) and the 12-carbon ester (DDNAP), which possess hydrophilic and lipophilic properties similar to that of azone. Results indicated that pre-treatment with UNAP decreased the lower lipid transition from 48.8 to 47.8 "C, while eliminating the higher transition at 71.7 "C. However, on treatment with DDNAP, both transitions disappeared, allowing the conclusion that DDNAP and azone work by a similar mechanism of action. For UNAP, elimination of only one lipid transition results in this enhancer having less of an effect on the SC and obviously less potential for adverse effects due to the penetration enhancer.

O C 0 0 R

I

R = -CllH23 (UNAP)

R = -C 12H25(DDNAP)

Figure 7. Molecular structures of the carbon ester UNAP and DDNAP of Nacetylproline The transdermal penetration of ibuprofen (shown in Figure 8) has been investigated with the obvious advantage for this drug of avoiding the GI side effects due to the primary insult of the acidic hnctionalities on the molecule. Saturated solutions of ibuprofen in various concentrations of disodium hydrogen phosphate (DHP) were prepared and a permeability study was conducted. An increase in drug flux across the skin was noted and was attributed to an interaction between the drug and the skin rather than an increased concentration of the drug. Surface tension measurements showed a decrease in the surface tension of the DHP which was attributed to the ability of ibuprofen to act as an ionic surfactant.

Figure 8. Molecular structure of ibuprofen. DSC studies supported this result showing lipid transitions in the untreated rat SC at 41.9, 55.1, 70.2, and 77.5 "C, while SC pre-treated with disodium hydrogen phosphate and ibuprofen showed only the last lipid transition. This confirms that ibuprofen disrupts the SC and, through its ability to act as an anionic surfactant, gives rise to self permeation enhancement [12]. Eutectic mixtures of ibuprofen and terpene penetration enhancers have also been studied. The eutectic mixture results in superior penetration compared to that of an aqueous solution when the skin was pre-treated with terpenes. In addition to these results with ibuprofen [13], screening by thermal analysis has shown a number of other drug classes, which because of the formation of eutectic mixtures or solid solutions and resulting melting point depression, may be suitable for enhanced drug delivery. These drugs include ACE inhibitors, Pblockers and other antiinflammatory drugs [14, 151.

Further work has been undertaken to investigate the effect of terpenes as penetration enhancers and DSC studies have provided information on their thermotropic behaviour. Three endothermic transitions were found in the SC, at 62, 79 and 95 "C. Pre-treatment with terpineol decreased the first transition, attributed to bilayer disruption, while all the terpenes decreased the second transition, which can be explained in terms of fluidization or extraction of lipids 1161. Iontophoresis is a technique which enhances the transport of drug molecules through the skin by creating a potential gradient with an applied electrical current [I], with a typical device consisting of a battery, microprocessor, drug reservoir and electrodes. This technique has obvious application in the increased migration of ionic drugs and small iontophoretic patches are becoming accepted by the pharmaceutical industry. DSC studies have also been undertaken to investigate the effect of iontophoretic transport of the P-blocker, propranolol hydrochloride on the lipid bilayers of the stratum corneum, pre-treated by two penetration enhancers, sodium lauryl sulfate (NaLS) and hexadecyl trimethylammonium bromide (CTAB). DSC results indicated that NaLS was included into the intercellular lipids during iontophoresis and also showed a slight decrease in the two higher transitions in the SC. Lowering of these lipid endotherms is an indication that the bilayer has increased in fluidity. In the case of CTAB, DSC revealed an increase in the initial skin transition, attributed to the insertion of CTAB into the lipids responsible for this transition. However, pre-treatment of the SC with NaLS proved to be successful in enhancing the flux of propranolol hydrochloride across the SC, while this was not the case for CTAB [17]. DSC studies undertaken to investigate the effect of propylene glycol (PG), glycerol and isopropyl myristate (IPM), and combinations thereof on the SC, showed a decrease in transition temperatures of the lipid fraction. The combination of PGIIPM, affected the SC microstructure, although the decreases in lipid transition temperatures were not as much as expected from the results for individual PG and IPM [ 181. Nicardipine, used in the treatment of hypertension, has proved to be an excellent candidate for transdermal delivery as its bioavailability is only 20-33% due the 'first pass' effect in the liver and the short half life requires frequent dosing resulting in poor patient concordance. The effect of menthol on the transdermal delivery of nicardipine hydrochloride from a hydroxypropyl cellulose gel formulation was investigated. The ideal penetration enhancer should be pharmacologically inactive, non-irritant and non-damaging to the skin, potent and cosmetically acceptable. DSC results showed increased penetration of nicardipine hydrochloride due to the ability of menthol to partially extract the lipids from the stratum corneum [19]. Extraction of lipids

leads to enhanced percutaneous absorption, with this enhanced absorption due to the increased solute diffusivity in the delipidized SC. DSC results in Figure 9 show transitions of lipids in the SC at 58.6, 70.6 and 86.2 "C, which has been treated both with water and the vehicle, ethanol. Broadening of the peaks is observed on treatment with menthol, with fbrther broadening associated with increasing concentrations of menthol, illustrating a pronounced effect on extraction of the lipids in the SC and the potential for improved drug penetration. The advantage of the use of menthol in this case as a penetration enhancer is that it is relatively safe, with these results confirmed by FTIR studies.

Water

70% vlv Ethanol

1% wlw Menthol

w I

I

30

1

1

1

1

53.8

l

1

72.5

1

I

,

91.3

12% W/W Menthol

,

,

)

,

110

Temperature ("C)

Figure 9. DSC curves of the SC treated with water, ethanol and menthol. (Adapted from [19]). Many other papers have described the use of DSC in the predication of penetration enhancement using the following penetration enhancers: oleic acidlpropylene glycol [20], straight chain fatty acids, monosaturated and polyunsaturated fatty acids [21], carvone [22], menthol [23], 5-aminolevulinic acid [24], and phospholipids [25].

3. APPLICATION TO DRUG DELIVERY 3.1. Introduction Pectins are soluble polysaccharides used not only as thickening agents but also for drug delivery. Polymeric materials have been extensively studied for the purpose of controlled release and delivery of drug substances, whether through the skin or to the gastrointestinal tract. Thermal analysis can be used to characterise the polymeric material and therefore to determine whether there are interactions with drug substances in order to control and predict drug delivery. 3.2 Thermoanalytical techniques used in drug delivery Pellets of calcium-alginate and pectinate and a combination of calciumalginate-pectinate were used to study the release of drugs in the gastrointestinal tract. The value of these pectins as drug carriers may be attributed to both their gelling and thickening properties. Modulated temperature differential scanning calorimetry (MTDSC) allowed changes in mobility to be detected and provided information on the effects of processing and any structural changes over a range of temperatures. For the calcium alginate pellets, a shallow transition at 50-100 "C was evident, followed by an endothermic event at 150-175 "C and a relaxation at 260 "C. On reversal of the heat flow signal, however, no glass transition was observed. The endothermic region was attributed to a depolymerisation process. For calciumpectinate, the endothermic event, and thus depolymerisation, occurs at the lower temperature of 125 "C, while the relaxation endotherm occurs over the range 125-168 OC, rather than as a distinct peak as was found for the alginate samples. This shows a highly complex structure. The results for the calcium-alginatepectinate samples were similar to those for calcium alginate. Thus MTDSC could be used to define the nature of the crosslinking by providing information on the pellet matrix. Selection of the components of a formulation [26] can be used to alter the drug release profiles. DSC may also be used to determine the properties of lipid delivery systems, to facilitate penetration into the skin. Ethosomes are composed of phospholipids, ethanol and water to facilitate transdermal delivery because ethanol is a well known penetration enhancer. DSC curves comparing ethosomes and liposomes without ethanol showed transitions at -15.2 "C and 6.3 "C, respectively. This suggests that the ethosomes could be in a more fluid state. Further investigation showed that the increased fluidity may be attributed to the particle size and the fact that the phospholipid is solubilized in the ethosomal system. Even small amounts of ethanol are capable of disrupting the lipid bilayers and increasing fluidity

These ethosomal systems showed improved performance both in terms of drug concentration in the skin and flux or penetration through the skin. Although ethanol is a proven penetration enhancer, there appears to be some synergy between the enhancing performance of ethanol, vesicles and skin lipids. Phospholipids may also interact with the SC lipids of the intercellular layers enhancing their permeability. Thus the activity may be explained in terms of the ethanol disturbing the lipid bilayer, the ethosomes penetrating these disturbed layers, or fusing with the skin lipids and subsequent drug release [27]. Biodegradable polymers are of use in therapy, because of their ability to control the release of drugs into the body. Many factors affect the degradation kinetics of these systems, including the influence of thermal processes. The effects of manufacture, whether by extrusion or injection moulding, on the degradation kinetics of aliphatic polyesters based on lactic acid (PLA), glycolic acid (PGA) and their copolymers (PGLA) have been investigated [28]. Different manufacturing processes may result in different molecular weights, crystallinities and thus different degradation rates. The polymer investigated was poly (L-lactic acid) (PLLA) and the drug used a somatostatin analogue vapreotide (shown in Figure 10) in the form of the pamoate derivative, with the goal of prolonging the release of the drug. Crystallinity was determined using DSC, and PLLA, which is a semicrystalline polymer, was shown to display crystalline areas as well as disordered amorphous regions. The degree of crystallinity can be altered as a result of manufacturing (of implants by extrusion or injection moulding), because of the conditions during the heating and cooling processes. DSC curves of both kinds of implants and of pure PLLA showed endotherms at 130-140 "C, with no difference in the area under the curve, showing that there was little difference in the degree of crystallinity. Results also showed that the PLLA before manufacture had crystalline regions and that this low molecular weight drug can be incorporated into the PLLA without any effect on the crystallinity. The limitation of this study is that DSC cannot be used to detect minor changes in crystallinity, which might have a significant effect on the degradation kinetics of the drug and thus the release profile.

Figure 10. Structure of the somatostatin analogue, vapreotide.

4. APPLICATION TO IMPLANTS

4.1. Introduction Thermoanalytical methods have application to implants for evaluating the mechanical properties prior to and after insertion into body cavities, and for determining the effects of these devices on biological systems. 4.2. Thermoanalytical techniques used in implants Thermal analyses have been used for over 20 years to characterise heart-valve poppets (silicone elastomer) implanted into humans [29] and to evaluate the in vivo absorption of lipids and related compounds from the body. Important deleterious effects on the implants, observed under in vivo conditions include changes in the hardness and mechanical properties, which appear to be associated with colour changes from white to yellow then brown. Because the thermal curves of the samples extracted fi-om new and unused implants were almost identical, the differences were attributed to materials absorbed into the polymer matrix. Thermal analyses may therefore be usefbl in understanding the behaviour of these poppets in vivo. Table 1 shows the thermoanalytical data for new and used poppets, with those used poppets differing in terms of the extent extracted materials in vivo. Table 1. Comparative data for heart poppets (before and after implantation)

Mass loss Apex

260-4 10 OC 390 "C

Poppet < 5% Extractables 150-160 "C 390 "C

Maximum exotherm Degradation complete

360 "C

360 "C

600 "C

600 "C

New poppet

Poppet >5% Extractables 150-160 "C 440-450 "C, 500-520 "C

600 "C

The silicone-rubber heart-valve poppets contain unreacted polymer which appears to be absent after the poppet has been exposed to body fluids. When the question was raised as to whether the new poppets contain a material subject to oxidation at a lower temperature than for the bulk of the poppet, nitrogen studies showed decomposition at the same temperature but to a lesser extent. For the new poppets in air a mass loss of about 9% was observed, while for those samples showing an apex at 390 "C, the percent of absorbed material was

usually between 3-7 %, and for those not showing this apex a relationship established allowed the estimation of the amount of absorbed materials. Gas chromatography and thin layer chromatography analyses of the extracts from the used poppets indicated the presence of CI1fatty acids and cholesterol. Comparison of their thermal profiles with that of the used poppets confirmed their presence as extractables. This study [30] provides important information on silicone heart valve poppets including what is absorbed and what may be lost on implantation. Although self-expandable metal stents have been used in the palliative care of oesophageal strictures, they have not been used for oesophageal malignancy in an emergency situation of varix bleeding. Benko et al. [30] investigated two stents (1- made of nickel-titanium and a polyurethane inner coating ; 2- made of corrosion resistant stainless steel with good radiopacity).

I

I

45

1

50

1

1

55

l

1

60

1

1

65

1

1

70

1

1

1

75

Temperature ("C)

Figure 11. Thermal profiles of stents 1 and 2 (oesophagus muscle). (Adapted from [30]). The surgical interventions resulted in both changes in the temperature-course DSC scans as well as changes in their thermal parameters. Although for stent 1 the results are closer to the healthy sample, greater calorimetric enthalpy and higher melting points were observed as compared to the control and stent 2 as shown in Figure 11. For stent 2, the calorimetric enthalpy is smaller than that of the control and stent 1 (Table 2), where the total sample refers to the oesophageal mucosa and muscle.

Table 2. Thermal parameters for the denatured oesophagus (mucosa and muscle) [30]. Stent 1

Stent 2

A W J ~ - 'TmI0C A W J ~ - ' TmI0C Total sample

2.0

64.1

1.34

64

Control AH/J~-' TmPC 1.6 62.9

As illustrated in Table 2, the DSC results show that the stent 1 has improved the thermal stability of the oesophagus and has assisted in its regeneration. This may be an effective intervention for oesophageal bleeding patients suffering from cirrhosis until endoscopic band ligation can be accomplished. Stent treatment has thus improved the thermal stability of the oesophagus. There has been increasing interest in polymeric stents, with those which are nondegradable gaining popularity over the metal stent due to the improved compatibility in vivo and the potential for drug loading. Bioresorbable stents also are only present in the body during the healing process and therefore do not require a second operation for removal. A common material for these stents is poly(L-lactic acid) (PLLA), and this is employed when high mechanical strength and toughness is required. It is thus important that the fibre and mechanical properties of these devices be evaluated, while also including a coating to incorporate drugs, which then undergo controlled release. DTA was used [31] to determine the melting temperature (T,), the heat of fusion (AH, ) and the degree of crystallinity (%C). PLLA with draw ratios of 3: 1 to 8: 1 exhibits good tensile properties. Crystallinity affects the mechanical properties and DSC was used [3 11 to consider the effect of the draw ratio on the crystallinity and, ultimately, the tensile strength of the fibres. Table 3. Differing crystallinity of fibres dependent on drawing ratio [3 11 Drawing ratio Undrawn fibre Drawing ratio 4: 1 Drawing ratio 8: 1

Crystallinity Low Increased Increased

%C 18 60 70

All fibres exhibited a major melting endotherm at 175-177 "C, attributed to the crystals created during the drawing process, and a minor one at 160 OC due to fast cooling, which decreases in size as the drawing ratio increases. Because fibre drawn at 8:l underwent a slow processing, no minor melting endotherm was observed. Generally an increase in the drawing ratio increases the tensile strength and modulus and decreases the ultimate strain. Crystallinity affects the mechanical properties of fibres at below 4:1, while at the higher ratios other

factors also become important. These PLLA stents proved to be applicable for supporting blood vessels for at least 20 weeks and, when microspheres were bound to these stents, protein loading was effective DSC has also been used to examine the thermal and thermo-oxidative properties of high molecular weight polyethylene (UHMW-PE) [32] used in components of an orthopaedic implant. These implants not only undergo changes during processing, but also during clinical use post-implantation, which may be different from those of the storage conditions. The changes likely during implantation may be due to temperature changes and oxidising compounds are present as a result of the inflammation in the patient. Changes in this material could be reflected in its thermal properties, melting, crystallinity and oxidation. DSC undertaken in an oxygen containing atmosphere thus may provide a topographic map of the acetabular cup and also makes it possible to compare melting, oxidation and the degree of crystallinity. These implanted cups were compared with the properties of controls. All control samples showed an endotherm at 136 OC, crystallinity varying from 5-7%, and two broad exotherms from 220-230 "C (lower oxidation temperature) and 245265 OC (higher oxidation temperature). However the implanted cups showed two exotherms, that at the higher temperature occurred in all samples, while that at the lower temperature occurred depending on the location of the sample. Both the heat of fusion and degree of crystallinity were higher for the retrieved cups. DSC results thus showed the heterogeneity of the polymer in the retrieved cups by the presence of double endo- and exothermic peaks. In addition the retrieved cups showed two melting peaks: the endotherm at the higher temperature was due to manufacture and the endotherm peak at the lower temperature was attributed to a change in texture of the UHMW-PE mass from the shear and compression forces in vivo, experienced by the person in conducting their normal life of walking etc. The fact that these two peaks coexist suggests that changes are ongoing in the implants Although coronary stents have been a breakthrough in the treatment of obstructive coronary disease, there have been adverse effects in patients, such as restenosis. Much research has gone into addressing this situation. Researchers have looked at coating stents for localized drug delivery and results have shown that delivery of a low dose of paclitaxel (PTx shown in Figure 12) has proved to be cytostatic but not cytotoxic [33]. The design of these stent coatings presents a challenge to the pharmaceutical scientist. The structural integrity must be preserved, while they provide the correct kinetics of dosing. The polymeric coatings need, in addition, to satisfy the following biological and physicochemical criteria:

i)

ii) iii) iv)

The need to withstand manufacturing and processing and the ability to withstand being deployed into the tissue without being physically compromised. The ability to provide modulated drug releases over a range of doses required in clinical practice. The coated stent should not produce any adverse biological response compared to the uncoated stent. The ability to meet ICH Guidelines and to be acceptable to the regulatory authorities.

Ranade et al. [33] describe the paclitaxel-eluting coronary stent, including the methodology to characterise drug delivery from this stent. The polymer chosen for this stent was poly (styrene-P-isobutylene- P-styrene) (SIBS shown in Figure 13) and showed good mechanical properties and stent performance.

Figure 12. Molecular structure of paclitaxel.

