Magnetism: Molecules to Materials II: Molecule-Based Materials: 2

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-V...

Author: Joel S. Miller | Marc Drillon

163 downloads 1478 Views 8MB 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

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

Magnetism: Molecules to Materials II

Edited by J. S. Miller and M. Drillon

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

Further Titles of Interest

J. S. Miller and M. Drillon (Eds.) Magnetism: Molecules to Materials Models and Experiments 2001. XVI, 437 pages Hardcover. ISBN: 3-527-29772-3

J. H. Fendler (Ed.) Nanoparticles and Nanostructured Films 1998. XX, 468 pages Hardcover. ISBN: 3-527-29443-0

P. Braunstein, L. A. Oro, and P. R. Raithby (Eds.) Metal Clusters in Chemistry 1999. XLVIII, 1798 pages ISBN: 3-527-29549-6

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

Magnetism: Molecules to Materials II Molecule-Based Materials Edited by Joel S. Miller and Marc Drillon

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

Prof. Dr. Joel S. Miller University of Utah 315 S. 1400 E. RM Dock Salt Lake City UT 84112-0850 USA

Prof. Dr. Marc Drillon CNRS Inst. de Physique et Chimie des Matériaux de Strasbourg 23 Rue du Loess 67037 Strasbourg Cedex France

This book was carefully produced. Nevertheless, editors, authors and publisher do not warrant the information contained therein to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Library of Congress Card No.: applied for A catalogue record for this book is available from the British Library. Die Deutsche Bibliothek - CIP Cataloguing-in-Publication-Data A catalogue record for this publication is available from Die Deutsche Bibliothek ISBN 3-527-30301-4 © WILEY-VCH Verlag GmbH, Weinheim (Federal Republic of Germany). 2001 Printed on acid-free paper. All rights reserved (including those of translation in other languages). No part of this book may be reproduced in any form - by photoprinting, microfilm, or any other means - nor transmitted or translated into machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition: EDV-Beratung Frank Herweg, Leutershausen. Printing: betz-druck GmbH, Darmstadt. Bookbinding: Wilh. Osswald + Co. KG, Neustadt Printed in the Federal Republic of Germany.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

Contents

1

Nitroxide-based Organic Magnets . . . . . . . . . . . . . . . . . . . 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Unconjugated Nitroxides . . . . . . . . . . . . . . . . . . . . . . 1.2.1 Mono-nitroxides . . . . . . . . . . . . . . . . . . . . . . 1.2.2 Oligo-nitroxides . . . . . . . . . . . . . . . . . . . . . . . 1.3 Conjugated Nitroxides . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 Mono-nitroxides . . . . . . . . . . . . . . . . . . . . . . 1.3.2 Oligo-nitroxides . . . . . . . . . . . . . . . . . . . . . . . 1.4 Nitronyl Nitroxides . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.1 Mono-nitronyl Nitroxides . . . . . . . . . . . . . . . . . 1.4.2 Oligo-nitronyl Nitroxides . . . . . . . . . . . . . . . . . 1.4.3 Co-crystallization of Nitronyl Nitroxides . . . . . . . . . 1.5 Imino Nitroxides . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Poly(nitroxides) . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

1 1 5 5 10 12 12 14 18 18 35 40 44 46 49 51

2

Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Aminoxyl Radicals . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Electronic Structure and Basicity . . . . . . . . . . . . 2.2.2 Aminoxyl Radicals with Another Basic Center . . . . 2.2.3 High-spin Di- and Poly(aminoxyl) Radicals . . . . . . 2.3 Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals . . . . . . . . . . . . . . . . . . . . . . 2.3.1 Indirect Coupling (Extended Superexchange) . . . . . 2.3.2 Complexes with Direct Metal-Aminoxyl Coordination 2.3.3 Cyclic Complexes . . . . . . . . . . . . . . . . . . . . . 2.4 Design Strategy for Various Crystyl Structures . . . . . . . . . 2.5 Preparation of 3d Transition Metal-Poly(aminoxyl) Radical Complexes . . . . . . . . . . . . . . . . . . . . . . . . 2.6 One-dimensional Metal-Aminoxyl Systems . . . . . . . . . . 2.6.1 Structure and Magnetic Properties of Ferrimagnetic 1D Chains Formed by Manganese(II) and Nitronyl Nitroxides . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

61 61 61 62 63 64

. . . . .

. . . . .

. . . . .

. . . . .

65 65 70 72 74

. . . . . . . .

77 80

. . . .

81

VI

Contents

2.6.2

Structure and Magnetic Properties of Ferrimagnetic 1D Chains Formed by Manganese(II) and Triplet bis-Aminoxyl Radicals . . . . . . . . . . . . . 2.7 Two-dimensional Metal-Aminoxyl Systems . . . . . . . . . . 2.7.1 Structure and Magnetic Properties of Ferrimagnetic 2D Sheets Formed by Manganese(II) and Nitronyl Nitroxides . . . . . . . . . . . . . . . . . . . 2.7.2 Structure and Magnetic Properties of Ferrimagnetic 2D Sheets Formed by Manganese(II) and High-spin tris-Aminoxyl Radicals . . . . . . . . . . . 2.7.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 2.8 Three-dimensional Metal-Aminoxyl Systems . . . . . . . . . 2.8.1 Crystal and Molecular Structure of the 3D System . 2.8.2 Magnetic Properties of the 3D System . . . . . . . . 2.8.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 2.9 Summary and Prognosis . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

4

. . . . . . . . . .

82 88

. . . . .

89

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. 90 . 97 . 97 . 98 . 99 . 104 . 104 . 105

Organic Kagome Antiferromagnets . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Inorganic Kagome Antiferromagnets . . . . . . . . . . . . . . . . . . 3.2.1 SrGa12-x Crx O19 (SCGO(x)) . . . . . . . . . . . . . . . . . . . 3.2.2 Jarosite, AM3 (OH)6 (SO4 )2 (A = Na+ , K+ , Rb+ , Ag+ , NH+ 4, H3 O+ , etc., and M = Fe3+ or Cr3+ ) . . . . . . . . . . . . . . . 3.3 Organic Kagome Antiferromagnet, m-MPYNN · X . . . . . . . . . . 3.3.1 Crystal Structure . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Magnetic Susceptibility . . . . . . . . . . . . . . . . . . . . . . 3.3.3 Heat Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4 Positive Muon Spin Rotation . . . . . . . . . . . . . . . . . . 3.3.5 Distorted Kagome Lattices . . . . . . . . . . . . . . . . . . . . 3.3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

109 109 110 110

Magnetism in TDAE-C60 . . . . . . . . . . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Synthesis and Structure . . . . . . . . . . . . . . . . . . . 4.2.1 Synthesis . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 The Lattice Structure . . . . . . . . . . . . . . . . 4.3 The Electronic Structure . . . . . . . . . . . . . . . . . . 4.4 The Magnetic Properties . . . . . . . . . . . . . . . . . . 4.4.1 The Bulk Magnetic Properties . . . . . . . . . . . 4.4.2 The Spin-glass Behavior of α -TDAE-C60 . . . . 4.4.3 Electron-spin Resonance . . . . . . . . . . . . . . 4.4.4 Ferromagnetic Resonance . . . . . . . . . . . . . 4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123 123 124 124 126 128 130 130 134 135 137 144 145

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

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

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

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

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

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

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

111 111 111 113 115 116 117 120 120

VII

Contents

5

Triarylmethyl and Amine Radicals . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 5.2 Monoradicals (S = 1/2) . . . . . . . . . . . . . . . . . 5.3 Diradicals (S = 1) . . . . . . . . . . . . . . . . . . . . 5.4 Triradicals (S = 3/2) . . . . . . . . . . . . . . . . . . . 5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10) . 5.6 High-spin Polymers (up to Sn = ca. 48) . . . . . . . . 5.7 Conclusions and Prospects (Beyond S = ca. 48?) . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

149 149 149 153 161 163 179 182 185

6

High-spin Metal-ion-containing Molecules . . . . . . . . . . . . . 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Self-assembly of Molecular Clusters . . . . . . . . . . . . . . 6.2.1 Competing Interactions and Spin Frustration . . . . 6.2.2 The Carboxylate Family . . . . . . . . . . . . . . . . 6.2.3 The Hydroxypyridonate Family . . . . . . . . . . . . 6.3 Host-Guest Approach . . . . . . . . . . . . . . . . . . . . . 6.3.1 Hexanuclear Iron(III) Rings . . . . . . . . . . . . . . 6.3.2 Hexanuclear Manganese Rings . . . . . . . . . . . . 6.4 Step-by-step Rationale Approach . . . . . . . . . . . . . . . 6.4.1 Complex as Ligand and Complex as Metal . . . . . . 6.4.2 Predicting the Spin Ground State . . . . . . . . . . . 6.4.3 Antiferromagnetic Approach . . . . . . . . . . . . . 6.4.4 Ferromagnetic Approach . . . . . . . . . . . . . . . . 6.4.5 Role of the Organic Ligand . . . . . . . . . . . . . . 6.4.6 Molecules with Two Shells of Paramagnetic Species 6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

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

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

189 189 190 190 192 197 201 202 203 204 206 207 208 211 214 217 223 223

7

Electronic Structure and Magnetic Behavior in Polynuclear Transition-metal Compounds . . . . . . . . . . . . . . 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Phenomenological Description of Exchange Coupling: the Heisenberg Hamiltonian . . . . . . . . . . . . . . . . . . . . 7.3 Qualitative Models of the Exchange Coupling Mechanism . . . 7.3.1 Orthogonal Magnetic Orbitals . . . . . . . . . . . . . . . 7.3.2 Natural Magnetic Orbitals . . . . . . . . . . . . . . . . . 7.4 Quantitative Evaluation of Exchange Coupling Constants . . . 7.4.1 Perturbative and Variational Calculations of State Energy Differences . . . . . . . . . . . . . . . . 7.4.2 Ab initio Calculations of State Energies . . . . . . . . . 7.4.3 Calculations using Broken-symmetry Functions . . . . . 7.5 Exchange Coupling in Polynuclear Transition-metal Complexes 7.5.1 Homodinuclear Compounds . . . . . . . . . . . . . . . . 7.5.2 Heterodinuclear Compounds . . . . . . . . . . . . . . .

. . . 227 . . . 227 . . . . .

. . . . .

. . . . .

228 230 230 234 235

. . . . . .

. . . . . .

. . . . . .

236 240 242 249 249 261

VIII

Contents

7.5.3 Polynuclear Compounds . . . . . . . . . . . . . . . . . . . . . 263 7.5.4 Solid-state Compounds: The Case of Cu2 (OH)3 NO3 . . . . . 266 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271 8

9

Valence Tautomerism in Dioxolene Complexes of Cobalt . . . . . . . . . 8.1 Introduction – Bistability, Hysteresis, and Electronically Labile Materials . . . . . . . . . . . . . . . . . . . 8.1.1 Bistability and Hysteresis . . . . . . . . . . . . . . . . . . . . 8.1.2 Electronically Labile Materials . . . . . . . . . . . . . . . . . 8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt . . . . . . 8.2.1 Valence Tautomerism – A General Chemical Description . . 8.2.2 Valence Tautomerism – A Simplified MO Description . . . . 8.2.3 VT Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . 8.2.4 Experimental Determination of Thermodynamic Parameters 8.2.5 Dependence of K V T Equilibrium on Ancillary Ligands . . . 8.2.6 Pressure-induced VT . . . . . . . . . . . . . . . . . . . . . . . 8.2.7 Light-induced VT and Rates of VT . . . . . . . . . . . . . . . 8.2.8 VT Complexes of Other Quinone Ligands and Redox Chemistry of VT Complexes . . . . . . . . . . . . 8.2.9 Polymeric VT Materials . . . . . . . . . . . . . . . . . . . . . 8.3 Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecule-based Magnets Derived from NiII and MnII Azido Bridging Ligand and Related Compounds . . . . . . . . . 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Synthetic Procedures . . . . . . . . . . . . . . . . . . . . . . 9.3 Exchange-coupling Parameter . . . . . . . . . . . . . . . . . 9.4 Molecular-based Magnetic Materials . . . . . . . . . . . . . 9.5 One-dimensional Systems . . . . . . . . . . . . . . . . . . . 9.5.1 With 1,3-Azido Bridging Ligands (AF, Uniform) . . 9.5.2 With 1,3-Azido Bridging Ligands (AF, Alternating) 9.5.3 With 1,1-Azido Bridging Ligands (Ferromagnetic) . 9.5.4 With 1,3-N3 and 1,1-N3 bridges . . . . . . . . . . . . 9.6 Two-dimensional Systems . . . . . . . . . . . . . . . . . . . . 9.6.1 With Only Azido as Bridging Ligand . . . . . . . . . 9.7 Three-dimensional Systems . . . . . . . . . . . . . . . . . . . 9.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

281 281 281 281 284 284 287 288 290 292 297 299 300 302 303 304

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

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

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

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

307 307 308 309 309 312 312 313 316 319 322 322 329 334 335

10 Oxalate-based 2D and 3D Magnets . . . . . . . . . . . . . . . . . . . 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Basic Principles of Specific 2D and 3D Network Configurations 10.3 Structural Studies on 2D Oxalato Bridged Compounds . . . . . 10.4 Magnetic Studies on 2D Oxalato Bridged Compounds . . . . . 10.5 Structural Studies on 3D Oxalato Bridged Compounds . . . . .

. . . . . .

. . . . . .

. . . . . .

339 339 339 344 346 349

IX

Contents

10.6 Magnetic Studies on 3D Oxalato Bridged Compounds . . . . . . . . 352 10.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355 11 Hybrid Organic-Inorganic Multilayer Compounds: Towards Controllable and/or Switchable Magnets . . . . . . . . . . . . 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Hydroxide-based Layered Compounds . . . . . . . . . . . . . . . 11.3 Anion-exchange Reactions . . . . . . . . . . . . . . . . . . . . . . 11.4 Influence of Organic Spacers in Hydroxide-based Compounds . . 11.4.1 The Cu2 (OH)3 X Series . . . . . . . . . . . . . . . . . . . . 11.4.2 The Co2 (OH)3 X Series . . . . . . . . . . . . . . . . . . . . 11.4.3 Dipolar Interaction and Long-range Magnetic Order . . . 11.5 Difunctional Organic Anions Connecting Magnetic Layers . . . . 11.6 Metal-radical-based Layered Magnets . . . . . . . . . . . . . . . 11.7 Controllable Magnetic Properties of Layered Copper Hydroxides 11.7.1 Solvent-mediated Magnetism . . . . . . . . . . . . . . . . 11.7.2 Photoisomerism of Azobenzenes in Layered Copper Hydroxides . . . . . . . . . . . . . . . 11.8 Layered Perovskite Ferromagnets . . . . . . . . . . . . . . . . . . 11.8.1 High-pressure Effects . . . . . . . . . . . . . . . . . . . . . 11.8.2 Spontaneous Magnetization in Layered Perovskite Ferromagnets . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

357 357 358 360 361 361 364 367 370 376 380 380

. . 384 . . 385 . . 386 . . 389 . . 393

12 Intercalation-induced Magnetization in MPS3 Layered Compounds . . . . . . . . . . . . . . . . . . . . . . . . 12.1 Introduction and Scope . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Structural and Electronic Aspects . . . . . . . . . . . . . . . . . . . 12.3 Magnetic Properties of the Pristine MPS3 Phases (M = Mn, Fe, Ni) 12.4 Ion-exchange Intercalation into the MPS3 Compounds . . . . . . . 12.5 The Magnetic Properties of the MnPS3 Intercalates . . . . . . . . . 12.5.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.5.2 X-ray and Neutron-diffraction Study of Selected Intercalates . . . . . . . . . . . . . . . . . . . . . 12.5.3 A Ferrimagnetic Model of the MnPS3 Intercalates: Imbalancing of Spins . . . . . . . . . . . . . . . . . . . . . . 12.6 The Magnetic Properties of the FePS3 Intercalates . . . . . . . . . 12.6.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.6.2 Spectroscopic Characterization of the FePS3 Intercalates . 12.6.3 Discussion on the Role of Intercalation into FePS3 . . . . . 12.6.4 Magnetic Properties of Iron-diluted Fe1−x Cdx PS3 Compounds . . . . . . . . . . . . . . . . . . . 12.6.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.7 The NiPS3 -Cobaltocene Intercalation Compound . . . . . . . . . .

. . . . . . .

397 397 397 399 401 402 402

. 405 . . . . .

407 410 410 413 415

. 416 . 417 . 419

X

Contents

12.8 Multi-property Materials: Associating Magnetism and Non-linear Optics . . . . . . . . . . . . . 420 12.9 Conclusion and Perspectives . . . . . . . . . . . . . . . . . . . . . . . 421 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421 13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets . . . . . . . . . . 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 13.2 Synthesis of the Ligands . . . . . . . . . . . . . . . 13.3 Vanadium Phosphonates . . . . . . . . . . . . . . . 13.3.1 Preparation . . . . . . . . . . . . . . . . . . 13.3.2 Crystal Structures . . . . . . . . . . . . . . . 13.3.3 Magnetic Properties . . . . . . . . . . . . . 13.4 Divalent Metal Phosphonates . . . . . . . . . . . . 13.4.1 Synthesis . . . . . . . . . . . . . . . . . . . . 13.4.2 Crystal Structures . . . . . . . . . . . . . . . 13.4.3 Magnetic Properties . . . . . . . . . . . . . 13.5 Metal(II) Diphosphonates . . . . . . . . . . . . . . 13.5.1 Synthesis . . . . . . . . . . . . . . . . . . . . 13.5.2 Crystal Structures . . . . . . . . . . . . . . . 13.5.3 Magnetic Properties . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . .

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

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

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

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

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

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

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

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

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

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

425 425 426 426 426 427 427 430 430 432 437 449 449 450 452 454

14 Magnetic Langmuir-Blodgett Films . . . . . . . . . . . . . . . 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 The Langmuir–Blodgett Technique . . . . . . . . . . . . 14.2.1 Fabrication of Langmuir–Blodgett Films . . . . . 14.2.2 Structural and Physical Characterization of Langmuir and LB Films . . . . . . . . . . . . . 14.3 Magnetic Systems: A Molecular Approach . . . . . . . . 14.3.1 Hybrid LB Films with Magnetic Clusters . . . . . 14.3.2 Spin-transition Systems . . . . . . . . . . . . . . . 14.3.3 Comments on Molecular Magnetism in LB Films 14.4 Extended Systems and Cooperative Effects . . . . . . . 14.4.1 “Literally Two-Dimensional Magnets” . . . . . . 14.4.2 Metal Phosphonate LB Films . . . . . . . . . . . 14.4.3 Organic and Inorganic “Dual Network” Films . . 14.4.4 Bimetallic Compounds . . . . . . . . . . . . . . . 14.5 Comparison with Other Lamellar and Colloidal Systems 14.5.1 Hybrid Lamellar Systems . . . . . . . . . . . . . 14.5.2 Self-organized Media . . . . . . . . . . . . . . . . 14.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

457 457 459 459

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

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

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

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

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

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

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

461 462 462 465 468 469 469 470 473 475 478 479 480 481 482

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

List of Contributors

Pere Alemany Center de Recerca en Qu´ımica Teorica ` (CeRQT) Departament de Qu´ımica Inorganica ` and Departament de Qu´ımica F´ısica Universitat de Barcelona Diagonal 647 08028 Barcelona Spain Santiago Alvarez Center de Recerca en Qu´ımica Teorica ` (CeRQT) Departament de Qu´ımica Inorganica ` and Departament de Qu´ımica F´ısica Universitat de Barcelona Diagonal 647 08028 Barcelona Spain

Kunio Awaga Department of Basic Science The University of Tokyo Komaba, Meguro Tokyo 153-8902 Japan Carlo Bellitto CNR-Istituto di Chimica dei Materiali Area della Ricerca di Montelibretti Via Salaria Km. 29.5 C.P.10 00016 Monterotondo Staz Rome Italy Robert Blinc Josef Stefan Institute P.O. Box 3000 Jamova 39 1001 Ljubljana Slovenia

David B. Amabilino Institut de Ciencia ` de Materials de Barcelona (CSIC) Campus Universitari de Bellaterra 08193 Cerdanyola Spain

R. J. Bushby School of Chemistry University of Leeds Leeds LS2 9JT UK

Denis Arcon Josef Stefan Institute P.O. Box 3000 Jamova 39 1001 Ljubljana Slovenia

Rene´ Clement ´ Laboratoire de Chimie Inorganique UMR 8613, Bt 420 Universite´ Paris XI 91405 Orsay France

XII

List of Contributors

Roberto Cortes ´ Departamento de Qu´ımica Inorganica ´ Universidad del Pais Vasco Apartado 644 48040 Bilbao and Apartado 450 01080 Vitoria Spain Silvio Decurtins Department of Chemistry and Biochemistry University of Berne Freiestrasse 3 3012 Berne Switzerland Pierre Delhaes Center de Recherche Paul Pascal CNRS Universite´ Bordeaux I Avenue du Dr A. Schweitzer 33600 Pessac France Marc Drillon Institut de Physique et Chimie des Materiaux ´ UMR 7504-CNRS 23 rue du Loess 67037 Strasbourg France Albert Escuer Departament de Qu´ımica Inorganica ` Universitat de Barcelona Diagonal 647 08028 Barcelona Spain Wataru Fujita Department of Basic Science The University of Tokyo Komaba, Meguro Tokyo 153-8902 Japan

Mohamed A.S. Goher Chemistry Department Faculty of Science Alexandria University Alexandria 21321 Egypt Tamotsu Inabe Division of Chemistry Graduate School of Science Hokkaido University Sapporo 060-0810 Japan Katsuya Inoue Institute for Molecular Science 38 Nishigounaka Myodaiji Okazaki 444-8585 Japan Hiizu Iwamura National Institution for Academic Degrees 4259 Nagatsuta-cho Midori Yokohama 226-0026 Japan Anne Leaustic ´ Laboratoire de Chimie Inorganique UMR 8613, Bt 420 Universite´ Paris XI 91405 Orsay France Luis Lezama Departamento de Qu´ımica Inorganica ´ Universidad del Pais Vasco Apartado 644 48040 Bilbao

List of Contributors

Talal Mallah Laboratoire de Chimie Inorganique UMR CNRS 8613 Universite´ Paris-Sud 91405 Orsay France

Ales Omerzu Josef Stefan Institute P.O. Box 3000 Jamova 39 1001 Ljubljana Slovenia

Arnaud Marvilliers Laboratoire de Chimie Inorganique UMR CNRS 8613 Universite´ Paris-Sud 91405 Orsay France

Melanie Pilkington Department of Chemistry and Biochemistry University of Berne Freiestrasse 3 3012 Berne Switzerland

Carlo Massobrio Institut de Physique et Chimie des Materiaux ´ 23 rue du Loess 67037 Strasbourg France Mark W. Meisel Departments of Physics and Chemistry University of Florida Florida 32611 USA Dragan Mihailovic Josef Stefan Institute P.O. Box 3000 Jamova 39 1001 Ljubljana Slovenia Christophe Mingotaud Center de Recherche Paul Pascal CNRS Universite´ Bordeaux I Avenue du Dr A. Schweitzer 33600 Pessac France Montserrat Monfort Departament de Qu´ımica Inorganica ` Universitat de Barcelona Diagonal 647 08028 Barcelona Spain

XIII

Yann Pouillon Institut de Physique et Chimie des Materiaux ´ 23 rue du Loess 67037 Strasbourg France Pierre Rabu Institut de Physique et Chimie des Materiaux ´ UMR 7504-CNRS 23 rue du Loess 67037 Strasbourg France Joan Ribas Departament de Qu´ımica Inorganica ` Universitat de Barcelona Diagonal 647 08028 Barcelona Spain Antonio Rodr´ıguez-Fortea Center de Recerca en Qu´ımica Teorica ` (CeRQT) Departament de Qu´ımica Inorganica ` and Departament de Qu´ımica F´ısica Universitat de Barcelona Diagonal 647 08028 Barcelona Spain

XIV

List of Contributors

Teofilo ´ Rojo Departamento de Qu´ımica Inorganica ´ Universidad del Pais Vasco Apartado 644 48040 Bilbao Eliseo Ruiz Center de Recerca en Qu´ımica Teorica ` Departament de Qu´ımica Inorganica ` and Departament de Qu´ımica F´ısica Universitat de Barcelona Diagonal 647 08028 Barcelona Spain Taketoshi Sekine Department of Basic Science The University of Tokyo Komaba, Meguro Tokyo 153-8902 Japan David A. Shultz Department of Chemistry North Carolina State University North Carolina 27695-8204 USA Daniel R. Talham Departments of Physics and Chemistry University of Florida Florida 32611 USA

Jaume Veciana Institut de Ciencia ` de Materials de Barcelona (CSIC) Campus Universitari de Bellaterra 08193 Cerdanyola Spain Ramon Vicente Departament de Qu´ımica Inorganica ` Universitat de Barcelona Diagonal 647 08028 Barcelona Spain Nobuo Wada Department of Basic Science The University of Tokyo Komaba, Meguro Tokyo 153-8902 Japan Isao Watanabe Muon Science Laboratory The Institute of Physical and Chemical Research (RIKEN) Hirosawa, Wako Saitama 351-0198 Japan

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

Subject Index

A3 Cu3 (PO4 )4 39 ab-initio calculations, spin distribution 325, 329–350, 360, 361, 363, 366, 368, 370, 372, 375 AgVIII P2 S6 59, 67, 69, 74, 78, 83–85 alkyl nitroxides 346 amorphous materials 257 anisotropy 3, 4, 9, 10, 41, 42, 53, 373 – coupling 3, 5, 44 – local 64, 193 – magnetocrystalline 44, 211, 216–232 – nonaxial 71 – single ion 50, 54, 55, 57, 71 – XY 50 antiferromagntic coupling prediction, orbital model 62 Ba2 CaMnFe2 F14 20, 21 Ba2 MnCoAl2 F14 42 BCPTTF salts 115, 118, 121 BDT 124 BEDT-TTF, see ET salts bielectronic exchange integral 62 bimetallic chain 1 – Co 32 – CoCu 32 – CoNi 18–19, 35, 36 – Cu 367–370 – CuGd 28 – CuIr 10, 12 – CuMn 24 – CuNi 359–361 – CuPt 10, 12 – MM 15, 16, 35 – Mn 22 – MnCo 34, 35 – MnCu 24, 26–28, 34, 361–367 – MnNi 15, 18, 370–372 – Ni 35 – ring 10

biphenyl radical anion 407 biquadratic exchange 28, 50–52 – spin 3 bis(acetylacetonato)nickel 417 bis(benzene)metal 409 bis-µ-(hydroxo)dinuclear Cu(II) unit 61 bis(2-phenyl-4,4,5,5-tetramethyl-4,5-dihydro-1H -imidazolyl-1-oxyl-3-oxide)copper(II) 351 Bonner-fisher 8, 96, 108, 117, 120, 121 C60 203, 252 Ca3 Cu3 (PO4 )4 40, 41 CaMnO3 315 carboxylate bridges 15 – ligands 1 (CH3 )4 NNi(NO2 )3 59, 70 chain, alternating 2, 9, 64, 95 – anisotropic 54 – bimetallic, see bimetallic chain – classical spin and quantum system 17, 27 – cluster chains 1 – connected 1 – 1-D 95 – ferro- and antiferromagnetic coupling 43 – infinite 43 – linear 2, 8, 43 – mixed valent 203–206 – nonuniform 57, 64 – spin 43 – triangular 1, 9, 22 – uniform 2, 4, 5, 63, 64 charge disproportionation 109 – delocalization 321 – ordering 300, 308–320 4-(4-Chlorobenzylideneamion)-2,2,6,6tetramethylpiperidin-1-oxyl) 342–343