Figure 13. Chemical structure of poly(styrene-P-isobutylene- P-styrene) - SIBS

DSC has been used [33] in the characterisation of the polymer, providing information on thermal properties such as the glass transition, crystallization and melting of the polymer. The DSC profile shows the appearance of two glass transitions, one at -63 to -71 "C, while the other occurs at 90-103 "C. There is a slight increase in the latter glass transition on inclusion of the paclitaxel into the polymer attributed to interactions between the aromatic electrons. The release of paclitaxel from the polymer after an initial burst follows a longer sustained release pattern. The polymer-drug system is a heterogeneous matrix. The presence of particulate matter in the SIBS suggests limited solubility of the drug in the polymer. DSC investigations, used to confirm any interactions between the drug and the polymer, suggest that there is no drug dissolved in the polymer as evidenced by a lack of any effect on the polymer glass-transition. Further research has continued into the development of drug eluting coronary stents for the treatment of cardiovascular restenosis. Because the miscibility of paclitaxel with SIBS as demonstrated by the previous authors [33] is low as measured by DSC, Sipos et al. [34] have reported on the synthesis of poly (ptertbutyldimethylsiloxy)styrene (TBDMS)-P-isobutylene (1B)-PTBDMS), which can be hydrolysed to poly(PH0S-P-PIB-P-PHOS) in order to modulate the solubility of PTx in the polymer, which, in turn, affects its release from the polymer. Further chemical modification of the polymer can be undertaken to effect changes in the polarity of the polymer, including acetylation, which gives rise to PAcOS-P-PIB-P-PAcOS. DSC analyses [34] of all the polymers showed two glass transitions at -67 "C for all the polymers and at 174 and 130 "C for poly(PH0S-P-PIB-P-PHOS) and PAcOS-P-PIB-P-PAcOS, indicating a phase separation for these polymers as expected. In investigating the miscibility of paclitaxel with these polymers [34], a shift in the glass transition towards that of PTx would be expected to be observed. Results showed a shift in the glass transition of the poly(PH0S-P-PIB-P-PHOS) to a temperature in between that of the polymer and the drug, indicating some miscibility between these two components. There was no evidence of any interaction occurring between the acetylated polymer and paclitaxel, because a separate transition at 150 "C was assigned to paclitaxel. These results may be explained in terms of an interaction between the electronrich aromatic ring of the PHOS and the electron-poor ring of PTx, due to further enhanced ability the carbonyls of PTx to interact with the phenoxy protons of the PHOS block. This property is not present in the acetylated polymer due to

the lack of availability of the phenoxy protons and the decreased electron donating effects of the acetoxy-substituted ring. Release of PTx, as previously described, is accompanied by a burst effect followed by sustained release. Because of the hydrophilicity of the PHOS polymer, the aqueous medium diffuses rapidly into the coating, allowing for effective dissolution from the surface. However, due to the decreased hydrophilicity of the acetylated polymer, the initial burst release as seen in the PHOS polymer is limited and a linear initial release was observed. The fact that PTx is less miscible in the acetylated polymer than in the PHOS polymer further explains the difference in release profiles. Research has continued into the use of polymer material to evaluate the release of PTx from coronary stents. Richard et al. [35] have investigated the use of acrylate-based block polymers, consisting of poly (butyl acrylate) or poly(laury1 acrylate) soft blocks and hard blocks composed of poly(methy1 methacrylate), poly(isoborny1 acrylate) or polystyrene. The initial burst effect followed by the sustained release of paclitaxel from these polymers does not appear to be associated with the polyacrylate or the hard block polymers at a 25% loading, but there may be some effect at lower percentage loadings. DSC results showed some miscibility of PTx with poly(buty1 acrylate). It is possible to tailor the release rate by altering the hydrophilic/lipophilic balance of the polymer. Paclitaxel, used in this study, is a lipophilic drug. DSC confirmed the phase separation of the block polymers by the appearance of two distinct glass transitions, and was also used to investigate the miscibility of PTx in the polymers. Changing the hard block polymer to a less polar form was found to increase the release rate of PTx initially, while the release rate was unaffected by the nature of the soft block polymer. Although the release of PTx from acrylate-based polymers appears to be independent of the soft mid-block polymers, the nature of the end blocks when altered to highly non-polar monomers, results in more than 50% of the PTx being released within the first few hours. As expected, much slower release was observed when the end block polymers are polar, and DSC results showed no miscibility between PTx and these polymers. Thus polar interactions between the hard polymers may affect the release of PTx from these matrices. An important development of orthopaedic implants has been the development of artificial composites, which when combined with biomolecules will induce osteogenesis. The properties of hydroxyapatite (HA) materials have been studied over the years with a view to increasing elasticity. Because biocompatibility has been a problem when HA has been combined with various polyethylenes and polysulphones, polyhydroxyalkanoates (PHAs) and its chemical composites, a polymer of hydroxybutyric acid (PHB), copolymers of hydroxybutyric acid and

hydroxyvaleric (PHBIPHV) acids have been suggested because of their successhl use as matrices for culturing osteogenic cells. The physicochemical properties and biocompatibility of a PHB/HA biodegradable composite have been reported [36]. DSC results indicated that the crystallinity of the composite was higher at 81-89 % than that of PHB at 67-73%. DTA results indicated that the HA fiaction did not influence the melting temperature of PHB at 166 "C, irrespective of the HA content, but did affect the temperature for the onset of decomposition. Composites with higher percentages of HA thus had lower thermal stabilities. When the results for the PHBIHA composite were compared with those for the the original PHB, the crystallinity was the same, while the surface proved to be more hydrophilic, due to increasing HA percentages causing increased wettability of the surface. In addition to their application to controlled drug release, biodegradable or absorbable implants have value over metal implants especially because no removal operation is required. Poly (lactide) PLA thus has application for bone plates and screws for bone fixation. Because PLA degrades in the body to lactic acid, a natural intermediate in carbohydrate metabolism, it is suitable for implants and blood vessels which are replaced eventually by the body tissue. In orthopaedics, PCL (poly E-caprolactone) is also suitable. Thermal analysis, specifically DSC can be used to investigate the effect of shear-controlled orientation injection moulding (SCORIM) on the properties of PCL and PLA polymers. SCORIM for the PLA samples promotes a higher degree of crystallinity as shown by the increased energy required to melt the crystalline phase, with the SCORIM processed samples requiring 46% more energy than those of conventional mouldings to melt the crystalline phase. These results were confirmed by XRD [37] where the degree of crystallinity was 21% for SCORIM and 4% for the conventional sample. For the PCL samples, there was no significant increase in crystallinity shown between the two methods (46 and 45%, respectively). The improved mechanical properties may thus be attributed to the shear-induced enhancement of molecular orientation and not crystallinity as was the case of the PLA samples. The improved mechanical properties may thus be associated with SCORIM for both PLA and PCL [37]. Because of their biological tolerance, silicone breast implants consisting of an elastic silicone envelope have been used. These implants are filled with a silicone gel, consist of an inner core and an outer shell filled with saline. The earliest manufactured implants have been reported to cause the most adverse effects. Later implants possess thinner envelopes but also a less-viscous gel. The ageing process of these implants is associated with gel bleeds, which is the removal of low molecular weight silicones by diffusion. Other ageing

processes are the penetration of body compounds into these envelopes, causing changes in the polymer networks and resulting in decreases of the tensile strength. The decrease in strength allows the diffusion of low molecular weight (LMW) silicones to local and distant sites in the body. DSC is useful in providing insight into the ageing process of these implants. 26 implants and 1 virgin control were studied [38] using DSC to determine glass transitions, crystallization points and melting. points and the following results were found: the two endothermic melting points TmIand T m 2(1 1 out of 23) ranged from -39.4 to -44.4 "C (control -46.3 "C) and -32.5 to -40.3 OC (control -39.2 "C), while the recrystallization point T, (15 out of 23) ranged from -82.8 to -88.8 "C compared to -88.1°C for the control and one glass transition Tg from -90.7 to -140.3 "C as compared to -125.2 "C for the control. The shell presented only one Tmand Tg. Migration of material from the gel to the shell was confirmed because 50% of the gels showed only Tm2 No defininite T, was detected in the shells and crystallization was suppressed in some of the shells. The shifting of Tmand Tg confirmed the increased mobility of the gels compared to the shells. Tm2values were different for different generations, with later generations showing behaviour similar to each other and presenting the highest values of T,,, and T,, Decreases in these values are indicative of increased molecular mobility as observed with the older generation implants. Implantation times and the generation affect the DSC parameters. Ruptures and gel bleeding are responsible for changes in the melting behaviours, with the Tm, disappearing and T m 2occurring at a higher temperature. Thus patients with implants should be monitored carefully because they have a high risk of silicone exposure. DSC has proved useful to monitor these processes, such as the migration of siloxanes from the gel to the shell and the manifestations of the ageing process [3 81. 5. APPLICATIONS TO PROSTHETICS 5.1. Introduction Thermoanalytical methods have application in the area of prostheses because it is possible to evaluate their mechanical properties prior to and after insertion into body cavities and to determine the effects of these devices on biological systems.

5.2. Bioprostheses used in heart valves Processing of animal tissues to reduce biodegradability and antigenicity for the bioprostheses used in heart valves has been most successfully achieved by glutaraldehyde (GA) fixation. Because tissue length may be shortened as much

as 80% within a fairly narrow temperature range, it is important to determine the shrinkage temperature of these tanned biological tissues. DSC is the obvious choice because shrinkage is accompanied by absorption of heat and hence gives rise to a measurable endotherm over the temperature range. The stability of the tissue is high if either the shrinkage temperature is relatively high, or the enthalpy of shrinkage is relatively large [39]. Loke et al. [39] used DSC to determine the shrinkage temperature of animal tissues for heart-valve applications. Previous DSC studies had been ambiguous and lacked experimental details. Loke et al. [39] present detailed information on the methodology, including instrumental considerations: calibration of the instrument, heating rate, type of sampling pan, tightness of the crimplseal, and sample considerations: condition of the biological sample, type of reference used and the duration of washing prior to analysis. Typically broad peaks were displayed in the DSC curves, with the porcine pericardium exhibiting a higher shrinkage temperature and thus a greater thermal stability than for the untreated samples. In the GA-fixed tissue, the porcine pericardium was the most thermally stable tissue (86.79 - 87.5 1°C) compared to the equine pericardium at 86.19-86.71°C. Instrument and method validation have played an important role in the reproducibility of these results. Temperature calibration in DSC experiments shows shifts to higher temperatures with increased heating rates. Based on considerations of time and reproducibility, the authors decided on a heating rate of 5 "C mid'. Aluminium pans coated with a non-anodic protective coating of alodine were recommended to decrease the possibility of reaction of water with aluminium. Pans were sealed and there was no special sample preparation. The authors noted that increased washing did increase the width of the endotherm. DSC was considered to provide suitable and reproducible information on thermal shrinkage of animal tissue and such information could be of value in a clinical setting.

5.3. Bioprostheses used in aortic valves The ideal aortic valve bioprosthesis would consist of a collagen-elastin matrix. Lack of success in achieving such a system has been attributed to factors such as calcification and degradation. Calcification may be due to the cell membrane and I or the glutaraldehyde treatment. It is important to be able to remove cellular remnants and maintain the structural integrity of this collagen-elastin matrix. To mimic the properties of the aortic valve, an extraction process was developed [40]. The authors [40] investigated the effect of sodium dodecyl sulphate (SDS) and Triton-X 100 and cholate on the structural parameters of this extracellular matrix. Thermogravimetry (TG - loss of sample mass over a range of 30-550 "C) and DSC (phase transition curves recorded over a range of 30-250 "C) were used to

characterise the biopolymers obtained after extraction from, porcine aortic valves. TG and DSC indicated the release of bound water from the biopolymer, while the DSC scans of the aortic wall proved to be complex. DSC revealed the effects of SDS on the complex, with SDS causing a shift to lower temperature values for all transitions. However, the denaturing action of SDS on collagen was related to changes in structural parameters, while for Triton-X 100, destabilization was not observed and the structural integrity of the collagenelastin networks in the aortic wall was maintained. The clinical use of bioprosthetic heart valves (BHV) has been limited because of calcification. Although ethanol has been suggested as a suitable anticalcification agent, both clinically and non-clinically, there is still need for further investigation into the anti-calcification activity of BHVs in the aortic valve. Chelating agents, such as EDTA, in addition to ethanol may decrease calcification in the aortic wall. DSC was used to determine the denaturation temperatures of the collagen-elastin matrix (CEM) and to elucidate the calcium binding properties of the CEM. The thermal denaturation temperatures for glutaraldehyde-treated CEM, pretreated with EDTA and ethanol, have been reported to be all in the narrow range of 95-105 "C. This suggests that these pre-treatments did not affect the stability of CEM. However, even within this range, there was a difference between the untreated CEM, that pre-treated with ethanol only and that pre-treated with ethanol and EDTA. The study found that sequential treatment with ethanol and EDTA decreased the calcification rate of CEM and this has value in the future therapeutic strategy of BHV application. 6. MISCELLANEOUS APPLICATIONS

6.1. DSC studies on albumins Because albumins can be acquired from various sources, it is important to be able to detect differences between these various albumins in terms of for example their thermal unfolding properties and to compare their thermodynamic parameters as an indication of their relative thermal stabilities. Although human serum albumins showed higher thermal stability than that of bovine albumin, their thermal profiles were affected by the content of bound fatty acids [41]. Albumins have applications in areas of medicine such as a fatty acid carriers for the biosynthesis of lens lipids 1421 and as solders [43] for laser welding. DSC is a sensitive technique allowing for the thorough investigation of the thermal behaviour of these proteins.

6.2. DSC studies on the human intervertebral disc Intervertebral disc degeneration (IVD) is one of the most common causes of lower back pain which is a common problem affecting Western society. In addition to this being an inconvenience, this has proved to be costly to governments worldwide. This degeneration is attributed to mainly environmental effects, including mechanical injuries as a result of the normal ageing process. The use of DSC represents a new approach in this field and is suggested as a means of characterising the different stages of IVD. In a study by Doman and Illes [44], DSC was used to examine specimens from patients in different age-groups. These groups were: Stage I (17-24), I1 (19-40), 111 (39-68), IV (43-84), V (70-88). DSC results showed a gradual increase in the main transition temperature from 60.5 "C to 62.7 "C from stages I to V attributed to mechanical overload resulting in the structure being more tightly packed. Because of this tight packing, extra energy is necessary for disintegration resulting in the transformation beginning at a higher temperature. The elevated transition in the degenerated disc is shown in Figure 14. The decrease in enthalpy from 0.87 to 0.42 J g-', due to loss of bound water, is associated with broadening of the transition period. When comparing the stages, distinct differences were only observed at stages I, I11 and V, confirming that IVD is a continuous process, rather than a disease with distinct stages.

--

/-

Degenerated ------_

.-0

E

5 0 X

W

L

I 45

I

, 50

,

I

I 55

I

I 60

I

65

I

70

I

I

I

75

Temperature ("C)

Figure 14. Transition temperatures of healthy and degenerated discs. (Adapted from [44]).

6.3. DSC studies of human skin from patients with diabetes mellitus (DM) Melling et al. [45] have compared the calorimetric parameters of skin taken from insulin-dependent diabetics with that of control patients. For patients with DM there are changes in the collagen structure and function. The skin from DM patients is more cross-linked and less soluble. These changes may lead to long term complications in diabetic patients. Collagen denaturation was investigated by DSC, with results showing clearly an elevation of the heat flow per unit mass in diabetic patients with a result of 65.7 "C as opposed to 64.2 "C. The majority of the DSC parameters, including temperature at onset of the phasetransition, temperature at peak heat flow and the peak heat flow per unit mass were increased in samples from diabetic patients compared to the samples from control patients. These changes in thermal characteristics can be explained in terms of improved collagen stability due to glucose-mediated crosslinking. The elevation of the peak heat flow shown in older patients may be attributed to the greater energy required to denature the triple helices which have been additionally stabilized by crosslinks. Clinical applications of these findings are limited because they do not provide any evidence of the severity of any complications and also all of the patients were suffering from long-term diabetes [45].

6.4. DSC studies on cartilage destruction by septic arthritis Treatment of bacterial arthritis is providing a challenge for the medical profession because this can result in severe osteoarthritis in patients in later years. DSC scans can be used [46] to monitor the destruction of the cartilage over a period of time. Results for control samples indicated melting temperatures of 49.7, 55 and 63.4 "C, which shifted to higher temperatures with increasing days of infection, with 57 and 63.15 "C observed for the first two transitions at day 1. At time 0, the total enthalpy change was 0.55 J g'l and this decreased to 0.375 J g-l on the first day. These findings highlight the importance of early treatment of infected joints and the usefulness of DSC in the determination of the effect of these infections [46]. 6.5. DSC studies on the effect of tetracaine on erythrocyte membranes The mechanism of interaction of local anaesthetics with membrane lipids and proteins has been investigated [47], because local anaesthetics such as tetracaine can affect both lipids and proteins of erythrocytes. DSC results confirmed that tetracaine affects not only lipids and proteins but also haemoglobin. The transition at 55-56 "C was attributed to the proteins and the transitions at 68-70 "C were attributed to the melting of haemoglobin. Although shifts in transition confirm a moderate involvement with proteins, the packing of the lipids is affected to a much greater extent [47].

6.6. DSC studies on modified poly(urethaneurea) blood sacs Liu et al. [48] used DSC to investigate the stability of modified poly(urethaneurea) blood sacs implanted for up to 160 days. For sacs tested both in vivo and in vitro, there was no significant difference which was confirmed by ATR-FTIR, indicating chemical stability. There were however variations in the molecular weight attributed to chain extensions in vivo. 7. CONCLUSIONS

Thermoanalytical techniques have, especially in recent years, become important in providing vital information on the effects of changes in biological systems and as such have a wide application to medicine, from drug penetration of the skin to measurements on the skin of patients suffering from diabetes. Thermoanalytical techniques have been used for some time in the study of the thermal transitions in the human stratum corneum (SC). Recent efforts to design novel dosage forms and routes of delivery, and the use of the skin for the systemic delivery of drugs, has made this information valuable to the medical profession. DSC has been used to elucidate the mechanism of action of many penetration enhancers designed to overcome the natural barrier effect of the skin. Because of the potential toxicity of these chemical enhancers, it is extremely valuable to be able to determine whether their mechanism of action involves relatively minor disruption of only the surface lipid bilayers, or a deeper penetration. Determination of the effects on these enhancers on the skin transitions by DSC allows understanding not only of the extent of interaction of these substances with the various components of the skin, but also the penetration of the drug substance and subsequent therapeutic outcome. DSC studies have also been valuable in characterising the interactions between drug substances and the polymeric materials used in controlled drug-delivery devices and in determining the release profile of these drug substances from the devices. Thermoanalytical methods have also found application in monitoring implants, whether they are heart-valve poppets, or aortic valves, or metal stents used in the palliative care of oesophageal strictures, or silicone breast implants. DSC studies have played a role in determining the mechanical strength of these devices and their behaviour in biological systems and thus the ability to affect clinical outcomes. Various miscellaneous applications of thermoanalytical techniques have included characterisation of albumins and their behaviour as fatty acids in the biosynthesis of lens lipids, while DSC has also provided on the age degeneration of human invertebral discs. Thus, the extent of the use of thermoanalytical techniques and their application to medicine will continue to increase as they are able to provide important

information on the effects of changes in biological systems which ultimately allows for the prediction of therapeutic and medical outcomes in patients. 8. REFERENCES

1. L.V. Allen, N.G. Popovich and H. C. Ansel, in Dosage Forms and Drug Delivery Systems, gthEdition, Lippincott Williams and Wilkins, New York, 2004, pp. 298-3 13. 2. M.Guia, B. S. Golden, B. Donald, B. S. Guzek, R. R Harris, J. E. Mckie and R.O. Potts, J. Invest. Dermatol., 86 (1986) 255. 3. H. Tanojo, J. A. Bouwstra, H. E. Junginger and H. E. Bodde, Pharm. Res., 11 (1994) 1610. 4. F. P. Bonina and L. Montenegro, Int. J. Pharm., 102 (1994) 19. 5. D.W. Osbone and J. J. Henke, Pharm. Tech., 2 1 (1997) 50. 6. D. Rolf, Pharm. Tech., 12 (1988) 130. 7. A. Aioi, K. Kuriyama, T. Shimizu, M. Yoshioka and S. Uenoyama, Int. J. Pharm., 93 (1993) 1. 8. K. Ghosh and A. K. Banga, Pharm.Tech., 17 (1993) 62. 9. S-C Shin, E-Y Shin and C-W Cho, Drug Dev. Ind. Pharm., 26 (2000) 563. 10. L. Kang, P.C. Ho and S.Y. Chan, J. Them. Anal. Cal., 83 (2006) 27. 11. S.N. Tenjarla, R. Kasina, P. Puranajoti, M.S. Omar and W.T. Harris, Int. J. Pharm., 192 (1999) 147. 12. S.M. Al-Saidan, J. Control. Release, 100 (2004) 199. 13. P.W. Stott, A.C. Williams and B.W. Barry, J. Control. Release, 50 (1998) 297. 14. W.L. Chiou and S. Riegelman, J. Pharm. Sci., 60 (1971) 1281. 15. S. Yakou, K. Umehara, T. Sonobe, T. Nagai, M. Sugihara and Y. Fukuyama, Chem. Pharm. Bull., 60 (1984) 4130. 16. H.K. Vaddi, P.C. Ho and S.Y. Chan, J. Pharm. Sci., 97 (2002) 1639. 17. S. Chesnoy, D. Durand, J. Doucet and G. Couarraze, J. Control. Release, 58 (1999) 163. 18. I. Brinkman and C.C. Muller-Goymann, Pharmazie, 60(2005) 2 15. 19. Y.S.R. Krishnaiah, V. Satyanarayana and P. Bhaskar, Pharmazie, 57 (2002) 12. 20. H.D.C. Smyth, G. Beckett and S. Mehta, J. Pharm. Sci., 9 l(2002) 1296. 21. H. Tanojo, J. A. Bousstra, H. E Junginger and H. E. Bode, Pharm. Res., 14 (1997) 42. 22. Y.S.R Krishnaiah, V. Satyanarayana and P. Bhasker, Drug Dev. Ind. Pharm., 29 (2003) 2 191.