432

Subject Index

circularly polarized light 211 classical approximation 3 classical spin 3–5, 12, 13, 17, 29, 30, 35, 41 CNMR 119 cobalt hexacyanoferrate 258, 261, 270, 272, 291 coercivity 286 Co[N(CN)2 ]2 250 [Co(NH3 )5 (OH2 )][Cr(CN)6 ] 358 Co(OH)(NO3 )H2 O 37, 38, 40 Co(phen)(3,5-DTBSQ)2]L 288–291 Compton scattering 211 conducting polymers 236, 245 conductivity, frequency-dependant 111 – optical 113 contact shift 388–390 cooperative effect 264–266, 269, 290 correlation length 5, 17, 29, 37, 44, 49, 53–54, 73, 77, 78, 98, 120 co-semiquinone 258 Coulomb exchange 340 – interaction 137, 138, 199, 203, 204, 205 – repulsion 121, 192, 193, 206 coupling, electron-nuclear 381 – electron-phonon 95, 117, 121, 122 – exchange coupling 219 – ferromagnetic 158 – Heisenberg, 2D 43 – Heisenberg 3, 4, 17, 21, 24, 28, 29, 42, 49, 50, 163, 175, 176, 189 – Ising coupling 3, 4, 17, 18, 30–41 – isotropic coupling 5, 164, 178 – spin-spin coupling 87 – through-bond 379–380 – through-space 379, 386 – XY 42 cluster 372–375 – CrNi 148–149 – Fe3 S4 186, 189, 190 – single molecule magnet, see single molecule magnet Cr(Benzyl)(C5 H5 )(PEt3 )Cl 403 Cr{(CN)Mn(TrispicMeen)}6 ](ClO4 )9 148 Cr{(CN)Ni(tetren)}6 148 III CrII 1.5 [Cr (CN)6 ] · 5H2 O 271, 272 III [Cr (CN6 ]3− 358

Cr y [Cr(CN)6 ] 257 critical domain 44 – exponent 249 – field 57, 80 crossover, see Spin crossover CsCr1−x Mgx Cl3 87 III Cs0.83 CrII 1.10 [Cr (CN)6 ]3 271–279 III Cs2 K[Cr (CN)6 ] 357 CsNiCl3 58, 59, 77–78, 86 Cs[NiII CrIII (CN)6 ]2H2 O 137–144 Cu(bipy)(OH)2 Cu(bipy) 368 Cu(bipy)(OH)2 Cu(bipy)(OSO3 ) 368 Cu(hfac)2 NitMe 350, 351 Cu(salen)Ni(hfac)2 360 Cu2 (t-Bupy)4 (N3 )2 ](ClO4 )2 369, 370 CuCl2 (NitPh)2 351–353 CuGeO3 87, 96–110, 118, 124 – impurity-doped 105–112 – phase diagram 103–105 – structure 98, 99 Cu2 O(SO4 ) 39–41 CuMn(S2 C2 O2 )2 23 CuII (salen)NiII (hfac)2 359 cyanophenyl nitronyl nitroxide 248 DAP(TCNQ) 115–118 DCNQI 252 density matrix renormalization 366 density wave 115 di-µ-azido copper(II) 367, 369 dihydrobenzotriazinyl radicals 406–408 1,2-Dihydro-2-methyl-2-phenyl-3H-indole-3-oxo-1-oxyl 331 dimerization-induced gap 97 dimers, asymmetric 164, 165 dioxygen 61 diphenylpicrylhydrazyl 405 dipolar coupling 385, 386 – shift 388–390, 419 direct exchange 157 direct transfer 157 DMR 367 domain 118, 286, 311 double chain 1, 2, 9, 37, 38, 43 double-exchange 155, 156, 158–166, 175–186, 188, 190–197, 199–201, 203–207, 300 – anisotropic 166 double spinon 99

Subject Index DPPH 406 DTDA 349 Dzialoshinski coupling 3, 42 easy magnetization axis 50, 373 elastic neutron scattering 83 electron spin resonance, see EPR electron transfer 159–160, 197, 199, 200, 207 – salt 252 ENDOR 382, 406 energy gap 110 entropy 288 EPR 72, 79, 81, 82, 86, 101, 107, 110, 124, 240, 245, 372, 382–383 ESR, see EPR ET salts 115, 116, 121, 123, 237, 251, 252 2-(6-Ethynyl-2-pyridyl)-4,4,5,5-tetramethylimidazoline-1-3-oxide 346 [Et3 NH]2 [Mn(CH3 CN)4 (H2 O)2 ] [Mn10 O4 (biphen)4 Br12 ] 373 exchange transfer 183, 184 Faraday effect 212 far-infrared 101–102, 113 [Fe(C5 Me5 )2 ][TCNE] 249 [Fex (Co1−x (btr)2 (NCS)2 ] 261, 262, 266, 267, 269 Fe2 O3 41 [Fe8 O2 (OH)12 (tacn)6 ]Br 374 Fe3 S4 cluster 186, 189, 190, 193, 194 ferrimagnetic chain 4, 5 – Ising 32–37 – linear 5–8, 14, 24, 39, 40 – random 14 ferrimagnetism, 1-D, topoligical 2, 20, 37, 38 ferrimagnets 5, 40 ferrites 257 ferromagnetic metal 310 Fisher model 12–14, 18 frustration 95–97, 181 g factor 217 [Gd2 (ox)][Cu(pba)]3 [Cu(H2 O)5 ] glavinoxyl radical 406–407 haldane spin chain 49–88 half-filled magnetic shells 41 hard axis 218

28–29

433

heat-capacity 100 Heisenberg antiferromagnet 49, 86 – chain 4, 5, 17, 43, 54, 72, 84, 87, 88, 95, 104, 108, 250 – alternating 8, 10, 11 – antiferromagnet 103 – classical-spin 12–22 – ferrimagnetic chains 23–29 – ferromagnetic 6, 7 – quantum spin 8–11 – random 19, 107 – uniform 8 – double chain 29 – ladder 88 – model 76 hexacyanometallates, spin distribution 420–421 high-field magnetization 80 4-hydroxy-2,2,6,6-tetramethyl-1-piperidinyloxy 346 hydroxyhenyl nitronyl nitroxide 248 hyperconjugation 402, 407 hyperfine couplings 382 – interactions 421 hysteresis 314 – light-induced optical 265, 268 – light-induced presure 265, 267 – light-induced thermal 265 2-imidazoline-1-oxide radicals 406–408 incommensurate structure 311 indirect transfer 157 indolinonic nitroxide 332 inelastic 70 – light scattering 99 – neutron scattering 54, 72–75, 79–82, 84, 100, 108, 110, 372 infrared 112, 117, 121 intervalence absorption 169, 179, 271, 258 iron(II)bis(hydrotris(pyrazol-1-yl)borate) 391 Ising anisotropy 50 – chains 30–41 – ferrimagnetic 32–37 – ferromagnetic 34 – model 3, 100 isotropic 42, 55 itinerant electron 156 Jahn-Teller effect 185–187, 194, 202, 308

434

Subject Index

Kagome antiferromagnet 251 K0.2 Co1.4 [FeII (CN)6 ] 270 Kerr effect 212 KFeIII [FeII (CN)6 ] 270 kinetic energy 183, 194 kinetic exchange 183 Knight shift 101, 421, 422 IV (Lax Ca1−x )(MnIII x Mn1−x )O3 156 Ladder chains 1, 9, 37, 38, 108, 109, 114 Langevin function 14, 15 Langmuir-Blodgett films 124, 264 Larmor energy 83 – precession 239 LB films, see Langmuir-Blodgett films LIESST 257, 261, 262, 268, 269, 289 light-induced – excited spin state trapping, see LIESST – ferrimagnetism 258 – magnetic pole inversion 258, 271 light-stimulated magnetic after effects 257–292 linear chain, see chain linear long-range order 44, 51, 52, 71

MAE, see magnetic after effects magnetic after effect, photo stimulated, see light-stimulated magnetic after effects magnetic after effects 257, 278 magnetic circular dichroism, X-ray 133–151, 211–232 magnetic dimers 155 magnetic dipole 215 magnetic domain 216 magnetic dots 216 magnetic exchange 193, 200 magnetic excitations 101 magnetic fluctuations 73, 84 magnetic force microscope 216 magnetic metastability 258, 291 magnetic orbitals 359, 379 magnetic scattering 211 magnetic sensors 124 magnetic storage 124 magnetic structures 325 magnetism, 1-D 1–45, 49–88

magnetization density 325 magnetocrystalline anisotropy, see anisotropic, magnetocrystalline magnetoelastic 95, 112, 117 magnetoresistance, colossal 156, 300–306 magnetostatic 13 magnetostriction 100, 104, 312 magnons 72 – phonon interactions 3 manganites 156, 300, 315 – change ordering 300, 308–320 – CMR 300–306 – doping 320–322 – rare earth 259, 300–322 maxent 328, 333, 334, 338 maximum of entropy 327, 359 – method 373, 374 McConnell mechanism 249, 381, 410 MEM(TCNQ)2 115, 116, 250, 251 metal alkyl 415 metal porphyrins 399, 400, 418 metal to insulator transition 315, 321, 322 metallocenes 392–396, 401, 403, 409–414 metallocenium tetracyanoethenides 249, 380 metallophthalocyanines 204–207, 357 metalloproteins 151 – MCD 150 2-(3-N -Methylpyridium)-4,4,5,5-tetramethyl 251 mexican hat 185 mixed valence 288 – clusters 197, 198 – dimers 156–180 – tetramers 190–196 – trimers 180–190 Mn12 O12 (CH3 CO2 )16 (H2 O) 253, 372, 373 [MnTEtOPP][TCNE] 258 molecular-beam epitaxy 224 motional narrowing 243 [MPYNN][BF4 ] 251 multilayers 216, 219 – Cox Pt1−x thin film 224–232 – Fe/Cu/Co magnetic multilayers 215 muon spin rotation 117, 235–253

Subject Index mixed valent – clusters 199, 201, 206, 207 – compounds 168 – dimers 158, 160–167, 169, 171, 173–176, 178, 180, 181, 183, 184, 196, 203–205, 207 – asymmetric 164 – orbital degeneracy 165 – tetramer 191–195 – trimers 182, 183, 185–189 muon spin resonance, see muon spin rotation muon spin rotation 235–254 naphthyl nitronyl nitroxide 248 NaV2 O5 96, 108–116, 124 NC(C6 F4 )(CNSSN) 249 Neel ´ state 6, 51, 56, 58, 71, 115 NENF 64, 70 NENP 62–65, 70, 73–76, 78–81, 83, 84, 86 – Cu-doped 85 neutron diffraction 139, 146 – scattering 70, 71, 74, 95, 99, 106, 107, 117, 325–353, 357-376 – spin polarized 150, 217, 325–353, 357–376 59, 62, 65–67 Ni(C2 H8 N2 )2 NO+ 2 Ni(C3 H10 N2 )2 N+ 59 3 59 Ni(C3 H10 N2 )2 NO+ 2 59 Ni(C5 H14 N2 )2 N+ 3 NiL(diamine)2 64 Ni[N(CN)2 ]2 250 Ni(NH3 )4 (NO2 )2 372 Ni1−x Mgx NiCl3 86 NiII (NH3 )4 (NO2 )2 357 Ni nitroprusside 258 NINAZ 63, 64, 69, 70, 78, 79 NINO 62, 63, 64, 70, 73 6-NitPy(C≡C−H) 346, 347 nitronyl nitroxides 246–249, 334, 336, 343–347, 350, 380, 383, 384, 389, 406–408 nitroxides 249, 331, 346, 394 NMR, see nuclear magnetic resonance nonaxial anisotropy, see anisotropy, nonaxial nonlinear optical solids 124 nonorthogonalized magnetic orbitals 62

435

m-NPIM, 2-(3-nitrophenyl)-4,4,5,5-tetramethyl4,5-dihydro-1H -imidazol-1-oxyl 338, 339 p-NPNN, 2-(4-nitrophenyl)-4,4,5,5-tetramethyl-4,5-dihydro-1H -imidazol-1oxyl-3-oxide, 246–248, 340, 341 nuclear magnetic resonance 72, 73, 83, 100, 101, 109, 116, 118, 120, 240, 383–392 – spin distribution determination 381–422 occupation correlation 20 one-dimensional ferromagnetism 205–206 optical 118 – information storage 257 – mesurements, under pressure 102 – spectroscopy 102, 121, 139, 169 orbital-ordering 300 order induced by disorder 87 organic radicals 406 orthogonalized magnetic orbitals 62 4-oxo-2,2,6,6-tetramethyl-1-piperidinyloxy 331 Peierls transition, electronic 122 Perovskite manganites 300 perpendicular magnetocrystalline anisotropy, PMA 224 perylene 122–123 2-phenyl-4,4,5,5-tetramethyl-4,5-dihydro1H -imidazol-1-oxyl-3-oxide 333 photobleaching 262 photochromic 258 photodarkening 261, 262 photoemission 108 photoexcited state 257 photoinduced excited spin state trapping, see LIESST photoinduced ferrimagnetism, see light-induced ferrimagnetism photoinduced magnetic after effects, see light-stimulated magnetic after effects photoinduced magnetic pole inversion, see light-induced magnetic pole inversion photoinduced optical hysteresis, see hysteresis, light-induced optical

436

Subject Index

photoinduced pressure hysteresis, see hysteresis, light-induced pressure photoinduced thermal hysteresis, see hysteresis, light-induced thermal photoisomerizable 258 photo-tunable magnet 271 Piepho, Krausz, and Schatz model 155, 166–168, 170, 171, 173–175, 178, 179, 185–187, 194, 201, 202, 208 Piepho model 169, 170, 176, 178, 186, 187, 194 piezochromic 258 point-dipole model 386 polarized neutron diffraction, see neutron diffraction 150, 217, 325–353, 357–376 polyacetylene 245 polyoxometalates 197–203 potential exchange-transfer 184 prussian blue structured magnets 137–146, 258, 261, 263, 270–280, 291 – analogs 258, 291 pyrene 406–407 pyrene2 {M[S2 C2 (CN)2 ]2 } 115, 116, 121–124 quantum field theory 79 quantum magnetic fluctuations 49 quantum spin 4, 5, 12, 17, 24, 26, 29, 30, 35, 158 – classical spin 24, 35, 36 – chains 49–88 – tunneling 253 3-quinolyl 248 Raman scattering 98, 104, 108, 112 Rbx Co[FeII (CN)6 ] y 263, 271, 280 relaxation rate 83 remanent magnetization 273 resistivity 122 RKKY 244 R2 NiO5 87 second nearest neighbors 10 Seiden model 23 semimagnetic semiconductors 257 short-range order 44 single ion anisotropy, see anisotropic, single ion single molecule magnet 253, 372–375

soliton 99, 120 – excitations 116 specific heat 33, 38, 69, 70, 80, 101, 116 – high field 80 spin, correlated 56 spin crossover 44, 257–261, 265–267, 269, 288, 289, 291 spin delocalization 330, 333, 357, 358, 363, 370, 399–405, 412, 417 spin density, see spin distribution spin density wave 252 spin distribution 151, 325, 332, 334, 335, 339, 340–346, 349–352, 353, 357–360, 362, 364, 366, 368–375 – negative 335, 343, 361, 362, 370, 381 – NMR 379–422 – wave 242, 251 – transition metal complexes 357 spin dynamics 74–85 spin echo 100 spin flop 106 spin fluctuations 49 spin frustration 2, 22, 39 spin glass 244 spin pairing 124 spin Peierls 57, 63, 87, 95–124, 250 spin polarization 333–336, 357, 358, 396–398 spin population, see spin distribution spin state trapping 257 spin transfer 359, 402 spin wave 6 spin zero defect 80 spinels 257 spin-change 258 spin-lattice relaxation 241, 385 spin-orbit coupling 28, 60, 137, 142, 143, 148, 216–219, 221–223 spin-phonon coupling 105, 111, 117 spin-spin correlation 14, 15, 20, 51 spin-spin relaxation 241 Sr3 CuPt1−x Irx O6 9 SrMnO3 315 structural relaxation 257 superconductors, organic 236, 237 superexchange interaction 157, 308, 379 superparamagnetism 253 superstructure 310 synchrotron radiation 131, 211

Subject Index tanol suberate 249, 250 TCNE 249, 258, 338, 410, 422 TCNQ 96, 115–118, 250, 251, 422 [TDAE][C60 ] 252 tempo 249, 331, 342, 346 – spin distribution 332 tempone 331, 346 – spin distribution 332 thermal conductivity 110, 111 thermochromic 258 thermodynamic 288 thin films 211, 216, 217, 219, 224–232 2-(4-thiomethylphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide 343 through-bond coupling, see coupling, through-bond through-space coupling, see coupling, through-space TMMC 85, 241 TMNIN 63, 64, 70, 83, 85, 86 TMTSF, superconductor 118, 251 [TMTSF]2 [PF6 ], superconductors 242 [TMTTF]2 PF6 115, 116, 118–120 topological 1D ferrimagnetism, see ferrimagnetism, 1-D, topological transfer frustration 181, 193 transfer integral 156, 157 transfer-matrix method 30–32 triangular chain, see chain, triangular triarylaminium cations 406–408 triazolinyl radicals 406–408 tricritical behavior 111 trimers 39 triple chain 2 TTF 96, 115, 116, 422

437

uniform chain, see chain, uniform valence-localized 175 valence tautomerics 258, 259, 288–291 Van Vleck 216 – equation 195 – paramagnetism 24, 27, 166 V[CrIII (CN)6 ]z magnets, MCD 144–146 Verdazyl radical 406, 407 vibronic coupling 168, 174–177, 179, 186, 195, 203, 208 – effects 167 – interaction 164, 166, 170, 185, 189, 194 – model 170 weak ferromagnetism 41 XANES 271 x-ray absorption spectroscopy 131–133, 140, 211 x-ray diffraction 120, 372 x-ray diffuse scattering 98 x-ray neutron diffraction 312 x-ray scattering 107 XY anisotropy, see anisotropy, XY XY coupling, see coupling XY XY magnet 249 Y2 BaNiII O5 59, 67, 69 YBANO 63, 74–76, 78, 79, 83, 84, 86 ZN-doped YBANO 86 zero-field splitting 30, 60

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

1

Nitroxide-based Organic Magnets David B. Amabilino and Jaume Veciana

1.1

Introduction

Nitroxides [1] are arguably the group of organic radical molecules [2] which have found most application in the study of molecular magnetism [3], as materials in their own right [4] or in combination with paramagnetic metal ions [5]. The reason for their appeal stems from their relative synthetic accessibility, stability, and versatility, which gives rise to a wide range of magnetic behavior, not to mention other properties [6]. In the vast majority of all the nitroxides reported to date, the NO group carrying the free electron is safeguarded sterically by methyl groups attached to the adjacent carbon atoms, endowing persistency upon the radical. The four most widely studied families of radicals (Fig. 1) are the unconjugated nitroxides (derivatives of TEMPO (1)), the conjugated nitroxides (such as phenyl N -t-butyl N -oxides (2)), the nitronyl nitroxides (3), and the imino nitroxides (4). These divisions of open-shell compounds differ dramatically in the distribution of the free electron in the molecule, as determined experimentally [7] and theoretically [8]. While in the simple nitroxides 1 the NO group bearing the free electron is comparatively isolated from the rest of the molecule by saturated hydrocarbon groups and therefore bears the vast majority of the spin density of the free electron (as represented by the singly occupied molecular orbital, or SOMO), in the nitronyl nitroxides

Fig. 1. Four of the general types of nitroxide: derivatives of TEMPO (1), phenyl N -t-butyl N -oxides (2), the nitronyl nitroxides (3), and the imino nitroxides (4), and the SOMOs of the last two cases.

2

1 Nitroxide-based Organic Magnets

(3) the electron is delocalized across the ONCNO group, with a node located at the central carbon atom. In contrast, in the imino nitroxides (4) the carbon atom of the NCNO moiety is not necessarily a node of the SOMO and in principle there is no reason for its spin population to be small or negative [9]. Experimentally, negative spin population is observed on this carbon atom [9, 10]. Nevertheless, the effect is less pronounced than in the radicals of type 3. In the latter, the effect is purely a result of spin polarization while in the imino nitroxides it is the result of competition between spin polarization and spin delocalization that determine the sign and magnitude of the spin on this atom. The degree of dispersal of electron density is clearly a key factor in the determination of magnetic interactions between the magnetic moments of the radicals in the condensed state because the most important interactions arise from the magnetic exchange mechanism. Exchange magnetic interactions between spins S1 and S2 located on organic units are described in this chapter by the effective Heisenberg Hamiltonian H = −2J S1 S2 , that has an isotropic nature and where J is the exchange interaction energy, traditionally given either in cm−1 or, as we shall use, in K as J/k (where k is the Boltzmann constant), this constant taking positive values for ferromagnetic interactions. Perhaps the most appealing (and elusive) goal in the area of organic magnetic materials is the preparation of a bulk ferromagnet [11]. The first indication that an organic radical might display bulk ferromagnetism was in the diradical tanol suberate (5) [12, 13], although it was later reported that the material is in fact a metamagnet [14]. The target was reached unequivocally for the first time by Kinoshita and colleagues with 4-nitrophenyl nitronyl nitroxide (4NPNN, 6) [15]. Interest in the magnetic properties of organic magnetic materials underwent a boom in the early part of the 1990s as a result of this revelation.

As in all magnetic materials, three main factors contribute to the overall magnetic behavior, two of them molecular, i. e. molecular topology [16] and molecular conformation, and the third supramolecular (beyond the molecule) [17], the packing of the molecules in the condensed phase. The first two factors govern the distribution of the free electron in the molecule and interaction between the spins if there is more than one free electron, and the third the magnetic interactions through space and/or non-covalent bonds. These themes will recur during the course of this essay. For unpaired electrons to interact ferromagnetically the overlap integral of their orbitals must be close to zero, a situation favored by orthogonal orbitals (according to Hund’s rule developed for atoms) [3, 18], while the exchange integral must be

1.1 Introduction

3

large. The effects of molecular topology [4, 16] on intramolecular magnetic interactions are adequately described employing either molecular orbital or valence bond theories. As far as intermolecular magnetic interactions are concerned, the most frequently-applied model is “McConnell I” which was first put forward to prophesy ferromagnetic interactions between π -conjugated hydrocarbon radicals [19] and implies the spin-polarization of adjacent nuclei. This model predicts a ferromagnetic interaction between two moieties if the π -orbital overlap between atoms with spin densities of opposite sign dominates over others with the same sign. However, this model or mechanism has been called into question very recently as a result of theoretical work [20] and statistical studies [21]. Another mechanism that has also (although less frequently) been used to explain interactions is the so-called “McConnell II” that implies charge transfer between different molecular units producing an admixture of ground and excited state configurations with distinct spin multiplicities [2–4, 22]. Once an intermolecular ferromagnetic interaction between molecules has been achieved, it is necessary to propagate it, or an interaction of similar sign and magnitude, in all three directions of space (Fig. 2) [3]. In order to comply with this condition, a three-dimensional packing of the molecules is essential. The tendency of organic molecules to form one-dimensional chains or two-dimensional sheets often acts as an impediment to achieving this feat, since although ferromagnetic interactions could be propagated within the chain or sheet (Fig. 2), often the interaction between the chains or sheets is antiferromagnetic. It is therefore necessary to control completely the packing of the molecules in the whole of the crystals, an obstacle which is largely unsolved at the present time, although considerable effort is being expended to overcome it [23]. The packing of organic molecules in the solid state is governed by non-covalent bonds acting between them, and the area of supramolecular chemistry which is concerned with this phenomenon has become known as crystal engineering [24]. The non-covalent bonds which are of particular importance in the realm of the nitroxide magnets are hydrogen bonds [25]. They have proven to be involved in the transmission of magnetic interactions in inorganic complexes [26], and, as we shall see, are often invoked to explain the magnetic behavior of purely organic materials. As far as the nitroxides are concerned, there are principally three important types of hydrogen bonds (Fig. 3): (i) those formed by the oxygen atom of the N–O group with donors of hydrogen bonds (R–X–H), be they hydroxyl groups, amines, or the like; (ii) those formed by the hydrogen atoms attached to the methyl groups “protecting” the radical and acceptors of hydrogen bonds, such as basic nitrogen or oxygen atoms; and (iii) those formed by the latter acceptor and hydrogen atoms attached to aromatic rings. The Csp3 –H· · ·O hydrogen bonds [27] often include those formed with the NO group of the radical. Since the hydrogen atoms of these methyl groups have negative spin density, because of spin-polarization, the McConnell I mechanism has been applied frequently to implicate this non-covalent interaction as one which encourages ferromagnetic interactions. As we shall see throughout the course of this chapter, there is a tight relationship between hydrogen bonding and ferromagnetic interactions in nitroxide-based organic molecular materials [4]. In the following sections, we will discuss the most outstanding results attained in the area of purely organic nitroxide magnets in general, and reflect upon these

4

1 Nitroxide-based Organic Magnets

Fig. 2. Schematic representation-the dots represent the radicals that contain the spins in the magnetic system-of magnetic interactions in a one dimensional system, wherein J1 is much bigger than J2 and J3 , a two dimensional system, in which the exchange couplings J1 and J2 are larger than J3 , and a three dimensional magnetic system, where J1 , J2 and J3 are of similar magnitude.

achievements, the limitations associated with their behavior and interpretation of it, and discuss the possible ways in which we foresee that research will advance. The results are divided into the four structural groups according to Fig. 1 and the materials will be introduced in this order for mono and oligo radicals. Finally the sticky problem of polymeric nitroxide materials will be treated, and we shall reflect upon the state of the art in the area.

1.2 Unconjugated Nitroxides

5

Fig. 3. Some of the hydrogen bonds which drive the assembly of crystals of nitroxide radicals.

1.2

Unconjugated Nitroxides

Interest in applications of the nitroxides of the TEMPO type (1) and related compounds initially revolved around their use as spin probes [28], as a result of their extremely simple EPR spectra – three lines resulting from the coupling of the unpaired electron with the nitrogen nucleus – and the development of theory explaining changes in their line shapes. They were also the first group of organic radicals to be studied in depth regarding their magnetic behavior in their condensed phases.

1.2.1

Mono-nitroxides

The radical 2,2,6,6-tetramethyl-4-piperidinol-1-oxyl, more colloquially known as TANOL (7) [29], was shown very early on in the history of magnetism of free radicals to follow a one-dimensional Heisenberg model in the temperature range 1.8 to 300 K, with an exchange constant J/k of −6.0 K [30]. The molecular material also orders

6

1 Nitroxide-based Organic Magnets

antiferromagnetically at lower temperatures as a result of inter-chain interactions, having a Neel ´ temperature (TN ) of 0.49 K [31], which rises upon application of pressure to the sample as a result of the softness of the material [32]. The structure of the compound is characterized by a one-dimensional chain of molecules with a hydrogen bond between the hydroxyl group of one molecule and the NO group of the next [33]. Acrylic and methacrylic ester derivatives of the compound have been claimed to show antiferro- and ferromagnetic intermolecular interactions, respectively [34]. The methacrylic ester is a metamagnet below 0.16 K [35], and the methacrylamide derivative also shows similar behavior [36].

A very similar radical to TANOL structurally is the 4-hydroxyimino derivative of TEMPO 8, which also forms hydrogen-bonded chains in its crystals [37], and shows a transition to bulk ferromagnet [38]. The critical temperature for the material is 0.25 K, and it has a Weiss constant of +0.43 K at higher temperatures [39]. A detailed study of the compound and its perdeuterated homologs while having no effect on the critical temperature, did allow a detailed analysis of the spin density distribution in the compounds through solid-state NMR studies [40]. The negative hyperfine coupling constants in both the NOD group and the axial methyl groups of the radical, which form hydrogen bonds to the NO group and thereby unite the chains of molecules, indicated to the authors that these two non-covalent interactions result in ferromagnetic interactions in the two directions [41].

Perhaps the group of organic radicals which have the largest “success rate” – judged by the relation between materials showing ferro- and anti-ferromagnetic interactions – are the arylmethyleneamino-2,2,6,6-tetramethylpiperidin-1-oxyl radicals (9, Table 1), because at the last count [42] 52 of 165 radicals exhibit ferromagnetic [43] interactions. In addition to that, six of these molecular materials (summarized in Table 1) show transitions to bulk ferromagnets in their crystalline states, and another six show metamagnetic behavior (Table 2) [42]. The first of the family to be characterized as a bulk ferromagnet was PhATEMPO (9, R = Ph) [44], which exhibits a typical hysteresis curve, and displays a relatively high coercive force of 106 Oe [45]. While both 4PhPhATEMPO (9, R = 4-PhPh) [46]

7

1.2 Unconjugated Nitroxides Table 1. Formulas and properties of ferromagnetic nitroxides 9. Code Space group Z

Curie temperature Weiss constant

Ref.

PhATEMPO P21 /c 4

TC = 0.18 K θ = +0.74 K

45

4ClPhATEMPO P21 /c 4

TC = 0.4 K θ = +0.69 K

47

4IPhATEMPO P21 /c 4

TC = 0.4 K θ = +0.71 K

42

4MeSPhATEMPO P21 /c 4

TC = 0.34 K θ = +52 K

52

4PhPhATEMPO P21 /c 4

TC = 0.4 K θ = +0.62 K

46

4PhOPhATEMPO Pbca21 8

TC = 0.20 K θ = +0.39 K

52

and 4ClPhATEMPO (9, R = 4-ClPh) [47] display slightly higher Curie temperatures than the first example, their hysteresis curves indicate only very small coercive forces (10 and 5 Oe, respectively). The more recently reported 4IPhATEMPO (9, R = 4-IPh) [42] appears to have a coercive force of approximately 100 Oe. Zero-field muon spin rotation (µ+ SR) experiments confirmed the spontaneous magnetization in 4ClPhATEMPO [48] and indicated a mean internal field of 75 G [49]. In their crystals, 4ClPhATEMPO, 4IPhATEMPO, and 4PhPhATEMPO are isostructural [50], and all the ferromagnetic radicals have certain features in common [51]. The hydrogen atoms of the methyl or methylene groups at β-positions with respect to the NO moiety form hydrogen bonds with the oxygen atom which bears the unpaired electron. The spin-polarization of these hydrogen atoms, giving rise to spin-alternation in the backbone as shown schematically in Fig. 4, was proposed [52] as the mechanism which operates in these nitroxides, as well as in TANOL suberate (5), the authors having ruled out the possibilities of direct magnetic interactions, such as direct exchange, or dipole-dipole interactions. The proposed hypothesis has been backed-up by theoretical studies [53].

8

1 Nitroxide-based Organic Magnets

Table 2. Formulas and properties of metamagnetic nitroxides 9. Code Space group Z

Neel ´ temperature Critical magnetic field for spin-flip induction (at temperature)

Ref.

2PyATEMPO n.r.*

TN = 0.26 K 10–20 Oe (0.05 K)

42

4PyATEMPO P21 /c 4

TN = 0.12 K 110 Oe (0.09 K)

42

2NATEMPO Pna21 4

TN = 0.12 K 180 Oe (0.04 K)

54

35ClPhATEMPO Pbca 8

TN = 0.12 K 20 Oe (0.05 K)

42

34ClPhATEMPO n.r.*

TN = 0.10 K 20 Oe (0.04 K)

42

26ClPhATEMPO n.r.*

TN = 0.20 K 20 Oe (0.04 K)

42

* Crystal structure not reported.

The 2-naphthyl derivative’s metamagnetic behavior was also described appealing to the spin polarization mechanism through the hydrocarbon network, since a stack of radicals forms along one axis of the crystals united by the weak hydrogen bonds described previously [54]. Particularly interesting is the magnetization curve for this compound – magnetic hysteresis was observed at 40 mK when the ferromagnetic phase was entered (Fig. 5), but on decreasing the applied field from this condition, magnetization remained then dropped suddenly as zero field was approached, presumably as a result of the formation of an antiferromagnetically ordered ground state. Similar behavior is observed when the field is applied in the opposite direction. While the exact mechanism of the change in magnetization was not clear, the authors suggested ferromagnetic interactions (J/k = +0.2 K) in the column or stack

1.2 Unconjugated Nitroxides

9

Fig. 4. Schematic representation of the proposed spin-polarization mechanism present along the b axis of crystals of PhATEMPO (9, R = Ph) [52].

Fig. 5. Illustration of the magnetization curve at 40 mK for the purely organic metamagnet, 2-naphthylmethyleneamino TEMPO [54].

of radicals, while interchain interactions, whose route is not clear, are weakly antiferromagnetic (J/k = −0.02 K). The formation of salts of nitroxides functionalized with carboxylate groups is proving to be an interesting way to try and influence magnetic interactions in these materials by using crystal engineering tools. The sodium and potassium salts of 4-carboxy-TEMPO (10) show ferromagnetic interactions present in their crystals, as determined by susceptibility (Fig. 6) as well as magnetization experiments [55], whereas the parent acid shows antiferromagnetic interactions. Fitting of the susceptibility data to the Bleaney–Bowers equation gave values of J/k of +0.17 and +0.18 K for the sodium and potassium salts, respectively, and Weiss constants of +0.5 and +0.6 K when the high temperature data were fitted to the Curie-Weiss law. The

10

1 Nitroxide-based Organic Magnets

Fig. 6. Magnetic susceptibility data for the sodium (crosses) and potassium salts (dots) of 4-carboxy-TEMPO (10) along with best fits using linear chain (dashed line) and Bleaney Bowers (solid line) models, and the network of sodium ions coordinated to carboxylates and NO groups [55]. Reproduced by permission of The Royal Society of Chemistry.

crystal structure of the sodium salt revealed that the shortest distance between NO groups is through Na2 O2 parallelograms (Fig. 6), and the authors proposed that they may play an important role in the magnetic coupling in the material. Crystallization of salts formed between 2,2,5,5-tetramethyl-3-carboxypyrroline-1oxyl (11) and benzamidinium cations (12) results in solids with sheet-like structures in which the components are linked by strong hydrogen bonds forming a salt bridge [56]. The magnetic interactions between the radicals are weakly antiferromagnetic. Incorporation of a water molecule into one of the salts reduced dramatically these interactions, implying that the hydrogen bonds aid the transmission of exchange interactions.

1.2.2

Oligo-nitroxides

In general, the unconjugated oligo-nitroxide radicals present little or no appreciable through-bond coupling of the free electrons in the molecule. The aforementioned tanol suberate (5) prepared by Rassat and colleagues [57] is a classic example of this phenomenon, and is representative of the majority of the oligo-nitroxides derived from TEMPO [58], since the magnetic interaction through the hydrocarbon skeleton

1.2 Unconjugated Nitroxides

11

is extremely small, all the spin density being located equally on the nitrogen and oxygen atoms [59]. The diester 5 has ferromagnetic interactions in the solid within planes [60] containing the radical moieties from different molecules, which interact antiferromagnetically between them [14]. Therefore, the covalent linker serves only to influence crystal packing and hence relative arrangements of spins. A more recent and interesting example is the tetraradical 13 which has recently been shown to have ferromagnetic interactions (θ = +0.71 K) present in its crystals, wherein two centrosymmetric molecules are present in the unit cell [61]. Relatively short distances are observed both inter- and intramolecularly, with near orthogonality of the SOMOs, which was taken as the motive for the ferromagnetic interactions in the material.

An efficient approach to the generation of ferromagnetic interactions within a molecule is to ensure that the SOMO orbitals are orthogonal to one another [62]. Rassat and Chiarelli designed a family of diradicals meeting this condition because of the incorporation of a rigid adamantane-type skeleton in the molecules [63]. In all of the family, intramolecular ferromagnetic interactions are present, as ascertained by studies in solution [64]. However, while radicals 15 and 16 show dominant antiferromagnetic interactions in the bulk state, the diradical 14 (which a priori is the most symmetric (D2d ) of the derivatives) is a bulk ferromagnet [65], with the highest Curie temperature (TC ) of all the reported compounds containing only carbon, hydrogen, oxygen and nitrogen, at 1.48 K, and having a Weiss constant of +10 K. As dictated by the small magnetic anisotropy of this compound no hysteresis could be observed, which is the expected magnetic behavior of a “soft” magnet. Two crystalline phases have been detected for the compound but only one – the α-phase – is

12

1 Nitroxide-based Organic Magnets

a bulk ferromagnet [66]. In the ferromagnetic α-phase the NO groups of the radicals are arranged intermolecularly in a head-to-tail manner along one direction with the intramolecular interaction being perpendicular to this chain. Polarized neutron diffraction studies revealed that the spin density is located mainly in the ( orbitals of the nitrogen and oxygen atoms, although some is also detected on the contiguous CH2 groups [67]. The alternation of the sign of spin density of the carbon atoms linking the two NO groups was taken as an indication that the intramolecular ferromagnetic coupling is a consequence of exchange through the weakly polarized carbon framework. Interestingly, in its crystals, the molecules of 14 are not D2d symmetric as result of a non-planar conformation of the NO group, which therefore finds itself in a chiral (C2 ) situation [68].

1.3

Conjugated Nitroxides

The family of nitroxides having an aromatic ring directly attached to the nitrogen atom of the radical have also provided a rich variety of magnetic behavior. One of the most important features of this family of radicals is the extensive delocalization of the free electron over the aromatic ring, because it provides a pathway for it to interact magnetically with its neighbors. The most representative examples of this type of conjugated radicals are summarized below.

1.3.1

Mono-nitroxides

The di- p-anisyl nitric oxide radical (DANO, 17) was the first member of this type of open-shell molecular material in which the magnetization was studied in detail. The radical has characteristics of an isotropic, nearly two-dimensional quadratic, magnetic system with antiferromagnetic interactions [69], with a ratio of interplane and intraplane exchange interactions of the order of 10−3 , in line with the solid-state structure of the compound which consists of sheets of NO groups, each with four nearest neighbors. The coupling within the molecular sheets J/k was estimated as −2.45 K, and the sample exhibited an antiferromagnetic spin ordering at 1.67 K [70]. The absence of an anomaly in the EPR spectrum at the Neel ´ temperature was taken as an indication of short-range magnetic ordering [70].

1.3 Conjugated Nitroxides

13

A radical which is closely related to DANO is 9,9-bis(4-tolyl)-9,10-dihydroacridin10-yloxyl (BTAO, 18), which forms a dimer in its crystals with the acridine parts of the molecule π-stacked in such a way that the expected magnetic coupling is ferromagnetic according to the McConnell I mechanism [71]. The observed magnetic behavior does indeed show intradimer ferromagnetic coupling (J/k = +8.85 K), while inter-dimer interactions are antiferromagnetic with J /k = −0.16 K, assuming that inter-dimer interactions occur along two directions only among the four nearest neighbors.

A very interesting recent development is the report by Reznikov and colleagues of vinyl nitroxides, owing to its exemplification of a manner to augment the strength of intermolecular magnetic interactions in purely organic compounds through enhanced delocalization of the free electron [72]. The compound 19 (Fig. 7) exhibits extremely strong antiferromagnetic interactions perhaps because it crystallizes forming chains with short distances between the oxygen atom bearing the spin and the vinyl bond in the heterocyclic ring. The authors implied this feature to account for the magnetic behavior of the material, fitted to a Heisenberg one-dimensional model, with J/k of −101 K. Calculation of the spin densities by ab initio methods suggested localized spin density on the vinyl carbon atom bearing the cyano group (Fig. 7), implying a pathway for the interaction.

Fig. 7. Views of the calculated spin density of a model for the vinyl nitroxide 19. (a) Truncated (two-radical) model system. (b) General view of the spin density distribution of the model system [72]. Reproduced by permission of The Royal Society of Chemistry.

14

1 Nitroxide-based Organic Magnets

Although much work has been performed on the solution-state properties of monophenyl N -t-butyl N -oxides (2), to the best of our knowledge, this depth of investigation has not been matched in the solid state. In contrast, the oligo-radicals have found great interest in the study of intramolecular magnetic coupling, an area which is discussed in the next section.

1.3.2

Oligo-nitroxides

One important goal in the field of organic molecular magnetism is to prepare synthetically a compound incorporating more than one unpaired electron with appreciably large ferromagnetic coupling between them, resulting in a molecule which has a triplet (S = 1) or higher (S > 1) ground state [73]. One of the principle interests in such robust high-spin organic molecules is to employ them as magnetically active ligands with paramagnetic transition metal ions with the purpose of preparing coordination compounds which exhibit magnetic ordering at the highest possible temperatures. A chapter concerning this pursuit will be presented in this book. The engendering of ferromagnetic coupling within a molecule relies on certain moieties capable of sustaining such an interaction, and more often than not the spacer is conjugated. The classic example of this type of ferromagnetic coupler is m-phenylene [74]. In 1969, the bis-nitroxide 20 was shown to possess a triplet ground state [75], although the compound is extremely unstable.

In attempts to improve the stability of this bis-nitroxide skeleton, the groups led by Rassat [76] and Iwamura [77] prepared independently the radicals 21 and 22, respectively. In both these radicals, the ground state of the molecule has been shown to be a singlet. The reason for the breakdown of the expected topological rule appears to be the extremely distorted conformation adopted by the NOtBu groups [78], which are bent out of the plane of the benzene ring, as revealed in the X-ray structure of 22 in which the two NO groups point to the same “side” of the molecule, i. e. in a syn conformation [77]. Indeed, Rassat and coworkers managed to separate conformational isomers of radical 21 thereby demonstrating unambiguously that both isomers have intramolecular antiferromagnetic interactions of J/k = −33 and −40.5 K [76], respectively, similar to that of the syn isomer of 22 (J/k = −36.9 K) [77].

1.3 Conjugated Nitroxides

15

In an extension of their work, Iwamura and coworkers prepared the triradical 23 with the aim of studying “competing magnetic interactions and/or spin frustration in the context of molecular magnetism” [79]. The molecule exhibits an EPR spectrum in which signals attributable to both the quartet and doublet states were observed, the former being thermally populated with a small energy gap between the two magnetic states. In the solid state, the NO groups are oriented virtually perpendicularly to the plane of the central benzene ring. Magnetic susceptibility data of the crystals indicated that the ground state is indeed a doublet, with dominant antiferromagnetic interactions between two pairs of neighboring free electrons which force the ferromagnetic alignment of the remaining pair. It was claimed as the “first demonstration of an organic triradical showing competing interactions” [79]. Ab initio calculations of model compounds showed an angular dependence of the magnetic interaction through the m-phenylene coupler, and correctly reproduced antiferromagnetic coupling when the NO group is highly twisted out of the plane of the benzene ring, implying disjoint molecular orbitals as the origin of this phenomenon [80]. Very recently, the same group reported a triradical which presents curiously a doublet ground state, and concluded that not only topology but number of π -electrons is important in the determination of ferromagnetic couplers [81].

One of the aims concerning the propagation of ferromagnetic interactions within a molecule is the preparation of polymers consisting of coupled spins [82]. Towards this goal, Ishida and Iwamura prepared and studied the triradical 24 as a model of poly[(oxyimino)-1,3-phenylenes] [83]. The molecule does indeed have a quartet ground state. While EPR spectral intensity followed the Curie-Weiss law, and no half-field ( m S = 2) or third-field ( m S = 3) signals were observed, magnetic measurements on a Faraday balance revealed that the effective magnetic moment

16

1 Nitroxide-based Organic Magnets

of the micro-crystalline solid presented a maximum 3.53 ÌB at approximately 140 K, consistent with an energy gap between the lowest excited doublet and the quartet ground state of +240 K. This value plummeted on lowering the temperature as a result of intermolecular antiferromagnetic interactions, the Weiss constant being −19 K. To the best of our knowledge, no information concerning the conformation of the molecule in the solid is available, while in solution, EPR spectra suggest the presence of several conformers in accord with the complex conformational space available to this molecule as a result of its four torsional degrees of freedom. Trimethylenemethane (TMM) has proved to be another extremely efficient ferromagnetic coupler of unpaired electrons [84], and the preparation by Iwamura and colleagues of a bis-nitroxide with diphenylethylene spacers between the radical centers proved its worth [85]. When the nitroxide groups are located in the 4-positions of the benzene ring as in 25 (predicted to be a triplet by the through-bond topology rules) magnetic susceptibility measurements of microcrystalline samples revealed an increase of the effective magnetic moment between 40 and 10 K, followed by its decrease. The authors ascribed this effect to intramolecular ferromagnetic interactions with antiferromagnetic interactions between molecules. The energy gap between the triplet and singlet states was estimated as 15.3 K from fitting of the magnetic data to the Bleaney–Bowers equation. The corresponding molecules with the phenyl rings substituted at the 3-positions revealed a singlet ground state, in consonance with the disjoint nature of the molecular orbitals of the two groups. More recently, Shultz and coworkers have studied [86] the properties of various derivatives with TMM couplers in order to study the effects of conformation on the electronic coupling, which are in line with the π -conjugation in the molecules. Solution state EPR measurements of the half-field signal ( m S = 2) at variable temperature on fluid and frozen solutions of the radicals showed that for compounds 25–28 and 30, the intensities obey the Curie-Weiss law, indicating that either the triplet is the ground state, or that there is triplet-singlet degeneracy. The completely planar compound 30 (expectedly) showed the highest conjugation as judged from its UV-visible absorption spectrum, while the other compounds all contain twisted conformations. Indeed, the compound 31 which does not contain the TMM-type moiety also obeys the Curie-Weiss law in the temperature range studied by the authors, and the through space dipolar interaction between the two electrons, measured by the zero-field splitting parameter D, was similar to the other compounds reflecting similar effective spin density distribution for these derivatives. The behavior of compound 29 is radically different. The intensity of the half-field signal drops rapidly upon decreasing the temperature, indicating a singlet ground state well separated from the triplet state. This observation serves to emphasize the dramatic effects that conformation can have on intramolecular magnetic interactions. Solid state magnetic data for these radicals has yet to be reported. Oligo(1,2-phenylenevinylene) has also shown to be an efficient spin coupler that might be interesting for preparing super-high-spin polymers. The diradicals 32 and 33 have triplet ground states, with J/k values of +48 and +2.1 K, respectively, and antiferromagnetic coupling between molecules in the solid state samples used for these estimates [87]. The value of the magnetic coupling in 32 is approximately 1.5 times that of the similar compound 34 [88]. The authors suggested that the motive

1.3 Conjugated Nitroxides

17

for the increased coupling is a result of the lowering of the potential energy gap between the non-bonding molecular orbitals as a result of conjugation, an assertion supported by theory. The authors implied that the presence of spin defects might not be so important in super-high-spin polymers containing these fragments than in other polyradicals. Experimental evidence for such a proposal will be highlighted later.

18

1.4

1 Nitroxide-based Organic Magnets

Nitronyl Nitroxides

This group of radicals is perhaps the one which has received most attention in the realm of molecular magnetism, from the point of view of purely organic as well as coordination compound materials. The synthesis of the radicals was first reported by Ullmann and co-workers in the late 1960s [89], and most of today’s studies use their established routes to the compounds [90], in spite of the problematic and erratic nature of the preparation of some of the radical precursors. The orders of spin density in the molecules determined by polarized neutron diffraction [91], different spectroscopic techniques (EPR [90, 92], NMR [93], ENDOR [94]) and ab initio calculations [7] agree that in the nitronyl nitroxide unit the free electron in the singly occupied molecular orbital (SOMO) is distributed mainly over the two oxygen and two nitrogen atoms of the ONCNO conjugated system. The central carbon atom of this moiety is a node in the SOMO, a situation which limits delocalization of this electron over the substituents located at the 2-position of the imidazolyl moiety but permits the spin polarization phenomenon thereby creating an alternating spin density on the pendant group. It is well known that the properties of the nitronyl nitroxides are determined principally by the nature of the substituent located at this position and more importantly by the spin density on their atoms As a result of this situation, great effort has been expended recently in order to study in detail by EPR and NMR spectroscopies the spin distribution on the substituents, and to elucidate how this delocalization is influenced by the molecular conformation, as well as the molecular surroundings (solvent, neighbors, etc.) both in solution and the solid state [7h, 95].

1.4.1

Mono-nitronyl Nitroxides

An extremely wide range of nitronyl nitroxides have been prepared and studied, the majority of them being aromatic derivatives, as a result of their high stability and crystallinity. All the bulk ferromagnetic nitronyl nitroxides described so far are presented in approximate chronological order in Table 3, along with crystallographic information and most pertinent magnetic data. As shall be appreciated during the discussion that follows, the balance between ferro- or anti-ferromagnetic intermolecular interactions is an extremely delicate one, which can be influenced drastically by small changes in crystal structures. For this reason, it is convenient to discuss the radicals in comparison with their most similar chemical cousins. The structurally most simple nitronyl nitroxides are those in which a single atom or small group is attached to the carbon atom at the 2-position of the imidazolyl ring, the parent molecule being HNN (35) [90]. In both its crystalline phases reported so far, the molecule forms non-covalent dimers in the solid state of the type shown in Fig. 8, in which the hydrogen atom at the 2-position of the imidazolyl ring forms a hydrogen bond with one oxygen atom [96]. In the α-phase [97], these dimers pack into sheets by virtue of trifurcated Csp3 –H· · ·O hydrogen bonds [98], and the sheets stack with the carbon atoms at the 2-positions of the imidazolyl ring of one plane located

19

1.4 Nitronyl Nitroxides

Table 3. Structures and magnetic properties of nitronyl nitroxides with ferromagnetic ordering.

†

Code Space group Z

Curie temperature Magnetic coupling constants (θ or J/k) Mean internal field

Ref.

4NPNN (6) (β-phase) Fdd2 8

TC = 0.60 K θ = +1.2 K 160 G

15, 100, 101

3QNN (37)† P21 2

TC = 0.21 K J/k = +0.28 K 60 G

107, 109

4PYNN (38) C2/c 4

TC = 0.09 K J/kB = +0.27 K 120 G

114, 116

4MeSPNN (46) P21 /a 4

TC = 0.20 K θ = +0.36 K n.a.*

127

2OHPNN (47) (α-phase) Pbca 8

TC = 0.45 K θ = +0.62 K 140 G

131, 133

2,5OHPNN (52) (α-phase) P21 /n 4

TC = 0.50 K J/k = +0.93 K θ = +0.46 K n.a.*

138, 139

2FPNN (56) (α-phase) Pbca 8

TC = 0.30 K θ = +0.48 K n.a.*

145

canted ferromagnet, * Data not available.

on top of the NO groups in another. The observed antiferromagnetic coupling was fitted to the Bleaney Bowers equation giving a J/k of −11 K, but which direction these interactions operate in is as yet unclarified. In the β-phase, the sheets contain bifurcated Csp3 –H· · ·O hydrogen bonds and the orientation between planes locates NO groups on top of and antiparallel to one another, in slightly different orientations for the four pairs of molecules formed by the eight crystallographically-independent molecules [97]. The magnetic data of the β-phase was fitted to a model with two

20

1 Nitroxide-based Organic Magnets

Fig. 8. Schematic representation of the non-covalent dimer formed by HNN (35) in the solid state and its packing in the α- and β-phases (only one of the two types of NO· · ·NO interaction is shown) [97, 98].

exchange constants taking into account these four interplanar dimers, and giving J/k of −33 K and J /k of −1.5 K, these values being considered averages. Given the considerable differences in the antiferromagnetic interactions between the two phases, the non-covalent dimer was judged to be magnetically irrelevant. The derivative radicals with either iodine or bromine atoms or a cyclopropane group at the central carbon atom of the ONCNO unit all have honeycomb structures, and thus present similar magnetic behavior to each other, with competing ferro- and antiferromagnetic interactions, the latter compound exhibiting an antiferromagnetic ordering at 1.5 K [99]. The magnetic susceptibility data was fitted to a dimer model with intra- (J/k) and interdimer (J /k) interactions, giving the following exchange interaction energies: J/k = −5.3 and J /k = +4.0 K for the iodo derivative, J/k = −4.5 and J /k = +2.6 K for the bromo compound, and J/k = −1.8 and J /k = 0.4 K for the cyclopropane derivative. As previously highlighted, the first fully characterized organic ferromagnet is from the family of nitronyl nitroxides, concretely the β-phase of 4-nitrophenyl derivative 4NPNN (6), and it sparked the initiation of many research programs on this type of radical. The magnetic behavior of this crystalline phase along with those of the other three known phases of this radical will be discussed in more detail in another chapter in this series by its discoverer. However, for comparison purposes, we mention here that the critical temperature of the β-phase is 0.60 K [100], which is given along with a small fraction of the detailed magnetic data obtained in Table 3 [101]. The structure in the crystals is three dimensional [102], with non-covalent interactions between the oxygen atoms of the two NO groups and the nitrogen atom of the nitro group (presumably of a Coulombic nature) in one direction and the hydrogen atoms of the phenyl ring in another. Structurally related to 4NPNN, in the sense that radicals are arranged in a headto-tail manner as a result of Coulombic interactions, is the 4-cyanophenyl nitronyl nitroxide (4NCPNN (36), Fig. 9) [103]. The oxygen atoms of one molecule are located over the NCN unit of an adjacent molecule, while a short distance between an oxygen atom of the second molecule is located close to the CH group in the 3-position of the

1.4 Nitronyl Nitroxides

21

phenyl ring of the first. However, unlike the β-phase of 4NPNN, this compound has a two dimensional structure formed by sheets of nitronyl nitroxides with their long axis perpendicular to the ac plane, forming a square-lattice-type magnetic structure in which each molecule interacts with its four nearest neighbors (Fig. 9). Magnetic susceptibility data indicated ferromagnetic interactions within the molecular sheets, which were fitted to a square-lattice Heisenberg model affording J/k = +0.75 K. The two dimensional nature was confirmed by plotting the experimental C/χ T value against z J/kT with those predicted by one-, two-, and three-dimensional (Curie– Weiss) models, using J/k = +3.0 K (Fig. 9). The authors suggested that the short NO to aryl or CN distances could be possible sources of the ferromagnetic interactions.

Fig. 9. Views of the two-dimensional solid state structure of 4NCPNN (36) and the plot of experimental C/χ T against z J/kT along with those predicted by one-, two-, and threedimensional (Curie–Weiss) models with J/k = 3.0 K [103]. Reproduced by permission of The Royal Society of Chemistry.

22

1 Nitroxide-based Organic Magnets

The material showed no evidence for a magnetic phase transition down to 0.5 K according to a. c. susceptibility measurements. More recently, however, zero-field Ì+ SR experiments revealed the presence of an ordered magnetic state below 0.17 K [104], although details of the nature of this state were not derived. A number of heterocyclic nitronyl nitroxide derivatives have been reported, many of which display interesting magnetic properties on their own, and especially when complexed with transition metal ions [105]. It has been shown by ENDOR and TRIPLE spectroscopy that correct positioning of the heteroatoms in the heterocyclic substituent can enhance negative spin density on the pendant group when compared with hydrocarbon analogs [106]. This enhanced negative spin density is produced if the heteroatom is at a π-site that is positively polarized. The radical 3QNN (37, Table 3) crystallizes [107] in a non-centrosymmetric space group, and has a three-dimensional structure which is maintained by hydrogen bonds between the hydrogen atoms of the methyl groups in the nitronyl nitroxide moiety and both oxygen atoms of the same fragment in neighboring molecules as well as the nitrogen atoms of the pendant heterocycle. Paramagnetic susceptibility data of the compound reveal ferromagnetic interactions which were fitted [108] between 300 and 5 K to the Curie–Weiss law, giving θ = +0.27 K. Below 0.21 K this radical exhibits an ordered magnetic state when observed by zero-field Ì+ SR experiments [109]. Close examination of a.c. susceptibility data show a shallow dip at approximately 0.21 K which indicated a magnetic transition that, along with the lack of fitting of magnetic data at low temperature to conventional models for purely ferromagnetic interactions implied the presence of some antiferromagnetic interactions, suggesting, therefore, that the system is a canted ferromagnet [110]. This hypothesis is supported by the fact that the internal field of the sample experienced by the muons in the Ì+ SR experiments (60 G) is approximately half that of the other known ferromagnets of this family (150 G), and seems to be parallel to the c axis of the crystals in accordance with the theoretical estimation [109]. In contrast, the homologous 4-quinoline derivative presents antiferromagnetic interactions, which were interpreted in terms of an alternating 1D antiferromagnetic Heisenberg model with J/k = −7.8 K and α = 0.5 [111]. These antiferromagnetic interactions were ascribed to the close approach of two NO groups within the dimers present in the crystals. The family of nitronyl nitroxide with either 2-, 3-, or 4-pyridyl as substituent show antiferromagnetic interactions in the case of the first two [112], of which the 3-derivative apparently orders at 1.35 K [113], and ferromagnetic interactions for the latter, 4PYNN (38) [114]. This latter compound crystallizes to form a sheet-like structure [115], in which chains of molecules pack in a head-to-tail arrangement, this disposition being maintained by hydrogen bonds between the oxygen atom of the radical unit and the hydrogen atoms of the pyridyl ring, as represented in Fig. 10. The magnetic data was initially fitted to a model which assumed a one dimensional ferromagnetic chain with a J/k = 0.27 K, but zero field Ì+ SR experiments later revealed the appearance of an ordered state at less than 0.1 K with an internal field of 120 G [116], whose origin is as yet uncertain. The extremely low temperature of the transition has been ascribed to weak π -orbital overlap in the direction perpendicular to the chains shown schematically in Fig. 10.

1.4 Nitronyl Nitroxides

23

Fig. 10. A schematic representation of the one-dimensional ribbons of molecules formed by 4PYNN (38) in the solid state [115].