23. Y.S.R. Krishnaiah, V. Satyanarayana and R.S. Karthikeyan, Pharm. Dev. Tech., 7 (2002) 3 305. 24. A Winkler and C.C. Muller-Goymann, Eu. J. Phar. Biopharm, 60 (2005) 427. 25. C. Valenta and M. Janisch, Int. J. Pharm., 258 (2003) 133. 26. V. Pillay and R. Fassihi, J. Control. Release, 59 (1999) 243. 27. E. Touitou, N. Dayan, L. Bergelson, B. Godin and M. Eliaz, J. Control. Release, 65 (2000) 403. 28. A. Rothen-Weinhold, K. Besseghir, E. Vuaridel, E. Sublet, N. Oudry, F. Kubel, R. Gurny, Eur. J. Pharm. Biopharm., 48 (1999) 113. 29. M.E. Biolsi, K.G Mayhan, E.M. Simmons, C.H. Almond and S. Koorajian, Thermochim. Acta, 9 (1974) 303. 30. L. Benko, J. Danis, M. Czompo, R. Hubmann and A. Ferencz, J. Them. Anal. Cal., 83 (2006) 715. 3 1. M. Zilberman, N.D. Schwabe, .C. Eberhart, J. Biomed. Mat. Res. Part B, Appl. Biomat., 69 (2004) 1. 32. H. Witkiewicz, M. Deng, T. Vidovszky, M.E. Bolander, M.G. Rock, B.F Morrey and S. W. Shalaby, J. Biomed. Mat. Res., (Applied Biomaterials) 33 (1996) 73. 33. S.V. Ranade, K.M. Miller, R.E. Richard, A. K. Chan, M. J. Allen and M.N. Helmus, J. Biomed. Mater. Res., 71A (2004) 625. 34. L.Sipos, A. Som and R. Faust, Biomacromolecules, 6 (2005) 2570. 35. R.E Richard, M. Schwarz, S. Ranade, A.K Chan, K. Matyjaszewski and B. Sumerlin, Biomacromolecules, 6 (2005) 34 10. 36. E.I. Shishatskaya, I. A. Khlusov, T.G. Volova, J. Biomat. Sci. Polymer Edn., 17 (2006) 48 1. 37. H. Alpeter, M.J. Bevis, D.W. Grijpma and J.Feijen, J. Mat. Sci. Materials in Med., 15 (2004) 175. 38. A.B. Birkefeld, H. Eckert and B. Pfleiderer, Biomaterials, 25 (2004) 4405. 39. W.K. Loke and E. Khor, Biomaterials, 16 (1995) 25 1. 40. V. Samouillan, J. Dandurand-Lods, A. Lamure, E. Maurel and C. Lacabanne, G. Gerosa, A. Venturini, D. Casarotta, L. Gheradini, M. Spina, Biomed. Mater. Res., 46 (1999) 53 1. 41. A. Michnik, K. Michalik, A. Kluczewska and Z. Drzazga, J.Therm. Anal. Cal., 84 (2005) 113. 42. J. Sabah, E. Mconkey, R. Welti, K. Albin and L.J. Takemoto, Exp. Eye Res., 80 (2005) 3 1. 43. C.B. Bleustein, M.Sennett, R.T.V. Kung, D. Felsen and P. Poppas, Laser Surg. Med., 27 (2000) 82. 44. I. Doman and T. Illes, J. Biochem. Biophys. Methods, 61 (2004) 207.

45. M. Melling, W. Pfeiler, D. Kariman-Teherani, M Schnallinger, G. Sobal, C. Zangerle and E. J. Menzel, Anat. Rec., 259 (2000) 327. 46. T. Sillinger, P. Than, B. Kocsis and D. Lorinczy, J.Therrn. Anal. Cal., 82 (2005) 221. 47. F. Konzxol, N. Frakas, T. Dergez, J. Belagyt and D. Lorincy. J. Therm. Anal. Cal., 82 (2005) 20 1. 48. Q. Liu, J. Runt, G. Felder, G. Rosenberg, A.J. Snyder, W.J. Weiss, J. Lewis and T. Werley. J. Biomat. Appl., 14 (2000) 349.

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry Vol. 5: Recent Advances, Techniques and Applications M.E. Brown and P.K. Gallagher, editors O 2008 Elsevier B.V. All rights reserved.

Chapter 19

QUALITY CONTROL Donald J. Burlett Gates Corporation, 2975 Waterview Drive, Rochester Hills, Michigan 48309 1. INTRODUCTION

Throughout the world, industrial or commercial organizations produce materials for use by consumers. For most of these operations, a starting material is secured, a process is conducted with this material and a final product is produced. For these steps to occur correctly, the nature of the starting material must be understood and how it responds to either processing or storage before use. Then the process being conducted must be maintained as designed. (If intermediates are produced, the characteristics of these must also be considered.) Finally, the product produced must meet certain criteria to satisfy the user of the product. Obviously, this simple approach is not always so simple and there are circumstances when the product is not being manipulated but simply distributed to users or when multiple materials are involved in the process. Once an understanding of the starting material(s) and its (their) reaction to processing are understood, it is possible to begin the process of establishing limits on the critical properties of the starting material, any intermediates produced along the way, and the final product. These limits allow the producer to ensure that the starting material will produce the desired product, will act normally during processing and will yield the desired product which will meet the expectations of the end user. The nature of the starting material for any product can vary widely, depending on source and the method of its manufacture. Consider that commercial products such as metals, food, pharmaceuticals, polymers, chemicals and others are produced worldwide and each of these uses starting materials with very different properties. In many cases, multiple starting materials are involved in the end product, further complicating the nature of the product and the processes used to produce them. However, for any material to be used in a process, certain properties will be crucial to the proper processing and end-product properties.

Quality control (or quality assurance) is a very important component of any commercial or industrial operation that produces or uses materials. Quality control (QC), or quality assurance (QA), is the name of the operation conducted to assure that the nature of a material is within certain specifications that are critical for subsequent processing or use. This is the only way to guarantee that the process proceeds as expected and the customer gets what is expected. Because there are often assurances or guarantees associated with the product, this makes legal considerations an integral part of the quality control operation. In some industries, regulatory compliance is relevant and documentation of the quality of the product is required. Accreditation systems are in place around the world (for example, I S 0 - International Standards Organization, etc.) and they provide standards of operation covering quality that must be considered. Finally, the financial aspects of product quality must also be considered. Financial losses or loss of commercial credibility, as a direct result of poor quality product, need to be minimized. For QC to be conducted, a set of parameters must be established. First, the materials to be used in the process and their characteristics must be understood. Then the parameters of the processing must be established. (Quality testing can also be part of the processing operation during product preparation.) Finally, the end-product should be specified with certain performance or property specifications. With the specifications set, it is simply a matter of conducting whatever testing is necessary to characterize the material or product and determine if it is within the limits specified. The testing used is dictated by the physical or chemical properties of the starting material, intermediates or final product. 2. GENERAL CONSIDERATIONS

When considering quality control options, simplicity and time required for testing are important considerations. Many traditional tests are not currently found in the modem quality control laboratory because of the time required to prepare the material and conduct the testing. It is also important to consider the amount of sample required for the testing. Analysis of materials is a very large field with a wide variety of chemical and physical techniques available. Thermal analysis and calorimetry have established themselves as important techniques offering not only fundamental characterizations of the chemical and physical nature of materials but also physical measurements that can be correlated back to more traditional measurements of product performance such as hardness, softening, etc. Many of the techniques in this area are quick, accurate and utilize very little material.

The list of thermal and calorimetric techniques available for characterizationof materials is quite extensive. Differential scanning calorimetry (DSC), differential thermal analysis (DTA), microcalorimetry, solution calorimetry, thermogravimetry (TG), thermomechanical analysis (TMA), dynamic mechanical analysis (DMA) and many others are among the more traditional techniques available. The principles of these techniques are described in Volume 1 of this Handbook. Additionally, there are many non-traditional techniques that can be employed to analyze materials. Many of these techniques are possible with commercially available instruments and most of these come with software for both controlling the instrument and analyzing the data. Whatever technique is selected for analyzing a material property, it is important to understand that the data generated in the laboratory are only as good as the input and subsequent calculations will allow [1,2]. The precision and accuracy of all inputs must be considered. Willcocks [I] has provided a list of parameters which will have an effect on output data and should be considered for any of these techniques: 1) temperature: calibration, standards, precision, rate effects, etc. 2) specimen environment: gas type, flow rate, furnace size, etc. 3) specimen sizeltype: thermal history, preparation, stability, etc. 4) enthalpy: calibration, standards, etc. 5) modulus: calibration, clamping, etc. 6) software validation: input vs. output, processing algorithms, version, etc. These are not exhaustive, but do point to the necessity of having good data for quality control applications. This is further reinforced by Richardson [2] who states that good quantitative data cannot be produced without the use of standards to correct the data. Corrections can be determined to compensate for instrument error, differences in software or any other errors that are out of the control of the operator. A very good example of calibration of instrumentation is the work on calibrating a DSC conducted by GEFTA [3]. When proper procedures are used for generating data, the data will be as precise and accurate as possible and can be compared with previous data from the same laboratory or from other organizations. Common sense is the key. The techniques chosen for quality control will depend on the materials being evaluated and the necessary information. It is not possible to provide information on techniques for every quality control situation, but this survey of the literature provides illuminating examples from across a variety of material types. This survey covers traditional analytical methods and some that are unusual and different. The examples that follow are organized by the class of material that they pertain to. ,

3. POLYMERS Polymers comprise a traditionally large class of materials that are industrially important. For this reason, there are a large number of examples of using thermal analysis and calorimetry for quality control purposes. The polymers cover a range from thermoplastics to elastomers to thermosets. In most cases, the polymer is not used by itself and quality control is extended from the polymeric materials (as feedstocks) to the hlly compounded composites with all the myriad of ingredients and the desired performance. As can be expected, many of the traditional thermal analysis and calorimetry techniques are usehl for quality control purposes. These can range from physical characteristics of the polymer (glass transition, melting point) to physical properties of the polymer (storage modulus, loss modulus, coefficient of thermal expansion, thermal conductivity) to the compound/composite that is made from it (composition, modulus, etc.) Castelli et al. [4] used DSC to characterize the conformity and technical quality of incoming thermoplastics. They stressed the need for proper selection of the temperatures for scanning and for calculating the heat of hsion. They stressed that analyzing the "as is" sample is important to take into account the thermal history of the sample. This is different fi-om the recommendation made in ASTM D 3417-83 where a preliminary melting and cooling cycle is used to destroy the "memory" of the sample. Combining both techniques provides the most information on the polymer to be used. They also stressed that scans should be started some 120 "C below any expected melting points and continued to at least 30 "C above the transition. Mohler [5] discussed the same approach looking at both scans to examine thermal history as well as the base characteristics for a polyarylamide. The appearance of additional peaks in the scan of the "as is" material can be as important as the main peaks, because they may affect performance such as paint adhesion. A shown in Figure 1, the first scan shows a relaxation enthalpy before the melt, providing information on the heat history of the sample. The second scan after a controlled cooling does not have this peak and provides a much cleaner analysis of the glass transition. Also, a second endothermic peak after the melt has disappeared in the second scan, indicating that a potential second crystalline phase might have been present in the material originally. This second phase may affect the performance of the material in its application. Duske [6] discussed the use of thermomechanical analysis (TMA) in assessing the quality of wire insulation materials. This is particularly important for characterization of the polymer glass transition temperature, T,, and for curing process control. TMA has been known to be a very sensitive technique for the

characterization of the polymer glass transition temperature, Tg, and for curing process control. TMA has been known to be a very sensitive technique for the glass transition and can establish the state of cure. This has the advantage over traditional tests that provide a passlfail verdict on whether enough cure has taken place but do not give the actual state of cure. This technique can also provide information on solvent penetration, electrical failure and mechanical/thermal failure of the wire insulation. 198.OoC[I]

-

Onset : 79.3 "C : 82.3% : lnflpt : 82.0% - Endpt : 85.4% -0,6 1 Del Cp: 0.10 J/(gNK) -0

k Midpt I

80

I

I

I

I

I

100

I

I

I

I

120

2312 OC I2] I

I

I

I

140

I

I

I

I

160

I

I

I

I

180

I

I

I

I

200

I

I

I

I

220

I

I

v I

I 238,30C I

240

I

I

I

I

260

I

I I

I

.

280

TIoC

Figure 1. DSC measurements on a polyarylamide, PA MXD6-GF30, in nitrogen (solid line, first scan; dashed line, second scan after controlled cooling) [ 5 ] . (With permission from Kunststoffe.) Duske [7] also discussed several other techniques of value for this material. DSC can be used to check the oxidation resistance of the wire insulation by determination of the oxidation induction time (OIT) - an isothermal technique to determine when the oxidation will occur. Thermogravimetry (TG) has also been used to examine the decomposition of the insulation and the effectiveness of flame retardants. TMA is also useful for measuring the thermal expansion of the wire insulation and this technique is also applicable to multi-layered coatings. Elastomeric systems are quite commonly monitored by thermal techniques. Mohler [ 5 ] has discussed how DSC is used to characterize the quality of blends of elastomers. Changes in polymer Tg can point to miscible phases that might otherwise require a microscopic technique to establish. The cold crystallization of a polymer in a blend can also be examined using the relaxation enthalpy of the sample. Vacuum TG can look at plasticizer content separate fi-om polymer degradation. Dynamic mechanical analysis (DMA) is always a powerful technique with elastomers. Dimensional stability, storage modulus, loss modulus and loss factor are all important for this class of materials and are

easily established using DMA. OIT determinations are also conducted using DSC. Riga [8] used TG and TMA for quality control of automotive elastomeric seals. TMA was used to confirm the Tg of the polymer, as well as the thermal expansion coefficient. TG was also used to provide data on the composition of the compounded seal material. Both of these techniques provide critical information for control of quality but it was pointed out that these techniques alone could not predict failure of the seals unless a very specific problem was pointed out. Knappe [9], in discussing quality control applications, provided a schematic for characterizing incoming polymeric materials, shown in Figure 2. Because traditional testing, such as melt flow index, ash content and tensile testing, may not provide enough information to avoid processing defects, the more complete analysis is desirable. Here, the data is entered into a "computer aided quality" (CAQ) system for analysis and comparison with specifications. The decision tree utilized is typical of what might be used for this class of materials. This same approach can be used for failure analysis of these materials in the product. Knappe [lo] described the use of DMA to check the plasticizer level of polybutadiene/natural rubber blends. DMA can also be used to look at coatings on elastomer parts, an example being a polyurethane coating on an EPDM (ethylene propylene diene monomer) bumper part, where the low temperature storage modulus can be a key to component toughness. Polymers are also used in adhesives and Avalon [ l 11 noted that TG can be used in combination with a database of information about the components of the product and the product itself to compare the maximum mass loss temperature and profile of the TG curve to provide a quantitative analysis of the product. The database of information can be either a strength or weakness of this approach, depending on the completeness of the data available for the comparative work. The use of DSC, DMAITMA and TG were also discussed [12] relative to adhesive applications. Among the properties examined were % crystallinity, Tg, cure state, environmental stability, blend ratios, modulus and damping, thermal expansion and shrinkage, composition, volatiles and plasticizer levels. For non-elastomeric thermosets, Carrasco [13] described the use of DSC to follow the conversion during cure of a trifunctional aminoepoxy resin. Here, the conversion was determined versus a non-cured sample. Sauerbrunn [14] also described the use of DSC to follow thermoset curing amongst other applications. Stephan [15] utilized dielectric analysis (DEA) as well as DSC, viscosity measurements and insolubles analysis to follow the curing, cure state and

delivew (e.a. delivew note, b ~ iof l freight, etc.)

quality agreement 1. Melt flow index 2. Yield stress 3. Elongation at yield 4. Notch impact strength

Enter test values into CAQ system

Ck goods, ctabins

to laboratory immediately

~~k~ goods and sample to laboratory

1

k4

DSC~TGmeasurement

DSC/TG measurement

measurements into CAQ

Accept material and fill silo

delivery required

Accept goods

Figure 2. Flow diagram for control of incoming materials at a supplier to the automotive industry [9]. (CAQ = computer-aided quality) (With permission from Kunststoffe.) physical properties of an epoxy composite. The relationship between T, t, viscosity, ionic conductivity, extent of reaction and T, is part of a developing field of chemorheology and DEA is a powerful tool in this area. The T,, as determined by DEA can be compared with the extent of reaction to provide a tool for characterizing the cure state of the product (Figure 3) The ionic conductivity of the sample is also related to the state of cure and can be used to examine products to ensure ample cure (Figure 4). This only requires establishing a calibration curve for any potential cure temperature and using a

-10 I 0

I

I

I

I

I

I

I

I

I

I

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Extent of reaction

Figure 3. Dependence of T, on degree of conversion of an epoxy composite at various temperatures [15]. (With permission from Wiley Interscience.)

- 1 2 ~ 1 ' 1 ' 1 1 1 ' 1 1 1 ' 1 1 1 " " " " ' ~ 0

10

20

30

40

50

time, min.

Figure 4. Log (ionic conductivity) versus time of an epoxy composite at various cure temperatures [15]. (With permission from Wiley Interscience).

single reading to determine the state of cure that has been achieved. Full analysis of the data from these experiments allows for use of equations relating Tg and conversion. Care must be used, however, in how the dielectrics are used. It is also important to consider the thickness of the sample being monitored and understand that several monitors may be necessary in thicker samples because of the exothermic nature of this chemistry. Another growing category of polymer applications is composites. A wide variety of fillers, many in fibre form, are being used in products worldwide. Characterizing the fibre content of mixed products is a vital characteristic of product performance. Moon [16] and Licea-Claverie [17] used TG in comparison with standard tests like ignition (time consuming) and digestion (toxic) techniques. The epoxy system described [16] had both glass and carbon fibres present. Using air as the atmosphere, the decomposition could be examined. With carbon fibre "calibrated" into the analysis, a formula could be generated that provided very good analysis of the fibre contents. This approach provided a faster analysis and added capability to the laboratory. A nylon composite with glass and carbon fibres [17] was characterized by simultaneous DTA-TG in air using a relatively fast scan rate (25 'Clmin.). Tg, T,,, and Td were characterized for various compositions and were shown to be very reproducible. The results also provide compositions that predicted impact test results very nicely. Fibreglass reinforced polycarbonate can also be examined for quality and statistical process charts (SPC) are used to chart incoming parts [18]. % fibreglass (from loss on ignition (LOI) and the-polymer Tg (from DSC) are used to monitor product quality (Figure 5). The control limits (UCL = upper control limit; LCL = lower control limit) are based on product performance criteria and a database of products examined. Constant monitoring of product provides valuable information on quality over a period of production.

Number of Observations

Figure 5. Statistical process chart (SPC) of T, by DSC for incoming parts [18]. (With permission from Elsevier.)

4. ORGANIC CHEMICALS Much like polymers, organic chemicals have their own set of properties that can be used to provide quality control techniques. These properties can be very similar to those used in polymer systems and can include glass transition temperatures, melting (temperature, enthalpy), crystallization (temperature, enthalpy), decomposition (thermal, oxidative), reaction (with various other materials) and more. Thus, the techniques that could be used for quality control of starting materials, process intermediates and final products are similar to those used for polymers. Oils are a class of organic chemicals that are very much like polymers, in that they can have glass transition temperatures and melting points, can oxidize and degrade thermally and can have other materials added to enhance performance. Quality control techniques are thus very similar. Riga and coworkers [19, 20, 211 have evaluated various aspects of lubricating oils. In examining the performance of lubricating oils at low temperatures [19], thermomechanical analysis (TMA) was used to study the viscometric properties. Here, crystallization kinetics (determined by TMA data at various temperatures) of waxy ingredients are important for determining whether the product will become too viscous at low temperatures. The effectiveness of "pour point" depressants was also analyzed in various samples. Another aspect of lubricating oil quality that was evaluated was oxidation induction time (OIT) using pressure differential scanning calorimetry (PDSC) [20]. Various oils were evaluated using an ASTM method for OIT evaluation. Variables used were heating rate,

temperature, pressure and sample size along with pan parameters (size, shape, surface topology and potential contaminants). The conclusions of the work included recommendations for pan type, because the pan type played a significant role in the variability of the oxidation results. Flat aluminum pans were recommended for this application to minimize that variability. Quality control laboratory issues have been addressed for oil monitoring involving TG and PDSC [21]. The importance of statistical quality control is discussed and examples of techniques used in an industrial laboratory are given to highlight the importance of these techniques. Figure 6 shows a typical quality control chart where data from a series of test samples are plotted to follow (in this case) the derivative peak temperature for a mass loss event in the TG. Also noted on the chart are points in the control process where other hboratory functions were performed so that changes in the data that were not product related could be accounted for in the plotting process. Figure 7 provides a flow chart for decision- making in the quality control process. In this example, the flow rate of purge gas in the analyzer is the aspect examined when a problem occurs but it is obvious that many other options could be included in the flow diagram to cover other reasons for laboratory problems. REF 638: 10 OC/min; 25-650 "C: AIR

y

o

340 335 -

IN-HOUSE SERVICE

.-c

$! 330 -

-E

-E

Y

325

UCL

X BAR

320 -

a w

'= .-S

z

315

-

LCL

310 305

I

0

I

I

I

I

I

I

10 20 30 Ordered date in 1992 and early 1993

I

I

I

40

Figure 6. TG data, control chart with laboratory procedure notations [21]. (With permission from Elsevier.) Wesolowski and coworkers [22] have used another approach to handling data from lubricating oils, namely the application of chemometric analysis. TG and DTG data from a variety of oil samples were gathered and utilized in a matrix analysis to determine which factors provide information related to oil

characteristics such as kinematic viscosity and flash point. After analysis of onset, end and successive mass losses in the TG curves and temperature range of peaks, peak temperature, peak height, peak width at half height from DTG, regression analysis was used to determine whether the temperature range of the DTG peak and peak height could be used to provide estimates of kinematic viscosities and flash points. Principal component analysis "makes possible the reduction of the dimensionality of data space and provide for a concentration of systematic information previously dispersed over many variables in a few common abstract factors". This is a powerful combination of thermal analysis data and mathematical manipulation to obtain correlations between common physical properties necessary for quality assessment and thermal analysis data from the laboratory. STANDARDIZATION

GENERATE DATA DSC, TGA & TMA ANALYSIS

I

CORRECTIVE MONITOR FLOW RATE DAILY

I

SCause W~I

6 CONTROL

Variation Only

FLOW RATE

Figure 7. Thermal analysis process flow chart: Statistical quality control [21]. (With permission from Elsevier.) Fuels form another class of organic chemical. Masson [23] examined TG methods to detail the composition of incoming bitumen samples. As shown in

Figure 8, using standard TGIDTG methods, the reproducibility of scans of a sample of bitumen was poor. Peak shifts and peak height discrepancies were noted between individual scans of a single sample. When a "high-resolution" TG technique was applied (using control of the heating rate by the rate of decomposition/volatilization of the sample), it was possible to control the resolution factor (R) and sensitivity setting (S) over the various portions of the overall scan to provide excellent resolution of the peaks. This can be seen in Figure 9 where a standard scan is compared with a high-resolution scan of the same material. Once the conditions were established through a series of experiments, it was possible to provide exctllent resolution and reproducibility as shown in Figure 10. The main factor is heating under the conditions of constant reaction rate versus under dynamic rate. The final technique established was a three-step high resolution TGIDTG technique customized such that R and S vary within specific temperature regimes during the analysis. Optimizing the sample size and shape completes the development of this quality control technique. This technique could be extended to studying cracking processes for heavy hydrocarbons. Some examples from biotechnology, pharmaceutical and inorganic chemicals provide illustrations of other techniques that might be applied.