Substitution at the 6-position of the 2-pyridyl nitronyl nitroxide radical with a bromine atom (6Br2PYNN, 39) and subsequently with an alkyne group (6A2PYNN (40), Fig. 11) gives crystalline materials which show, according to fitting to the CurieWeiss law, dominant antiferro- (θ = −0.18 K) and ferromagnetic (θ = +1.24 K) interactions, respectively [117]. Both molecular structures have a high twist angle between the component rings, presumably because of repulsive electrostatic interactions between the oxygen and nitrogen atoms of the radical and pyridyl units, respectively. Interestingly, in 6A2PYNN, a strong intermolecular hydrogen bond between the alkyne proton and an NO group leads to the formation of zigzag chains (Fig. 11) which are pulled together by Csp2 –H· · ·O hydrogen bonds from the 3position of the pyridyl ring to the other NO group. The packing results in an angle of 98◦ between nearest radical moieties. The ferromagnetic interactions, confirmed

Fig. 11. The radicals 6Br2PYNN (39) and 6A2PYNN (40) and a schematic view of the hydrogen-bonded chains formed by the latter [117].

24

1 Nitroxide-based Organic Magnets

by magnetization experiments, were analyzed using a one-dimensional Heisenberg chain model, with a J/k of +1.40 K and a weak coupling between the neighboring chains of z J /k of −0.27 K, where z is the number of interacting chains. The spin density distribution in the crystals was probed using polarized neutron diffraction, revealing a significant spin population on the alkyne hydrogen atom involved in the hydrogen bond to the NO group, whose oxygen atom has depleted spin density [118]. It was therefore proposed by the authors that the hydrogen bond is involved in the transmission of the ferromagnetic interaction, though no detailed pathway was suggested. The 5-pyrimidinyl nitronyl nitroxide displays ferromagnetic interactions in its crystals at higher temperatures, while antiferromagnetic interactions compete strongly below 14 K, giving rise to a maximum in the plot of χ T against T [119]. These data were fitted nicely to a regular one-dimensional Heisenberg chain model, with a ferromagnetic exchange constant J/k of +26.2 K, corrected with a mean molecular field approximation J /k of −1.1 K assuming that the number of nearest neighboring chains is 4. The correlation of these interactions with the structure indicated that the chains of molecules formed by the radicals are responsible for the ferromagnetic interaction, since an oxygen atom of one NO group is located very close (2.92 Å) to the central carbon atom of the ONCNO moiety of another molecule, an intermolecular arrangement which is favored by π –π stacking of the pyrimidine moieties. This interpretation based on spin-polarization arguments was appealed to since analysis of the structure shows that direct interaction between NO groups is not likely because the associated orbitals are far from being orthogonal. Interestingly, the stacking of the pyrimidinyl rings could also be invoked using the McConnell I theory because of the staggered arrangement between them. The authors claimed that the magnitude of the magnetic interaction was much smaller than expected given these two pathways for interaction, and implied that direct interaction between NO groups cancels a large part of the ferromagnetic coupling. The family of nitronyl nitroxide radicals with five-membered heterocyclic rings containing nitrogen atoms as substituents display a wide range of particularly interesting magnetic properties (Table 4). In particular, Kahn’s group has reported the preparation and properties of three triazole-derived nitronyl nitroxides. The 4methyl-1,2,4-triazole derivative 4MTNN (41) behaves as a weak ferromagnet [120]. The field-cooled (0.2 Oe) magnetization shows a break at 0.6 K, characteristic of a spin canting phenomenon which is compatible with the space group to which the crystals pertain. The molecule packs so that weak hydrogen bonds form between the methyl groups and oxygen atoms bearing the spin. These bonds lead to a zigzag chain in which the molecules are arranged in a head-to-tail fashion, which unite giving a parquet-like structure. The magnetic susceptibility data obtained for this compound were quantitatively interpreted in line with the structural features using a one-dimensional Heisenberg model, corrected with a mean field approximation accounting for interchain antiferromagnetic interactions. The intrachain interaction J/k was found to be +0.65 K while the interchain one (J /k) was −0.14 K assuming four interacting neighboring chains. The nitrogen atom at the 2-position of the triazole ring has negative spin density (according to polarized neutron diffraction data) and is located very close to one of the nitroxide oxygen atoms with positive spin

25

1.4 Nitronyl Nitroxides

Table 4. Properties of nitronyl nitroxides bearing five-membered nitrogen-containing heterocycles. Code Space group Z

Magnetic behavior and data

Ref.

4MTNN (41) P21 21 21 4

Weak ferromagnet J/k = +0.65 K J /k = −0.14 K

120

4,5DMTNN (42) P21 21 21 4

Metamagnet θ = +0.45 K TN = 330 mK HC = 0.7 kOe

121

5MTNN (43) P21 /c 4

Strong ferro- and weak antiferromagnetic interactions J/k = +10.6 K z J /k = −1.4 K θ = +8.9 K

122

2BimNN (44) Pbca 8

Strong ferro- and weak antiferromagnetic interactions J/k = +17.3 K* θ = +8.2 K

123

2ImNN (45) P21 /a 4

Antiferromagnetic interactions J/k = −89 K

123– 125

* Value of weak antiferromagnetic interactions not determined.

density, and was therefore implied as the route for the dominant ferromagnetic interactions, although the authors insisted that it is not clear if this negative spin density is a cause or a consequence of these interactions. The short contact between NO groups of neighboring chains could be responsible for the antiferromagnetic interactions. The decrease in χ T exhibited by this compound at very low temperatures is also shown by the metamagnetic 4,5-dimethyl-1,2,4-triazole derivative 4,5DMTNN (42) [121], which packs in a very similar way to 4MTNN, although the distances between nitroxide groups are somewhat longer. The presence of Csp3 –H· · ·O hydrogen bonds

26

1 Nitroxide-based Organic Magnets

Fig. 12. Crystal structure and magnetic susceptibility curve for 5MTNN (43) [122].

which put atoms of opposite spin density close to one another may be important in the transmission of the ferromagnetic interactions. Unlike the aforementioned triazoles, the 5-methyl-1,2,4-triazole derivative 5MTNN (43) has extremely strong hydrogen bonds – between the NH group at the 2-position of the triazole ring and a spin-bearing oxygen atom – linking the molecules in the crystal so as to form a one-dimensional molecular chain (Fig. 12) [122]. The molecule is a lot flatter than the other triazoles mentioned (angle between heterocycle planes of 23.3◦ ) and the nitroxide groups are twisted by 34◦ between molecules, i. e. they are not orthogonal. The material has susceptibility data which follow the Curie–Weiss law with θ = +8.9 K (above 5 K) showing the presence of extremely strong ferromagnetic interactions, with weak antiferromagnetic interactions superimposed. The magnetic behavior can be interpreted by a one-dimensional Heisenberg model with J/k = +10.6 K and z J /k = −1.4 K using the appropriate mean-field correction. While the origin of the strong interaction was considered by either spin-polarization effect or by an admixture of ground and excited state configurations, the exact reasons for the tremendous ferromagnetic interaction are as yet unclear [122]. A similarly strong magnetic interaction accompanied by strong hydrogen bonds of a similar type has been observed in the crystals of the 2-benzimidazole nitronyl nitroxide 2BimNN (44) [123]. In this case however, the symmetry of the chain which forms is somewhat different, since the molecules are related by a translation and not by a screw axis. The magnetic susceptibility measurements (which gave a maximum of χ T at 3.2 K, consistent with the presence of dominant ferromagnetic interactions along with weak antiferromagnetic interactions) indicated a magnetic coupling similar in magnitude to that of 5MTNN (Table 4). In addition, magnetization isotherms at 2.8 and 4 K were fitted to Brillouin function curves of a system with an effective S of 9/2. In contrast, the 2-imidazolyl derivative 2ImNN (45) has strong antiferromagnetic interactions [123, 124] (Table 4) as a consequence of the closeness of the spin-carrying oxygen atoms of two radical molecules which form a dimer. In the

1.4 Nitronyl Nitroxides

27

crystals of this compound, the NH group does not form a hydrogen bond with this oxygen atom, but rather with the basic nitrogen atom in a neighboring imidazolyl ring [125]. Finally, it is worth mentioning that this family of nitronyl nitroxides bearing five-membered heterocyclic rings on the whole shows relatively strong magnetic interactions, which could be a consequence of the enhanced spin polarization provoked by the heterocyclic ring. Given that the interaction between NO groups having a positive spin density and phenyl groups having opposite spin density was calculated to give rise to ferromagnetic interactions [126], Gatteschi and coworkers prepared the compound 4MeSPNN (46, Table 3) – a bulk ferromagnet with a TC of 0.20 K – with the idea that the SMe group would increase the spin density on the phenyl ring and have a relatively large spin density on the sulfur atom [127]. The solid state structure of the compound reveals a twisted conformation for the molecule which packs into dimers that form two-dimensional sheets of molecules. Evidence of “NO-phenyl” interactions, as well as short distances between the NO groups and the SMe substituent in addition to S· · ·HCsp3 interactions between planes were highlighted by the authors [128]. The magnetic susceptibility data at high temperature were fitted to the Curie–Weiss law (θ = +0.36 K), given that EPR studies implied a three-dimensional propagation of interactions between spins. Alternative fitting of the data to a square planar Heisenberg model for S = 1/2 ferromagnetically coupled spins with an additional correction for interplanar interactions gave J/k = +0.35 K in the plane and J /k = +0.04 K between them. The low dipolar interaction energy provoked the researchers to suggest the interplanar interactions as the driving force for the magnetic transition. Interest in the group of hydroxyphenyl nitronyl nitroxides was spurred initially by the suggestion that hydrogen bonds are implicated in the transmission of magnetic interactions in inorganic complexes [26], as well as for the potential controlling role that hydrogen bonds could fulfil in the crystallization of the molecules [129]. The family has provided an extremely interesting assortment of magnetic behavior [130] – the monohydroxyphenyl derivatives, 2OHPNN (47, Table 3) in the α-phase is a bulk ferromagnet [131], 4OHPNN (48) presents ferromagnetic interactions in two dimensions [132] and 3OHPNN (49) has a three-dimensional network of predominantly antiferromagnetic interactions (Fig. 13) [131]. The crystal structure of the α-phase of the bulk ferromagnetic radical 2OHPNN reveals a molecular structure in which the OH group forms a strong intramolecular hydrogen bond with one of the oxygen atoms of the nitronyl nitroxide moiety, causing a high twist angle between the planes formed by the ONCNO moiety and the phenyl ring (ca. 40◦ ). The molecule packs to generate a three dimensional structure in which the NO groups of the open shell moiety form weak hydrogen bonds with hydrogen atoms in the aliphatic part of the molecule as well as the aromatic ring. The Curie– Weiss susceptibility above 4 K gave θ of +0.62 K. Hysteretic M–H behavior with a very small coercive field was observed below 0.41 K, a feature characteristic of a soft ferromagnet [133]. Zero-field muon depolarization experiments (Fig. 14) support the three-dimensional model of magnetism, and provide a value of the internal field (140 G) similar to that of the radical 4NPNN. A similar conclusion about its magnetic dimensionality was accomplished by low temperature heat capacity measurements. The β-phase of 2OHPNN has a completely different structure that is dominated

28

1 Nitroxide-based Organic Magnets

Fig. 13. Magnetic properties of the mono-hydroxyphenyl nitronyl nitroxides 3OHPNN and 2OHPNN, as revealed in the plots of χ T against T and the a. c. susceptibility of 2OHPNN.

Fig. 14. Time evolution of the muon polarization in zero-field muon spin rotation experiments on 2OHPNN [133].

1.4 Nitronyl Nitroxides

29

Fig. 15. A view of the crystal structure of 4OHPNN (48).

by a chain of molecules, and accordingly with this feature its magnetic behavior is reproduced by a one-dimensional Heisenberg chain model corrected with a mean field approximation [130b]. The intrachain interaction J/k was −1.31 K while the interchain one J /k was +0.70 K, assuming that there are four interacting neighboring chains. The positioning of the hydroxyl group in the 4-position of the phenyl ring in 4OHPNN (48) results in a completely different crystal structure to that of the αand β-phases of 2OHPNN. Strong intermolecular hydrogen bonds dominate, leading to the formation of chains of molecules in which hydrogen bonds are given by the hydroxyl group and received by one of the oxygen atoms of the radical moiety (Fig. 15) [132]. In turn, these molecular chains are united through two Csp3 –H· · ·O hydrogen bonds from diametrically opposite methyl groups in one radical moiety and the otherwise free NO group of a molecule in the adjacent chain. The resulting two-dimensional sheet is virtually flat. Indeed, the ferromagnetic interactions are quasi-two-dimensional, as revealed by EPR studies on oriented single crystals, and are suggested to be transmitted through these hydrogen bonds [134]. The compound shows two successive magnetic transitions at 700 and 100 mK, as revealed by zero-field Ì+ SR experiments [133]. The first is suggested to correspond to the ferromagnetic ordering in the plane of the sheets, while the second could come about because of ordering between them although the nature of the final magnetic state is as yet unclear. The radical 3OHPNN (49) forms non-covalent dimers (Fig. 16) in both the solution [135] and solid states [131]. In the latter it presents predominantly antiferromagnetic

30

1 Nitroxide-based Organic Magnets

Fig. 16. Schematic representation of parts of the crystal structures of 3OHPNN (49), 3,4OHPNN (50), and 3,5OHPNN (51).

interactions as a result of closeness of SOMOs of radicals in different dimers. While the placement of chlorine atoms in the aromatic ring of the hydroxyphenyl nitronyl nitroxides has so far given materials with a variety of antiferromagnetic interactions [136], additional hydroxyl groups do give materials with ferromagnetic interactions. Two of these constitutional isomers – 3,4OHPNN (50) and 3,5OHPNN (51) – contain the same “supramolecular synthon” [24] as that present in 3OHPNN: that is the head-to-tail hydrogen bonded dimer, as one can appreciate in Fig. 16. This motif is therefore a robust one which can be manipulated in a variety of structures, but it is not apparently very significant magnetically, since the three materials which incorporate the unit do not have the same dominant features. Thus, 3,4OHPNN presents competing ferro- and antiferromagnetic interactions [137], which were fitted to a one-dimensional Heisenberg spin chain with alternating signs of exchange interactions JF /k = +9.3 K, JAF /k = −0.38 K with weak interchain interactions (J /k = +0.34 K). The strong ferromagnetic interaction arises most likely from the head-to-tail overlapping of molecules in contiguous layers, appealing to the charge transfer mechanism. On the other hand, the radical 3,5OHPNN presents dominantly antiferromagnetic interactions as a result of overlapping between parallel SOMOs of molecules in stacked hydrogen-bonded ribbons [130b, 138]. The 2,5-dihydroxyphenyl nitronyl nitroxide (2,5OHPNN, 52) also has two known polymorphs, the first of which presents bulk ferromagnetism [139]. The molecular structure of the α-phase is dominated by a strong intramolecular hydrogen bond between the hydroxyl group at the 2-position of the phenyl group and one of the NO oxygen atoms, which results in a lengthening of the latter bond compared with

1.4 Nitronyl Nitroxides

31

Fig. 17. Schematic representation of part of the crystal structure of 2,5OHPNN (52) [138].

its usual state, and consequently has a non-planar conformation (angle between rings is 37◦ ). Presumably for this reason the crystalline synthon observed in 3OHPNN, 3,4OHPNN, and 3,5OHPNN is not present in these crystals. The molecules are pulled together in all three dimensions by significant intermolecular hydrogen bonds: (i) between the “free” hydroxyl group acting as a donor to the other hydroxyl group which acts as an acceptor forming molecular chains (Fig. 17); (ii) dimerization of these chains through a bifurcated hydrogen bonded system involving the hydroxyl groups at the 2-position of the phenyl ring (Fig. 17); and (iii) union of these sheets by weak Csp3 –H· · ·O–N bonds. Magnetic susceptibility data gave a plot of χ T against T which increased monotonously as the temperature decreased, and was fitted to the singlet-triplet (ST) model, which suggested two ferromagnetic interactions (Table 3). A magnetization curve at 80 mK showed hysteresis, with a small coercive force (200 Oe). The bulk nature of the transition was also confirmed by heat capacity measurements [138]. Deuteration of the OH groups in 2,5OHPNN afforded crystals (70% deuteration) which had slightly lower χ T at 1.8 K than the protonated sample, and heat capacity measurements showed a decrease in the Curie temperature. This result was interpreted as a lengthening of the distances between non-covalently interacting groups, thereby supporting the contention that these bonds assist the transmission of ferromagnetic interactions. The dimer formation between the sheets is considered to result in ferromagnetic interactions, and McConnell’s theory was also invoked to explain the magnetic interactions in all three directions. Semi-empirical and ab initio calculations indicate that the hydrogen bonds present in the structure act as accomplices in the transmission of the ferromagnetic interactions in the crystals [140]. When the OH group at the 2-position is absent in the calculation, an antiferromagnetic interaction was predicted between the electrons on the two proximal NO groups. The phenyl boronic acid radical 4BAPNN (53, Fig. 18) also has a crystalline structure dominated by strong hydrogen bonds, and low dimensional ferromagnetic interactions exist between the unpaired electrons [141]. The molecules form chains linked

32

1 Nitroxide-based Organic Magnets

Fig. 18. Schematic representation of the crystal structure of 4BAPNN (53) which has the four possible pathways of spin-spin interaction: AB, AC, AD, and BC.

by complementary hydrogen bonds in the manner schematized in Fig. 18. Thus, molecules form head-to-tail dimers by virtue of BOH· · ·ON hydrogen bonds, and these dimers are linked in turn by dimerization of the boronic acid moieties. The magnetic susceptibility data was fitted to a dimer model, giving J/k of +0.71 K. Three possible routes for this coupling were considered by the authors, all of which involve strong hydrogen bonds: between the spins A and B in the head-to-tail dimer, between spins A and C, and then the longer pathway A to D and B to D (the alternative route A to D was not considered). The first of these pathways was suggested as the active one based on distance criteria and charge transfer arguments. The corresponding 3-substituted boronic acid isomer follows the Curie–Weiss law (θ = −0.82 K) indicating the presence of dominant antiferromagnetic interactions [141], although its structure has not yet been reported. Several halophenyl nitronyl nitroxides have been reported, many of them showing ferromagnetic interactions. A three-dimensional structure is formed by 4FPNN (54) radicals in the solid state in such a way as to favor relatively strong ferromagnetic interactions [142]. The molecules form head-to-tail dimers (Fig. 19) by virtue of Csp2 –H· · ·O hydrogen bonds (the H–O distance is less than the sum of the van der Waals radii), similar to 4PYNN and 4NPNN, except that a polymeric chain does not result. Instead, the dimers come together about a fourfold screw axis (Fig. 19), which leads to a quite short distance (3.54 Å) between the free oxygen atoms and the carbon in the center of the ONCNO moiety [143]. A strong ferromagnetic interaction is observed, which above 25 K can be described by a θ of +2.5 K. The formation of a triplet species below approximately 10 K was implied by the value of the susceptibilities, which were near to those predicted by the Curie law (C = 0.5 emu K mol−1 ). However, in the plot of χ T against T below 4 K, the value exceeds that for a triplet, a situation indicative of further ferromagnetic interactions. A model involving two ferromagnetic interactions was derived to fit the data, which gave J/k = +5.0 K, and J /k = +0.02 K, and were assigned to intra- and inter-dimer exchange, respec-

1.4 Nitronyl Nitroxides

33

Fig. 19. Structural formulas of 4FPNN (54) and 4BrNN (55) and a view of the crystal structure of projected on the bc plane, with the plot of magnetization M against H T with the predicted Brillouin function for ideal S = 1/2 and S = 1 paramagnetic systems [143]. Reproduced by permission of The Royal Society of Chemistry.

tively. The weaker interaction was said to be characteristic of the approach of one of the oxygen atoms to the central carbon atom of the ONCNO moiety. It should be noted that variable temperature EPR and X-ray diffraction experiments indicated that there is a lowering of crystal symmetry at temperatures below 100 K, which is a relatively gradual change [143]. It was concluded that the dimer structure was still present, since the cell parameters do not change drastically. The bromo analog

34

1 Nitroxide-based Organic Magnets

4BrPNN (55) presents similar ferromagnetic interactions to the fluoro compound, which were ascribed to the dimer structure in the solid state [142], which is a distorted form of the ones formed by 4FPNN. The halogen atoms have not been implied to play any significant role in the packing or transmission of magnetic interactions. Indeed, the chloro-derivative presents antiferromagnetic interactions [142]. The corresponding 2-halophenyl nitronyl nitroxides have twisted molecular conformations resulting from steric interactions between the halogen and oxygen atoms [144]. The angles between the phenyl and imidazolyl rings is 55◦ for the fluoroderivative and 60◦ for the chloro-derivatives which show positive (θ = +0.48 K) and negative (θ = −2.00 K) Weiss constants, respectively, while the bromo- and iodo-derivatives (whose X-ray structures are not available) both present antiferromagnetic interactions (θ = −3.32 and −3.36 K, respectively) [145]. The 2-fluorophenyl nitronyl nitroxide 2FPNN (56, Table 3) is a bulk ferromagnet, with a Curie temperature of about 0.3 K, as revealed by ac susceptibility data and magnetic heat capacity measurements at very low temperatures [145]. The latter results show a broad hump distinctive of a one-dimensional Heisenberg ferromagnet, the model for which gave an intrachain J/k = +0.6 K. The ratio of interchain to intrachain coupling was estimated from mean field theory, using the TC and J/k value, to be between 1/4 and 1/10. The structure of the radical reveals the typical Csp3 –H· · ·O hydrogen bonds between the head-to-tail packed radical moieties, which was used to argue for a spin-polarization mechanism that would explain the intermolecular ferromagnetic interaction. Also present in the structure are two Csp2 –H· · ·O hydrogen bonds proceeding from the 4- and 3-positions of the aromatic ring. This latter non-covalent bond is also present in the structure of the 2-chloro-derivative, in which the molecules are arranged head-to-head, and was used to argue the case for the antiferromagnetic interaction in the crystals [145]. Antiferromagnetic interactions are observed in the crystals of the multiply-halogenated phenyl nitronyl nitroxides that have been reported – 3,5-difluoro- and pentafluoro [144, 146]. In the vast majority of solid state structures of phenyl nitronyl nitroxides, an appreciable torsion angle exists between the ONCNO plane and the adjoined aromatic ring. However, in the radical 2-phenylbenzimidazol-1-yl N ,N -dioxide [147] (PBIDO (57), Fig. 20), in which the –CMe2 CMe2 – group is replaced by an ortho-substituted benzene ring, this angle is only 10.3◦ , and the molecule has a practically planar shape [148]. The absence of a twisting force in the imidazolyl unit in this structure would seem to favor a low inter-ring angle, and indicate that the reason for the favored pseudo-eclipsed geometry is the presence of a molecule with a flat shape in the crystal, in line with the ideas put forward about three decades ago by Kitaigorodskii [149]. Dimers of the molecules form in the crystal, and these dimers are in turn arranged in a herringbone-type manner, giving a largely two-dimensional system, a fact borne out by EPR measurements, that has a similar form to the κ-phase formed in some molecular superconductors. The magnetic behavior of the compound is quite complex, with two maxima in the plot of susceptibility against temperature, one at 45 K and the other at approximately 3 K (Fig. 20). The magnetic susceptibility data were fitted in a limited temperature range to give intradimer (J/k ≈ −40 K) and interdimer (J/k ≈ −10 K) antiferromagnetic exchange constants, although the values are approximate because of the complexity of the system. The number of spins

1.4 Nitronyl Nitroxides

35

Fig. 20. Chemical formula of PBIDO (57) and the temperature dependence of the magnetic susceptibility of the compound represented as a log plot of χ T against T [148]. Reproduced by permission of The Royal Society of Chemistry.

responsible for the anomaly at the lowest temperatures was ascribed to one fiftieth of the spins, and EPR revealed (fine structure satellites and half field signal) that it results from spin-multiplet states [148]. A group of radical salts derived from nitronyl nitroxide radicals which have generated much interest are the alkyl pyridinium salts of 4PYNN (38) and its constitutional isomer with the pyridyl ring substituted at the 3-position [150], especially because one of their exotic magnetic behavior. One of the latter salts is a possible kagom´e antiferromagnet [151], while another salt has been claimed as a molecular spinladder, in which the radical component is not part of the ladder (which is formed by nickel(II)dithiolthionethiolate anions), but which interacts ferromagnetically within the layers it forms [152]. The exotic properties of these compounds will form the subject of another chapter in this series presented by their inventor.

1.4.2

Oligo-nitronyl Nitroxides

As for the simple nitroxides, ferromagnetic coupling units have been employed to prepare molecules with triplet or higher ground states [153]. The diradical 3phenylene bis(nitronyl nitroxide) (1,3PBNN, 58A) is expected to have, and in reality exhibits, a triplet ground state, as determined by variable temperature EPR spectroscopy [154]. Very recently, Turek and Catala have determined by EPR spectroscopy that the exchange coupling in this diradical (and other related ones), both in frozen solution and in a polymeric matrix, is J/k = +30 K [155]. In the complicated magnetic behavior of this compound in crystalline form, however, antiferromagnetic interactions dominate ferromagnetic intramolecular interactions [156]. The complicated behavior is partially a result of a reversible phase change in the crystals below 100 K, which causes non-equivalence of the centrosymmetric dimers formed by the

36

1 Nitroxide-based Organic Magnets

Fig. 21. A representation of the molecular dimers formed by 1,3PBNN (58A) in the solid state and the magnetic model used to explain the magnetic behavior of the material [156].

molecules in the solid (Fig. 21). After a fitting of the magnetization data at 1.8 K to an eight spin model, which allows for two kinds of dimers, intramolecular ferromagnetic interactions (J/k) of the order of +30 K were found. The related diradical 2,6-pyridyl bis(nitronyl nitroxide) (2,6-PyBNN, 58B) also shows complicated magnetic behavior resulting from competing ferro- and antiferromagnetic interactions [157]. More recently a bis-nitronyl nitroxide based on phenyl pyrimidine PPyrBNN (59) was characterized as a triplet ground state compound which has antiferromagnetic interactions between the molecules in its crystals as a result of a close O· · ·O contacts [158]. The estimated intramolecular ferromagnetic coupling was of J/k = +3.5 K. An extremely interesting compound (preceding these examples) is that reported for the first time by Dulog and Kim, in which three α-nitronyl nitroxide moieties are located at the 1,3, and 5 positions of a benzene ring (PTNN, 60) [159], and which (in principle at least) has a quartet ground state. Very little EPR data concerning the compound has appeared in the literature. A broad spectrum was presented in the original article which implied “strong intramolecular spin–spin interaction”, but no information on the ground state was presented. The magnetic susceptibility data of a powdered sample, which was repeated by Sugawara’s group [160], showed predominantly antiferromagnetic interactions in the material, whose crystal structure has not been solved. The latter group formulated an ingenious way of revealing the intramolecular magnetic interactions using a supramolecular chemical approach [161]

1.4 Nitronyl Nitroxides

37

(See co-crystallization of nitronyl nitroxides below), showing that the intramolecular exchange coupling, mediated by 1,3-phenylene units, is J/k = +23 K. Building on their work concerning nitroxides attached to thienothiophene rings [162], Iwamura and co-workers reported the preparation of the diradicals TTBNN (61) and TTTBNN (62), which both show singlet ground states, with very small singlet-triplet gaps [163]. On the other hand, 2,4-thiophene bis-nitronyl nitroxide (2,4TBNN, 63) has a triplet ground state [164], with a coupling determined by magnetometry of J/k = +40 K, while the corresponding 2,5-analog is a ground state singlet. The crystals of both materials are dominated by strong antiferromagnetic interactions at low temperatures, as are those of the 2,2 -bithienyl derivatives [164]. Very recently, metallocenes have been shown to behave as magnetic couplers between radical centers [165]. In solution, the bis(nitronyl nitroxide)s MCBNN (64, M = Fe; 65, M = Ru) have singlet ground states, as revealed by a study of the temperature dependence of the m S = 2 transition in the EPR spectra of the radicals in dilute solution, with exchange coupling constants J/k of −29 (M = Fe) and −27 K (M = Ru), respectively (Fig. 22). The magnetic susceptibility data for crystals of the com-

38

1 Nitroxide-based Organic Magnets

Fig. 22. Formula of MCBNN (64, M = Fe; 65, M = Ru) diradicals, the X-ray structure of the ferrocene derivative, and the temperature dependencies of the m S = 2 transition in the EPR spectra of the two [165].

pounds confirmed the presence of antiferromagnetic interactions both within and beyond the molecule, the exchange interactions being Jintra /k of −3.2 and Jinter /k of −4.2 for the ferrocene derivative, whose crystal structure was determined. Unusually, the ferrocene unit adopts a syn geometry in which the two substituents are located on the same “side” of the molecule, held in this position by two complementary Csp3 –H· · ·ON hydrogen bonds, a situation also pertains in solution, as determined from the zero-field splitting parameters from EPR spectra. In the crystal, chains of molecules are formed by virtue of further Csp3 –H· · ·ON hydrogen bonds. Iwamura’s group has recently reported the synthesis of triradicals exhibiting quartet ground states. The molecular materials incorporate two 4-phenyl nitronyl nitroxides coupled to either a nitroxide [166] or a carbene group [167]. In the case of the nitroxide coupled phenyl nitronyl nitroxide NOBPNN (66, Fig. 23) the χ T value of the microcrystalline sample at room temperature was characteristic of three uncoupled spins, but rose upon cooling to reach a maximum at approximately 80 K before decreasing as a result of intermolecular antiferromagnetic couplings. The intramolecular coupling in NOBPNN is aided by the possibility of conjugation, as exemplified by the existence of quinoid resonance structures NOBPNNA and NOBPNNB, resulting in an extremely strong exchange coupling between the central nitroxide and the terminal nitronyl nitroxides of J/k = +231 K. The mixed nitronyl-imino nitroxide and bis-iminonitroxides were also prepared and shown to exhibit quartet states. These results build on previous studies by the same group on a diradical [168] with a triplet ground state because of strong coupling (lower limit of J/k of +450 K) between a nitronyl nitroxide and a 4-substituted phenylnitroxide through a quinoid state, although this state is not very evident in the solid state. Antiferromagnetic intermolecular interactions were also observed in this compound. All the bis-nitronyl nitroxides mentioned up to this point have had appreciable intramolecular magnetic coupling between them, even though without exception the intermolecular interactions are antiferromagnetic. There are also examples of nonconjugated nitronyl nitroxides, for example the diradicals BABPNN (67) [169, 170]

1.4 Nitronyl Nitroxides

39

Fig. 23. The resonance forms of triradical NOBPNN [166].

and PBIMBNN (68) [171]. The former presents competing intermolecular ferroand antiferromagnetic interactions in what can be described as a an alternating onedimensional Heisenberg chain with JAF /k of −3.9 K, and JF /k of +1.2 K [170]. The compound PBIMBNN has both short intra- and intermolecular distances between SOMOs, and its magnetic behavior in the solid state can be described as a one dimensional antiferromagnetic alternating Heisenberg chain, with J/k of −158 K

40

1 Nitroxide-based Organic Magnets

and α of 0.22 [171]. The magnetic interactions are considered not to be throughbond because of the large torsion angles within the molecules.

1.4.3

Co-crystallization of Nitronyl Nitroxides

While the crystallization of pure radicals may result in somewhat disappointing properties, an interesting supramolecular approach is to take advantage of the interactions of the open shell molecules with other compounds – of diamagnetic or paramagnetic nature – to generate new relative orientation of the radical units and therefore altered magnetic behavior. The area of supramolecular chemistry which applies most to this objective is that of crystal engineering [24] – the use of non-covalent intermolecular bonding interactions to influence solid state arrangements of molecules. Perhaps the most simple approach appeals to simple acid-base chemistry. For example, the radicals 3PYNN (69, Fig. 24) and 4PYNN (38) when treated with HBr(g) form micro-crystals which incorporate two molecules of radical for each hydrogen bromide ion pair [172]. The (3PYNN)2 HBr adduct presents ferromagnetic interactions (while the parent radical has antiferromagnetic ones), which according to the authors is a result of the type of stack shown in Fig. 24, in which two pyridine nitrogen atoms complex one proton, and these units pile up, although there is no hard structural evidence for this argument. According to this model, the fairly strong ferromagnetic interactions (J/k = +5.7 K) in the stack are counteracted by weak antiferromagnetic interactions (J /k = −0.24 K) between them [173]. In contrast, the salt of 4PYNN of the same stoichiometry shows antiferromagnetic interactions that seem to extend in three dimensions (θ = −2.3 K) while in the parent compound they are ferromagnetic. Co-crystallization of 4PYNN independently with three diacids has produced zerodimensional supramolecules in the form of 2:1 radical/acid hydrogen bonded complexes of different topologies depending on the hardness of the acid (Fig. 25) [174]. When the acid is soft, as in the case of hydroquinone, the acidic hydrogen atoms form hydrogen bonds with the spin-bearing oxygen atoms, while in the case of harder acids the pyridinic nitrogen atom accepts the hydrogen bond. All the complexes showed antiferromagnetic interactions of varying degrees of strength which were correlated

Fig. 24. A representation of the molecular dimer formed in the (3PYNN)2 HBr salt and the stacks of dimers assumed to exist in the solid state [172].

1.4 Nitronyl Nitroxides

41

Fig. 25. Zero-dimensional hydrogen-bonded complexes of pyridine-derived nitronyl nitroxides [174, 175].

with the packing of the complexes, which were governed by weak interactions. The combination of the aforementioned basic nitronyl nitroxides (3PYNN and 4PYNN) with the complementary 4-benzoic acid-derived nitronyl nitroxide (4HOOCPNN, 70) also leads to similarly zero-dimensional complexes (Fig. 25), materials which also show antiferromagnetic coupling between the spins [175]. In turn, this acid has been crystallized as its lithium, sodium and potassium salts [176]. While the pure 4HOOCPNN [177] and the sodium and potassium salts show antiferromagnetic interactions, the lithium salt, which crystallizes with two molecules of methanol, shows strong ferromagnetic interactions (J/k = +15.9 K) which take place between the dimers formed in the crystals wherein the nitroxide oxygen atom of one molecule is in close proximity to the carbon atom between the same residues in another molecule [176]. The ability of the spin-bearing oxygen atoms of the nitroxide radicals to accept hydrogen bonds has been exploited by Kobayashi and coworkers for the preparation of three 1:1 complexes of nitronyl nitroxides with boronic acids (Fig. 26) [178]. All the complexes reported show moderate ferromagnetic couplings accompanied by weaker antiferromagnetic ones. The structures of the two crystals incorporating phenyl nitronyl nitroxide have been solved, and are almost isostructural [178], the magnetic behavior and structure of the phenyl boronic acid complex [179] being shown in Fig. 27. The complexes consist of hydrogen bonded one-dimensional helical chains of molecules with alternation of the two molecular components. The

42

1 Nitroxide-based Organic Magnets

Fig. 26. One-dimensional hydrogen bonded 1:1 complexes of phenyl nitronyl nitroxides with boronic acids and their magnetic exchange constants, where J/k and J /k are the intra- and inter-chain exchange coupling constants [178].

Fig. 27. The magnetic susceptibility (presented as plots of χ T against T ) and views of the crystal structure (helical chain and close approaches between radicals in different chains) of phenyl nitronyl nitroxide and phenyl boronic acid [179]. Reproduced by permission of The Royal Society of Chemistry.

1.4 Nitronyl Nitroxides

43

NO· · ·HOB(R)OH· · ·ON unit was proposed as the ferromagnetic interaction route. However, the shortest NO to NO distance (approximately 4.6 Å) is between helical chains (Fig. 27). This approach was assigned to the weak antiferromagnetic interaction, with the rational that the SOMOs are relatively close and there is some overlap between them. The orientation of this packing was slightly different in the two structures, and was inferred to explain the small differences in the exchange interactions. Buchachenko proposed, twenty years ago, an ingenious strategy for the preparation of materials with the potential behave as ferrimagnets that is based on cocrystallization of radicals with different spin multiplicities [180]. Following this idea, Sugawara and coworkers prepared several very interesting complexes [181]. Firstly, they managed to co-crystallize the triradical PTNN (60) with trinitrobenzene, which forms a material with molecular stacks containing the two components in alternating fashion, and thereby confirm that the open shell molecule has a quartet ground state, with an intramolecular exchange coupling J/k of +23 K [161, 182]. At low temperatures, weak intermolecular antiferromagnetic interactions become evident. The same driving forces used in the crystallization, which are Coulombic and π –π stacking interactions [183], were used for the preparation of a material containing stacks of compounds with spin multiplicities of S = 1 and S = 1/2 [184]. The bisnitronyl nitroxide 1,3PBNN (58A) with S = 1 and the mononitroxide 3,5BNO2PNN (71) with S = 1/2 are the components, which pack in alternating fashion forming a column (Fig. 28), with intercalated benzene molecules of crystallization between the stacks. The ferrimagnetic material which was the goal of the work was not achieved, because the monoradical interacts magnetically with one of the two radical moieties in the diradical preferentially, giving rise to a doublet pair. This phenomenon was witnessed in the magnetic susceptibility data, which at room temperature has a χ T value corresponding to independent three spins, a value which decreases on lowering the temperature down to 10 K as a result of antiferromagnetic interactions. The magnetic data in this temperature range is well reproduced by a model which has an intramolecular coupling J1 /k of +20 K within the diradical and an intermolecular coupling J2 /k of −30 K between neighboring 1,3PBNN and 3,5BNO2 PNN components along one direction. Between 4 and 6 K the χ T value is almost constant with a value somewhat larger than that expected for an S = 1 system, suggesting a short-range magnetic spin coupling. Below 3 K further antiferromagnetic interactions are evident. Surprisingly, the spin remaining at low temperature behaves basically independently of the other radical moiety in the molecule, as evidenced by EPR measurements which show two broad lines below 10 K as a result of the change in spin state [185]. One line corresponds to the antiferromagnetically coupled spins, and is most intense in the direction of the molecular columns, while the other has the same intensity in all directions and corresponds to the other spin in the diradical. The conclusion of this work is that the magnetic degree of freedom in the triplet molecule precludes the success of this approach to ferrimagnetic materials, and the criteria for equal coupling of the two spins in the triplet to the singlet are not met [186].

44

1 Nitroxide-based Organic Magnets

Fig. 28. Structure of the 1,3PBNN (58):3,5BNO2 NN (71) co-crystal and its magnetic susceptibility behavior [181].

1.5

Imino Nitroxides

The imino nitroxides are directly derived from the nitronyl nitroxides, and are often obtained by accident as a consequence of over-oxidation of the latter, but have received more interest for their coordination chemistry than for their inherent magnetic behavior, perhaps because many of the radicals in their pure state show predominantly antiferromagnetic interactions [119, 125, 162-164]. This situation is somewhat disparaging, given that in principle the spin density is able to delocalize more in these compounds onto the substituent at the 2-position of the imidazolyl ring than in the nitronyl nitroxides since there is no node at the carbon atom at the 2-position of the imidazolyl ring (Fig. 1) [9, 10].

1.5 Imino Nitroxides

45

Fig. 29. A representation of the molecular chains formed by EPIN (72) and its low temperature dependence of the susceptibility (M/H) cooled in a 3.7 G field [187].

Ferromagnetic order has been observed in the 2-ethynylpyridine imino nitroxide EPIN (72, Fig. 29) [187]. As in the corresponding nitronyl nitroxide derivative [117], the compound packs forming hydrogen-bonded chains linked by the alkyne hydrogen atom and the nitroxide oxygen atom, an interaction which was implicated in the ferromagnetic exchange. Fitting of the magnetic susceptibility data to a onedimensional Heisenberg chain model gave a J/k of +0.50 K. Application of a mean field correction didn’t improve the fit. Spontaneous magnetization was observed below 0.2 K (Fig. 29), with a saturation value of magnetization close to 1 ÌB /molecule. No hysteresis was observed, field dependence of the magnetization shows that the sample is very easily saturated, which prompted the authors to infer a bulk ferromagnetic order. The 4- and 3-benzoic acid imino nitroxides have also recently been reported [188]; in these the interactions are ferromagnetic and, it has been proposed, are propagated by the intermolecular hydrogen bonds which unite the molecules in the crystals. The same compounds have also been used in interesting ferromagnetic organic–inorganic composites with layered cobalt hydroxides [189]. Several of the bis(nitronyl nitroxides) described above were also used to prepare the corresponding bis(imino nitroxide) and combined imino-nitronyl bis nitroxide

46

1 Nitroxide-based Organic Magnets

derivatives by deoxygenation. In general the magnetic properties of the resulting compounds do not differ dramatically from their precursors [155, 163, 190], although the strength of the exchange interaction is somewhat lower, an effect ascribed to the lower extent of spin polarization occurring in imino nitroxides [155].

1.6

Poly(nitroxides)

One of the most conceptually-appealing ways to generate an organic ferromagnet is in the form of a polymer [191], in which radicals are spaced equally along the polymer skeleton with ferromagnetic coupling units between them (Fig. 30), as in this way ferromagnetic interactions within the covalent skeleton can in principle be extended over extremely large distances. Nevertheless, the extension of the ferromagnetic interactions in two- or three-dimensions, a fundamental requirement for achieving a bulk ferromagnet [4a], can only be attained with two- or three-dimensional polymer networks containing millions of radical centers and in which each spin-containing unit is ferromagnetically coupled with at least three (preferably more) nearest neighbor units. The apparently simple objective of achieving a super-high-spin macromolecule, as laid out in ideal manner here, has received a great deal of attention, and yet has met with extremely limited, not to say no, practical success. Following the controversy surrounding a family of polymerized diacetylene-linked bis(TEMPO) derivatives [192], which in principle might present ferromagnetic interactions through space within the across the monomeric units in the polymer chains as in A in Fig. 30, work has concentrated on the preparation of polymers in which a ferromagnetic coupler is used to link radical moieties, as in B in Fig. 30. Iwamura and coworkers presented a cunning approach to poly(1-phenyl-1,3butadiyne)s based on the solid state polymerization of co-crystals of diacetylene derivative 73 and radical 74, given that the pure radical had a solid state structure

Fig. 30. Two approaches to polymer radicals with ferromagnetic interactions between spins: A in which the spins interact principally through-space in the polymer side-chains and B in which the spins are coupled through-bonds by a ferromagnetic coupler (FC).

1.6 Poly(nitroxides)

47

unsuitable for polymerization [193]. The resulting material suffered drastically from loss of spin, the surviving radicals being predominantly paramagnetic, whereas 10% appeared to be coupled ferromagnetically, but which showed loss of spin below 250 K, perhaps as a result of contraction of the crystals causing recombination.

Various poly(1,3-phenyleneethynylene) radicals have been prepared (Fig. 31), all showing less than perfect number of spins per repeat unit of the polymer, and all presenting antiferromagnetic interactions in their condensed phases. The polymers 75–77 all show weak antiferromagnetic interactions [194–196], for example in the polymer 77 which incorporates two distinct radical moieties, the Weiss constant was −1.5 K, and was attributed to through-space interactions between polymers, since the in-chain magnetic interaction was said to be extremely weak [197]. Given that all

Fig. 31. Structure and properties of poly(1,3-phenyleneethynylene) radicals.

48

1 Nitroxide-based Organic Magnets

Fig. 32. Structures and properties of poly(1,2-diethynylphenyl) and poly(1,3-diethynylphenyl) radicals.

these polymers were prepared using the palladium catalyzed coupling if acetylenes with aromatic iodides, one has to ask whether some of the nitronyl nitroxide units had been converted to imino nitroxides, since this process has precedent. It is known that these polymers, in the absence of radical groups, have high conformational flexibility, but this aspect was not highlighted in these polymers, unlike the group of polymers which follows. The poly(1,2-diethynylphenyl) and poly(1,3-diethynylphenyl) radicals 78–81 (Fig. 32) show either paramagnetic or slightly antiferromagnetic coupling between radicals. Nitronyl nitroxide polymers 78 [198] and 79 [199] have elevated spin contents, but in the former case shows paramagnetic or weak antiferromagnetic coupling between spins according to the processing conditions. Indeed, one might expect this kind of morphology-dependent behavior (relatively common in polymers in general) since it is analogous to polymorphic behavior of crystals. Polymer 79 showed a large rise in the χ T curve above 100 K in one sample, probably indicative of paramagnetic impurity. The polymer 80 was paramagnetic in the temperature range studied (like the precursor monomer), while 81 showed weak antiferromagnetic interac-

1.7 Outlook

49

tions which had Weiss constants varying between −6.2 and −5.0 K depending on the preparation and precipitation conditions [200]. Following their studies concerning phenylenevinylenes [87], Nishide, Tsuchida and co-workers prepared the polyradical 82, which contains nitronyl nitroxide radicals at the 4-position with respect to the linking alkene unit in the stereoregular macromolecule [201]. The spin content of this radical was claimed to be 97% of the maximum possible value, and in addition the polymer shows ferromagnetic interactions with a J/k of +21 K (using the Bleaney–Bowers equation) in solution, although there are antiferromagnetic interactions present. The value of the ferromagnetic coupling far exceeds that of a model diradical and points to cooperative interactions. However, in the powder form, the radical presents antiferromagnetic coupling pointing to significant through-space interactions which were also seen to a lesser degree in solution.

Among the various causes that thwart the valiant efforts at preparing polymeric magnets are: (i) the preparation of the polymers can result in either the loss of radical centers or incomplete radical formation during the synthesis, giving rise to diamagnetic defects incapable of sustaining the required long-range ferromagnetic interactions; (ii) the conformation of the polymer chain is generally irregular, giving rise to occasional null or to antiferromagnetic interactions, thereby generating domains which can cancel one another; (iii) the polymer chains have ferromagnetic interactions between the spins, but between polymer chains the interaction is usually antiferromagnetic; (iv) usually, a modest number (no more than a few tens) of ferromagnetically-coupled spin units is obtained instead of the gigantic number (millions) required; and (v) the difficulties in reproducing the synthesis, the conformational stereoregularity, and characterizing completely the end products.

1.7

Outlook

Nitroxide magnets’ prime limitation is their uniformly low critical temperatures (below 2 K), which results from the extremely weak magnetic exchange interactions within and between the organic molecules currently being prepared. In turn, these

50

1 Nitroxide-based Organic Magnets

open-shell molecules show a weak spin–orbit coupling which results in very isotropic Heisenberg-type spin systems in the purely organic molecules incorporating only light elements. This characteristic makes the presence of ferromagnetic interactions between radicals in three-dimensions essential, otherwise no ferromagnetic ordering is possible. The strict electronic and structural requirements implicit in this condition undoubtedly hinder enormously the development of this kind of magnetic material. The low magnetic anisotropy of nitroxide units also makes it difficult to chance upon canted ferromagnets, since when an antiferromagnetic interaction is established between two units it leads to a complete compensation of the two spins because of their isotropic nature. It is clear that there is relative dominance over intramolecular magnetic interactions, whatever their strength. Synthetic organic chemistry has guided the preparation of various high-spin molecules with triplet or higher ground states. However, the situation beyond the molecule is not so controlled, and intermolecular magnetic interactions are as yet a relatively untamed beast. It is also clear from the work discussed in this chapter that in order to progress in this area efforts in the following directions are necessary: – Development of experimental methods for characterizing the spin density distribution of open-shell organic molecules both in solution and the solid states, and for determining the mechanisms, pathways, and dimensionalities of intermolecular magnetic interactions. – Development of theoretical models that reproduce and correctly interpret the magnitude and dimensionality of experimental magnetic interactions, and for calculating with high precision intermolecular magnetic interactions and magnetic anisotropy of organic molecules. – Preparation of new organic molecules with highly delocalized spins that does not imply a loss of their persistence, and/or that present a large spin polarization over the whole structure. – Alternative methods to enhance the magnetic anisotropy in molecules containing only light elements, and the discovery of new tricks for increasing the strength of magnetic exchange interactions between molecules united by either covalent or non-covalent bonds. – New ways of organizing the compounds to form new molecular materials, for example in the form of thin films by Langmuir-Blodgett techniques [202] or chemical vapor deposition [203]. – Synthesis of new molecules incorporating new crystal engineering synthons that control relative dispositions of molecules in the solid state which have strong ferromagnetic interactions. While all of these aspects are unlikely to raise the transition temperatures of the nitroxide materials the orders of magnitude which would be necessary for applications in today’s technology, they are bound to provide important messages for chemistry and physics concerning the interactions of free electrons [204].

References

51

Acknowledgments We thank our coworkers who have contributed to our own research: Joan Cirujeda, Merce´ Deumal, Robert Feher, Esteve Hernandez, ` Oriol Jurgens, ¨ Maria Minguet, Concepcio´ Rovira, J. Vidal-Gancedo, and especially Juan J. Novoa for his enthusiastic collaboration. This work was supported by grants from the DGICyT, Spain (Proyecto no. PB96-0862-C0201), and the 3MD Network of the TMR program of the E.U. (Contract ERBFMRXCT 980181).

References [1] The nitroxides are more properly named aminoxyls, according to IUPAC, although we have maintained the former term because of its prevelence in the literature. For a few of the many general reviews concerning nitroxides, along with references cited therein, see: (a) A.R. Forrester, J.M. Hay, R.M. Thomson, Organic Chemistry of Stable Free Radicals, Academic Press, New York 1968; (b) J.F.W. Keana, Chem. Rev., 1978, 78, 37–64; (c) M. Dagonneau, E. S. Kagan, V. I. Mikhailov, E. G. Rozantsev, V. D. Sholle, Synthesis, 1984, 895–916; (d) L. B. Volodarsky, Janssen Chim. Acta, 1990, 8, 12–19; (e) M.-E. Brik, Heterocycles, 1995, 41, 2827–2873. [2] P. M. Lahti, (Ed.), Magnetic Properties of Organic Molecules, Marcel Dekker, New York 1999. [3] (a) Magnetic Molecular Materials, D. Gatteschi, O. Kahn, J. S. Miller, F. Palacio (Eds.): NATO ASI Series E Vol. 198, Kluwer, Dordrecht 1991. (b) O. Kahn, Molecular Magnetism, VCH, Weinheim, 1993. [4] For reviews concerning the application of nitroxides in the realm of molecular magnetism, see: (a) F. Palacio, in Magnetic Molecular Materials, D. Gatteschi, O. Kahn, J. S. Miller, F. Palacio (Eds.): NATO ASI Series E Vol. 198, Kluwer, Dordrecht 1991, p. 1–34; (b) M. Baumgarten, K. Mullen, ¨ Top. Curr. Chem., 1994, 169, 1–103. (c) J. Veciana, J. Cirujeda, C. Rovira, J. Vidal-Gancedo, Adv. Mater. 1995, 7, 221–225; (d) S. Nakatsuji, H. Anzai, J. Mater. Chem., 1997, 7, 2161–2174. [5] (a) A. Caneschi, D. Gatteschi, R. Sessoli, P. Rey, Acc. Chem. Res., 1989, 22, 392–398. (b) D. Gatteschi, R. Sessoli, J. Magn. Magn. Mater., 1992, 104–107, 2092–2095. (c) A. Caneschi, D. Gatteschi, R. Sessoli, Mol. Cryst. Liq. Cryst., 1996, 279, 177–194. (d) H. Iwamura, K. Inoue, N. Koga, T. Hayamizu, in Magnetism: A Supramolecular Function, O. Kahn (Ed.), pp. 157–179, NATO ASI Series C Vol. 484, Kluwer Academic Publishers, Dordrecht, 1996. (e) K. Fegy, K. E. Vostrikova, D. Luneau, P. Rey, Mol. Cryst. Liq. Cryst., 1997, 305, 69–80; (f) P. Rey, D. Luneau, in Supramolecular Engineering of Synthetic Metallic Materials, J. Veciana, C. Rovira, D.B. Amabilino (Eds.), pp. 145-174, NATO ASI Series C Vol. 518, Kluwer Academic Publishers, Dordrecht, 1999. [6] For example, NLO properties: (a) J.-F. Nicoud, C. Serbutoviez, G. Puccetti, I. Ledoux, J. Zyss, Chem. Phys. Lett., 1990, 175, 257–261; (b) G. Puccetti, I. Ledoux, J. Zyss, NATO ASI Ser., Ser. E, 1991, 194, 207–213. (c) S. Yamada, M. Nakano, S. Kiribayashi, I. Shigemoto, K. Yamaguchi, Synth. Met., 1997, 85, 1081–1082; (d) S. Yamada, M. Nakano, I. Shigemoto,

52

[7]

[8]

[9] [10] [11]

[12] [13] [14] [15] [16]

[17] [18] [19] [20] [21]

1 Nitroxide-based Organic Magnets S. Kiribayashi, K. Yamaguchi, Chem. Phys. Lett., 1997, 267, 438–444; (e) S. Yamada, M. Nakano, K. Yamaguchi, Chem. Phys. Lett., 1997, 276, 375–380; (f) M. Nakano, S. Yamada, K. Yamaguchi, Bull. Chem. Soc. Jpn., 1998, 71, 845–850. (a) J.W. Neely, G.F. Hatch, R.W. Kreilick, J. Am. Chem. Soc., 1974, 96, 652–656; (b) H.G. Aurich, K. Hahn, K. Stork, W. Weiss, Tetrahedron, 1977, 33, 969–975; (c) I.A. Grigor’ev, A.Z. Gogolev, R.Z. Sagdeev, L.B. Volodarskii, Izv. Akad. Nauk SSSR, Ser. Khim., 1989, 1029–1036; (d) A. Zheludev, M. Bonnet, E. Ressouche, J. Schweizer, M. Wan, H. Wang, J. Magn. Magn. Mater., 1994, 135, 147–160; (e) M. Bonnet, D. Luneau, E. Ressouche, P. Rey, J. Schweizer, M. Wan, H. Wang, A. Zheludev, Mol. Cryst. Liq. Cryst., 1995, 271, 35–53; (f) T. Takui, Y. Miura, K. Inui, Y. Teki, M. Inoue, K. Itoh, Mol. Cryst. Liq. Cryst., 1995, 271, 55–66; (g) D. Shiomi, S. Honbu, K. Sato, T. Takui, K. Itoh, T. Sugano, Mol. Cryst. Liq. Cryst., 1997, 306, 199–208; (g) J. Schweizer, Physica B, 1997, 234–236, 772– 779; (h) H. Heise, F.H. Kohler, ¨ F. Mota, J.J. Novoa, J. Veciana, J. Am. Chem. Soc., 1999, 121, 9659–9667. (a) B. Gillon, Y. Ellinger, Mol. Phys., 1988, 63, 967–979; (b) V. Barone, A. Grand, D. Luneau, P. Rey, C. Minichino, R. Subra, New J. Chem., 1993, 17, 545–549; (c) M. Okumura, K. Yamaguchi, M. Nakano, W. Mori, Chem. Phys. Lett., 1993, 207, 1–8; (d) A. Zheludev, V. Barone, M. Bonnet, B. Delley, A. Grand, E. Ressouche, P. Rey, R. Subra, J. Schweizer, J. Am. Chem. Soc., 1994, 116, 2019–2027; (e) J. J. Novoa, F. Mota, J. Veciana, J. Cirujeda, Mol. Cryst. Liq. Cryst., 1995, 271, 79–90. (f) S. Yamanaka, T. Kawakami, H. Nagao, K. Yamaguchi, Mol. Cryst. Liq. Cryst., 1995, 271, 19–28. (g) V. Barone, A. Bencini, A. di Matteo, J. Am. Chem. Soc., 1997, 119, 10831–10837. A. Zheludev, M. Bonnet, D. Luneau, E. Ressouche, P. Rey, J. Schweizer, Physica B, 1995, 213–214, 268–271. ¨ A. Zheludev, M. Bonnet, B. Delley, A. Grand, D. Luneau, L. Orstr om, ¨ E. Ressouche, P. Rey, J. Schweizer, J. Magn. Magn. Mater., 1995, 145, 293–305. (a) J. S. Miller, A. J. Epstein, W. M. Reiff, Science, 1988, 240, 40–47; (b) H. Iwamura, Adv. Phys. Org. Chem., 1990, 26, 179–253; (c) L. Dulog, Nachr. Chem., Tech. Lab., 1990, 38, 445–448, 451. (d) T. Nogami, Shikizai Kyokaishi, 1990, 63, 685–693; (e) J. S. Miller, A. J. Epstein, W. M. Reiff, Chem. Rev., 1988, 88, 201–220; (f) T. Nogami, Oyo Butsuri, 1994, 63, 674–681; (g) J. S. Miller, A. J. Epstein, Angew. Chem. Int. Ed. Engl., 1994, 33, 385–415. H. Lemaire, A. Rassat, The Triplet State, Beirut Symposium, Cambridge University Press 1967, p141. M. Saint-Paul, C. Veyret, Phys. Lett., 1973, 45A, 362–364. (a) G. Chouteau, C. Veyret-Jeandey, J. Physique, 1981, 42, 1441–1444; (b) A. Benoit, J. Flouquet, B. Gillon, J. Schweizer, J. Magn. Magn. Mater., 1983, 31–34, 1155–1156. (a) M. Kinoshita, Mol. Cryst. Liq. Cryst., 1993, 232, 1–12 (b) M. Kinoshita, Jpn. J. Appl. Phys., 1994, 33, 5718–5733. (a) H. C. Longuet-Higgins, J. Chem. Phys., 1950, 18, 265; (b) W.T. Borden, E. R. Davidson, J. Am. Chem. Soc., 1977, 99, 4587–4594; (c) D. A. Dougherty, Acc. Chem. Res., 1991, 24, 88–94. J.L. Atwood, J.E.D. Davies, D.D. MacNicol, F. Vogtle ¨ (Eds.) Comprehensive Supramolecular Chemistry, Pergamon, Elsevier Science Ltd., 1996, Oxford. L. Salem, C. Rowland, Angew. Chem., Int. Ed. Engl., 1972, 11, 92–111. H. M. McConnell, J. Chem. Phys., 1963, 39, 1910. M. Deumal, J.J. Novoa, M.J. Bearpark, P. Celani, M. Olivucci, M.A. Robb, J. Phys. Chem. A, 1998, 102, 8404–8412. (a) M. Deumal, J. Cirujeda, J. Veciana, J.J. Novoa, Adv. Mater., 1998, 10, 1461–1466; (b) M. Deumal, J. Cirujeda, J. Veciana, J.J. Novoa, Chem. Eur. J., 1999, 5, 1631–1642.