Run 1

I;

- - - - - - . Run 2

. 8

I I I I I I "I I

P

I;

I

.-C

a (51

5

0.5

-

L 0

5

.+

0 a +

m

n:

0

I

I

I

I

I

I

150

250

350

450

550

650

Temperature / "C

Figure 8. DTG results fiom duplicate runs on bitumen heated at 20 OCImin [23]. (With permission from Elsevier.)

High-resolution

-

- - - - - -- Standard

Heavy fraction

\4

/:

,

I \

-

_--*

_--------

-

1

- I

f4

::

I

,'

\\

'.f\-I

!\

'\J

\

\\ \

\.'

\\ '\----

Temperature I o C

Figure 9. Standard and high-resolution DTG for bitumen (dotted curve has been shifted up for clarity) [23]. (With permission from Elsevier.)

150

250

350

450

550

650

Temperature / OC

Figure 10. Reproducibility of the DTG profile of five replicates of a PetroCanada bitumen [23]. (With permission from Elsevier.)

5. PHARMACEUTICALS

Quality control in pharmaceuticals is an extremely important area because poor quality can have a critical impact on the consumer. For this reason, characlerization and analytical procedures undergo great scrutiny and considerable regulatory control. Because most pharmaceuticals are combinations of the drug with other additives that provide delivery and handling assistance, thermoanalytical techniques are useful in quality control applications. Quality of incoming materials for synthesis/fabrication, intermediates in the process, composition of final products and final product characteristics are all quality control concerns. Figure 11 provides a good summary of the broad range of thermal analysis and calorimetry techniques that can be used for pharmaceutical applications [24]. These examples show the application to characterization of polymorphs, glass transition temperatures, hydrates, solvates, hygroscopicity, purity, stability, processability, interactions between components and interactions in aqueous media. These properties have impact on important pharmaceutical characteristics such as activity, solubility, homogeneity and stability. With this collection of properties to be monitored, it is obvious that DSC (and MDSC) and TG will be the primary techniques utilized both alone and coupled with other techniques such as spectroscopy.

Application

Techniques Polymorphism DSC, solution calorimetry, microcalorimetry, sub- ambient^^^, TG,

Raw materials: characterization, control of crystallization, drying, milling

DSC-spectroscopy, DSC-X-ray, Thermomicroscopy

Raw materials: storage conditions Drug products: control of processes, granulation, mixing, milling, spray-drying, kneading, melting, lyophilization

DSC, TG, adsorption isotherms DSC, TG, DSC-IR, DSC-Raman

Amorphous state Temperature Tgof glass transition of single components and influence of moisture, excipients

DSC, MDDSC

Study of polymers, copolymers

DSC, MDDSC

Optimization of formulations: microspheres, lyophilization, coating

DSC, MDDSC

Quantitation

DSC, microcalorimetry Purity

Raw materials: purification, stability

DSC Stability

Thermal decomposition, kinetics

DSC, TG, TG-MS, TG-IR

Compatibility

Microcalorimetry Drug products

Physical interactions, phase diagrams

DSC

Process optimization: solid dispersions, solid solutions, microspheres, modified release, lyophilisates

DSC, DSC-spectroscopy

Melting point of liquid formulations Identification, quantitation

Sub-ambient DSC DSC Water interaction

Gels, creams, polymers

DSC, sub-ambient DSC, DSC-microscopy

Determination bound water

DSC-TG, sub-ambient DSC

Liposomes Characterization hydrated phospholipid bilayer DSC, microcalorimetry Phospholipid-drug interactions DSC, microcalorimetry

Figure 11. Main applications of thermal analysis and combined techniques in pharmaceutical development/quality control [24]. (With permission from Springer.)

An example of the use of DSC to characterize the three polymorphs of a drug molecule [24] is shown in Figure 12. Form A converts to B upon heating (both enantiotropic forms) and C (the stable form) can be formed from A or B given the correct conditions. The DSC allows characterization of the pharmaceutical and differentiation of the various polymorphs. This characterization is more difficult by other techniques. In Figure 13, the malonate salt of a drug substance is characterized by DSC to show melting of the salt, decomposition of the malonic acid and melting of the base drug. The TG analysis showed the appropriate mass loss associated with the decomposition of malonic acid to C02. This characterization allows quantification of the salt prepared and determination of the melting temperatures of the product.

I

I

I

I

I

50

75

100

125

60

Temperature1"C

I

I

80

I

I

I

100

I

I

120

Temperature1"C

Figure 12. Example of kinetic behaviour of enantiotropic and monotropic transformations. A and B are monotropes to C; a- DSC at 10 Klmin., A: Form A, B: Form B, C: Form C; b - DSC at 5 K/min. of a mixturel: 1 of Forms A and C [24]. (With permission from Springer.) Figure 14 shows a diagram of the melting point of various compositions of the two materials, polyethyleneglycol 6000 and daropidine. From this diagram, examination of a sample will provide compositional information for quality control and control of the manufacturing process. In developing standard DSC methods for characterization of pharmaceuticals, use was made of experimental design to determine the best protocol [25]. Sample size, heating rate, atmosphere (air or nitrogen), crucible type (open, crimped, pinhole, crimped with pin hole) and relative humidity were included in an evaluation of technique and recommendations were made for the best procedure. In this investigation, data derived from benzoic acid and vanillin

were compared with literature values for melting temperatures and heats of fusion.

75

125

175

225

275

Temperature / OC

Figure 13. Interpretation of DSC curves. Use of combined methods: DSC and TG curves of a malonate salt; 1 - melting of the salt; 2 - decomposition of malonic acid and evolution of COz; 3 - melting of the base of the drug substance. TG: 21.5% loss of mass; theoretical amount of malonic acid - 21.1% [24]. (With permission from Springer.)

0

20

40

60

80

100

Drug substance 1 %

Figure 14. Phase diagram of polyethyleneglycol 6000 and darodipine. The temperature of melting is fixed according to the phase diagram [24]. (With permission from Springer.)

TG has been used to examine starting materials, intermediates and finished products, as well as to characterize the interactions with excipients used in drug formulation [26, 27, 281. Thermal stability of the formulated drug is important for shelf-life determinations as well as for understanding of interactions with excipients. Different combinations of the drug formulations were examined to determine which components were involved in decreasing the thermal stability of the primary component of the drug formulation. Another aspect of pharmaceutical science is the delivery system used for the drug. Many different approaches are used: solid tablets, liquids, suspensions, emulsions and fatty suppositories. Characterization of the materials used in these systems is also very important for proper delivery of the drug [29]. Figure 15 shows a DSC scan of a water/oil/water (w/o/w) emulsion used in controlled drug delivery. The sample was scanned during cooling and the crystallizationof water in the sample reflects both internal and external water. These characterizations are important for assuring the quality of the emulsion via quantification of the water levels in each phase (the composition of the different phases affects the osmotic pressure and water migration from one phase to another). The temperature of freezing of the internal water and the shape of the curve provide information on the internal water (which is affected by the stirring rate, viscosity modifiers, nature of the oil phase and presence of active substances used during preparation of the emulsion). DSC can also be used to monitor the stability of the emulsion over time.

-10

-14

-18

-22

-26

-30

-34

TemperatureIoC

Figure 15. Typical DSC curve of a water/oil/water multiple emulsion [29]. (With permission fiom Springer.)

Fatty suppositories rely on the nature of the solid fats both for delivery of the drug and during the manufacturing process [30]. In use, these fats need to disintegratelsoften within a timeframe (30-60 minutes) at 37 "C in water. The melting and solidification character of the fat is important to production control and proper drug delivery. DSC can be used to characterize the melting and crystallization processes, with a "heat then cool cycle" providing the crystallization exotherms for analysis. Figure 16 shows the crystallization isotherms for a sample obtained at different temperatures. When the peak temperatures from these crystallizations (t,,,,) are plotted using a semilogarithmic hnction versus crystallization temperature, Figure 17 shows a straight line indicating an exponential hnction. A plot like that shown in Figure 16 provides key information for both manufacturing and use of the suppository. These data can then be used to check the quality of solidification and the ultimate quality of the product.

Time in min

Figure 16. DSC isotherms for Suppocire AT obtained at various crystallization temperatures [30]. (With permission from Elsevier.) Thermal analysis techniques can be also be used as in-line monitors for both quality control and process optimization tools [3 11. In bioreactors, heat is associated with the reactions occurring and calorimetric data are used to monitor the reactions and help keep the reactor in the appropriate temperature zone for producing the desired product. Figure 18 shows the heat effects observed in an isothermal reaction calorimeter as different reactants are exhausted. These data can be used to control the quality of the product during manufacturing.

xHs t, max

298

299

300

301

302

303

304

Crystallization temperature in K

Figure 17. Crystallization parameters for Suppocire AT obtained by isothermal DSC. T,,, ,time of maximal solidification rate; t,,, nucleation time; AH, ,heat of solidification released [30]. (With permission from Elsevier.)

exhaustion

> characteristic perturbation

No 3

exhaustion V

I

ammonium exhaustion

I I I I I I I I I I I I I I I I I I I I I I I

0

10

20

30

40

50

60

70

80

timelh

Figure 18. Heat production rate (qR)in an isothermal reaction calorimeter of a batch culture of Saccharopolyspora erythraea after correction for baseline and torque variation. Metabolic changes due to ammonium, nitrate or glucose exhaustion are clearly visible [3 11. (With permission from Elsevier.) 6. FOODS

Whether food is being grown on a farm or produced in a factory from other starting materials, the nature of the product is vital to the manufacturer. It must be of the best quality, be safe and stable until the consumer uses it and it must

meet the consumer's expectations. One of the fastest growing groups of analytical techniques being used in process and quality control of foods are thermal and calorimetric techniques. Foods can be classified into four major component categories: carbohydrates, lipids, proteins and water. Although there are other minor components, these categories dominate the make-up of most foods. Raemy [33] has discussed these characteristics and outlined the techniques to be used, as shown in Table 1. Also of importance for foods are the thermal behaviours of minor constituents such as caffeine or vitamins. Application to reconstituted and raw food as well as microbiological study of microorganism growth in food and process safety is also discussed. Table 1. Major food components, their properties and the thermal/calorimetric techniques used to characterize them [33]. Techniques Phenomenon Major Component crystallization of water DSC, TG Carbohydrates melting DSC DSC, TG decomposition gelatinization of starch in DMA the presence of water retrogradation of the gel Isothermal calorimetry glass transition of DSC, MDSC amorphous samples relaxation of amorphous DSC, MDSC

As can be seen from Table 1, DSC (or MDSC) is very commonly used and is a very powerhl technique. It provides a direct estimate of overall enthalpy

change of transitions without requiring knowledge of the thermodynamic mechanism. Little sample preparation is necessary and the heat processing conditions encountered in food treatment can be closely simulated. An example of the characterization of one of these food components is the effect of an emulsifier on the crystallization of a fat [33]. As shown in Figure 19, different emulsifiers can change the crystallization temperature to higher or lower than the fat alone would exhibit. Here, DSC is used to measure the crystallization in a cooling scan fi-om above the normal melting temperature. These temperatures are important to the processing and ultimate utilization of the fatty material.

Figure 19. Cooling curves of a fat in the absence of emulsifier and in the presence of emulsifier A or B, showing the effect on the fat crystallization temperature [33]. (With permission from Springer.) Another example is characterization of a protein and how it is affected by water content [34]. Samples of a soy protein were isolated with water contents varying between 15% and 40%. To produce textured protein, it is necessary to cook the protein at an elevated temperature and allow it to rehydrate after cooking. The temperature needed for texturization is that of the denaturation of the protein. The DSC scans in Figure 20 show the denaturation temperatures as a function of initial water content and provide processing and quality control options.

A microcalorimeter scan can also be used to provide a determination of the nature of meats [34]. As shown in Figure 21, samples of chicken breast and thigh show different profiles which may be due to differences in their pH and contraction state as well as different content of red and white muscle fibres. It is also important to consider whether the material is examined in isolation versus in situ because the thermal properties may be different.

60

80

100

120

140

160

180 200

Temperature ("C)

Figure 20. DSC curves of a soy-protein isolate at various water contents [34]. (With permission from Springer.)

Figure 21. Microcalorimetric curves of chicken breast and thigh meats [34]. (With permission from Springer.)

Oillwater emulsions are essential to creams and the crystallinelhydrophilic gel phase, lyophilic gel phase, dispersed oil phase and bulk water make up much of these systems. Understanding the surfactants, their effect on the cream (thermally and rheologically) and the microstructure of the system are all important in optimization of the formulation and manufacturing. TGIDTG has been used to characterize these systems [35]. In this example, a mixed surfactant system has been employed and the ratio of the two surfactants dictates the structure and stability of the cream being produced. As shown in Figure 22, as the ratio is changed, the DTG profiles show the stability changing with composition. The proportion of interlamellarly-fixed water decreases with an increase in amphiphile content. This information is used for quality control of both the processing and aging of creams.

Figure 22. Influence of the composition of a mixed emulsifier. DTG curves of creams prepared with Eucarol AGEJEC, the ratio of surfactant and amphiphile being 1.4, 1.6, 1:12 [35]. (With permission from Springer.) Much in the same manner, TG can be used to fingerprint food materials for quality control. Milks used for cheese making are very important for the final product (both in quality and meeting claims for the product). An example is the use of sheep's milk versus cow's milk [36]. As shown in Figure 23, there is a marked difference in the profiles of the two milks. This can be used to determine if the milk that is being provided for cheese making is really sheep's milk or a combination of the two (which is the case in Figure 24). This can be done very simply with a 20 'Clminute scan in nitrogen using 40 mg of material (ideal for quality control). TG and DSC can be used to characterize other foods. Rice and its by-products were characterized using both DSC (gelatinization of the starch, loss of water) and TG (ash content, moisture content) [37]. TG has also been used to study corn for quality control via examination of the mass losses of the food [38]. The

TG typically shows three mass-loss steps: water loss (room temperature to 150 OC), thermal decomposition of the carbohydrates - starch, amylase, amyl pectin (250-400 "C) and secondary thermal decomposition of carbohydrates of low molecular mass and fibres (400-600 "C). Corn and its derivatives (hominy, gritz, vitamilho (pre-cooked flour) and bran) were characterized and their kinetics of decomposition evaluated. Once activation energies and preexponential factors were provided, these could be correlated with properties of the food materials and used for quality control purposes.

Figure 23. TGDTG scans of sheep's milk (left) and cow's milk (right) [36]. (With permission from Springer.)

Figure 24. TGDTG curve of fraudulent milk [36]. (With permission from Springer.)

Many foods are dried as part of the processing before use. This is common for many fruits. The drying process can be simple sun drying or oven drying. In either case, properties including thermal conductivity (k), specific heat capacity (c,) and thermal diffusivity (6)need to be determined for each material so that temperature and time of drying can be determined based on moisture content. An example of the determination of these properties for fmit utilizes slabs of fruit prepared to different moisture contents followed by measurement of the thermal conductivity using a probe [39]. A plot of thermal conductivity versus moisture content is provided in Figure 25. The heat capacities of these samples were determined using DSC. The porosity, content of other chemicals (such as sucrose) and several other factors can affect the results for various fruits. The results fiom a fmit sample can be used to control the quality of the product being produced via different drying processes.

0 10

20

30

40

50

60

70

80

Moisture Content, M (%, wet basis)

Figure 25. Thermal conductivity of cassava as a function of water content [39]. (With permission fiom Elsevier.) A major concern for foods is unwanted microbial growth that can lead to loss of the food. This is especially true for areas where inadequate refrigeration occurs. Calorimetric techniques can be used to assess microbial growth and be done in much less time and more conveniently than traditional techniques. An example is the assessment of microbial growth in milk [40]. Here, three samples are examined: raw milk at 30 OC, milk pasteurized at 72 "C for 15 s, then taken to 30 "C and milk treated with a lactoperoxidase system (LPS)

activation then held at 30 O C . Isothermal calorimetry was conducted using the samples in a hermetically sealed cell at the desired temperature for 1 hour. The measured heat was compared to the rate of microbial growth of the sample. As seen in Figure 26, there is a good correlation between metabolic heat rate and microbial growth. This is a good method for assessment of the quality of product treatment to avoid microbial growth. C U

P

uLPS

0

8 Time (h)

16

Figure 26. Microbial growth (a) and metabolic heat rate (b) from milk kept at 30 "C as affected by pasteurization (P), lactoperoxydase system (LPS) activation and untreated raw milk (C). A highly significant interaction (P < 0.001) was found between milk treatment and incubation period [40]. (With permission from Elsevier.) 7. INORGANIC CHEMICALS

The use of thermal analysis and calorimetry for inorganic chemicals is more restricted to TG because of the nature of the materials and what is important for their use. These materials typically only show degradation processes within the

temperature range attainable by TG. DSC or calorimetric techniques could be used to look at the energetics of these decompositions if the instruments were capable of reaching these temperatures. However, that is typically not possible. An example is the quality control of commercial plasters. Plasters are calcium sulfate of varying degrees of hydration. The quality of these materials will depend on the hydration state (which depends on the temperature and water vapor pressure) and composition. TG, both dynamic and isothermal, has been used to characterize different plaster samples [41]. The mass-loss steps seen in a sample during a dynamic scan, shown in Figure 27, allow the direct calculation of the compositional levels of hemihydrate and dihydrate in the sample. At the same time, isothermal scans can be used to evaluate samples by comparison of the original sample versus sample having undergone a hydrationldrying step. Although dynamic scans are faster than isothermal scans, isothermal scans were recommended because they do not depend on an estimation of mass loss.

92

1

I

I

I

I

I

I

I

0

50

100

150

200

250

300

350

TemperatureIO

C

Figure 27. Typical dynamic TG curve of 2.503 g of plaster A in air. Heating rate = 5 "Clmin [41]. (With permission from Elsevier.) Another example of inorganic materials and their quality evaluation involves the mineral impurities of industrial talc [42]. Talc is commonly used for plastic reinforcement. Pure talc (determined by XRD analysis) only shows a mass loss in TG at about 900 "C. When samples of talc are analyzed using TG, other mass losses can be seen in the temperature ranges of 500-600 "C and 600-700 "C, as shown in Figure 28. By comparing the XRD results of the impurities with the

data from TG, it was shown that the impurities being degraded at 500-600 "C are chlorites and the impurities being degraded at 600-700 "C are carbonates from calcite and dolomite. The TG can be used to check for the presence of these impurities but cannot be used to look for quartz because quartz does not degrade within the temperature range accessible in TG.

94 300

400

500

600

700

800

900

Temperature/OC

Figure 28. TG curves of different talc samples [42]. (With permission from Elsevier.)

8. METALS

Metals as a class of materials are limited in the thermal analysis and calorimetric techniques that can be used for quality control purposes. Some high temperature DTA work can be done, but DSC, DEA, DMA, and most calorimetry techniques are not found in quality control labs working with metals, but there are some alternative approaches. Because metals are usually formed at very high temperatures, there is little to be done once the metal is formed and cooled. However, there are some interesting techniques that are used to study the influences of the cooling process on the formation of the various phases of the metal in a product. The solidification process is a non-linear heat conduction problem complicated by the release of latent heat during the solidification process. Equations are available to describe the general process [43]. Because the latent heat of

solidification is the most important factor for characterizing the solidificationof an alloy, measurements of the temperature change of the metal, by placing thermocouples in the mass of the metal and applying different algorithms to the cooling curves obtained, enable the solidification mechanism to be evaluated. A single thermocouple can be used, or an array of thermocouples to examine the material in more than one dimension. Figure 29 shows a cooling curve for a sample of cooling metal. The temperature difference is a complicated effect that is attributed to the heat transfer coefficient defined between the thermocouple and the quartz tube. The discussion in this reference involves a comparison of different models applied to this cooling phenomenon. 1300

I

I

I

I

40

I

-0

0

100

200

-20

300

Time [s]

Figure 29. Cooling curve of a thermocouple in a molten metal and the temperature difference between the thermocouple and the migrating hot spot [43]. (With permission from Elsevier.) Another example of this application to metal cooling emphasizes that the quality of the metal is affected by production control and cooling needs to be determined and analyzed [44]. Because it is believed that faster cooling rates lead to refinement of the microstructural features (including grain size, dendrite arm spacing (DAS) and intermetallic phases, it important to characterize these cooling curves throughout the entire process. Typical analysis of the metal has involved use of the DAS as a non-destructive and predictive tool for property evaluation. However, thermal analysis can provide information about composition of the alloy, latent heat of solidification, the formation of the solid fraction, the types of phases that are created during cooling and possibly

dendrite coherency. Also, the liquidus and solidus temperatures, characteristic temperatures related to eutectic regions and intermetallic phase formation. The same approach mentioned in the previous reference is used here. The thermocouples are introduced into the sample and the cooling curve is recorded. A cooling curve for a 3 19 aluminium alloy is shown in Figure 30, along with the rate of cooling and indications of where the liquidus and solidus points are, as well as the eutectic points for Si and Cu. Figure 31 shows the effect of different cooling rates on the solidification process. The liquidus point is being reached at different times for each cooling curve. Figure 32 shows the effect of the cooling rate on the nucleation temperature (liquidus temperature) for the same 3 19 alloy. This will affect the characteristics of the metal alloy produced. Using this technique, a prediction of the quality of the metal produced can be made from the cooling rate of the process.

Cu eutectic

Figure 30. Cooling curve, first derivative curve and representation of characteristic parameters used and analyzed in this evaluation of a 3 19 aluminium alloy [44]. (With permission from Elsevier.)

, .,.,

-

- 0.52 "CIS - 1.3 "C/S - 2.1 "CIS

480 0

I

I

I

I

I

200

400

600

800

1000

1200

Time (s)

Figure 3 1. Cooling curves of a 3 19 aluminium alloy under various solidification conditions [44]. (With permission from Elsevier.)

605

1

I

I

I

I

0

0.5

1

1.5

2

2.5

Cooling ratercis)

Figure 32. Effect of the cooling rate on the nucleation temperature of a 3 19 aluminium alloy [44]. (With permission from Elsevier.)

Computer-aided cooling curve analysis (CA-CCA) can be used to analyze the solidification of metals [45]. On the foundry floor, this technique (cooling curve analysis) is used instead of DSC. Along with determination of various metal alloy parameters affected by cooling rate, it is possible to characterize the

properties of the alloy as a function of composition. As shown in Figure 33, as the Ti and B levels are increased, the cooling curve is shifted upwards and this reduces the undercooling during the process. The approach is simple, inexpensive and suitable for commercial use. Both Newtonian and Fourier techniques were used in the analysis of the data. The Fourier method was found to be more complicated but also more accurate than the Newtonian method. This approach is useful for calculation of latent heat, solid fraction and dendrite coherency calculations, much like the previous references discussed.

Figure 33. Effect if Ti and B content (in mass%) on the cooling curve for A356 alloy (Al-7% Si) [45]. (With permission from Springer.)

9. OTHER REFERENCES In addition to the references discussed in the previous sections, there are a number of other references that provide examples of the use of thermal analysis and calorimetry in quality control and assurance situations. These are listed at the end of the list of other references. They cover polymers [46-711, organic chemicals [72-731, foods [74-761, inorganic chemicals [77] and metals [78-811. Even this list of references is not nearly a complete bibliography of the use of these techniques in this area.

10. FUTURE OPPORTUNITIES Potential applications of thermal analysis and calorimetry to quality control is not limited in any way to those discussed in this chapter. Once some physical or chemical characteristic of a material or process is known and can be examined and/or characterized by these techniques, it is only the imagination that limits the possibilities for quality control applications. Both traditional techniques (DSC, TG, DMA, isothermal calorimetry, etc.) and non-traditional techniques (temperature modeling, etc.) have been shown to have potential uses for quality control. With the introduction of many new techniques (fast scanning DSC, sample controlled thermal analysis (SCTA), modulated and other temperature programmed techniques, etc.), many more new opportunities will arise for providing quality control tools. 11. REFERENCES

1. P. H. Willocks, Thermochim. Acta, 256 (1995) 1. 2. M. J. Richardson, J. Therm. Anal. Cal., 56 (1999) 1401. 3. S. M. Sarge, W. Hemminger, E. Gmelin, G. W. H. Hohne, H. K. Cammenga, W. Eysel, J. Therm. Anal. Cal., 49 (1997) 1125. 4. R. Castelli, M. Deplano and P. M. Rota, J. Therm. Anal., 47 (1996) 117. 5. H. Mohler, Kunststoffe, 84(6) (1994) 736. 6. A. M. Duske, Wire J. Int., 29(9) (1996) 64. 7. A. M. Duske, Wire J. Int., 31(3) 1998) 96. 8. A. Riga, Termochim. Acta, 357-358 (2000) 217. 9. S. Knappe and C. Mayo, Kunststoffe, 85(12) (1995) 2066. 10. S. Kanppe and R. Oemichen, British Plast. Rubber, May (1995) 18. 11. G. A. Avalon and M. A Bradshaw, Paper Film & Foil Converter, 77(2) (2003) CL18. 12. Adhesives Age, 38(11) (1995) 18. 13. F. Carrasco, P. Pages, J. Lacorte and K. Briceno, J. Appl. Polym. Sci., 98 (2005) 1524. 14. S. Sauerbrunn, N. Jing and R. Riesen, Am. Lab., 34(16) (2002) 15. 15. F. Stephan, A. Fit and X. Duteurtre, Polym. Eng. Sci., 37(2) (1997) 436. 16. C-R. Moon, B-R. Bang, W-J. Choi, G-H. Kang and S-Y. Park, Poly. Testing, 24 (2005) 376. 17.A. Licea-Claverie and F. J. U. Carrillo, Polym. Testing, 16 (1997) 445. 18. R. M. Rush, Thermochim. Acta, 2 12 (1992) 5 1. 19. A. T. Riga, Thermochim. Acta, 2 12 (1992) 227. 20. P. L. Stricklin, G. H. Patterson and A. T. Riga, Thermochim. Acta, 243 (1994) 20 1.

21. G. H. Patterson, A. T. Riga and B. M. Bomback, Thermochim. Acta, 243 (1994) 277. 22. M. Wesolowski, M. Czenvonka and P. Konieczynski, Thermochim. Acta, 323 (1998) 159. 23. J-F. Masson and S. Bundalo-Perc, Thermochim. Acta, 436 (2005) 35. 24. D. Giron, J. Therm. Anal. Cal., 56 (1999) 1285. 25. S. Roy, A. T. Riga and K. S. Alexander, Thermochim. Acta, 392-3 (2002) 399. 26. R. 0. Macedo, A. G. de Souza and A. M. C. Macedo, J. Therm. Anal., 49(2) (1997) 937. 27. R. 0. Macedo, C. F. S. Aragao, T. G. do Nascimento and A. M. C. Macedo, J. Therm. Anal. Cal., 56 (1999) 1323. 28. R. Macedo, T. G. do Nascimento, C. F. S. Aragao and A. P. B. Gomes, J. Therm. Anal. Cal., 59 (2000) 657. 29. A. Kovacs, I. Csoka, M. Konya, E. Csanyi, A. Feher and I. Eros, J. Therm, Anal. Cal., 82 (2005) 491. 30. K. Simon and R. Suverkrup, Thermochim. Acta, 248 (1995) 271. 3 1. U. von Stockar, P. Duboc, L. Menoud and I. W. Marison, Thermochim. Acta, 300 (1997) 225. 32. L. Menoud, PhD thesis No. 1480, Swiss Federal Institute of Technology, Lausanne, 1996. 33.A, Raemy, J. Therm. Anal. Cal., 71 (2003) 273. 34. J. Farkas and C. Mohacsi-Farkas, J. Therm. Anal., 47 (1996) 1787. 35. M. Konya, M. Sorrenti, F. Ferrari, S. Rossi, I. Csoka, C. Caramella, G. Bettinetti and I. Eros, J. Therm. Anal. Cal., 73 (2003) 623. 36. J. M. Resa and T. Quintana, J. Therm. Anal., 41 (1994) 1667. 37. J. A. Amorim, S. A. Elizario, D. S. Gouveia, A. S. M. Simoes, J. C. 0. Santos, M. M. Conceicao, A. G. Souza and M. F. S. Trindade, J. Therm Anal. Cal., 75 (2004) 393. 38. M. M. Conceicao, A. M. L. Melo, N. Narain, I. M. G. Santos and A. G. Souza, J. Therm. Anal. Cal., 67 (2002) 373. 39. D. N. Njie, T. R. Ramsey and R. P. Singh, J. Food Eng., 37 (1998) 63. 40. A. A. Gardea, E. Carvajal-Millan, I. Higuera-Ciapara, C. Figueroa, J. Molina-Corral, A. Rascon, A. Orozco and A. Inda, Thermochim. Acta, 394 (2002) 179. 41. J. Dweck and E. I. P. Lasota, Thermochim. Acta, 3 18 (1998) 137. 42. M. Alonso, A. Gonzalez and J. A. De Saja, Thermochim. Acta, 184 (1991) 125. 43. A. Dioszegi and I. L. Svensson, Mat. Sci. Eng. A, 413-414 (2005) 474. 44. S. G. Shabestari and M. Malekan, Canadian Metallurgical Quarterly, 44(3) (2005) 305.

45. D. Emadi, L. V. Whiting, S. Nafisi and R. Ghomashchi, J. Therm. Anal. Cal., 81 (2005) 235. 46. J. Parker, M. J. Rich and A. R. Wolfe, Int. SAMPE Symp. Exhib., 50 (2005) 199. 47. B. R. Whiteside, M. T. Martyn and P. D Carter, Int. Polym. Process., 20(2) (2005) 162. 48. Anon., Brit, Plast. Rubb., Feb. 2005, p. 7. 49. Anon. Plast. Rubb. Asia, 19(123) (2004) 48(1). 50. T. N. Prudskova, L. N. Gurovich, S. V. Arenina and A. L. Kachalina, Int. Polym. Sci. Tech., 3 l(8) (2004) Tl61. 51. A. Frick, A. Rochman and C. Stem, Kunstst. Plast. Eur., 94(8) (2004) 132. 52. D. S. Chaudhary, M. C. Jollands and F. Cser, Polym. Polym. Compos., 12(5) (2004) 3 83. 53. M. Schmid, A. Ritter and S. Affolter, Polimery, (2004) 333. 54. S. Cook, S. Groves and A. J. Tinker, J. Rubb. Research, 6(3) (2003) 121. 55. S. Agren, S. Karlsson and F. Strangenberg, Recycl. Reuse Waste Mater. Proc. Int. Symp., (2003) 325. 56. S. C. Shit, Popular Plast. And Packag., 48(2) (2003) 55. 57. M. Wacker and G. W. Ehrenstein, SPE 2002 Paper 395, Session T33 Thermoset, Advances in Thermoset Mat'ls., p. 5. 58. A. Frick, Kunstst. Plast. Eur., 91(8) (2001) 63, 149. 59. H. W. Kramer, Medical Device Technology, 12(9) (2001) 16. 60. A. Sanchez-Blazquez, Revista de Plasticos Mod4ernos, 81 (538) (2001) 46 1. 61. A. B. Kummer, RAPRA 2000 May 3-5,2000, p. 25. 62. W. A. Wortman, K. P. Beardsley and J. D Caldwell, RAPRA 1999 April 13-16, 1999, Paper 5, p. 33. 63. F. Hoffman, Kunstst. Plast. Eur. 89(8) (1999) 27. 64. A. Sass, Kunstst. Plast. Eur., 88(2) (1998) 34. 65. R. Brummer, F. Hetzel, C. Kimmerl, V. Lenuck, J. W. Kenndoff and R. Knieler, Adhasion Klebe & Dichten, 42(3) (1998) 26. 66. E. Rudnick, Z. Doblowski and A. Winiarska, Polim. Tworz. Wielk., 42(5) (1997) 327. 67. T. Nitschke and B. Benzler, Kaut. U. Gummi Kunstst., 50(6) (1997) 492. 68. A. Prasad and D. Mowery, Materials Annual Technical Conference, ANTEC Conf. Proceedings, V.2 1997 SPE, p. 23 10. 69. M. P, Divito and I. M. Hamadeh, Modern Plastics Encyclopedia, 1996, p. E70. 70. H. Deckmann and V. Peterson, Kaut. U. Gummi Kunstst., 48(4) (1995) 287.

71. E. Ramirez, B. M. Huerta and A. Ledezma, Ind. D. Gomma, 39(11) (1994) 43. 72. F. Piccinini, CFI Cream. Forum Int., 82(12) (2005) E28. 73. M. J. Pohlen, Tribotest, 6(3) (2000) 241. 74. R. C. Rao and M. C. Wang, Bioeng. Proc. Northeast Conf., (2005) 108. 75. D. B. Bechtel, 0. K. Chung, R. A Flores and Y. S. Kim, J. Food Process. Eng., 26(5) (2003) 469. 76. M. Oziemblowski and T. Trziszka, Proc. SPIE Int. Soc. Opt. Eng., 4410 (2002) 43. 77. D. J. Broton, L. Graf, L. M. Hills, M. Morrison and E. Shkolnik, World Cem., 34(11) (2003) 179. 78. R. Sitlen, Foundrymen, 96(10) (2003) 245. 79. L. Gomez-Bejar, S. D. De la Torre and F. Espinoza-Magana, J. Metastab. Nanocryst. Mater., 15-16 (2003) 507. 80. M. K. Saxena, S. B. Deb, K. L. Ramakumar and V. Venugopal, Indian J. Chem. Tech., July 2002, p. 324. 81. Y. X. Li and Q. Wang, Cailiao Kexue yu Gongyi, 9(2) (2001) 215.

INDEX A absorption, 183 AC calorimetry, 49,67, 332 accelerating rate calorimeters (ARC), 33 ACE inhibitors, 672 acetamide, 602 acetaminophen, 620 acetonitrile, 397, 399,400 acetylene adsorption, 421 acid-base properties of catalyst surfaces, 401 acidic solids, 41 8 acoustic load impedance, 144 - 146 acoustical impedances, 153 acoustical phase shift, 145 activation energies of desorption, 389 activation energy, 447,456,473,482, 487,503,504,508 - 51 1,515 - 522, 524,525,527 - 529,531,535 activation entropy, 229,253 active carbon, 345,375 active pharmaceutical ingredient, 598, 602 adamantane as calibrant, 277 Adam-Gibbs (AG) theory, 234 additives, 593,594 adhesives, 700 adiabatic calorimeters, 29,30,32, 33, 38 - 40,42,43,50,547,551,552 admittance, 2 17,2 18 adsorbate - adsorbate interactions, 387 adsorbate, 345,346,355,356,360,361, 363,372,374,382 adsorbate-adsorbent interactions, 356, 382 adsorbents with deposited proteins, 345 adsorption calorimetry, 387,388,390, 395,396,401,405,413,414,416,417, 421,422,425,427,430 adsorption capacity, 344, 345,35 1, 356, 363,372 adsorption complexes, 397 adsorption energy, 345,346,362 adsorption isotherms, 389 adsorption model, 353

adsorption of ammonia, 124, 125 adsorption of C02,396,409,410,412 adsorption of oxygen, 422 adsorption, 94,124 - 130 adsorption-desorptionnitrogen isotherms, 363 Ag/A1203catalysts, 416 ageing of polymorphs, 614 ageing process, 684,685, 688 air pollutants, 4 14 albumins, 687,690 Al-Cu-Mg-Mn alloys, 446 algorithms for fractal coefficients, 358 alkaline-earth cuprates, 462 alkaline-earth oxides, 410 alloys, 11,299, 300,302,303,311,312,315, 316,334,440,445,447,448,490,493,494, 725 - 728 alumina, 345,371,404,408,409,411,413 418,422,425,430 y-alumina, 106 aluminium foil for full sample encapsulation, 277 aluminium powder, 463 alunite, 179 - 181 AlZnMg alloy, 449 ammonia as catalytic probe, 397,4 11,4 13, 428 ammonium hydrogen carbonate, 45 1 ammonium jarosite, 478 ammonium nitrate (AN), 454, 532, 533 ammoxidation, 4 15 amorphism, 622 arnorphization, 7,269 amorphous phases, 514,598,615 amphoteric oxides, 396,409 - 41 1,418 ampicillin trihydrate, 609 aneroid calorimeters, 545 anion exchange, 450,491 annealing, 7,269,447,449,453 anthranilic acid, 600,602, 603 anti-aromaticity, 562, 563 anti-inflammatory drugs, 672 antimicrobial activity, 492 antisymmetric stretching vibration, 187 anti-tumour drugs, 488 aortic valve, 686, 687 appearance energy, 555,556 aragonite, 612

archwires, 641,643,647 argentojarosite, 478 Argonne rotating bomb calorimeter, 30 Arrhenius equation, 447, 5 14, 519, 520 Arrhenius parameters, 504, 532 Arrhenius plot, 5 10, 5 13, 5 14,589 Arrhenius-type temperature dependence, 225,237,240,244 artificial photosynthesis, 158 ash content, 700, 7 19 asphaltene, 579, 580, 591 aspirin, 602,603, 617 ASTM, 13 atom-atom potentials, 622 atomic force microscopy (AFM), 6, 56 61,65,67,84,274,358,367 - 369,381 - 384 atomic layer deposition, 41 5 atomic oxygen, 453,460 atomic polarization, 210 atomization, 562 Auger electron spectroscopy, 200 austenite, 63 1 automatic sorption apparatuses, 347 automotive elastomeric seals, 700 automotive exhaust catalysts, 116 automotive exhaust gases, 121 average pore radius, 350 Avrami equation, 447 Avrami-Erofeev (AE) equation, 620 Avrami-Erofeev model, 533 azone, 669 - 671 azoxyanisole as calibrant, 277 - 279

B bacteria, 157, 158, 164, 173 bacterial arthritis, 689 band component analysis, 201 band fitting, 201 barbital, 620 barium aluminates, 111 barium carbonate, 111, 121,452,453, 465 barium cerates, 111 barium stannate, 465 barium-zirconium diorthophosphate, 465