References

53

[22] For more detailed discussions of this model and other related ones developed by other authors see reference 4, along with papers cited in these publications. [23] Supramolecular Engineering of Synthetic Metallic Materials, J. Veciana, C. Rovira, D.B. Amabilino (Eds.), pp145-174, NATO ASI Series C Vol. 518, Kluwer Academic Publishers, Dordrecht, 1999. [24] (a) G.R. Desiraju, Angew. Chem., Int. Ed. Engl., 1995, 34, 2311–2327. (b) G.R. Desiraju (Ed.) The Crystal as a Supramolecular Entity: Perspectives in Supramolecular Chemistry Volume 2, John Wiley & Sons, New York, 1996. [25] G.A. Jeffrey, An Introduction to Hydrogen Bonding, Oxford University Press, Oxford, 1997. [26] B.N. Figgis, E.S. Kucharski, M. Vrtis, J. Am. Chem. Soc. 1993, 115, 176–181. [27] (a) R. Taylor, O. Kennard, J. Am. Chem. Soc., 1982, 104, 5063–5070; (b) G.R. Desiraju, Acc. Chem. Res., 1996, 29, 441–449; (c) T. Steiner, Chem. Commun., 1997, 727–734; (d) T. Steiner, Chem. Commun., 1999, 313–314. [28] L.J. Berliner (Ed.), Spin Labeling Theory and Applications, Vol. 1 & 2, Academic Press, New York 1976 & 1979. [29] R. Briere, ` H. Lemaire, A. Rassat, Bull. Soc. Chim. Fr., 1965, 3273–3283. ´ [30] (a) Y.S. Karimov, E.G. Rozantsev, Soviet Physics-Solid State, 1967, 8, 2225–2226, (b) J. Yamauchi, T. Fujito, E. Ando, H. Nishiguchi, Y. Deguchi, J. Phys. Soc. Jpn., 1968, 25, 1558–1561; (c) J. Yamauchi, Bull. Chem. Soc. Jpn., 1971, 1971, 2301–2308. [31] (a) H. Lemaire, P. Rey, A. Rassat, A. de Combarieu, J.-C. Michel, Mol. Phys., 1968, 14, 201–208; (b) J.-P. Boucher, M. Nechtschein, M. Saint-Paul, Phys. Lett., 1973, 42A, 397–398. [32] K. Takeda, N. Uryu, M. Inoue, J. Yamauchi, J. Phys. Soc. Jpn., 1987, 56, 736–741. [33] (a) J.-P. Lajzerowicz-Bonneteau, ´ Acta Cryst., 1968, B24, 196–199.; (b) L.J. Berliner, Acta Cryst., 1970, B26, 1198–1202. [34] H. Sugimoto, H. Aota, A. Harada, Y. Morishima, M. Kamachi, W. Mori, M. Kishita, N. Ohmae, M. Nakano, M. Sorai, Chem. Lett., 1991, 2095–2098. [35] M. Kamachi, H. Sugimoto, A. Kajiwara, A. Harada, Y. Morishima, W. Mori, N. Ohmae, M. Nakano, M. Sorai, T. Kobayashi, K. Amaya, Mol. Cryst. Liq. Cryst., 1993, 232, 53–60. [36] A. Kajiwara, W. Mori, M. Sorai, K. Yamaguchi, M. Kamachi, Mol. Cryst. Liq. Cryst., 1995, 272, 67–74. [37] P.D. Bordeaux, J. Lajzerowicz, ´ Acta Cryst., 1977, B33, 1837–1840. [38] T. Nogami, T. Ishida, M. Yasui, F. Iwasaki, H. Iwamura, N. Takeda, M. Ishikawa, Mol. Cryst. Liq. Cryst., 1996, 279, 97–106. [39] (a) T. Nogami, R. Imachi, T. Ishida, N. Takeda, M. Ishikawa, Mol. Cryst. Liq. Cryst., 1997, 305, 211–216; (b) R. Imachi, T. Ishida, T. Nogami, M. Yasui, F. Iwasaki, N. Takeda, M. Ishikawa, Synth. Met., 1997, 85, 1743–1744. [40] G. Maruta, S. Takeda, R. Imachi, T. Ishida, T. Nogami, K. Yamaguchi, J. Am. Chem. Soc., 1999, 121, 424–431. [41] G. Maruta, S. Takeda, T. Kawakami, W. Mori, R. Imachi, T. Ishida, T. Nogami, K. Yamaguchi, Mol. Cryst. Liq. Cryst., 1997, 306, 307–314. [42] K. Togashi, R. Imachi, K. Tomioka, H. Tsuboi, T. Ishida, T. Nogami, N. Takeda, M. Ishikawa, Bull. Chem. Soc. Jpn., 1996, 69, 2821–2830. [43] (a) T. Ishida, K. Tomioka, T. Nogami, H. Iwamura, K. Yamaguchi, W. Mori, Y. Shirota, Mol. Cryst. Liq. Cryst., 1993, 232, 99–102; (b) K. Tomioka, T. Ishida, T. Nogami, H. Iwamura, Chem. Lett., 1993, 625–626. (c) M. Yasui, H. Yoshikawa, H. Yamamoto, T. Ishida, T. Nogami, F. Iwasaki, Mol. Cryst. Liq. Cryst., 1996, 279, 277–285. [44] K. Tomioka, S. Mitsubori, T. Ishida, T. Nogami, H. Iwamura, Chem. Lett., 1993, 1239– 1242.

54

1 Nitroxide-based Organic Magnets

[45] T. Nogami, K. Tomioka, T. Ishida, H. Yoshikawa, M. Yasui, F. Iwasaki, H. Iwamura, N. Takeda, M. Ishikawa, Chem. Lett., 1994, 29–32. [46] T. Ishida, H. Tsuboi, T. Nogami, H. Yoshikawa, M. Yasui, F. Iwasaki, H. Iwamura, N. Takeda, M. Ishikawa, Chem. Lett., 1994, 919–922. [47] (a) T. Nogami, T. Ishida, H. Tsuboi, H. Yoshikawa, H. Yamamoto, M. Yasui, F. Iwasaki, H. Iwamura, N. Takeda, M. Ishikawa, Chem. Lett., 1995, 635–636. (b) T. Nogami, K. Togashi, H. Tsuboi, T. Ishida, H. Yoshikawa, M. Yasui, F. Iwasaki, H. Iwamura, N. Takeda, M. Ishikawa, Synth. Met., 1995, 71, 1813–1814. [48] T. Nogami, T. Ishida, Mol. Cryst. Liq. Cryst., 1997, 296, 305–322. [49] R. Imachi, T. Ishida, T. Nogami, S. Ohira, K. Nishiyama, K. Nagamine, Chem. Lett., 1997, 233–234. [50] F. Iwasaki, J. H. Yoshikawa, H. Yamamoto, E. Kan-Nari, K. Takada, M. Yasui, T. Ishida, T. Nogami, Acta Cryst. B, 1999, 55, 231–245. [51] F. Iwasaki, H. Yamamoto, J. H. Yoshikawa, K. Takada, E. Kannari, M. Yasui, T. Ishida, T. Nogami, Mol. Cryst. Liq. Cryst., 1998, 313, 223–228. [52] T. Nogami, T. Ishida, M. Yasui, F. Iwasaki, N. Takeda, M. Ishikawa, T. Kawakami, K. Yamaguchi, Bull. Chem. Soc. Jpn., 1996, 69, 1841–1848. [53] T. Kawakami, A. Oda, S. Takeda, W. Mori, T. Ishida, M. Yasui, F. Iwasaki, T. Nogami, K. Yamaguchi, Mol. Cryst. Liq. Cryst., 1997, 306, 141–150. [54] T. Ishida, K. Tomioka, T. Nogami, H. Yoshikawa, M. Yasui, F. Iwasaki, N. Takeda, M. Ishikawa, Chem. Phys. Lett., 1995, 247, 7–12. [55] A. Misiolek, R. Huang, B. Kahr, J. E. Jackson, Chem. Commun., 1996, 2119–2120. [56] D. Papoutsakis, J. P. Kirby, J. E. Jackson, D. G. Nocera, Chem. Eur. J., 1999, 5, 1474– 1480. [57] R. Briere, ` R.-M. Dupeyre, H. Lemaire, C. Morat, A. Rassat, P. Rey, Bull. Soc. Chim. Fr., 1965, 3290–3297. [58] For a review of oligonitroxides derived from TEMPO, see Ref. [1e], and for a selection of recent examples see: (a) L. Dulog, W. Wang, Adv. Mater., 1992, 4, 349–351; (b) L. Dulog, W. Wang, Liebigs Ann. Chem., 1992, 301–303; (c) J. S. Kim, L. Dulog, Bull. Korean Chem. Soc., 1993, 14, 465–468. (d) M.-J. Tien, Y. Wang, T.-I. Ho, Y. Chiang, Synth. Met., 1997, 85, 1675–1676. [59] P. J. Brown, A. Capiomont, B. Gillon, J. Schweizer, Mol. Phys., 1983, 48, 753–761. [60] A. Capiomont, Acta Cryst., 1972, B28, 2298–2301. [61] K. Grieser, W. Haase, I. Svoboda, F. Fuess, Acta Cryst., 1997, C53, 1488–1490. [62] A. Rassat, Pure & Appl. Chem., 1990, 62, 223–227. [63] (a) R. Chiarelli, A. Rassat, P. Rey, J. Chem. Soc., Chem. Commun., 1992, 1081–1082. (b) R. Chiarelli, A. Rassat, Bull. Soc. Chim. Fr., 1993, 130, 299–303. [64] (a) A. Rassat, R. Chiarelli in Magnetic Molecular Materials, D. Gatteschi, O. Kahn, J. S. Miller, F. Palacio (Eds.): NATO ASI Series E Vol. 198, Kluwer, Dordrecht 1991, p. 191– 202. (b) R. Chiarelli, A. Rassat, Y. Dromzee, Y. Jeannin, M. A. Novak, J. L. Tholence, Phys. Scr., 1993, T49B, 706–710. [65] R. Chiarelli, M. A. Novak, A. Rassat, J. L. Tholence, Nature (London), 1993, 363, 147– 149. [66] Y. Dromzee, R. Chiarelli, S. Gambarelli, A. Rassat, Acta Cryst., 1996, C52, 474–477. [67] A. Zheludev, R. Chiarelli, B. Delley, B. Gillon, A. Rassat, E. Ressouche, J. Schweizer, J. Magn. Magn. Mater., 1995, 140–144, 1439–1440. [68] P. Fluekiger, J. Weber, R. Chiarelli, A. Rassat, Y. Ellinger, Int. J. Quantum Chem., 1993, 45, 649–663. [69] W. Duffy Jr., D. L. Strandburg, J. F. Deck, Phys. Rev., 1969, 183, 567–572. [70] O. Takizawa, Bull. Chem. Soc. Jpn., 1976, 49, 583–588.

References

55

[71] R. Imachi, T. Ishida, M. Suzuki, M. Yasui, F. Iwasaki, T. Nogami, Chem. Lett., 1997, 743–744. [72] V. A. Reznikov, I. V. Ovcharenko, N. V. Pervukhina, V. N. Ikorskii, A. Grand, V. I. Ovcharenko, Chem. Commun., 1999, 539–540. [73] H. Iwamura, Mol. Cryst. Liq. Cryst., 1993, 232, 233–250. [74] P. G. Wenthold, J. B. Kim, W. C. Lineberger, J. Am. Chem. Soc., 1997, 119, 1354–1359. [75] A. Calder, A.R. Forrester, P.G. James, G.R. Luckhurst, J. Am. Chem. Soc., 1969, 91, 3724–3727. [76] M. Dvolaitzky, R. Chiarelli, A. Rassat, Angew. Chem. Int. Ed. Engl., 1992, 31, 180– 181. [77] F. Kanno, K. Inoue, N. Koga, H. Iwamura, J. Am. Chem. Soc., 1993, 115, 847–850. [78] R. Chiarelli, S. Gambarelli, A. Rassat, Mol. Cryst. Liq. Cryst., 1997, 305, 455–478. [79] J. Fujita, M. Tanaka, H. Suemune, N. Koga, K. Matsuda, H. Iwamura, J. Am. Chem. Soc., 1996, 118, 9347–9351. [80] K. Yoshizawa, T. Kuga, M. Hatanaka, K. Tanaka, T. Yamabe, Synth. Met., 1997, 85, 1765–1766. [81] T. Itoh, K. Matsuda, H. Iwamura, Angew. Chem., Int. Ed. Engl., 1999, 38, 1791–1793. [82] H. Iwamura, N. Koga, Acc. Chem. Res., 1993, 26, 346–351. [83] T. Ishida, H. Iwamura, J. Am. Chem. Soc., 1991, 113, 4238–4241. [84] P.G. Wenthold, J. Hu, R R. Squires, W.C. Lineberger, J. Am. Chem. Soc., 1996, 118, 475–476. [85] T. Matsumoto, T. Ishida, N. Koga, H. Iwamura, J. Am. Chem. Soc., 1992, 114, 9952–9959. [86] D.A. Shultz, A.K. Boal, H. Lee, G.T. Farmer, J. Org. Chem., 1999, 64, 4386–4396. [87] T. Kaneko, S. Toriu, E. Tsuchida, H. Nishide, D. Yamaki, G. Maruta, K. Yamaguchi, Chem. Lett., 1995, 421–422. [88] N. Yoshioka, P.M. Lahti, T. Kaneko, Y. Kuzumaki, E. Tsuchida, H. Nishide, J. Org. Chem., 1994, 59, 4272–4280. [89] J. H. Osiecki, E. F. Ullman, J. Am Chem. Soc., 1968, 90, 1078–1079. [90] E. F. Ullman, J. H. Osiecki, D. G. B. Boocock, R. Darcy, J. Am. Chem. Soc., 1972, 94, 7049–7059. [91] See, for example; (a) A. Zheludev, E. Ressouche, J. Schweizer, P. Turek, M. Wan, H. Wang, Solid State Commun., 1994, 90, 233–235; (b) A. Zheludev, M. Bonnet, E. Ressouche, J. Schweizer, M. Wan, H. Wang, J. Magn. Magn. Mater., 1994, 135, 147–160. [92] (a) R.W. Kreilick, J. Becher, E.F. Ullman, J. Am. Chem. Soc., 1969, 91, 5121–5124; (b) J.A. D’Anna, J.H. Wharton, J. Chem. Phys., 1970, 53, 4047–4052. [93] (a) M.S. Davis, K. Morokuma, R.W. Kreilick, J. Am. Chem. Soc., 1972, 94, 5588–5592; (b) J.W. Neely, G.F. Hatch, R.W. Kreilick, J. Am. Chem. Soc., 1974, 96, 652–656; (c) J. Goldman, T.E. Petersen, K. Torssell, J. Becher, Tetrahedron, 1973, 29, 3833–3843. [94] (a) M.F. Ottaviani, J. Chem. Soc., Faraday Trans., 1990, 86, 3211–3219; (b) T. Takui, Y. Miura, K. Inui, Y. Teki, M. Inoue, K. Itoh, Mol. Cryst. Liq. Cryst., 1995, 271, 55–66. [95] J. Cirujeda, J. Vidal-Gancedo, O. Jurgens, ¨ F. Mota, J.J. Novoa, C. Rovira, J. Veciana, J. Am. Chem. Soc. 2000, 122, 11393–11405. [96] Y. Hosokoshi, M. Tamura, K. Nozawa, S. Suzuki, H. Sawa, R. Kato, M. Kinoshita, Mol. Cryst. Liq. Cryst. 1995, 271, 115–122. [97] Y. Hosokoshi, M. Tamura, K. Nozawa, S. Suzuki, M. Kinoshita, H. Sawa, R. Kato, Synth. Met., 1995, 71, 1795–1796. [98] (a) J. J. Novoa, M. Deumal, M. Kinoshita, Y. Hosokoshi, J. Veciana, J. Cirujeda, Mol. Cryst. Liq. Cryst., 1997, 305, 129–141; (b) J.J. Novoa, M. Deumal, Mol. Cryst. Liq. Cryst., 1997, 305, 143–156. (c) M. Deumal, J. Cirujeda, J. Veciana, M. Kinoshita, Y. Hosokoshi, J. J. Novoa, Chem. Phys. Lett. 1997, 265, 190–199.

56

1 Nitroxide-based Organic Magnets

[99] Y. Hosokoshi, M. Takahashi, T. Goto, K. Inoue, J. Magn. Magn. Mater., 1998, 177–181, 713–714. [100] M. Tamura, Y. Nakazawa, D. Shiomi, K. Nozawa, Y. Hosokoshi, M. Ishikawa, M. Takahashi, M. Kinoshita, Chem. Phys. Lett., 1991, 186, 401–404. [101] Y. Nakazawa, M. Tamura, N. Shirakawa, D. Shiomi, M. Takahashi, M. Kinoshita, M. Ishikawa, Phys. Rev. B: Condens. Matter, 1992, 46, 8906–8914. [102] (a) K. Awaga, T. Inabe, U. Nagashima, Y. Maruyama, J. Chem. Soc., Chem. Commun., 1989, 1617–1618. (b) K. Awaga, T. Inabe, U. Nagashima, Y. Maruyama, J. Chem. Soc., Chem. Commun., 1990, 520. (c) K. Awaga, T. Inabe, T. Yokoyama, Y. Maruyama, Mol. Cryst. Liq. Cryst., 1993, 232, 79–88. [103] Y. Hosokoshi, M. Tamura, H. Sawa, R. Kato, M. Kinoshita, J. Mater. Chem., 1995, 5, 41–46. [104] S.J. Blundell, P.A. Pattenden, K.H. Chow, F.L. Pratt, T. Sugano, W. Hayes, Synth. Met., 1997, 85, 1745–1746. [105] For a selection of metal complexes of heterocyclic derivatives of nitronyl nitroxides, see: (a) D. Luneau, P. Rey, Mol. Cryst. Liq. Cryst., 1995, 273, 519–225; (b) Z. N. Chen, J. Qiu, J. M. Gu, M. F. Wu, W. X. Tang, Inorg. Chim. Acta, 1995, 233, 131–135; (c) K. Fegy, D. Luneau, E. Belorizky, M. Novac, J.-L. Tholence, C. Paulsen, T. Ohm, P. Rey, Inorg. Chem., 1998, 37, 4524–4532; (d) C. F. Huang, H. H. Wei, G. H. Lee, Y. Wang, Inorg. Chim. Acta, 1998, 279, 233–237; (e) D. Luneau, F. M. Romero, R. Ziessel, Inorg. Chem., 1998, 37, 5078–5087; (f) J.-P. Sutter, M. L. Kahn, S. Golhen, L. Ouahab, O. Kahn, Chem. Eur. J., 1998, 4, 571–576. [106] D. Shiomi, S. Honbu, K. Sato, T. Takui, K. Itoh, T. Sugano, Mol. Cryst. Liq. Cryst., 1997, 306, 199–208. [107] T. Sugano, M. Tamura, M. Kinoshita, Y. Sakai, Y. Ohashi, Chem. Phys. Lett., 1992, 200, 235–240. [108] T. Sugano, M. Tamura, T. Goto, R. Kato, M. Kinoshita, Y. Sakai, Y. Ohashi, M. Kurmoo, P. Day, Mol. Cryst. Liq. Cryst., 1993, 232, 61–68. [109] (a) F.L. Pratt, R. Valladares, J. Caulfield, I. Deckers, J. Singleton, A.J. Fisher, W. Hayes, M. Kurmoo, P. Day, T. Sugano, Synth. Met., 1993, 61, 171–175; (b) P.A. Pattenden, R.M. Valladares, F.L. Pratt, S.J. Blundell, A.J. Fisher, W. Hayes, T. Sugano, Synth. Met., 1995, 71, 1823–1824; (c) T. Sugano, M. Kurmoo, P. Day, F.L. Pratt, S.J. Blundell, W. Hayes, M. Ishikawa, M. Kinoshita, Y. Ohashi, Mol. Cryst. Liq. Cryst., 1995, 271, 107–114. [110] T. Sugano, F.L. Pratt, M. Kurmoo, N. Takeda, M. Ishikawa, S.J. Blundell, P.A. Pattenden, R.M. Valladares, W. Hayes, P. Day, Synth. Met., 1995, 71, 1827–1828. [111] T. Sugano, M. Tamura, M. Kinoshita, Synth. Met., 1993, 56, 3305–3310. [112] K. Awaga, T. Inabe, Y. Maruyama, T. Nakamura, M. Matsumoto, Synth. Met., 1993, 56, 3311–3316. [113] J. G. Park, C. Paulsen, J. L. Tholence, P. Rey, B. Barbara, J. Phys.: Condens. Matter, 1996, 8, 745–749. [114] K. Awaga, T. Inabe, Y. Maruyama, Chem. Phys. Lett., 1992, 190, 349–352. [115] (a) K. Awaga, T. Inabe, T. Nakamura, M. Matsumoto, Y. Maruyama, Mol. Cryst. Liq. Cryst., 1993, 232, 69–78. (b) H.-H. Wei, H.-Y. Wong, G.-H. Lee, Y. Wang, J. Chin. Chem. Soc. (Taipei), 1996, 43, 253–259. [116] S. J. Blundell, P. A. Pattenden, R. M. Valladares, F. L. Pratt, T. Sugano, W. Hayes, Solid State Commun., 1994, 92, 569–572. [117] F. M. Romero, R. Ziessel, A. De Cian, J. Fischer, P. Turek, New J. Chem., 1996, 20, 919–924. [118] Y. Pontillon, E. Ressouche, F. Romero, J. Schweizer, R. Ziessel, Physica B, 1997, 234– 236, 788–789.

References

57

[119] F. Lanfranc de Panthou, D. Luneau, J. Laugier, P. Rey, J. Am. Chem. Soc., 1993, 115, 9095–9100. [120] Y. Pei, O. Kahn, M. A. Abersold, L. Ouahab, F. Le Berre, L. Pardi, J. L. Tholence, Adv. Mater., 1994, 6, 681–683. [121] J.-P. Sutter, A. Lang, O. Kahn, C. Paulsen, L. Ouahab, Y. Pei, J. Magn. Magn. Mater., 1997, 171, 147–152. [122] A. Lang, Y. Pei, L. Ouahab, O. Kahn, Adv. Mater., 1996, 8, 60–62. [123] N. Yoshioka, M. Irisawa, Y. Mochizuki, T. Kato, H. Inoue, S. Ohba, Chem. Lett., 1997, 251–252. [124] T. Sugano, M. Kurmoo, P. Day, Synth. Met., 1997, 85, 1729–1730. [125] N. Yoshioka, M. Irisawa, Y. Mochizuki, T. Aoki, H. Inoue, Mol. Cryst. Liq. Cryst., 1997, 306, 403–408. [126] (a) K. Yamaguchi, M. Okumura, J. Maki, T. Noro, H. Namimoto, M. Nakano, T. Fueno, K. Nakasuji, Chem. Phys. Lett., 1992, 190, 353–360. (b) K. Yamaguchi, M. Okumura, M. Nakano, Chem. Phys. Lett., 1992, 191, 237–244. [127] A. Caneschi, F. Ferraro, D. Gatteschi, A. Le Lirzin, M. A. Novak, E. Rentschler, R. Sessoli, Adv. Mater., 1995, 7, 476–478. [128] A. Caneschi, F. Ferraro, D. Gatteschi, A. le Lirzin, E. Rentschler, Inorg. Chim. Acta, 1995, 235, 159–164. [129] (a) M.C. Etter, J. Phys. Chem., 1991, 95, 4601–4610; (b) O. Ermer, A. Eling, J. Chem. Soc., Perkin Trans. 2, 1994, 925–944; (c) E. Fan, C. Vicent, S.J. Geib, A.D. Hamilton, Chem. Mater., 1994, 6, 1113–1117; (d) C.B. Aakeroy, ¨ K.R. Seddon, Chem. Soc. Rev., 1994, 22, 397–407; (e) J.C. MacDonald, G.M. Whitesides, Chem. Rev., 1994, 94, 2383–2420; (f) J. Bernstein, R.E. Davis, L. Shimoni, N.-L. Chang, Angew. Chem., Int, Ed. Engl., 1995, 34, 1555–1573; (g) V.A. Russell, M.D. Ward, Chem. Mater., 1996, 8, 1654–1666. [130] (a) J. Veciana, J. Cirujeda, C. Rovira, E. Molins, J. J. Novoa, J. Phys. I France, 1996, 6, 1967–1986. (b) J. Cirujeda, Ph.D. Thesis, Universitat Ramon Llull, 1997. [131] J. Cirujeda, M. Mas, E. Molins, F. Lanfranc de Panthou, J. Laugier, J.G. Park, C. Paulsen, P. Rey, C. Rovira, J. Veciana, J. Chem. Soc., Chem. Commun., 1995, 709–710. [132] E. Hernandez, ` M. Mas, E. Molins, C. Rovira, J. Veciana, Angew. Chem., Int, Ed. Engl., 1993, 32, 882–884. [133] J.L. Garcia-Munoz, ˜ J. Cirujeda, J. Veciana, S.F.J. Cox, Chem. Phys. Lett. 1998, 293, 160– 166. [134] J. Cirujeda, E. Hernandez-Gasio, ` C. Rovira, J.-L. Stanger, P. Turek, J. Veciana, J. Mater. Chem., 1995, 5, 243–252. [135] J. Cirujeda, O. Jurgens, ¨ J. Vidal-Gancedo, C. Rovira, P. Turek, J. Veciana, Mol. Cryst. Liq. Cryst., 1997, 305, 367–384. [136] O. Jurgens, ¨ J. Cirujeda, M. Mas, I. Mata, A. Cabrero, J. Vidal-Gancedo, C. Rovira, E. Molins, J. Veciana, J. Mater. Chem., 1997, 7, 1723–1730. [137] J. Cirujeda, L. E. Ochando, J. M. Amigo, ´ C. Rovira, J. Rius, J. Veciana, Angew. Chem., Int. Ed. Engl. 1995, 34, 55–57. [138] M.M. Matsushita, A. Izuoka, T. Sugawara, T. Kobayashi, N. Wada, N. Takeda, M. Ishikawa, J. Am. Chem. Soc. 1997, 119, 4369–4379. [139] T. Sugawara, M.M. Matsushita, A. Izuoka, N. Wada, N. Takeda, M. Ishikawa, J. Chem. Soc., Chem. Commun., 1994, 1723–1724. [140] A. Oda, T. Kawakami, S. Takeda, W. Mori, M. M. Matsushita, A. Izuoka, T. Sugawara, K. Yamaguchi, Mol. Cryst. Liq. Cryst., 1997, 306, 151–160. [141] T. Akita, K. Kobayashi, Adv. Mater., 1997, 9, 346–349. [142] M. Tamura, D. Shiomi, Y. Hosokoshi, N. Iwasawa, K. Nozawa, M. Kinoshita, H. Sawa, R. Kato, Mol. Cryst. Liq. Cryst., 1993, 232, 45–52.