Barrett-Joyner-Holend (BJH) method, 364, 378 baseline calibrations, 65 1 basic electrical quantities, 308, 328 BaTiO-,, 441 battery calorimeter, 36 benzene, 399,400 benzoic acid, 544, 547 - 549,557,558,711 Berthelot calorimeter, 30,38 P-blockers, 672 p-relaxation, 229,230,237 bimetallic sensors, 6 1 binaphthol, 622,623 binary and ternary copper catalysts, 423 Bingham and Casson plastic model, 589 Bingham fluid, 589 bioavailability, 600,604 biochemical systems, 566 biocompatibility, 647 biodegradable polymers, 676 bioelectronics, 158 biofilms, 656 bioinorganic complexes, 469 biomacromolecules, 154, 157 biomedical polyurethanes, 655 biomembranes, 159 biomimetic materials, 158 biomolecular reactions, 157 biopolymers, 157 bioprosthetic heart valves, 687 bioreactors, 714 biosensors, 156, 157, 158 birefringent materials, 58 1 bismuth halides, 469 bismuth oxalate, 459,460 bismuth phosphonates, 469 bismuth titanate, 46 1 bitumen, 706, 707, 708 black-body radiators, 173, 174, 175, 178, 317 Boltzmann superposition principle, 21 1 bomb-calorimeter, 543 bond additivity corrections, 562 bond dissociation energy, 16, 17,21,25,554, 555,559,563 bond dissociation enthalpy, 555 - 557 bone plates, 684 borates, 450 boron oxide, 522

bound water, 349,668, 687,688 British Laboratory of the Government Chemist, 68 broadband dielectric spectroscopy, 217 bromine species, 566 Brijnsted acid sites, 391, 397, 402,404, 416 Brunauer-Emmett-Teller (BET) method, 364,367 bubble formation, 608 buckminsterfullerene, 563 Budrugeac kinetic method, 507, 537 bulk moduli, 323 Burger and Ramberger rules of polymorhism, 606 Butterworth-Van Dyke equivalent circuit, 143 n-butylamine, 399,409 C C60,563,564 CaC204.2H20,186,187,191 CaC204.H20,187,201,202 cadmium-naphthalenedicarboxylate, 469 caesium nitrate, 456 caffeine, 7 16 calcification, 686 calcination, 106, 107, 126,202,405 calcite, 6 12 calcium carbonate, 106, 107, 111, 187, 188, 192,194,201,522,523,612 calcium oxalate, 187, 188, 192,201, 525 calcium phosphate precipitates, 656 calcium sulfate, 723 calibrants, 277 calibration for thermal lag, 277 calibration in cooling, 277 calibration materials, 68, 150 calibration matrices, 278 calibration of gases, 95 calibration of instrumentation, 697 calibration of liquids, 99 calibration of pyrometers, 3 17 calibration of spectrometric signals, 95, 183

calibration of the calorimetric system, 547, 551 calixarenes, 609 calorimetry, 440,446,448,453,454,465, 468,472,473,475,483,492 Calvet calorimeters, 552 capillary condensation, 352,360,363, 372 capillary configurations, 378 capillary transfer line, 80 carbamazepine, 602,620 carbamazepine-benzoquinone, 602 carbohydrates, 716,720 carbon fibres, 76, 703 carbon nanotubes, 349,485 carbonates, 45 1 cartilage destruction by septic arthritis, 689 cassava, 72 1 casting, 300 catalysis, 7,9,94, 118, 387,388,390,395, 414,426,430 catalyst supports, 408, 417 catalysts used in environmental protection, 344 catalysts, 172, 173, 202,445,488 catalytic activity, 395,402,403,407,412, 413,416,423,425 - 430 catalytic applications of transient metal sulfides, 566 catalytic conversion of 4-methylpentan-2-01, 427 catalytic reaction of 2-propanol conversion, 41 1 catalytic selectivity, 394,395,403,407,425 430 cavity perturbation technique, 454 CBS models, 561 C-C bond scission, 525 CdTi03, 470 cells, 157 - 159 cellular biology, 61, 82 Ce02,410, 415,419-422 cephadrine dihydrate, 609 ceramic pigments, 462 ceramics, 76,344,445,461 ceramides, 670 ceria-lanthana catalysts, 4 17 ceria-zirconia catalysts, 4 17 cerium oxide, 116

cesium hexafluoroarsenate, 449 chain folding, 272 chalcogenides, 566 channel hydrates, 610 cheese making, 7 19 chemical flames, 299 chemical heterogeneity, 345 chemical sensors, 156 chemical vapour deposition, 491 chemisorption, 395,398,401,421,423, 426 chemometric analysis, 705 chemorheology, 701 chip calorimetry, 275,276,288 cholesterol, 566,670, 671,678 cholic acid, 609 chromates, 453 chromatography, 347,349 chromia, 4 17 chromium steels, 449 Circular 502, 558 classification of pulse-heating systems, 302 classification of resin composites, 657 clathrates, 473 Clausius-Clapeyron equation, 389, 552 clay nanocomposite, 522 clay, 198, 199,202,203,205,211,418 clinoptylolite, 350 cloud point, 579,586, 593, 594 cluster oxalate complexes, 460 CO adsorption, 410,412,421,422,423 CO disproportionation, 123 CO stretching vibration, 186 CO symmetric stretching mode, 187 coatings, 270, 294 Coats and Redfern kinetic method, 506 cobalt amino complex, 111 cobalt oxalate, 111,458 cobalt phosphonates, 471 cobalt(I1) sulfate(V1) hydrate, 467 cocoa butter, 6 12 co-crystallization, 287,288 CODATA, 559 coefficient of thermal expansion, 698 coke formation, 403 cold crystallization, 7,269 Cole-Cole plot, 21 5

Cole-Davidson plot, 21 5 collagen denaturation, 689 collagen-elastin matrix, 686,687 combustion calorimetry, 542 - 545,547,549, 548,550,557,563 combustion, 108, 109, 121, 123 compensation effect, 532 complex formation between aspartic acid and the bismuth ion, 469 complex of nitrilotriacetic acid and bismuth trichloride, 469 compliance, 8, 146 - 148, 150, 153, 155, 156, 161, 164 components in a pulse-heating experiment, 305 composites, 684, 698, 703 compressibilities, 323 computational methods, 599,601 computational thermochemistry, 561, 562 computer aided quality (CAQ) system, 700 computer hard discs, 344 computer-aided cooling curve analysis, 727 conductivity measurements, 447,468,469 congealing temperature, 593 constancy of the scan rate, 282 constant rate cooling, 507 constant rate thermal analysis, 389 constant-volume heat capacities, 548 contact resistances, 309 contact thermometry, 305,316 controlled-rate thermal analysis (CRTA), 17, 460 conversion function, 504 COO, 108,111,118,119 cooling rates, 271,276,277,282,284 - 287 cooperativity, 225,238 - 240 coordination chemistry of manganese, 486 coordination compounds, 10,439,441 coordination polymers, 460,469,470,472, 476,479,480,490 copper acetate complexes, 476 copper carbonate, 44 1 copper sulfate, 466,468 copper(1) halide-alkali-metal halide systems, 477 copper(1) 8-diketonates, 474 copper(I1) carbonate hydroxides, 45 1 copper(I1) oxalate, 460

copper(I1) trans-l,4-cyclohexane dicarboxylate, 473,474 copper-nickel-phosphorusalloys, 446 copper-silver alloy, 490 corn, 719 coronary stents, 680,682,683 correction factors for indium, 443 correction factors, 280,28 1,284 correction of crystallization peak temperatures, 285 correction tables, 280 correlation gas chromatography, 553, 560,563,565 corrosion, 154,35 1 coupled techniques, 95,598,611,612 cow's milk, 719, 720 cracking, 350 creams, 7 19 creep compliance, 647 criteria for nanotechnology, 344 critical parameters, 300, 307, 322 critical point measurements, 304,323 critical temperature, 5 18, 5 19 crosslinking, 273, 503, 526, 527 crossover region, 236 crucible type, 7 11 crude oils, 8,579 - 595 crystal engineering, 469,472,621 crystallinity, 270,284,291,292,700 crystallization kinetics, 704 crystallization of polymer melts and glasses, 503 crystallization, 440,447,450,461,462, 467,470,487, 512, 579, 581, 582,584, 586,587,589,591, 593, 594, 599,601, 602,612,613,614,619,649,651,652 crystallographicthermal displacement parameters, 6 10 crystallographicallyequivalent voids, 609 crystal-seeding, 600 CsZnP04, 464 CTAB, 673 CuIZn nanocomposites, 462 CuAlNiMn alloy, 446 cubane, 563 cuneane, 563 cuprate superconductor, 110

Curie temperature, 223,35 1,5 18 Curie-Weiss law, 481 curing, 155,156,165,503,526 - 528,531, 699 - 701 cyclic voltammetry, 556 cyclobutadiene, 562 cyclobutene, 563 cyclodextrins, 609 cyclohexane, 400 cyclo-metallated compounds, 488

D damping, 144, 148, 156, 165 daropidine, 71 1 DDNAP, 671,672 de-alumination, 403,404 deamination, 476 Debye model, 211,213 - 215,218,219,236, 245 ,246,248,257,258 Debye temperature, 3 15 decarbonisation, 172 decarboxylation, 476 decationization, 405 decompositions of oxalates, 118 decompositions of solids, 93, 130 deconvolution of overlapping peaks, 6 14 deep-cooling accessory, 284 definition of the field of Thermal Analysis, 15 degradation of polymers, 503,525 degradation, 294 degree of crystallinity, 676,679,680,684 degree of surface coverage, 352 dehydration, 172, 189, 191, 192,201,202 dehydrogenation,417,427,430 dehydroxylation, 172,203,405,478 density function, 353 density measurement, 58 1 density, 300,304,306,311,312, 314, 315, 317,323,325 dental alloys, 650 dental materials, 12, 63 1 dentures, 650 deoxycholic acid, 600,609 Derivatograph, 35 1 desensitizing, 454 desolvation, 599, 606,608,610,620 - 624 desorption distribution, 353

desorption energy, 352 - 354,361 - 363, 365,373,374 desorption of water, 362 desorption rate, 352 desorption studies, 9,388 determination of the energy profile, 387 devitrification, 69, 70 Dewar calorimeter, 30 DFT methods, 561 diabetes mellitus, 689 diamond content in grinding tools, 108 Diamond DSC of PerkinElmer, 277 diamond, 59,6 1 diastereomers, 471 diathermal calorimeters, 29, 34, 35 dielectric analysis, 700 dielectric materials, 209 dielectric properties, 441,454 dielectric relaxation spectroscopy, 213, 216,222,223,225,226,241,250,259, 260 dielectric strength, 230,249,25 1 dielectric thermal analysis (DEA), 16,17,700,701,724 diesel fuels, 583,587 differential ebulliometric method, 553 differential heat of adsorption, 387, 392, 413,418,421 differential scanning calorimeters, 269 271,275 - 282,284,286,288,291,293, 55 1,552,560,631 - 644,647 - 650,654 - 658 differential scanning calorimetry (DSC), 7-12, 15, 17, 19,20,27,29,34,46,49, 51, 55,67,68,71,75, 172,310,329, 330,388,426,439,440,442 - 491,509, 516, 518,520, 527,528, 532, 598,600, 602 - 625,663,665 - 667,671 - 680, 682 - 690,697 - 704,709,711 - 721, 723,724,727,729 differential thermal analysis (DTA), 9, 11,15,17,51,95,108,109,113,118,

124, 172,388,416,439,443,446,449, 450,455,458,460 - 466,468,470,471, 476,477,482,484 - 487,492,669, 679,684,697,703,724 diffuse reflectance spectra, 4 16

diffusion coefficients, 352,355,371,380, 381,384 diffusion, 388,392,395,397, 509,510, 511, 513,514,527,532 diglycidyl ethers, 526, 528 dilatometers, 3 11 dimethylated P-cyclodextrin, 601 2,6-dimethylpyridine,397,412,417 2,4-dinitroaniline, 528 dinuclear metal carboxylates, 473 dip coating, 152 dipotassiurn tetraborate tetrahydrate, 450 direct calorimetric methods, 552 Dirichlet tessellations, 45 1 dispersive IR spectroscopy, 82 dissipation, 143, 158, 159 dissociative adsorption, 408 distinguishing between physical and chemical adsorption, 391 distorted linear heating, 507 Division-of-AmplitudePhotopolarimeter (DOAP), 321,326 DNA, 157,159 documentation, 696 Dollimore-Heal method, 377 doped oxides, 41 1 doping of catalysts, 41 1 double helical coordination metal compounds, 480 Doyle's approximation, 506 draw ratio, 679 DRIFT spectra, 458, 609 driving circuits, 147, 148 drop calorimetry, 465,552,560 drop coating, 152 drug delivery, 12,663,665,672,673,675, 680,681,690,713 drug hydrate analysis, 598 drug penetration, 12,663,665,667,668,674, 690 drug, 598,599,601,602,604,606,609,616 621,624,625 dry reforming, 429 drying, 151,155,165,721 dynamic mechanical analysis (DMA), 15 - 17, 20,67,657,658,697,699,700,716,724,729 dynamic oxygen exchange capacity, 117 dynamic pulse calorimetry, 299 - 301

dynamic resistance, 144 E effective activation energy, 5 11 , s 17, 524,525 elastic modulus, 63 1, 64 1, 654 elastic scattering, 182 elastomeric impression materials, 650, 654 elastomers, 698, 699 electric displacement, 2 10,2 11,212 electrical double-layer, 153 electrical impedance tomography, 66 electrical impedance, 144 electrical resistivity measurements, 644 electrical resistivity, 300, 307, 31 1, 313, 328,329,333 electrical techniques, 8 electrochemical cells, 153 electrochemical deposition, 152 electrochemistry, 141, 149, 153, 158 electron affinities, 554 electron microscopy, 55,349,358 electron paramagnetic resonance (EPR), 456 electron photodetachment spectroscopy, 556 Electron Spectroscopy for Chemical Analysis (ESCA), 200 electron transfer, 556 electronlneutronheating, 299 electron-electron interactions, 3 15 electronic materials, 76 electronic polarization, 210,218 electron-phonon interactions, 3 15 electropolymerization, 158 Eley-Rideal type mechanism, 129 ellipsometry, 3 19, 321, 322 emanation thermal analysis (ETA), 16, 18,484 emission spectra, 173, 175, 177, 192 emissivity, 174 emittance, 174, 175, 178, 300, 307,317 - 322,325 - 327,329,331 - 333 emulsifiers, 717 emulsions, 713,719 enantiotropic forms, 71 1 enantiotropy and monotropy, 598,605

enclathration selectivity, 623 endodontic instruments, 631,632,636,641 endodontics, 632 energetic heterogeneity, 345,352, 361,372, 373,402 energy barrier, 509,512, 513,515,517,521 energy dispersive X-ray spectroscopy (EDX), 194, 358,367, energy distribution function, 352 energy equivalent of the calorimeter, 547 energy of dilution, 548 energy-temperaturediagram, 606,617 - 619 enthalpy of adsorption, 9, 389,391 enthalpy of combustion, 549,550, 559 enthalpy of desolvation, 610 enthalpy of dilution, 450 enthalpy of dissolution, 605 enthalpy of formation of aqueous sulfuric acid, 548 enthalpy of formation, 549, 542, 544, 548, 550,554 - 557,559,562 - 567 enthalpy of fusion, 55 1,606,617,619 enthalpy of mixing, 565 enthalpy of precipitation, 583 enthalpy of solution, 598,606 enthalpy of sublimation, 551,552, 564 enthalpy of transition, 598,604,605,606, 608,612,618,620 enthalpy of vaporization of water, 548 enthalpy of vaporization, 548,551 - 553,565 entropy coefficient, 374 entropy factor, 352 entropy of adsorption, 393 entropy, 441,449,481,487,492,542 environmental applications, 414 environmental scanning electron microscope (ESEM), 188 environmental stability, 700 EPON828,241 epoxidation, 111, 125 epoxy resin, 155 equation-of-state (EOS) parameters, 300,323 equilibrium constant, 542,543 equine pericardiurn, 686 errors in combustion calorimetry, 549 errors in enthalpy measurements, 554 erythrocyte membranes, 689

estimation of kinetic and thermodynamic parameters, 353 ethosomes, 675 ethylene adsorption, 421,424,425 ethylene-1-octene copolymer, 273 ethylene-1-pentene copolymer, 293,294 ethylene-propylene copolymer, 273, 274,275 Eucken and Nernst's calorimeter, 30 Euclidean, nonporous, completely flat surfaces, 348 EuP04, 465 European Symposium on Thermal Analysis and Calorimetry (ESTAC), 3,4 evaporation of liquids, 99 evaporation of water, 100 evaporation, 599,622 evolved gas analysis (EGA), 16, 18,20, 439,440,455,461,490 excess heat capacity, 465 excipients, 74,75, 608, 713 explosives, 294,454,463,598 extent of adsorption, 394 extent of conversion, 504, 508,510, 51 1,529,532,535 extinction coefficient, 174,300,320 extrapolated onset temperature, 280 Eyring equation, 556 F fast cooling, 5 14 fast framing CCD-cameras, 3 12 fast pulse-heating technique, 489 fast scanning calorimetry, 7 fast scanning DSC, 729 Fe(II1) complexes with nicotinamide, 478 Fe203-Ti02solid acid catalysts, 41 8 Fe-Mn-Si-based alloys, 447 ferroelectric crystals, 223 ferroelectric properties, 450 ferroelectric transition, 44 1 ferromagnetic to paramagnetic transition, 5 18 fertilizer, 454 FeSO4.6H20,466 fibre coatings, 462

Fick's law of diffusion, 371 fillers, 703 film blowing, 286,287 film formation, 155, 158 film-thickness monitors, 152 filter plugging, 594 finite-element analysis (FEA), 3 18,331 first law of thermodynamics, 545 first-order model, 530, 533 Fischer-Tropsch reactions, 118, 120,429 fissionlfusion, 299 fixed-point cells, 3 17 flame retardants, 699 flame spray pyrolysis, 452 flash point, 706 flowing afterglow, 556 fluid cracking catalysts, 406,407 fluorescence, 183 fluorides, 111 Flynn-Wall-Ozawa kinetic method, 471,506, 508 food, 76,612,695,715 - 721 For Better Thermal Analysis and Calorimetry, 25 forbidden Raman bands, 188 forensic science, 61,294 fossil fuels, 579 Fourier method, 728 Fourier spectrum integral method, 358,359 Fourier transform infrared emission spectroscopy (FTIR ES), 172, 173, 176, 177 Fourier transform infrared spectroscopy (FTIR), 7,93 - 101,104 - 106,110,129,388, 410,412,413,416,443,445,458,460,464, 465,471,473,487,490,613 Fourier transform techniques, 220 fractal coefficients, 352, 358, 367,381,382, 384 fractal dimension, 348, 355,357 - 359, 377, 381,383 fractal geometry, 348 fractal objects, 348 fragility, 447 fragmentation, 96, 101, 103 - 105 framework solids with chiral structures, 469 free fatty acids, 670 freezing point of silver, 3 16 frequency counter, 147, 162

Fresnel equations, 321 Friedman kinetic method, 467,505 508,515,537 fructose dehydration, 429 fruit, 721 fuels, 706 fullerenes, 344 2-furancarboxylic acid, 550 G Ga203,410,414 Ga-Ge-Y alloys, 448 gallium nitrate hydrate, 455 gas - solid reactions, 7,93,94, 104, 113, 114,118,120,123,127,130 gas calorimeter, 560 gas chromatography, 556,557,560, 563.581 gas ionization, 102 gas pulse device, 94 gas storage, 491 Gaussian n-theories, 559 GC-MS, 78,79,80,81 gelatinization of starch, 716 gelling temperature, 588 gelling, 587, 594 geometrical heterogeneity, 345,372 Gibbs free energy, 512, 542, 556,604, 606 Gibbs-Thomson relation, 274 glass fibres, 703 glass transition, 69,70, 83, 150,333, 447,450,503,514,520 - 522,527,698, 699,704,709,716 glasses, 445 glass-transition dynamics, 224,249 251,253,259 glass-transition temperature, 225,234, 237,249,259,582,647 glassy polymer, 150 glassy state, 226 glucose, 522 glycerol, 673 glycine, 600,601,620 graphene, 349,364 graphite supports, 423 graphite, 550 gravimetric and Sauerbrey masses, 151

gravimetry, 133,349 gravitational collapse of a liquid, 302 grey-body, 174 Group 6 metals, 566 growth of nuclei, 594 Griineisen parameters, 323 guest exchange, 61 1,625 guest molecules, 609,610,621,623 guest-loss processes, 608 guided ion beam mass spectrometry, 556, 566 Guinier-Preston (GP) zones, 449