58

1 Nitroxide-based Organic Magnets

[143] Y. Hosokoshi, M. Tamura, M. Kinoshita, H. Sawa, R. Kato, Y. Fujiwara, Y. Ueda, J. Mater. Chem., 1994, 4, 1219–1226. [144] S.i. Nakatsuji, M. Saiga, N. Haga, A. Naito, M. Nakagawa, Y. Oda, K. Suzuki, T. Enoki, H. Anzai, Mol. Cryst. Liq. Cryst., 1997, 306, 279–284. [145] S.i. Nakatsuji, M. Saiga, N. Haga, A. Naito, T. Hirayama, M. Nakagawa, Y. Oda, H. Anzai, K. Suzuki, T. Enoki, M. Mito, K. Takeda, New J. Chem., 1998, 22, 275–280. [146] Y. Hosokoshi, M. Tamura, D. Shiomi, N. Iwasawa, K. Nozawa, M. Kinoshita, H. Aruga Katori, T. Goto, Physica B, 1994, 201, 497–499. [147] (a) A.T. Balaban, P.J. Halls, A.R. Katritzky, Chem. Ind., 1968, 651. (b) H.G. Aurich, W. Weiss, Chem. Ber., 1973, 106, 2408; (c) H.G. Aurich, K. Stork, Chem. Ber., 1975, 108, 2764; (d) T.E. Gough, R. Puzic, J. Magn. Reson., 1976, 23, 31. [148] Y. Kusaba, M. Tamura, Y. Hosokoshi, M. Kinoshita, H. Sawa, R. Kato, H. Kobayashi, J. Mater. Chem., 1997, 7, 1377–1382. [149] A. I. Kitaigorodskii, Organic Chemical Crystallography, Consultants Bureau, New York, 1961. [150] (a) K. Awaga, T. Inabe, U. Nagashima, T. Nakamura, M. Matsumoto, Y. Kawabata, Y. Maruyama, Chem. Lett., 1991, 1777–1780. (b) K. Awaga, T. Inabe, Y. Maruyama, T. Nakamura, M. Matsumoto, Chem. Phys. Lett., 1992, 195, 21–24. (c) K. Awaga, A. Yamaguchi, T. Okuno, T. Inabe, T. Nakamura, M. Matsumoto, Y. Maruyama, J. Mater. Chem., 1994, 4, 1377–1385. (d) K. Awaga, A. Yamaguchi, T. Okuno, Mol. Cryst. Liq. Cryst., 1995, 271, 97–105. [151] (a) K. Awaga, T. Okuno, A. Yamaguchi, M. Hasegawa, T. Inabe, Y. Maruyama, N. Wada, Phys. Rev. B: Condens. Matter, 1994, 49, 3975–3981. (b) N. Wada, T. Kobayashi, H. Yano, T. Okuno, A. Yamaguchi, K. Awaga, J. Phys. Soc. Jpn., 1997, 66, 961–964. [152] (a) H. Imai, T. Inabe, T. Otsuka, T. Okuno, K. Awaga, Synth. Met., 1997, 85, 1705–1706. (b) H. Imai, T. Inabe, T. Otsuka, T. Okuno, K. Awaga, Phys. Rev. B: Condens. Matter, 1996, 54, R6838–R6840. [153] P. Wautelet, A. Bieber, P. Turek, J. Le Moigne, J.-J. Andre, ´ Mol. Cryst. Liq. Cryst., 1997, 305, 55–67. [154] (a) D. Shiomi, H.A. Katori, T. Goto, H. Sawa, R. Kato, M. Tamura, M. Kinoshita, Mol. Cryst. Liq. Cryst., 1993, 232, 109–216. (b) D. Shiomi, M. Tamura, H. Sawa, R. Kato, M. Kinoshita, J. Phys. Soc. Jpn., 1993, 62, 289–300. (c) D. Shiomi, M. Tamura, H. Sawa, R. Kato, M. Kinoshita, Synth. Met., 1993, 56, 3279–3284. [155] L. Catala, Ph.D. Thesis, Universite´ Louis Pasteur, Strasbourg, 1999. [156] D. Shiomi, M. Tamura, H. A. Katori, T. Goto, A. Hayashi, Y. Ueda, H. Sawa, R. Kato, M. Kinoshita, J. Mater. Chem., 1994, 4, 915–920. [157] T. Sugano, M. Tamura, M. Kinoshita, Synth. Met., 1993, 56, 3305–3310. [158] H. Oshio, M. Umeno, T. Fukushi, T. Ito, Chem. Lett., 1997, 1065–1066. [159] L. Dulog, J. S. Kim, Angew. Chem., Int. Ed. Engl., 1990, 29, 415–417. [160] A. Izuoka, M. Fukada, T. Sugawara, M. Sakai, S. Bandow, Chem. Lett., 1992, 1627–1630. [161] A. Izuoka, M. Fukada, T. Sugawara, Mol. Cryst. Liq. Cryst., 1993, 232, 103–238. [162] T. Akita, Y. Mazaki, K. Kobayashi, N. Koga, H. Iwamura, J. Org. Chem., 1995, 60, 2092–2098. [163] (a) T. Akita, Y. Mazaki, K. Kobayashi, K. Ushida, Tetrahedron Lett., 1995, 36, 5543–5546. (b) T. Akita, K. Kobayashi, Tetrahedron, 1996, 52, 6893–6902. [164] T. Mitsumori, K. Inoue, N. Koga, H. Iwamura, J. Am. Chem. Soc., 1995, 117, 2467–2478. [165] O. Jurgens, ¨ J. Vidal-Gancedo, C. Rovira, K. Wurst, C. Sporer, B. Bildstein, H. Schottenberger, P. Jaitner, J. Veciana, Inorg. Chem., 1998, 37, 4547–4558. [166] M. Tanaka, K. Matsuda, T. Itoh, H. Iwamura, J. Am. Chem. Soc., 1998, 120, 7168–7173. [167] K. Matsuda, G. Ulrich, H. Iwamura, J. Chem. Soc., Perkin Trans. 2, 1998, 1581–1588.

References

59

[168] K. Inoue, H. Iwamura, Angew. Chem. Int. Ed. Engl., 1995, 34, 927–928. [169] D. Zhang, W. Zhou, D. Zhu, Solid State Commun., 1993, 86, 291–294. [170] E. Hernandez-Gasio, ` M. Mas, E. Molins, C. Rovira, J. Veciana, J. J. Borras-Almenar, ´ E. Coronado, Chem. Mater., 1994, 6, 2398–2411. [171] R. Akabane, M. Tanaka, K. Matsuo, N. Koga, K. Matsuda, H. Iwamura, J. Org. Chem., 1997, 62, 8854–8861. [172] T. Okuno, T. Otsuka, K. Awaga, J. Chem. Soc., Chem. Commun., 1995, 827–828. [173] T. Okuno, T. Otsuka, K. Awaga, Mol. Cryst. Liq. Cryst., 1996, 279, 57–63. [174] T. Otsuka, T. Okuno, K. Awaga, T. Inabe, J. Mater. Chem., 1998, 8, 1157–1163. [175] T. Otsuka, T. Okuno, M. Ohkawa, T. Inabe, K. Awaga, Mol. Cryst. Liq. Cryst., 1997, 306, 285–292. [176] K. Inoue, H. Iwamura, Chem. Phys. Lett., 1993, 207, 551–554. [177] C. Batz, ¨ P. Amann, H.-J. Deiseroth, L. Dulog, Liebigs Ann. Chem., 1994, 739–740. [178] (a) T. Akita, Y. Mazaki, K. Kobayashi, Mol. Cryst. Liq. Cryst., 1996, 279, 39–45. (b) T. Akita, K. Kobayashi, Mol. Cryst. Liq. Cryst., 1997, 306, 257–264. [179] T. Akita, Y. Mazaki, K. Kobayashi, J. Chem. Soc., Chem. Commun., 1995, 1861–1862. [180] A.L. Buchachenko, Dokl. Akad. Nauk SSSR Engl. Ed., 1979, 244, 107–109. [181] A. Izuoka, R. Kumai, T. Sugawara, Adv. Mater., 1995, 7, 672–674. [182] A. Izuoka, M. Fukada, T. Sugawara, M. Sakai, S. Bandow, Chem. Lett., 1992, 1627– 1630. [183] (a) C.A. Hunter, Chem. Soc. Rev., 1994, 23, 101–109. (b) T. Dahl, Acta Chem. Scand. 1994, 48, 95–106. (c) D.B. Amabilino, J.F. Stoddart, D.J. Williams, Chem. Mater. 1994, 6, 1159–1167. [184] A. Izuoka, M. Fukada, R. Kumai, M. Itakura, S. Hikami, T. Sugawara, J. Am. Chem. Soc., 1994, 116, 2609–2610. [185] M. Nishizawa, D. Shiomi, K. Sato, T. Takui, K. Itoh, H. Sakurai, A. Izuoka, T. Sugawara, H. Sawa, R. Kato, Mol. Cryst. Liq. Cryst., 1997, 306, 191–197. [186] D. Shiomi, M. Nishizawa, K. Sato, T. Takui, K. Itoh, H. Sakurai, A. Izuoka, T. Sugawara, J. Phys. Chem. B, 1997, 101, 3342–3348. [187] F. M. Romero, R. Ziessel, M. Drillon, J.-L. Tholence, C. Paulsen, N. Kyritsakas, J. Fisher, Adv. Mater., 1996, 8, 826–829. [188] C. Stroh, F. M. Romero, N. Kyritsakas, L. Catala, P. Turek, R. Ziessel, J. Mater. Chem., 1999, 9, 875–882. [189] V. Laget, C. Hornick, P. Rabu, M. Drillon, P. Turek, R. Ziessel, Adv. Mater., 1998, 10, 1024–1028. [190] P. Turek, P. Wautelet, J. Le Moigne, J.-L. Stanger, J.-J. Andre, ´ A. Bieber, P. Rey, A. de Cian, J. Fischer, Mol. Cryst. Liq. Cryst., 1995, 272, 99–108. [191] (a) J.S. Miller, Adv. Mater., 1992, 4, 298–300; (b) J.S. Miller, Adv. Mater., 1992, 4, 435– 438; (c) A. Rajca, Chem. Rev., 1994, 94, 871–893. (d) Y. Miura in Magnetic Properties of Organic Molecules, (P. M. Lahti, Ed.), Marcel Dekker, New York 1999, pp. 267–284. (e) N. Nishida, T. Kaneko in Magnetic Properties of Organic Molecules, (P. M. Lahti, Ed.), Marcel Dekker, New York 1999, pp. 285–303. [192] (a) J. S. Miller, J. C. Calabrese, D. T. Glatzhofer, A. J. Epstein, J. Appl. Phys., 1988, 63, 2949–2951; (b) J.S. Miller, D. T. Glatzhofer, J. C. Calabrese, A. J. Epstein, J. Chem. Soc., Chem. Commun., 1988, 322–323; (c) J.S. Miller, D.T. Glatzhofer, R. Laversanne, T. B. Brill, M. D. Timken, C. J. O’Connor, J.H. Zhang, J.C. Calabrese, A.J. Epstein, S. Chittipeddi, P. Vaca, Chem. Mater., 1990, 2, 60–69. [193] K. Inoue, N. Koga, H. Iwamura, J. Am. Chem. Soc., 1991, 113, 9803–9810. [194] Y. Miura, Y. Ushitani, M. Matsumoto, K. Inui, Y. Teki, T. Takui, K. Itoh, Mol. Cryst. Liq. Cryst., 1993, 232, 135–242.

60

1 Nitroxide-based Organic Magnets

[195] Y. Miura, Y. Ushitani, K. Inui, Y. Teki, T. Takui, K. Itoh, Macromolecules, 1993, 26, 3698–3701. [196] Y. Miura, Y. Ushitani, Macromolecules, 1993, 26, 7079–7082. [197] Y. Miura, T. Issiki, Y. Ushitani, Y. Teki, K. Itoh, J. Mater. Chem., 1996, 6, 1745–1750. [198] P. Swoboda, R. Saf, K. Hummel, F. Hofer, R. Czaputa, Macromolecules, 1995, 28, 4255– 4259. [199] (a) R. Saf, D. Hengg, K. Hummel, K. Gatterer, H. P. Fritzer, Polymer, 1993, 34, 2680– 2683. (b) R. Saf, K. Hummel, K. Gatterer, H. P. Fritzer, Polymer Bull., 1994, 32, 83–90. [200] R. Saf, P. Swoboda, A. Lechner, K. Hummel, R. Czaputa, Macromol. Chem. Phys., 1996, 197, 1439–1447. [201] H. Nishide, Y. Hozumi, T. Nii, E. Tsuchida, Macromolecules, 1997, 30, 3986–3991. [202] (a) M. Clemente-Leon, ´ C. Mingotaud, B. Agricole, C.J. Gomez-Garc´ ´ ıa, E. Coronado, P. Delhaes, Angew. Chem. Int. Ed. Engl., 1997, 36, 1114–1116; (b) N. Yoshioka, Hyomen Kagaku, 1998, 19, 244–248. [203] J. Caro, J. Fraxedas, O. Jurgens, ¨ J. Santiso, C. Rovira, J. Veciana, A. Figueras, Adv. Mater., 1998, 10, 608–610. [204] During the writing and proof stages of this Chapter, many articles concerning nitroxidebased organic magnets have appeared, among them: (a) F. Iwasaki, J.H. Yoshikawa, H. Yamamoto, K. Takada, E. Kan-Nari, M. Yasui, T. Ishida, T. Nogami, Acta Cryst. B., 1999, 55, 1057–1067; (b) Y. Pontillon, A. Caneschi, D. Gatteschi, A. Grand, E. Ressouche, R. Sessoli, J. Schweizer, Chem. Eur. J., 1999, 5, 3616–3624; (c) S. Hayami, K. Inoue, Chem. Lett., 1999, 545–546; (d) J.R. Ferrer, P.M. Lahti, C. George, G. Antorrena, F. Palacio, Chem. Mater., 1999, 11, 2205–2210; (e) B. Gillon, M.A. Aebersold, O. Kahn, L. Pardi, B. Delley, Chem. Phys., 1999, 250, 23–34; (f) Y. Pontillon, T. Akita, A. Grand, K. Kobayashi, E. Lelievre-Berna, J. Pecaut, ´ E. Ressouche, J. Schweizer, J. Am. Chem. Soc., 1999, 121, 10126–10133; (g) M. Mitani, D. Yamaki, Y. Yoshioka, K. Yamaguchi, J. Chem. Phys., 1999, 111, 2283–2294; (h) C. Stroh, F.M. Romero, N. Kyritsakas, L. Catala, P. Turek, R. Ziessel, J. Mater. Chem., 1999, 9, 875–882; (i) T. Sugano, M. Kurmoo, H. Uekusa, Y. Ohashi, P. Day, J. Solid State Chem., 1999, 145, 427–442; (j) O. Felix, ´ M.W. Hosseini, A. De Cian, J. Fischer, L. Catala, P. Turek, Tetrahedron Lett., 1999, 40, 2943–2946; (k) H. Oka, T. Tamura, Y. Miura, Y. Teki, Polym. J. (Tokyo), 1999, 31, 979–982; (l) A.F.C. Alcantara, ˆ H.F. Dos Santos, W.B. De Almeida, M.G.F. Vaz, L.M.M. Pinheiro, H.O. Stumpf, Struct. Chem., 1999, 10, 367–374; (m) L. Catala, P. Turek, J. Le Moigne, A. De Cian, N. Kyritsakas, Tetrahedron Lett., 2000, 41, 1015–1018; (n) F.M. Romero, R. Ziessel, M. Bonnet, Y. Pontillon, E. Ressouche, J. Schweizer, B. Delley, A. Grand, C. Paulsen, J. Am. Chem. Soc., 2000, 122, 1298–1309; (o) N.L. Frank, R. Clerac, ´ J.-P. Sutter, N. Daro, O. Kahn, C. Coulon, M.T. Green, S. Golhen, L. Ouahab, J. Am. Chem. Soc., 2000, 122, 2053–2061; (p) K. Matsuda, M. Irie, J. Am. Chem. Soc., 2000, 122, 7195–7201; (q) H. Sakurai, A. Izuoka, T. Sugawara, J. Am. Chem. Soc., 2000, 122, 9723–9734; (r) M. Minguet, D.B. Amabilino, J. Cirujeda, K. Wurst, I. Mata, E. Molins, J.J. Novoa, J. Veciana, Chem. Eur. J., 2000, 6, 2350–2361; (s) M. Nishizawa, D. Shiomi, K. Sato, T. Takui, K. Itoh, H. Sawa, R. Kato, H. Sakurai, A. Izuoka, T. Sugawara, J. Phys. Chem. B, 2000, 104, 503–509; (t) L. Ding, D.Q. Zhang, B. Zhang, D.B. Zhu, Chin. Chem. Lett., 2000, 11, 749–752; (u) P. Gerbier, C. Guerin, J. Le Bideau, K. Valle, Chem. Mater., 2000, 12, 264–267; (v) R. Feyerherm, S. Welzel, J.P. Sutter, O. Kahn, Physica B (Amsterdam), 2000, 276–278, 720–721; (w) Metal-Organic and Organic Molecular Magnets (Eds. P. Day, A.E. Underhill), RSC, London, 2000; (x) S. Pillet, M. Souhassou, Y. Pontillon, A. Caneschi, D. Gatteschi, C. Lecomte, New J. Chem., 2001, 25, 131–143, as well as several articles in Mol. Cryst. Liq. Cryst., 1999, 334 (Proceedings of VIth ISMM).

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

2

Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals Hiizu Iwamura and Katsuya Inoue†

2.1

Introduction

Several different approaches are currently used in order to design and synthesize free radical-based materials exhibiting spontaneous magnetization. Obviously there are three conditions necessary for its appearance at finite, preferably high, temperature: (i) assemblage of unpaired electrons in high concentration; (ii) operation of strong exchange interactions aligning the spins, and (iii) formation of magnetic domains in which all the spins are ordered in two- or three-dimensional network in mesoscopic scale. To satisfy these conditions, a purely organic approach has a number of drawbacks. There is a density limitation; a number of non-magnetic atoms are necessary to stabilize the unpaired electrons kinetically and/or thermodynamically. A network structure may not be a problem if an appropriate crystal design is made. However, since the exchange coupling between neighboring molecules through van der Waals force, hydrogen bonds, hydrophobic interaction, etc. is not necessarily strong, it is difficult to expect strong intermolecular magnetic coupling. It is therefore a strong design strategy to make extended structures by assembling free radicals by means of magnetic metal ions. Such polymeric complexes satisfy the above three conditions [1–3]. In this chapter are discussed magnetic materials made up of coordination complexes of magnetic metal ions with aminoxyl radicals.

2.2

Aminoxyl Radicals

To make free radicals as ligands for a magnetic metal ion, it is necessary for the radicals to have enough coordinating ability. Such radicals are classified into two groups: those having a basic radical center or amine and phosphine donors having a radical center as a substituent. In the former the interaction with the meat ions is direct, while it is indirect, through-bonds, and the mechanism may be called “extended superexchange” in the latter. Aminoxyl radicals alias “nitroxide radicals” are representatives of the former examples. o-Semiquinone radicals serve as bidentate ligands but are less stable under ambient conditions and have not been used to prepare coordination complexes having extended structures.

62

2.2.1

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Electronic Structure and Basicity

Aminoxyl radicals are isoelectronic with ketyl radicals. Whereas the ketyl of di-tbutyl ketone is red and stable in deoxygenated solution, di-t-butylaminoxyl is a red oil stable under ambient conditions. Aminoxyl radicals happened to be the first organic ferromagnets ever discovered in the early nineties. The above three conditions for molecular-based magnets appear to be realized by special arrangement of the radical molecules in crystals [4, 5], although TC is quite low (<1.4 K). They are discussed in Chapter 5 by Veciana. The unpaired electron resides on the antibonding π ∗ orbital in aminoxyl radicals. They have resonance structures as depicted by Scheme 1 in which contribution of the two structures are roughly the same. The fluid solution EPR spectra of aminoxyl radicals therefore consist of a triplet due to the interaction with the 14 N nucleus. The g tensor is only slightly anisotropic. Most rigorous demonstrated of the distribution of the spin density has been performed by neutron diffraction studies on 2,2,6,6tetramethylpiperidin-1-oxyl (TEMPO) and some other aminoxyl radicals [6].

Scheme 1

Reactivity of aminoxyl radicals is kinetically attenuated by bulky substituents, e.g., one or more t-butyl group(s). The carbon atoms β to the aminoxyl nitrogen must be tertiary; otherwise (hydrogen is abstracted to form disproportionation products: a nitrone and a hydroxyamine (Scheme 2). The exception to this rule is an aminoxyl

Scheme 2

2.2 Aminoxyl Radicals

63

unit next to bridgehead carbons. In this case, the radicals are stable as formation of the corresponding nitrones would violate Bredt rule. Aminoxyl radicals show the weak donating ability [7–12]. Calorimetric and spectrometric studies show that the enthalpy (−H ) of formation of the 1:1 mixed-ligand complexes of [Cu(hfac)2 ] (in cyclohexane) [7] and [VO(hfac)2 ] (in CCl4 ) [11] with TEMPO is 11.7 and 10.7 kcal mol−1 , respectively. The formation constants of the 1:1 complexes of Cu(hfac)2 with di-tert-butylaminoxyl (DTBN) [8] and TEMPO [7] in CCl4 are 1.6 × 103 and 4.2 × 103 M−1 , respectively. These values are two orders of magnitude smaller than that with pyridine (7.0 × 105 M−1 ) [8, 9]. Thus relatively strong Lewis-acidic metal centers are necessary to make stable complexes. Metal anhydrous halides, perchlorates, hexafluoroacetylacetonates (hfac), and halogenated carboxylates satisfy these conditions. The Mn(II)-O bond lengths in complexes [Mn(hfac)2 .(aminoxyl)2 ] are reasonably short: 2.150 and 2.127 Å for 2,2,5,5-tetramethylpyrrolidin-1-oxyl (PROXYL) and TEMPO, respectively [13]. However, since the electron pairing energy estimated from an energy gap of the spin parallel and antiparallel states amounts to no more than 1000 K ≈ 2 kcal mol−1 as will be amply seen in later sections of this chapter, the contribution of a pair of an unpaired electron of the aminoxyl group and a 3d electron on the magnetic metal ion to the coordination bond is regarded to be fractional and insignificant.

2.2.2

Aminoxyl Radicals with Another Basic Center

Any aminoxyl radical may serve as a ligand for a Lewis-acidic metal ion, but it cannot necessarily form an extended structure of solid state interest. Another basic center including additional aminoxyl group is necessary for the ligand to form chain polymer and/or network structures. Earlier work used spin-labeled ligands for spin-labeling biological molecules to determine the structures and dynamics of substrates and biological active centers. These studies are excellently reviewed [14, 15]. 4-Hydroxy-2,2,6,6-tetramethylpiperidin-N -oxyl (TEMPOL) [16–18]. and N -t-butylN -(4-pyridyl)aminoxyl ( pNOPy) [19] are some of the first spin-labeled ligands. Ullman’s “nitronyl nitroxides” (4,4,5,5-tetramethyl-imidazolin-1-oxyl 3-oxides, NIT) have a nitrone chromophore that is in resonance with the aminoxyl; the radical serves as a symmetric bis-monodentate ligand [20]. Imino nitroxides (Im) [20] and 3-imidazolin-1-oxyl [21] are other favorite ligands. A number of nitronyl nitroxides having chelating amine bases as substituents have been explored [22, 23]. They include 2-(2-pyridyl)-, 6-(2,2 -bipyridinyl)-, and 2,2 -bipyridine-6,6 -diyl- derivatives of NIT [24, 25], and 2,6-bis(4,4,5,5-tetramethyl-imidazolin-1-oxyl-2-yl)pyridine (= Bisimpy) [26]. In these ligands, the pyridyl nitrogen having stronger donor property (see Section 2.2.1) coordinates readily to metal ions, thereby creating favorable conditions for the coordination of the oxygen atom of the nitronyl nitroxide moiety.

64

2.2.3

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

High-spin Di- and Poly(aminoxyl) Radicals

Several aliphatic and alicyclic bis-aminoxyl radicals have been reported. They are favorite molecules to study through-bond exchange and distance-dependent dipolar couplings. Many of them may be found in a book and a review article [27, 28]. The magnitude of the exchange coupling in such systems is quite small and the sign is antiferromagnetic. They can form chain polymers with coordinatively doubly unsaturated metal ions but the magnetic coupling is isolated within the coordination units as in Cu(hfac)2 complex with bis-PROXYL [22, 23]. 1,3,5,7-Tetramethyl-2,6diazaadamantane-2,6-dioxyl is an exception in that it has D2d symmetry and the two π ∗ orbitals at the aminoxyls are orthogonal to each other: the ground state is triplet with a singlet-triplet energy gap of ca. 10 K (1 K = 0.695 cm−1 = 1.987 cal mol−1 ) [4]. These bis-aminoxyl radicals seem not to have been used as ligands for magnetic metal ions. As in benzyl radicals, positive spin density is distributed over the ortho and para carbons of the phenyl ring in N -phenyl-N -tert-butylaminoxyl. The spins are more localized in the latter; whereas about 50% the spin density reside on the ring in benzyl radicals, it is only 20–30% in the phenylaminoxyls. Still there is a tendency of free-radical reactivity at the p-position. Electron donating substituents stabilize the phenylaminoxyls thermodynamically.

2.3 Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals

65

When two unpaired electrons are placed in proximity and allowed to interact as in diradicals and radical pairs, the Coulombic repulsion between the two electron lifts the zeroth-order degeneracy and gives rise to singlet and triplet states of different total energy for these chemical entities. It is singlet states which are more often stabilized relative to triplets by the overlap of the two singly occupied orbitals. Thus the unpaired electrons of the two aminoxyl groups couple antiferromagnetically through the p-phenylene ring to give a p-benzoquinonimine N ,N -dioxide structure as in p-benzoquinodimethanes [3, 29, 30]. However, when sterically restricted from assuming a planar structure, they become uncorrelated diradical [31]. In m-quinodimethanes, two unpaired electrons reside one each on the two degenerate non-bonding π-molecular orbitals [3, 29, 30]. Therefore m-phenylenebis(N tert-butylaminoxyl)s have triplet ground states. Various bis-aminoxyl radicals with triplet ground states and tris-aminoxyl radicals with quartet ground states have been designed on the basis of the m-phenylene topology and prepared by oxidation of the corresponding hydroxyamines. Here again through-bond interaction becomes weaker as the number of π-bonds between the aminoxyl groups increases. Highspin oligo-aminoxyl radicals that have been used as ligands for magnetic metal ions are summarized in Table 1 [32–37].

2.3

2.3.1

Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals Indirect Coupling (Extended Superexchange)

In spin-labeling studies of metal-containing biomolecules such as heme proteins, Vitamin B12, various metal-enzymes, etc., with aminoxyl radicals, distances of 10– 20 Å between the two paramagnetic centers are estimated by measuring the EPR relaxation time perturbed by the dipolar interaction [14, 15]. These studies are not discussed here as they are performed in fluid and solid solutions and little information is available on their solid states. The exchange coupling through many σ bonds is limited in magnitude. Coupling between the aminoxyl radical and copper(II) through pyridine base PyImTEMPO is determined to be J/kB = 6.07 mK [10]. In the [Cu(hfac)2 .TEMPOL]n polymer chain [19], a pair of S = 1/2 centers couple ferromagnetically (2J/kB = 18 ± 7 K) in the repeating unit Cu(II)· · ·O–N<, but the inter-unit coupling amounts only to 2J /kB = −78 ± 2 mK. Coordinatively unsaturated Cr(III), Mn(II) and Cu(II) ions and complexes form with N -tert-butyl-N -(3- and 4-pyridyl)aminoxyls (mNOPy and pNOPy, respectively) discrete mixed-ligand complexes in which the pyridyl nitrogen atoms are coordinated to the metal ions [38]. A number of them are obtained as single crystals amenable to X-ray crystal structure analysis. ORTEP drawings for two representative complexes are given in Fig. 1. The metal ions are typically in the center of elongated and/or distorted octahedra and the two pyridyl nitrogen atoms are attached to the metal

66

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Table 1. The magnitude of the intra-ligand magnetic coupling (Jintra ) between the radical centers in bis- and tris-aminoxyl radicals. Aminoxyl radical

Jintra /kB , K

Sample morphology

Ref.

BNO BNOCl BNOBr ThBNIT BNOP TNOPB TNOPB TNOB TNO TNOP

>300 >300 >300 +80 ± 4 ≥300 +6.8 ± 0.1 +5.3 ± 0.1 ∼300, +67 ± 5 ≥300 +240 ± 20

Orange crystals Orange crystals Orange crystals Black block crystals Red-purple crystals Red crystals Isolated in Tween 40 Isolated in PVC Reddish yellow crystals Orange crystals

32, 33 57, 80, 82 57, 80, 82 35 37 34 34 36 34 33

ion in the cis or trans configuration. An examination of the bond lengths around the metal ions reveals that the N(1)–Mn-N(1 ) bonds are on the elongation axes and correspond to the dz orbital of Mn(II). The elongation axes lie along O(1)–Cu–O(1 ) bonds in three copper complexes, indicating that the lobes of the magnetic orbital are directed toward the O(2)s of the hfac units and N(1)s of the pyridine units. The magnetic susceptibility data of these metal-radical complexes are obtained in the temperature range 2–300 K at a constant field of 100–800 mT on a SQUID susceptometer/magnetometer. Typical results of the temperature-dependence of the molar paramagnetic susceptibility χmol and their analysis are exemplified by [M(hfac)2 ( pNOPy and mNOPy)2 ] in Fig. 2. A χmol T value of 1.29 emu K mol−1 obtained at 300 K for [Cu(hfac)2 ( pNOPy)2 ] is close to a theoretical 1.13 emu K mol−1 calculated for three isolated S = 1/2 spins {χmol T = 0.125 × g × 2[3S(S + 1)] = 0.125 × 2 × 2(3/2(1/2 + 1))}. As the temperature is decreased, χmol T values (open circles in Fig. 2a) increase gradually, reach a maximum at 36 K, and rapidly decrease

2.3 Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals

67

Fig. 1. ORTEP drawings of [Cu(hfac)2 .( pNOPy)2 ] and [Mn(hfac)2 .( pNOPy)2 ].