H HACOT, 458 haemoglobin, 689 Hammett titrations, 401 Hamon's transformation, 220 hardness, 35 1 Havriliak-Negami, 21 5,236,250,25 1 HDPE, 291,292 head-space analysis, 553 heart-valve poppets, 677, 690 heat capacity of the system, 551 heat capacity, 270,291,292,300,306,310, 311,315,326, 329, 332,333,444,449,450, 464,465,481,490, 520, 527,582,583,620 heat conduction calorimetry, 162 heat loss corrections, 302 heat of adsorption, 387,388,391,393,397, 399,413,423,524 heat of fusion rule, 606, 6 19 heat of fusion, 300,310, 333,513,516 heat of solidification, 3 10 heat processing conditions, 717 heat transport, 300 heat-compensating principle, 46 heat-conduction calorimeters, 545, 552 heat-flow sensors, 152, 162 heat-flowmeter calorimeters, 34,42,44 heat-flux DSCs, 282 heating stage scanning electron microscopy, 172 heating stage spectroscopy, 171 heating stage transmission electron microscopy, 172 heating stage X-ray diffraction, 172 heating stage X-ray photoelectron spectroscopy, 172

hectorite, 203,204,205 Hedvall effect, 45 1 hemispherical emittance, 300,302,322, 332 Hess' law, 544,550 heterogeneity of surfaces, 345,346 heterolysis, 556 heteropolyacids, 401 heteropolyanions, 41 8 hetero-poly-compounds,449 high heating rates, 271,277 high performance differential scanning calorimetry (HPer DSC), 7,269,270, 275,277,278,281,282,283,284,286, 288,290,291,293,294,295,443 high-density polyethylene, 291 high-pressure mass spectrometry, 556 high-resolution TG, 707 high-speed calorimetry, 7,269,270, 275 - 278,282 high-speed DSC techniques, 598 high-speed scanning microcalorimeter, 332 high-temperature thermometry, 320 high-throughput experimentation, 294 high-throughput screening, 60 1 historical development of pulse-heating, 301 history of thermochemistry, 543,544 Hoffman - Lauritzen theory, 5 16 homogeneous surfaces, 345,346 homolysis, 543, 555, 556 horny cell membranes, 665 host molecules, 609,609 host-guest stoichiometry, 608 hot crystallization, 7, 269 hot stage microscopy (HSM), 603,608, 609,611 - 613,617 - 619 hybrid composites, 657 hydrates of d-metal nitrates, 455 hydrates, 709 hydrides, 447 hydrocracking catalysts (HC), 430 hydrocracking, 350 hydro-demetallation (HDM), 430 hydro-denitrogenation (HDN), 430 hydro-desulfurization (HDS), 430

hydrogen adsorption, 391,398,399,401,409, 421,423 - 425,429,43 1 hydrogen bond distances, 181 hydroperoxy and peroxy structures, 525 hydrotalcites, 401, 41 8 hydrotreating, 430 hydroxyapatite, 683 hydroxyprocaine hydrochloride, 6 18 hygroscopicity, 709 hyperconjugation, 565 Hyper DSC, 277,284,295 hyphenated techniques, 439 hysteresis, 357, 363,378 I ibuprofen, 672 ice calorimeter, 34,37, 38, 39,41 ICTA, 13,22,24,51,52 ICTAC Kinetics Project, 503 ICTAC Nomenclature Committee, 14,23, 24,51 ICTAC, 557,575 image contrast enhancement, 66 image processing, 63,64,65 immunosensors, 157 impedance analysis, 148 impedance, 2 17,2 18 implants, 12 implants, 663,676,677,680,683,684, 685,690 indium as calibrant, 277,278,280,281 indium oxide (111203) catalysts, 425 inductance, 306,309, 310 infrared emission spectroscopy (IES), 171, 172, 173, 176, 180, 181, 182, 191, 192, 194, 195.196.478 , , infrared spectroscopy, 171, 173, 181, 182, 183 infrared thermography, 57 initiation, 525, 528 injection-mouldedproduct, 284 InN, 111 instantaneous nucleation model, 620 insulating materials, 508 integral methods, 505,506, 507 integrating sphere reflectometry, 3 18 intercalation, 450,473,483 interferometers, 2 17

interferometry, 157,311, 323 intermetallic phases, 448, 725 intermolecular energies, 549, 55 1 internal energy of the system, 545 International Confederation of Thermal Analysis and Calorimetry (ICTAC), 3, 4,5,6, 10, 13, 14,23,24,25,26,51, 52,71 International Congress on Thermal Analysis and Calorimetry (ICTAC), 3 International Critical Tables, 558,573 International Standards Organization (ISO), 696 intervertebral disc degeneration, 688 intrinsic dissolution rate, 617 iodides, 453 ion associated hydrates, 6 10 ion cyclotron resonance mass spectrometry, 556, 566 ion exchange, 350 ionic conductivity, 701, 702 ionic liquids, 247,454 ionization in the mass spectrometer, 487 ionization potentials, 554, 556 iontophoresis, 673 iridium - expansion of, 314,325,326, 327,328,333,334,337 iridium, 325,326, 327,328,337,340 iron oxide, 4 12 isoconversional kinetics, 503,534 isoconversional methods of kinetic analysis, 10 isoconversional principle, 504 isodesmic reactions, 562, 563 isomeric compounds, 549,542 isomerization, 350,470 isoperibol calorimeters, 545, 546, 547 isoperibol, 30,41,43 - 47,5 1 isopropyl myristate, 673 isopropylamine, 389, 398, 399 isosteres, 389,429 isostructural crystals, 609 isostructurality, 11,597, 609 isotacticity, 6 13 isoteniscope, 553 isothermal calorimetry, 598 isothermal differential calorimetry, 27 isothermal gravimetry, 27

isothermal jacketed calorimeter, 550 isothermal microcalorimetry, 162 isothermal titration calorimetry (ITC), 663, 670 isothermal titration calorimetry, 12 isothermally jacketed calorimeter, 39 IUPAC, 13,22,24,46,53,557,573 IVD, 688 J JANAF Thermodynamic Tables, 559 jarosites, 179 Joule effect, 36,41,49, 50 Junkers' flame calorimeter, 35

K K20,414 KD2P04, 44 1 keratin, 667 Kerr cells, 3 12 keto-en01 tautomerism, 565 KH2P04,464 kinetic energy release distributions, 556 kinetic models, 623 kinetic predictions, 503, 529 - 531, 535 kinetic triplet, 504,529 kinetics of polymorphic transformation, 614, 620 kinetics, 9, 10, 503 Kirchhoff s law, 173, 174,318 Kissinger kinetic method, 471, 508, 515, 530, 53 1 Knudsen cell mass spectrometry, 556 Knudsen cells, 552, 560 Kohlrausch-Williams-Watts (KWW) model, 215 Kolrausch-Williams-Watts (KWW) relaxation function, 236 Kolsky bar, 330 L La2(C20& .10HzO,458 La203,410,414,418,420,422,430 lactoperoxydase system (LPS) activation, 722 La-Mn complex, 460 Landau theory of phase transitions, 5 18 Landbolt-Bijrnstein, 558, 573 Langmuir-Blodgett monolayer films, 158

Langmuir-Hinshelwoodtype mechanism, 129 lanthanide benzenedicarboxylates, 479 lanthanides, 478 lanthanum citrate trihydrate, 480 lanthanum hexaluminate, 462 lanthanum hydroxide (La(OH)3),482 lanthanum hydroxide oxide (LaOOH), 482 lard, 612 laser flash (LF) technique, 3 16,331 laser heating, 299 lateral force microscopy, 58 latexes, 9,344 lattice energies, 441,474, 616,622, 623 lattice heat capacity, 465 laurocaprarn, 670 LDPE, 273,286 - 288 lead as calibrant, 277,287 lead(I1) complexes, 482 lead-acid battery, 482 levetiracetam, 602 levitation, 299,300, 321,324,333,334 Lewis acid sites, 392, 397,402,404, 405,410,413,416,427 lifibrol, 617, 618 Lightspeed dental instrument, 633,635, 636,638 - 640 Li-ion batteries, 483 linear polyethylenes (LPEs), 271 linoleic acid, 669 LiPF6,464,465,483 lipids, 344, 665,716 liposomes, 157 liquid crystals, 443,488 liquid films, 150 liquid-solid interactions, 347 liquidus point, 726 literature of thermal analysis and calorimetry, 2 lithium hexafluorophosphate, 464 lithium sulfate monohydrate, 525 lithium-ion batteries, 462 local anaesthetics, 618,689 localised pyrolysis-evolved gas analysis, 78

localised thermal analysis, 6, 58, 67 - 76, 78, 82,84 localized adsorption, 345 Lorentz-Gauss cross-product function, 201 Lorenz function, 3 15 Lorenz number, 307,3 15 loss factor, 213,259 loss modulus, 698, 699 low density polyethylene, 273 low-density polyethylene / Ziegler-Natta linear low-density polyethylene, 286 low-temperature calorimeters, 33 low-temperature nitrogen adsorptiondesorption isotherms, 356 low-temperature transmission electron microscopy, 644 L-threonic acid, 475 lubricating oil, 580,704,705 lumped-circuit technique, 217 lutidine, 622,623

M macro-crystalline waxes, 579 macropores, 345,363 magnetic materials, 448 magnetic susceptibilities, 481 magnetocaloric effect, 449 malonic acid, 71 1,712 maltitol, 520, 522 mammalian cells, 157 Mampel model, 487 manganese-cerium mixed oxides, 123, 127, 128 manometers, 553 Mars, 179 martensite, 63 1 martensitic transformation, 445,448 mass spectrometry (MS), 7, 1l,78, 80, 93 110, 118, 119, 122, 129,439,440,443,451, 455,458 - 461,466,469,476 - 478,482,485, 487,556,566 mass spectrometry-kineticmethod, 556 mass-loss effusion, 553, 560 maxillofacial materials, 647 maximum bubble-pressure method, 3 11 McBain balance static adsorption method, 372 mean effective wavelength, 3 17

meats, 718 mechanical-conduction calorimeters, 39 mechanism of catalytic action, 426 mechanism of dissociative evaporation, 453 medical applications, 663 medical implants, 158 Medicta, 4 melt crystallization, 507,s 14 - 5 18 melt flow index, 700 melt spinning, 446 melt viscosity, 5 13 melting of benzyl, 441 melting of indium, 441 melting point, 698, 71 1 melting, 55,68, 70,72,74, 83,271, 273,293,306,310 - 316,319 - 322,325 menthol, 673,674 mercury calorimeter, 45 mercury porosimetry, 356 mesopores, 345, 353,372, 375, 377, 378,402,407 metabolic heat rate, 722 metal oxides, 388,414,421,427 metal surfaces, 42 1 metal-ceramic restorations, 650 metallic films, 141 metallo-proteins, 475 metals, 59, 68, 76, 299, 724 metastability, 269 - 271,275 metastability, 7,604 metastable drug polymorphs, 598 methods for producing nanomaterials, 344,347, 349,350, 352,354,359,364, 375,376,381,382,384 metrological investigations, 302 Mg(I1) complexes of nitro-substituted benzoic acids, 484 Mg-RE-Y-Zr (RE=rare earth) alloy, 449 micellar-polymer film systems, 158 microbalance, 8, 133 - 136, 141, 148, 149,152 - 154,157,161 - 163, 165 microbial growth, 72 1 microcalorimetry, 9,388, 389,393,394, 396,400,402,403,406,407,410 - 418, 421 - 423,427,430,441,481,559,598, 61 1,697

micro-crystalline waxes, 579 micro-DTA (pDTA), 71 micro-EGA, 79, 80 microelectronics, 61,77,294 microfilled composites, 657 microgravity, 324 microhardness, 449 microorganism growth, 7 16 micropores, 345,360,363,364,367,378, 382,389 microscopic robots, 344 microscopy, 274, 598,603 microsecond experiments, 302,303 microspheres, 680 microstructural features, 725 micro-thermal analysis (FTA), 6, 56,61,71, 74, 78, 84 micro-TMA (pTMA), 71 microwave-processed denture base resins, 658 microwaves, 441,454 milk fat, 6 12 milks, 719 millisecond experiments, 302,305, 3 12,318, 330 mimicking of rates during processing, 270 minerals, 172, 173, 179, 188,200,206,211, 345,348,386 miniaturization of calorimeters, 559 mixed cobalt(I11) complexes, 471 mixed oxalates, 459 mixed-oxide compounds, 462 mode coupling theory, 236 model-free kinetics, 451, 503, 530,531,625 modulated temperature differential scanning calorimetry, 663,675 modulated temperature programmed techniques, 729 modulated temperature thermogravimetry, 560 modulation conditions, 65 1,652,654 modulus and damping, 700 molecular inclusion, 11, 597 molecular mechanics, 563 molecular sieves, 344,388,396,400,402, 403 molecular wires, 485 molybdenum as a test material, 332 molybdenum(V1) complexes, 457

monomolecular desorption kinetics, 352 montrnorillonites, 198,349, 370,371, 462 morphologies, 11, 597 motional inductance, 146 motional resistance, 144, 146, 147, 148, 154,155, 164 Mueller bridge, 543 multilayer saturation of the surface, 375 multiple heating programmes, 503 multi-step processes, 503,504, 530 multi-wavelength pyrometry, 3 18 N NaHC03, 45 1,490 NaHS04-KHS04,468 nanocalorimetry, 275,441 nanocomposites, 61 nanocrystalline ceriatzirconia, 123 nanocrystalline CuS particles, 467 nanocrystals, 343 nanomaterials, 9, 161, 172, 343, 344, 363,364,367,381,382,384 nanomolecules, 343,344 nanoparticles, 343,441,452,460,461 nanopowders, 344 nanostructures, 294,343, 344,364 nanotechnology, 165, 343,344,385, 485 nanotubes, 9,343,344,345,349,359, 360,362,363,364,367,369,383,384 nanowires, 343,344 National Bureau of Standards (NBS), 543,548,558,568,573 near-field photothermal spectroscopy, 78 nedocromil sodium trihydrate, 534 Neo Sentalloy, 643,644, 645 Netzsch STA 41 1 simultaneous thermal analyzer, 94 neural networks, 63,467 neutron irradiation, 446 Newton's law of cooling, 547 Newtonian fluid, 593 Newtonian techniques, 728 NH4HC03,45 1 NiC204.2H20,458 nicardipine, 673

nickel boride, 423 nickel complexes, 486 nickel formate, 5 11 nickel phosphide, 423 nickel powder, 423 nickel-cobalt alloys, 108 nickel-titanium alloys, 63 1 nickel-zinc ferrites, 461 nifedipine, 534 niobic acid (NbzOs*nH20),429 niobium phosphate (NbOP04), 429 NIST Chemistry WebBook, 559 Nitinol, 641,643,644,646,647 nitrates, 122,454 nitrided Ti02/In203,111 nitrites, 454 nitrofurantoin, 609 3-nitro-l,2-phenylenediamine,528 5-nitro-2-anthranilatesof rare earths, 478 nitro-substituted phenylenediamine, 528 n-methyl cyclohexyl methacrylates, 228 NMR, 237,253 NMR, 29 1 NMR, 450,457,469,475,487,492 NMR, 556,557,566 NMR, 581,591 NO as a catalytic probe, 397,414,425,428, 429 noble metals, 303 nomenclature of calorimetry, 6 nomenclature, 6, 13, 14, 15,44,46,47,48, 51,52,71 non-Arrhenius behaviour, 5 14 non-crystalline forms of CaC03, 612 non-ideality of gases, 548 non-Newtonian behaviour, 584,593 nonreversing heat flow curve, 644,647,65 1 North American Thermal Analysis Society (NATAS), 4 NOx storage-reduction (NSR), 111, 121 NOX, 111,120 - 123,127 - 129,407,414, 416,425,426,428 nucleation temperature, 726,727 nucleation, 275,289,294 nucleation, 512 - 5 14, 594 numerical integration, 506 Nylon 6,508

0 oesophagus, 663,678,679 OH stretching bands, 180, 181 oil ignition temperature (OIT), 699, 700,704 oils, 704 optical imaging techniques, 31 1 optical microscopy, 55 optimization of HPer DSC measurements, 277 optimum choice of the purge gas(es), 284 optimum conditions for measurement, 280 order parameter (n), 447 ordinary calorimeter, 30 organic iron coordination compounds, 462 organic matter in rock samples, 111 organic peroxide initiator, 657 organic residue, 107, 109, 111 organometallic compounds, 554, 558, 566 organo-smectite complexes, 462 orthodontic elastomeric modules, 654 orthodontic wires, 63 1 - 635,641,642, 644 orthodontics, 641,655 orthopaedic implant, 680 osmotic pressure, 713 osteoarthritis, 689 osteogenesis, 683 oxalates, 186, 187, 188, 191, 192, 194, 195,201,456 oxanes, 564 oxidation induction time, 699,704 oxidation of butadiene, 123 oxidation of magnetite, 46 1 oxidation potentials, 55 1,554,556 oxidation resistance, 699 oxidation, 108, 109, 112, 113, 116118,123,126,127,446 oxidative degradation, 165 oxynitrides, 401,418 Ozawa kinetic method, 506, 508, 537 ozone layer, 566

P paclitaxel, 680 - 683 paints, 79 palladium complexes, 488 paracetamol, 64,65, 74, 606 paraffin content, 582, 583 paraffin waxes, 580,585 parallel reactions, 5 10 parallel-plate capacitor model, 209 parchment, 76 partition coefficients, 154 passive microwave devices, 351 pasteurization, 722 patchwise surface, 346,347 PDMS, 648,649 peak temperature, 508, 520 pectins, 675 Peltier effect, 36, 41 PEMA, 237,238,263 penetration enhancers, 664,665,667 - 674, 690 2,4-pentanedione (acetylacetone), 565 peptides, 566 perchlorates, 463 periodic surface, 346,347 permeability, 35 1 permittivity, 210,212,213,216,218,220, 221,223,224,241,242,244,246 - 248,254, 260 petroleum refining, 430 petroleum wax, 579 pharmaceutical hydrates, 609,609 pharmaceutical science, 61,64,74,75 pharmaceuticals, 11,597, 695,709,711 phase changes, 34,551,552,554,560 phase contrast microscopy, 581 phase diagrams, 440,448,477,712 phase separation, 58 1 phase transition temperatures, 300,302,3 18 phase transitions, 453,462,464,491 phosphates, 418,464 phospholipids, 566,676 photoacoustic calorimetry, 557, 563 photochemical calorimetry, 48 photodynamic therapy (PDT), 486 photoelectron spectroscopy, 200 photoinitiator camphoroquinone, 657 photoionization mass spectrometry, 556

photopolymerization, 155 photosensitizer, 486 photo-thermal effect, 60, 82 physical aging processes, 226 physisorption, 351, 391, 395,410 Picker liquid mixing calorimeter, 35 piezoelectricity, 134, 170 pigments, 647,655,656 pillared clays, 202 pipeline waxes, 580 piracetam-gentisic acid, 602 piroxicam monohydrate, 625 piroxicam, 669,670 Pitzer ion-interaction model, 450 planar hydrates, 610 Planck's law, 173,307, 3 16,318,319, 329 plant leaves, 76, 79 plasma sources, 304 plasters, 723 plasticizer content, 699 platinum powder, 421 platinum resistance thermometer, 543, 546 PLLA, 676,679,680 PMMA, 237,263 PnBMA, 237,238,263 pneumatic compensation calorimeter, 35,36 poisoning of catalysts,9, 387, 388,427 polarizability, 182, 183 polarization, 210,211,212,217 - 220, 241,242,244,246 - 249,256 - 260 polarized light microscopy, 581 poly (styrene-P-isobutylene- P-styrene), 681 poly(buty1ene terephthalate), 333 poly(buty1ene-2,6-naphthalate), 613 poly(ethy1ene 2,6-naphthalate), 5 16, 518 poly(ethy1ene oxide), 5 16, 5 18 poly(ethy1ene terephthalate), 66,70, 516,517,518,522,613 poly(ethy1ene-2,6-naphthalate),613 poly(L-lactic acid), 679 poly(n-alkyl methacrylates), 230 poly(n-butyl methacrylate), 522 poly(styrene), 521, 525

poly(urethaneurea) blood sacs, 690 poly(viny1 chloride), 230,237, 522 polyarnide, 275,276 polyarylamide, 698,699 polybutene-1,614 polycarbonate, 225,230,703 polycrystalline samples, 152 polydimethylsiloxane, 648 polyether, 65 1,652,654 polyethylene, 68, 72,73, 74 polyethyleneglycol, 71 1, 712 polymer blends, 72 polymer degradation, 699 polymer films, 72, 148, 153, 154, 157 polymer raw materials, 271 polymer shear modulus, 154 polymer transitions, 654 polymerization, 650,656,657 polymers, 698,700 polymorph identification, 597 polymorph prediction, 60 1 polymorphic purity, 619 polymorphic transitions, 56, 598, 604, 606, 608,613,614 polymorphism, 551, 554 polymorphism, 597 - 599,602 - 604,606,611 - 622 polymorphs, 75, 709,711 polyoxometalates, 456 polyoxometalate-supported transition metal complexes, 457 polyoxovanadiurn borate, 456 polypropylene (PP), 284,285,286,287,289, 290,612 polysaccharides, 229,244 polysulfide, 651,652,654 polyurethane, 647,650,656,700 polyvinyl alcohol (PVA), 202,203,204,205 porcine pericardium, 686 pore-size distribution functions, 348, 351, 357,366,370,378,384 pore-size effect, 400 porosity, 345,350,356, 385 porphyrins, 486 potassium perchlorate, 463 potential barrier for surface diffusion, 345 pour-point-depressants, 591 powder samples, 152

power compensation calorimeter, 60 power compensation design, 276 power compensation, 36 precision required in combustion experiments, 549 prediction of crystal structures, 11, 597, 60 1 prediction of the thermal decomposition pathway, 487 predictions of microstructures, 300 pre-exponential factor, 503, 504,529, 530,532,533 pressure DSC, 704 pressure gradients, 304 pressure/volume scanning, 48 principal component analysis, 706 principle of Thermal Analysis in general, 27 Privalov's calorimeter, 49 probe molecules, 396 ProFile dental instrument, 633 - 640 programmed liquid thermodesorption, 347,352 propagation, 525, 528 propellants, 454,463 propylene glycol, 667,668,670,671, 673,674 prosthetics, 12,663 prosthodontic reconstruction, 647 protein-ligand interactions, 164 proteins, 157, 158, 159,716, 566 proton affinities, 397,398,399 PrPO4,465 pseudopolymorphism, 606 pulse flow method, 394 pulse thermal analysis, 7,93,94,96, 98, 101,102,104,109 - 130,443,458 pulse-calorimeter, 277,278,333 pulsed high-pressure mass spectrometry, 556 pulsed light sources, 304 pulse-heated copper sample, 314 pulse-heating microcalorimetry, 332 pulseheating, 301, 308 purification of waste water, 350 purity, 709 pyridine, 394,397,398,399,400,403, 407,408,409,412,413,415,417

Pyris 1,276,278 pyrolysis of urea, 196 pyrometry, 303,316,317,318,334,335 pyrotechnics, 56,455

Q

Q-TG, 9,351 - 355,360 - 372,373,375,379, 380,382,383,384 quality assurance (QA), 12,696 quality control (QC), 12, 696,704,709 quality factor, 143, 148, 159 quality testing, 696 quantum points, 343 quartz crystal microbalance (QCM), 8,34, 134 - 166,172,173 quartz crystal microbalancelheat conduction calorimetry, (QCMrHCC), 8, 161 quartz crystal sensor (QCS), 149 quartz plate resonators, 135 quartz resonator, 553, 560 quartz thermometer, 546 quartz-crystal oscillators, 134 quasi-adiabatic, 30, 38,42,43 quasi-elastic incoherent neutron scattering (QINS), 371

R radiance temperatures, 302,3 18,325 radiation thermometry, 3 16,320 radical generation, 566 radical initiation, 525 radiometry, 302 Rarnan effect, 182 Raman spectroscopy, 8, 171, 172, 182,183, 186,192,598,601,611,618,620 random surface, 346,347 rat skins, 669 Rayleigh scattering, 182 reaction calorimetry, 542, 544, 550 reaction model, 451,503 - 505, 529, 530, 532 - 534 re-crystallization of nickel sulfide, 467 recrystallization, 7,269,270,273,275,289, 290,293,295 Redhead's equation, 353 redox properties of Ce02, 116 redox properties of catalyst surfaces, 421 reduction of CuO, 112, 114, 115

reduction of Ir02, 111, 113 reduction of Mnz03,115, 116 reduction of NO by hydrocarbons, 123 reduction of zeolite-supported iridium oxide, 113 reduction potentials, 554 reference materials, 62, 68, 547, 549, 557,558 reference pressure, 544 reference temperature, 549, 544 reflectivity, 174,318, 3 19,321,326 refractive index, 174, 300, 320, 321 refractory metals, 303, 335 regulatory compliance, 696 rehydration, 478,486 relative humidity, 189, 198,711 relaxation time, 445, 5 18 - 520 remelting, 290 removal of toxic gases, 350 reorganization, 270,271,273,275,288, 290,295 residual stress, 300 resin composites, 656,657 resistance thermometry, 59,67 resolution factor, 707 resonance frequency, 134, 142, 143, 145 - 148,153,159,162 reversibility, 605 reversible decompositions, 503, 522, 524 reversing heat flow curve, 646,647 rheological behaviour, 579, 582, 584 rheology, 579 rheology, 8 rheometry, 582, 595 rheometry, 8 ribavirin, 620 rice, 719 Richards' calorimeter, 37 ring strain, 562, 563 RNA, 157 rotary drills, 632, 633 rotating bomb calorimetry, 472 rotational viscometers, 584 R-phase, 632 - 644 rubber blends, 700 rubbery polymer, 150, 155 Ru-hydroxyapatite, 123

ruthenium catalysts, 423 S salicaine hydrochloride, 618 salmeterol xinafoate, 6 19 sample controlled thermal analysis (SCTA), 729 sample thickness, 175, 177 Sauerbrey relation, 146 SAXS, 612 scanning electron microscopy (SEM), 358, 367,654 scanning microscopes, 344 scanning probe microscopy, 56 scanning rate, 224 scanning thermal microscopy (SThM), 6, 57 62,66,67,69,71,74 - 77, 82, 84 scanning tunnelling microscope, 57,344 Schlieren photography, 3 12 secondary liquid crystalline calibrants, 281 second-order solid-solid transitions, 503 selective catalytic reduction (SCR), 127,415 selectivity, 349, 350, 351 self-assembled monolayer (SAM) chemistry, 152 self-assembly; 450,470 self-cooling, 507 self-heating, 507 self-seeding, 294 semiconductors, 9,345,448 semi-empirical calculations, 563 sensitivity in mass and heat detection, 162 sensitivity, 707 shade matching, 647 shadowgraph techniques, 3 12 shape memory effect (SME), 447 shape memory, 632,635,641,643 shape-memory alloys, 76,445,446 shear loss modulus, 145 shear modulus, 527,591 shear storage compliance, 146 shear storage modulus, 145, 155 shear viscosity, 155 sheep's milk, 719,720 shelf-life determinations, 713 shock waves, 299 short chain branching content, 293 shrinkage, 686

Si/Al ratios, 404 SIBS, 681,682 silica gels, 9, 344, 345, 350 silica, 408,409,411,413,415,417, 41 8,422,430 silica-aluminas, 4 17 silicone breast implants, 690 silicone elastomers, 647,650 silicone, 647 - 65 1, 654 silver carbonate decomposition, 111 silver complexes, 453,488 silver iodide, 453 silver loading, 416 silver production, 478 silver(1) carboxylate complexes, 489 silver(1) complexes of hexakis(tolylsulfanyl)benzene, 488,489 Simon kinetic method, 507, 537 simultaneous Q-DTA-Q-TG, 443 single heating rate methods, 532 sintering, 449,460 size-exclusion chromatography (SEC), 293 skin, 12 skin, 663 - 665,667 - 676,689,690 skin-effects, 303 small samples, 293 smart surfaces, 9,344 SmP04,465 snakes, 62,75 Sn02 loading, 414 sodium alkoxides, 490 sodium hydrogen carbonate, 96, 100, 451,490 solar filters, 344 solar heating, 299 soldering materials, 448 sol-gel formation of films, 152 sol-gel synthesis, 462 solid catalysts in contact with a liquid phase, 394 solidification process, 724,726 solidification shrinkage, 300 solid-solid phase transitions, 554 solid-solid transformations, 270 solid-state phase transition in silver iodide, 441 solubility trends, 11,597

solution calorimetry, 697 solvates, 599,603 - 61 1,615,621 - 626,709 solvatomorphism, 11, 597, 598,606,611, 621,622 solvent exchange, 599,611,625 solvent-drop grinding, 600 sorptomatic apparatus, 35 1 sorptometry, 363,383 specific enthalpy curve, 291,292 specific enthalpy, 300,3 10,328, 329 specific heat capacities, 276,448,721 specific surface area, 350 specific-heat ratios, 323 spectrometry, 347 spectroscopic techniques, 598 speed of sound, 323 sphere reflectometer, 3 18 spin coating, 152 split-Hopkinson pressure bar apparatus, 330 spray coating, 152 spray pyrolysis, 466 stability relationships, 604 stainless-steel alloys, 631 standard pressure, 549 standard-state conditions, 544 standard-state corrections, 544, 547, 549 stannates, 465 static parallel capacitance, 147 statistical process charts, 703 steaming, 403,404 steels, 449 stent implantation, 663 Stokes and anti-Stokes lines, 182 storage modulus, 698,699,700 stratum corneum, 664,665,667,668,673, 690 streak cameras, 3 12 stress-strain functions, 33 1 strontium bismuth niobate, 460 strontium complexes, 490 strontium metatitanate, 456 strontium nitrite, 455 strontium titanyl oxalate, 456 structural function method, 358 structural transformation, 632 subambient temperatures, 633 sublimation of urea, 196 sublimation, 599

submicrosecond experiments, 302,304, 333 sulfates, 466 sulfathiazole, 62 1 sulfation, 41 1,412 sulfides, 466 sulfites, 466 sulfone and sulfoxide derivatives, 564 sulfur dioxide, 406,409,411,415,417 sulfur, 111 superacids, 4 18 superconductors, 9,344,345,350 supercooling, 7,269,28 1,289, 5 12,s 13 supercritical fluids, 600 superelastic behaviour, 632 superheating, 44 1 super-ionic conductivity, 450 Suppocire AT, 714,715 supported gold catalyst, 107 supported metal catalysts, 401 supported Pd catalysts, 422 supported Pt surfaces, 421 supported tin oxide catalysts, 429 suppositories, 713, 714 supramolecular chemistry, 469, 621 surface acidity and basicity, 412,418 surface active sites, 349 surface energies, 345, 5 12 surface fractal dimension, 359 surface heterogeneity, 9,387,392,395, 402,409,430 surface liquid films, 382 surface plasmon resonance spectroscopy, 157 surface properties, 343 surface roughness, 62,63, 149, 154, 157 surface tension, 299, 300,306,307, 319,321,324,340,353 surface wettability, 349 surface-perimetermethod, 358 surfactant, 76,719 suspensions, 713 symbols used specifically in Thermal Analysis, 22 symmetry of the DSC, 281 synchrotron radiation, 473,612

T tablets, 64,65,74,75,713 talc, 723,724 tank waxes, 580 tantalum, 320 TeEGDMA, 232,233,234,235 Temkin isotherm, 402 temperature calibration, 222 277,278,443 temperature change calorimeters, 47 temperature correction, 547,548, 549 temperature definition, 303, 305,3 16,338, 339,340 temperature modulated differential scanning calorimetry(TMDSC), 12,439,440, 586,587,631,634,642 - 654 temperature modulation, 66 temperature of reference, 549,544 temperature variation in E, 5 14 temperature-programmeddesorption (TPD), 20,388,389,401,407,416,417 temperature-programmedoxidation (TPO), 20,388 temperature-programmed reduction (TPR), 20,388,415,425,429 tensile testing, 700 terfenadine, 615 ternary alloys, 448 ternary chlorides, 440 terpenes, 672,673 tertiary amine accelerator, 657 tetracaine, 689 tetra-ethyleneglycol dimethacrylate, 232 tetroxoprim, 624, 625 TG-DTA, 461 TG-FTIR, 11,439,443,458 - 460,473,486, 488 TG-MS, 11,439,458,462 thermal annealing of thin films, 332 thermal conduction calorimeters, 34 thermal conductivity, 57,60 - 66,75 - 77, 300,307,311,315,326,328,333,698,721 thermal curve, 20 thermal decomposition of CaC03, 451,456 thermal decomposition of hydrates of nitrates of d-electron metals, 455 thermal decomposition of Me(N03)n.qHz0, 455

thermal decomposition, 195 thermal degradation of vinyl polymers, 525 thermal degradation, 69,79 thermal dehydration, 524 thermal desorption gas chromatography-mass spectrometry, 78 thermal diffusivity, 61,66, 82, 300, 307, 311,315,316,326,328,331 -333,340, 72 1 thermal expansion and shrinkage, 700 thermal expansion, 61,66, 189, 190, 300,305,311,312,314,322,337,441 thermal expansivity imaging, 66 thermal expansivity, 66, 77 thermal history, 7,269,697,698 thermal lag, 278,280,281,284,285, 293 THERMAL list-server, 5 thermal probes, 61,67,76, 84 thermal resistance, 28,30, 32,40,42, 43,49,444 thermal runaway, 454 thermal stress resistance, 35 1 thermally stimulated current (TSC), 16, 20 thermistor, 546 thermoacoustimetry, 16 thermobalance, 26 thermobarometry, 16 thermochemical cycle, 555 thermochemical data bases, 558 thermochemistry, 549 - 544,549,551, 557 - 559,561 - 563,566,567 thermocouple, 57,59,60,67,316 thermodesorption, 348,350,351,354, 355,360 - 362,370 - 373,379,380, 382,384 thermodiffractometry, 16 thermodilatometry (TD), 15,21 thermodynamic stabilities, 604 thermogravimetric analysis (TGA), 15, 16,18 - 21,603 thermogravimetric analysis, 552, 560 thermogravimetry (TG), 11, 15,21,67, 95,96,98- 100,106- 108,113,115, 118, 127, 133, 172, 192, 196,203,350, 388,401,416,424,426,439,443,445,

449 - 466,470 - 492,505,509,525,526, 598,603,604,608 - 612,617,618,621 - 623, 697,699,700,703, 705,706,709,711,712, 713,716,719,720,722 - 724,729 thermogravimetry under quasi-isothermal conditions, 35 1 thermogravimetry, 663,686 thermokinetic parameter, 392, 393 thermoluminescence (TL), 16 thermomagnetometry, 16,461 thermomanometry, 16 thermomechanical analysis (TMA), 15, 19, 21,22,61,67,68,71,72,657,697,698,699, 700,704 thermometer, 543, 546, 547 thermometry, 15 thermomicroscopy, 8,21,56, 581, 584, 587, 591,592,595,598 thermo-oxidative degradation, 525 thermophysical properties, 11,299 - 306,310, 31 1,324,325,328,334 thermopile, 30,45,48, 545, 552 thermoplastics, 698 thermoptometry, 16 thermosets, 698,700 thermosonimetry, 16 thermospectrometry, 16 thermotropic lipid transitions, 665 thianes, 564 thin films of zinc sulfide, 466 thin films, 294 thin layers, 72 thin-film (chip) calorimeter, 275 third-law method, 453 Thomsen calorimeter, 29 - 31, 37 three-terminal cell, 220 threshold photoelectron photo-ion coincidence (TPEPICO), 566 Tian-Calvet calorimeter, 38 time domain reflectometry, 2 17 time-domain measurements, 219 tin tetrahydroxide, 465 Ti-Ni alloy, 445 TiNiCu alloy, 445,446 TiOz thin films, 461 titania, 203,409,413 - 415,418,422 titania1PVA expanded hectorite, 202,203 titania-silica aerogels, 125

titanium-aluminides, 303 titration calorimeters, 551, 566 tolbutamide, 600, 601 tomographic imaging, 66 topological isomerism, 470 torsional braid analysis (TBA), 22 torsion-effusion method, 553 total pore volume, 350,367,384 transdermal administration, 664 transmission electron microscopy (TEM), 451,591 transmission spectra, 173 transpiration or gas saturation techniques, 553,560 transverse shear mode (TSM) resonators, 134 TRC Thermodynamics Tables, 559 triacylglycerols, 612 triazole-based salts, 454 triglycine sulfate, 223 trimorphic system, 605 tulobuterol, 615,616,617, 618 tungsten ribbon lamps, 3 17 twin calorimeters, 43,49 twinning, 632,637,638,644,647

u ultra-fast calorimeters, 30,277 UNAP, 67 1,672 uncertainties in measurements, 333 undercooling, 728 uranium-cerium mixed oxides, 463 uranium-plutonium mixed oxides, 463

v VzOs catalysts, 424 V205/Ce02catalysts, 41 5 V205/y- A1203catalysts, 41 5 vacuum deposition, 141, 152, 162 van't Hoff equation, 447 van't Hoff plots, 605,606, 617 vanadia, 4 15 , 4 16 vanadium-chromium oxide compounds, 462 vanillin, 71 1 vaporization, 7 16 vapour diffusion, 599 vapour pressure sorption analysis, 618

vapour pressures, 14, 15, 16,22, 552 - 554, 560 vapour sensing, 157 vapreotide, 676 variable activation energy, 508 variable-temperature PXRD, 608 vaterite, 612 vibrational selection rules, 183 vibrational spectroscopy, 171,172 Vickers hardness measurements, 640 vinyl polysiloxane, 650,651,654 virus, 157 viscoelastic liquid, 144 viscoelastic response, 647 viscoelastic solids, 146 viscometric properties, 704 viscometry, 584, 585, 589, 591, 592 viscosity, 11,299, 300,306,324,340, 526, 528,447,700,701,706,713 viscous flow, 5 16,528 vitamins, 716 vitrification, 271,293,295, 527 VMgO catalysts, 417 Vogel temperature, 234,239 Vogel-Fulcher-Tammann-Hesse equation, 233 volatile oils, 609 volatiles, 700 volume expansion, 3 11 , 314 Vyazovkin kinetic methods, 503, 506, 507, 516,518,520,521,527,528,537

W Washburn equation, 356 Washburn's corrections, 547 - 549 water as a catalytic probe, 401 water calorimeters, 30,45 water content, 608 water vapour calorimeter, 37 water vapour pressure, 723 water-gas shift, 118, 119, 120 wax appearance temperature (WAT), 9,582, 584 - 587,589,591,595 wax content, 581, 586 wax crystal modifiers, 580 wax precipitation temperature (WPT), 591 wax precipitation, 579 wax separation, 579

waxes, 658 WAXS, 612 websites, 5 weddellite, 186, 187 wetting phenomena, 347,382 wettingldrying of montmorillonite, 198, 199 whewellite, 187,201,202 Wiedemann-Franz-Lorenz(WFL) relation, 3 15 Wien's law, 3 17 Williams-Landel-Ferry (WLF) equation, 234, 519 wire explosion experiments, 303 wire insulation materials, 698 Wollaston process wire, 59 work done by the system, 545 work hardening, 637, 640,641 X X-ray computerized tomography, 594 X-ray diffraction (XRD), 198,274, 349, 446,451,452,457,458,460,462,465, 468 - 482,486,487,489 - 492,597, 598,601,612,613,618,620,632,647, 654

X-ray photo-electron spectroscopy (XPS), 200 - 203,388,394,397,414

Y yield strength, 580, 593 yttrium-aluminate liquids, 46 1

z

zeolites, 9,344,350, 388 - 393,396 - 408, 417,418,426,427,430 zero-order kinetic model, 534 Zhuravlev equation, 456 zinc as calibrant, 277 zinc complexes, 49 1,492 zinc oxalate dihydrate, 104 zinc phosphates, 491 zirconium oxide catalysts, 428 zirconium-based alloys, 447 Z-match method, 153 Z~[CU(CN)~], 462 Zn4C03(OH)6.H20,45 1 ZnCz04.nH20,445 Zr02, 109,408,410,411,412,417,418,419, 420,422,425

This page intentionally left blank

Handbook of Thermal Analysis and Calorimetry, Volume 4 : From Macromolecules to Man (Handbook of Thermal Analysis and Calorimetry)

Theory of Calorimetry (Hot Topics in Thermal Analysis and Calorimetry)

Modulated Temperature Differential Scanning Calorimetry: Theoretical and Practical Applications in Polymer Characterisation (Hot Topics in Thermal Analysis and Calorimetry)

Streaming Media Architectures, Techniques, and Applications: Recent Advances

Glassy, Amorphous and Nano-Crystalline Materials: Thermal Physics, Analysis, Structure and Properties (Hot Topics in Thermal Analysis and Calorimetry)

Techniques in Thermal Analysis: Hyphenated Techniques, Thermal Analysis of the Surface, and Fast Rate Analysis (ASTM special technical publication, 1466)

Handbook of Thermal Analysis and Calorimetry: Recent Advances, Techniques and Applications

Handbook of Thermal Analysis and Calorimetry, Volume 4 : From Macromolecules to Man (Handbook of Thermal Analysis and Calorimetry)

Theory of Calorimetry (Hot Topics in Thermal Analysis and Calorimetry)

Handbook of Thermal Analysis and Calorimetry: From Macromolecules to Man

Introduction to Thermal Analysis: Techniques and Applications