Fig. 2. Plots of χmol T against T for crystalline samples of: (a) [Cu(hfac)2 .( pNOPy)2 ] (◦) and [Cu(hfac)2 .(mNOPy)2 ] (•) and (b) [Mn(hfac)2 .( pNOPy)2 ] (◦) and [Mn(hfac)2 .(mNOPy)2 ] (•). Solid curves are the best theoretical fit.

below 10 K. A maximum χmol T value of 1.69 emu K mol−1 is slightly smaller than a theoretical 1.87 emu K mol−1 calculated for S = 3/2. A linear three-spin model suggested by the X-ray crystal and molecular structure was adopted for analyzing the temperature-dependence of the observed χmol T values more quantitatively. Its spin Hamiltonian is written as Eq. (1): H = −2J (S1 S M + S M S2 )

(1)

68

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

A Boltzmann distribution among the spin states given by the Eigenvalues of Eq. (1) for three spins with S1 = S2 = SM = 1/2 was assumed and the theoretical equation thus derived was fitted to the experimental data by means of a least-squares method. The Weiss constant, θ , was used to represent by a mean-field theory a weak intermolecular interaction expected from the short contact of 5.85 Å in the crystal packing of the complex. The best-fit parameters are: J/kB = 60.4 ± 3.3 K, g = 2.048 ± 0.0091, and θ = −3.58 ± 0.09 K, where all symbols have their usual meaning. The fitted theoretical curve is presented by a solid curve in Fig. 2. A χmol T value for [Cu(hfac)2 (mNOPy)2 ] was ca. 0.6 emu K mol−1 at 300 K and gradually decreased to 0.43 at 10 K with decreasing temperature (filled circles in Fig. 2a). The observed values are close to a theoretical 0.37 emu K mol−1 for S = 1/2. Cancellation of the two spins of the aminoxyl radicals by a strong antiferromagnetic interaction between the neighboring complexes is suggested by the X-ray crystal structure of [Cu(hfac)2 (mNOPy)2 ]. It appears from plots of χmol T against T that the copper complex has no aminoxyl radical units. In the case of pairs of 1:1 mixed-ligand complexes of (tetraarylporphyrinato)chromium(III) chloride with NOPy, the complementarity between the mNOPy and pNOPy is obvious (Table 2). The results of the other magnetic measurements and analyses are summarized in Table 2. The observed exchange interaction of the magnetic metal ions with N -tertbutylaminoxyl radicals (Table 2) through the 4-pyridyl ring clearly demonstrates that its sign is determined by the kind of the metal ions: positive for copper(II) and negative for manganese(II) and chromium(III). Nature of the magnetic orbitals of the metal ions appears to play an important role in governing the sign of the coupling. In 4NOPy the spin density at the pyridyl nitrogen atom due to the presence of the aminoxyl radical at the 4-position is estimated to be 0.09 [19]. Whereas there are π-type 3d magnetic orbitals in Cr(III) and Mn(II) that can overlap with the 2pπ orbital at the pyridyl nitrogen, the magnetic orbital is dx 2 −y 2 and orthogonal to this Table 2. The exchange coupling parameters J/kB (K) in M(NOPy) [38, 54]. M

Mn(II) S = 5/2 Cu(II) S = 1/2 Cr(III) S = 3/2

−12.4 K in [Mn( pNOPy)2 (hfac)2 ] −10.2 K in [Mn( pNOPy)2 (acac)2 ] −11.3 K in [Mn(pNOPy)(hfac)2 ]2 60.4 K in [Cu( pNOPy)2 (hfac)2 ] 58.6 K in [Cu(pNOPy)(hfac)2 ]2 4.3 K in [Cu(4NOIm)2(hfac)2 ] −77 K in [Cr(TPP)( pNOPy)Cl] −86 K in [Cr(TAP)( pNOPy)Cl]

? in [Mn(mNOPy)2 (hfac)2 ] <−160 K in [Cu(mNOPy)2 (hfac)2 ] −2.7 K in [Cu(mNOIm)2 (hfac)2 ] 12.3 K in [Cr(TPP)(mNOPy)Cl] 16.0 K in [Cr(TAP)(mNOPy)Cl]

Plus and minus signs correspond to ↑ and ↓ spins, respectively, at M. TPP and TAP stand for tetraphenylporphyrinate and tetraanisylporphyrinato ligands, respectively.

2.3 Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals

69

Fig. 3. (a) Antiferro- and ferromagnetic coupling for the π - and σ -type magnetic orbitals, respectively, at M in the complex pNOPy-M. (b) Antiferro- and ferromagnetic interactions in directly coordinated copper(II)-aminoxyl pairs.

nitrogen orbital in the Cu(II) complexes. As in pQDM and pBNO in Scheme 3, the unpaired electrons should couple antiferromagnetically when the two orbitals concerned overlap at the pyridyl nitrogen. Hund’s rule would predict the coupling to be ferromagnetic when they are orthogonal in the Cu(II) complexes (Fig. 3a). The magnitude of the exchange coupling in the manganese(II) complexes [Mn(hfac)2 ( pNOPy)2 ] may be compared with those having aminoxyl oxygens attached directly to Mn(II) in trans configuration (see Section 2.3.2). The former value is only ca. 10% of the latter, suggesting the reduced π-spin density at the pyridyl nitrogen. In the case of the ferromagnetic coupling in the copper(II) complexes, the magnitude is in the range 30–100 K for the direct coupling in the NO axial configuration and ca. 60 K on an average through the pyridine ring. If the analogy of the manganese complexes is applicable, the value for the NO• -Cu direct coupling should have been much greater. It is highly likely that the orthogonality of the magnetic orbitals may not be ideal when directly attached to copper(II). The discussion should be taken only as qualitative as it is known that the electron is withdrawn to the metal ions when coordinated with pyridyl nitrogen of NOPy as evidenced by the decrease of aminoxyl nitrogen hyperfine coupling constant from 10.5 G to 9.24 G when coordinated with diamagnetic ZnBr2 [19].

Scheme 3

70

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Limited data on the sign of the coupling through the 3-pyridyl ring (Table 2) are in line with the idea that they can be opposite to those in pNOPy. The sign of the polarization of the π-electrons due to the aminoxyl radical alternates on the pyridine ring and is therefore out of phase by one atom as in isomeric QDM and BNO in Scheme 3. Bis(hexafluoroacetylacetonato)copper(I) forms with 2-(4pyridyl) derivative of nitronyl nitroxide a complex that shows an EPR spectrum characteristic of a triplet species [24b]. From the temperature-dependence of the area of the EPR spectrum, the ground state is concluded to be a singlet with 2J/kB = −450 K. Since NIT carries a positive spin density at the two imidazoline nitrogens and a negative spin density at 2-carbon atom, the result on the 2-(4-pyridyl) derivative is in line with the m-phenylene topology.

2.3.2

Complexes with Direct Metal-Aminoxyl Coordination

Synthesis and analysis of complexes having metal-aminoxyl coordination started in the early seventies as spin labels for metal complexes. In principle, when the overlap between the magnetic orbitals is large, a normal covalent bond will be formed and the spins are paired. In practice, however, the bond is only partial as judged from the observed pairing energy of the electrons. On the other hand, when the overlap is small or zero, direct exchange will be operative, giving either weak antiferro- or ferromagnetic coupling. The coupling modes are classified in terms of the geometry of the two magnetic orbitals concerned. There are quite a few metal complexes with aminoxyls reported. Their structures are classified on the first approximation into two limiting structures: the aminoxyl oxygen occupying an axial or an equatorial coordination site. In both limits the magnetic orbital on copper(II) can be loosely described as x y, lying in the equatorial plane of the pyramid with the lobes pointing toward the ligands. When the aminoxyl ligand occupies an axial position, its π ∗ magnetic orbital is essentially orthogonal to the magnetic orbital on copper(II) and the coupling is expected to be ferromagnetic. Indeed, this has been found to be the case of a number of complexes in which magnetic susceptibility data showed the existence of a ferromagnetic coupling in the range 30–100 K [39-43]. On the other hand, when the aminoxyl oxygen atom is in the equatorial (or basal) position in the copper (II) complexes, the exchange interaction tends to be antiferromagnetic (Fig. 3b) [39–41]. A 1:1 complex of Cu(hfac)2 with TEMPO has the coordination geometry intermediate between square pyramidal and trigonal bipyramidal, and best described as distorted square pyramidal. The aminoxyl oxygen is in the basal position and the interaction is antiferromagnetic [11]. Dinuclear trichloroacetates of copper(II) are formed when coordinated with TEMPO and PROXYL under basic conditions. The coordination geometry is trigonal bipyramid and magnetically diamagnetic [12, 44, 45]. When the aminoxyl oxygen atom is in the axial position in the copper(II) complexes, the exchange interaction has a tendency to become ferromagnetic [14–18]: J/kB = 13.8 K for intermolecular apical approach, and J/kB = 18.7 ± 7 K and g = 2.08 for an infinite chain complex of Cu(II) with TEMPOL. An EPR spectrum characteristic of a triplet species is obtained for a linear chain made up of Cu(hfac)2

2.3 Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals

71

with TEMPOL, suggesting that the magnetic properties are better represented by a localized picture of one copper(II) and one aminoxyl group. Magnetic susceptibility measurements revealed the weaker coupling to be 7.8 mK [19]. A reaction of [Cu(hfac)2 ] with 2-phenyl derivative of nitronyl nitroxide results in simultaneous crystallization of 1:1 and 1:2 mixed-ligand complexes that were separated mechanically. Only one oxygen atom of the nitronyl nitroxide ligand is coordinated with copper(II) in both complexes. In the 1:1 5-coordinated complex, this oxygen atom lies in the equatorial plane of the coordination polyhedron. Since this configuration leads to antiferromagnetic interaction, the complex is diamagnetic. In the 1:2 hexacoordinated complex, the aminoxyl oxygen atoms occupy apical positions. Analysis of the temperature-dependence of magnetic susceptibility reveals that it has S = 3/2 ground state [46]. In the pentacoordinated mononuclear Cu(hfac)2 complex with 2methyl derivative of nitronyl nitroxide, the NO ligand occupies an apical position and a record high ferromagnetic coupling of 94 K is observed [46]. There is a third mode of complexation: η2 -edge coordination to Cu(II). The interaction is antiferromagnetic and the S–T gap is estimated to be 7200 K [47]. Unusually large magnitude of the antiferromagnetic coupling is justified by a theoretical study [48]. In the case of high spin d5 manganese(II), at least one of five magnetic orbitals must have the correct symmetry for overlapping with those of the aminoxyls. As a result, the interaction is antiferromagnetic and a ground S = 3/2 (= 5/2 − 1/2 − 1/2) state will result in complexes of the general formulas [Mn(hfac)2 (aminoxyl)2 ]. For TEMPO and PROXYL ligands, the J/kB values are found to be −227 and −302 K, respectively [13, 19, 39–41]. 3-Imidazoline nitroxides and Ullman’s nitronyl nitroxides have two basic centers and therefore serve as good bridging ligands for extended structures with coordinatively doubly unsaturated metal ions. These will be discussed in Section 6. At the same time, since Cu(II) has the ability to form complexes with coordination numbers of both 5 and 6, trinuclear complexes with a 3:2 stoichiometry [3Cu(hfac)2 · 2aminoxyl] are obtained [49, 50]. The magnetic coupling in these complexes is ferromagnetic except one with 2-ethyl derivative of nitronyl nitroxide [50]. It is generally the case that the exchange coupling of Mn(II), Ni(II), and Co(II) with the ligand nitronyl nitroxide is antiferromagnetic in character. The magnitude of the exchange coupling in the manganese(II) complexes [Mn(hfac)2 (aminoxyl)2 ] may be compared with those having aminoxyl groups attached to Mn(II) through the pyridine ring in trans configuration (see Section 2.3.1). The latter value is only ca. 10% of the former, suggesting the reduced p-spin density at the pyridyl nitrogen. In the case of the ferromagnetic coupling in the copper(II) complexes, the magnitude is in the range 30–100 K for the direct coupling in the NO axial configuration and ca. 60 K on an average through the pyridine ring. If the analogy of the manganese complexes is applicable, the value for the direct NO• -Cu coupling should have been much greater. It is highly likely that the orthogonality of the magnetic orbitals may not be ideal when directly attached to copper(II). The discussion should be taken only as qualitative since it is known that, upon coordination with ZnBr2 , the hyperfine coupling constant (aN ) of the aminoxyl nitrogen decreases from 10.52 to 9.24 gauss, while aN due to pyridyl nitrogen increases from

72

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

1.23 to 1.60 gauss. These correspond to the decrease in the spin density from 0.21 to 0.13 at the aminoxyl nitrogen and the increase from 0.091 to 0.105 at the pyridyl nitrogen [19].

2.3.3

Cyclic Complexes

There are cyclic oligomers in-between zero-dimensional and one-dimensional extended structures. The cyclic complexes are obtained during the course of attempts at synthesizing infinite chain complexes. Since connectivity is just as in the infinite chains and the number of spins is finite, the systems are amenable to rigorous solution of the spin Hamiltonian to give the sign and magnitude of the exchange coupling concerned. These values serve as good measures for designing and analyzing new extended systems containing two or more kinds of exchange coupling parameters. When a 2-phenyl derivative (NITPh) of the nitronyl nitroxide is used as a bridging ligand for Mn(II)(hfac)2 , a 1:1 cyclic hexanuclear manganese(II) complex is formed. The ring is 36-membered and each Mn(II) ion is hexacoordinated with two aminoxyl oxygens from two different NITPh molecules in cis configuration. Magnetization measurements at 6 K agreed nicely with a Brillouin curve of S = 12 [51]. Two cyclic dimer complexes of bis(hexafluoroacetylacetonato)manganese(II) with 5-tert-butyl-1,3-phenylenebis(N -tert-butylaminoxyl) (BNOt -Bu )[52, 53] and 4(N -tert-butyl-N -oxyamino)pyridine ( pNOPy) [54] have been analyzed to obtain two sets of two intramolecular exchange coupling parameters J1 and J2 (Scheme 4). The crystal structure (Fig. 4) of the first complex [Mn(hfac)2 BNOt -Bu ]2 reveals itself to be a cyclic analog of 1D complex polymers made of similar components without a ring tert-butyl group or with halogen substituents in its place (see Section 2.6)

Fig. 4. ORTEP drawing of [Mn(hfac)2 pNOPy]2 .

2.3 Magnetic Interaction between Transition Metal Ions and Aminoxyl Radicals

73

Fig. 5. Temperature-dependence of χmol T for the cyclic dimer complex [Mn(hfac)2 pNOPy]2 .

[55–58]. The molecule consists of two Mn(II) ions (S = 5/2) and four aminoxyl radicals (S = 1/2) arranged in a cyclic array as shown in Scheme 4. The molar paramagnetic susceptibility χmol of the complex was investigated in the temperature range 2–300 K. The χmol T values exhibit a continuous increase from room temperature down to ca 50 K, reaching a plateau at 5.71 K cm3 mol−1 (Fig. 5). This value agrees well with the theoretical spin-only value for a S = 3 spin ground state, which can be explained from a ferromagnetic (F) interaction between two S = 3/2 pseudospins (5/2 − 2 × 1/2), obtained by assuming a predominant antiferromagnetic (AF) coupling (J2 ) between the manganese(II) ions and the two coordinated radicals. This high-spin ground state is stabilized with respect to upper energy levels by a few hundred wave numbers, as deduced from the decrease of χmol T upon increasing temperature. A decrease of χmol T at lowest temperatures may be analyzed in terms of intermolecular AF coupling.

Scheme 4

Plots of χmol T against T have been analyzed in two ways: one rigorous solution of the spin Hamiltonian for the six-spin system [59–61] and the other on a model of

74

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

coupling of a cyclic array of two effective spins S = 3/2. The first method together with the molecular field contribution taken into account by Curie–Weiss law gives J1 /kB = +1005 ± 11 K, J2 /kB = −539 K ±19 K, and Weiss constant λ = −0.10 ± 0.01 K, with g = 1.953 [62], gives a very good description of the experimental data over the whole temperature range. The second method gave J/kB = +23 ± 5 K for repeating S = 3/2 units and the Lande factor g = 1.955. Simulation of the observed data is quite good except that the high-temperature data are poorly reproduced by the second model. This is due to the lack of precision in the approximate method in locating correctly the higher energy levels where spins are more populated at higher temperature (see inset of Fig. 5). The binuclear complex [Mn(hfac)2 pNOPy]2 obtained by reaction of bis(hexafluoroacetylacetonato)manganese(II) with radical pNOPy in a 1:1 ratio has been analyzed similarly to give: J1 = −22.6 K, J2 = −372 K and g = 1.980 [52]. This and the foregoing J2 values fall within typical values of the direct coordination between Mn(II) and aminoxyl oxygen discussed in Section 2.3.2. The J1 value of −22.6 K corresponds to the typical value for indirect Mn–NO• coupling through a pyridine ring (see Section 2.3.1). Both 3- and 4-(N -tert-butyl-N -oxylamino)imidazoles also form 1:1 cyclic dinuclear complexes with Mn(hfac)2 [63].

2.4

Design Strategy for Various Crystal Structures of Given Dimensions from Metal Complexes with Poly(aminoxyl) Radicals as Bridging Ligands

As we have already seen, a set of two conditions make the necessary and sufficient conditions for realizing metal-radical complexes having interesting magnetic properties. – An aminoxyl radical must have at least an additional basic center including another aminoxyl unit. Any aminoxyl group may form a metal complex having this aminoxyl as a ligand, but there is no chance of forming extended structure without additional coordinating site. – The additional basic center must have certain amount of the spin polarization, because of the first aminoxyl radical center. The two centers must communicate with each other magnetically. These are reasons why Ullman’s “nitronyl nitroxides” and π -conjugated high-spin oligo-aminoxyl radicals are of importance for the design. When an organic free radical carries two basic centers as Ullman’s “nitronyl nitroxides”, extended structures that are often chain polymers or macrocycles are formed with coordinatively doubly unsaturated metal ions. This is schematically shown in Fig. 6a. Depending on the nature of the additional inter-chain interactions, the chain polymers become antiferromagnets, metamagnets, or ferri/ferromagnets. Extension of this design strategy to bis- and tris-aminoxyl radicals having triplet and quartet ground states, respectively,

2.4 Design Strategy for Various Crystyl Structures

75

Fig. 6. Schematic representations of the formation of: (a) 1D chains or macrocycles from bis(monodentate) monoradicals and (b) triplet diradicals. Formation of (c) ladder polymers from triplet bis{bis(monodentate)} diradicals, (d) 2D network sheets from quartet trisaminoxyls, and (e) 2D networks and/or (e ) 3D crossed parallels from quartet tris-aminoxyls, all with coordinatively doubly unsaturated 3d transition metal ions.

has led to the construction of one-dimensional (1D) chain, two-dimensional (2D) and three-dimensional (3D) network structures in which both the organic 2p and metallic 3d spins have been ordered in macroscopic scales [1, 2, 64]. π-Conjugated oligo-aminoxyl radicals in Table 1 have been employed as bridging ligands in which the spins of the unpaired electrons interact ferromagnetically {J (intraligand) > 0}. Dimension of the complexes as well as sign and magnitude of the exchange coupling between the adjacent spins may be readily tuned in this strat-

76

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Fig. 6. Continued.

egy [1, 2, 64]. A bis-monodentate diradical with a triplet ground state (S = 1), e. g., BNO, would form with coordinatively doubly unsaturated metal ions a 1:1 complex having a 1D infinite chain structure (Fig. 6b). Figs. 6a and 6b differ in the occurrence of one and two kinds of exchange coupling parameters, respectively. Since the exchange coupling between the ligands and the directly attached transition metal ions is typically antiferromagnetic {J (coordination) 0} and the 2p and 3d spins tend to cancel each other out, a residual spin would be established for the repeating unit unless the spin of the latter is 1/2. A ferromagnetic exchange coupling is not impossible if the magnetic orbitals become orthogonal to each other as found in

2.5 Preparation of 3d Transition Metal-Poly(aminoxyl) Radical Complexes

77

discrete complexes of copper(II). Anyway, such a 1D array of spins cannot order at finite temperature without interchain interaction. It would become an antiferro-, meta-, or ferromagnet depending on the nature of the interchain interaction. Since this interaction between the 1D chains is much weaker compared with the intrachain interaction, the critical temperatures (TC ) for exhibiting macroscopic ordering of the spins will consequently be very low. For a triplet diradical such as bis-nitronyl nitroxide ThBNIT in which each radical center can serve as a bis-monodentate bridging ligand [35], complexation would give rise to a ladder polymer as in Fig. 6c. The spin ordering in these systems should be less vulnerable to defects than that in purely 1D systems because there will be a detour available for the exchange coupling through bonds between the two parts of the polymer molecule separated by a chemical defect. Tris-monodentate diradical BNOP with a doublet ground state and triradicals TNOs with quartet ground states (S = 3/2) in which the radical centers are arranged in a triangular disposition would form 3:2 complexes with a coordinatively doubly unsaturated 3d metal ions M. In an ideal case, a 2D hexagonal network structure would be generated (Fig. 6d). A Tshaped quartet triradical carrying two inequivalent ligating sites, e. g., TNOP, would form a 1D chain by using two terminal aminoxyl groups. The middle aminoxyl group might then be used to cross-link the chains to form a 2D (Fig. 6e) or a 3D network structure (Fig. 6e ) depending on whether the second bridging takes place between the same chains as cross-linked by the first bridging. The spin alignment in these systems would be very much stabilized and is expected to give higher-TC magnets.

2.5

Preparation of 3d Transition Metal-Poly(aminoxyl) Radical Complexes

The complex formation is a kinetically and/or thermodynamically controlled self assemblage of the reactants. Typical procedures are as follows. A suspension of manganese(II) bis(hexafluoroacetylacetonate) dihydrate, [Mn(hfac)2 2H2 O], in nheptane is refluxed to remove water of hydration by azeotropic distillation. To the resulting cooled solution is added BNO in n-heptane. The mixture is concentrated under reduced pressure and the concentrated solution is allowed to stand to give black needles of [Mn(hfac)2 BNO] from a deep brown solution. It is preferable to carry out the reaction in inert atmosphere and anhydrous conditions, sometime in a refrigerator. The reaction is completed by precipitation. Some can be recrystallized but others are dissociated in solution. Excess of either one of the components can give complexes of different composition. For example, the reaction of [Mn(hfac)2 ] with tris(aminoxyl) TNOP is complex; while an equimolar mixture in ether containing n-hexane at −10◦ C gives black blocks of 1:1 complex [{Mn(hfac)2 }TNOP] · n-C6 H14 , a mixture containing [Mn(hfac)2 ] in 1.7 molar excess in n-heptane-ether gives black blocks of 3:2 complex [{Mn(hfac)2 }3 TNOP2 ] in ten days at 0◦ C. The complex [{Mn(hfac)2 }3 TNOPB2 ] · n-

78

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

C7 H16 is obtained by dissolving [Mn(hfac)2 2H2 O] in a mixture of diethyl ether, n-heptane and benzene followed by addition of TNOPB in benzene. Black blocks are formed from a deep violet solution. While [Mn(hfac)2 ] gave similar black violet 3:2 complexes with tris(aminoxyl) TNOB and bis(aminoxyl) BNOP, TNO did not form any complex probably because of steric congestion around the ligand molecule. Diradical ThBNIT gave with [Mn(hfac)2 ] dark green powders of complex [{Mn(hfac)2 }3 ThBNIT2 ]CH2 Cl2 · the expected 2:1 complex was not obtained. A 1:1 complex [Mn(hfac)2 TNOP] was obtained as orange bricks from a solution of [Mn(hfac)2 ] and TNOP in n-heptane/CH2 Cl2 containing a small amount of methanol. Dark greenish brick-like crystals of [Cu(II)(hfac)2 TNOP] were obtained similarly in benzene/CH2 Cl2 /CH3 OH. Recently a notorious side reaction has been elucidated leading to undesired byproducts that have unique [3 + 3] benzene-dimer structures [65]. 1D ferrimagnetic complexes, [Mn(hfac)2 BNOR ]n (R = Cl or Br), are typically obtained by the reaction of Mn(hfac)2 with BNOR . When it takes a few days for crystallization, however, black solutions often turn yellow in about one day and do not afford the expected, black polymer complexes. Instead yellow crystalline precipitates are obtained under these conditions (Scheme 5). An X-ray structure analysis revealed that [Mn(hfac)2 BBNOR H2 O]2 · CH2 Cl2 has a [3 + 3] benzene-dimer structure {R = Cl or Br; BBNOR = 3,10-dihalo-5,8,11,12-tetrakis(N -tert-butylimino)tricyclo[5.3.1.12,6 ]dodeca-3,9-diene N ,N ,N ,N -tetraoxide} (Fig. 7). Two crystal-

Fig. 7. A ball-and-stick X-ray structure of the [3 + 3] benzene dimer.

2.5 Preparation of 3d Transition Metal-Poly(aminoxyl) Radical Complexes

79

lographically equivalent manganese(II) ions have an octahedral coordination and are coordinated with four oxygen atoms of two hfac ligands, one oxygen atom of water, and one oxygen atom of O(1) of BBNOBr . Whereas the dimer complex is chiral, both enantiomers are contained in each unit cell (Scheme 6). The BBNOCl complex is isostructural to the bromine derivative.

Scheme 5

Scheme 6

80

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Resonance structure BNOBr must be responsible for the reaction leading to the dimer complex (Scheme 7); either dimerization of BNOBr or attack of BNOBr to free and complexed BNOBr . Whereas the 1D ferrimagnetic complex [Mn(hfac)2 BNOBr ]n is a kinetic product and precipitates out of the solution at the earlier stage of the reaction, the yellow crystal of [Mn(hfac)2 BBNOBr H2 O]2 . CH2 Cl2 appears to be a thermodynamic product.

Scheme 7

Another interesting feature of this work is the liberation of dimer ligand BBNOBr free from manganese ions by dissolving the complex in ether. Two water molecules of hydration appears to be crucial for the stability of BBNOBr · 2H2 O; it is stable in water at 100◦ C but starts to decompose by dissociation even at −78◦ C in CH2 Cl2 when dehydrated by molecular sieves.

2.6

One-dimensional Metal-Aminoxyl Systems

One of the first one-dimensional polymeric transition metal complexes with aminoxyls was [Cu(hfac)2 TEMPOL] [19]. Coupling between the aminoxyl radical and copper(II) is ferromagnetic with 2J/kB = 19 ± 7 K and this S = 1 pair couple with the adjacent pairs through rather lengthy superexchange paths of the σ -bonds of TEMPOL by 2J = −78(2) mK. Since Gatteschi et al. reported a polymeric transition metal complex with nitronyl nitroxide (NIT) in 1986, quite a few one-dimensional polymeric transition metal complexes with nitronyl nitroxides [39, 46, 66–78] and with 1,3-phenylenebis(aminoxyl) derivatives (BNO) [1, 55–58, 64, 79–84] have been documented (Table 3). Bis-monodentate nitronyl nitroxides with doublet states (S = 1/2) and bis-monodentate 1,3-phenylenebis(aminoxyls) with triplet ground states (S = 1) form with coordinatively doubly unsaturated paramagnetic metal ions 1:1 complexes having one-dimensional infinite chain structures. The coupling is such as to align the neighboring spins either parallel or antiparallel to each other along the chain, thereby producing one-dimensional ferromagnets or ferrimagnets, unless the size of the spins matches each other in the latter. The paramagnetic susceptibility values of both kinds of materials are expected to diverge at low temperatures, as a result of lengthening of the correlation of the spins along the

81

2.6 One-dimensional Metal-Aminoxyl Systems Table 3. Metal-aminoxyl-based one-dimensional complexes. 2J/kB (J /kB )

Formula

Type of Sequence

Cu(hfac)2 (NITR)

37.0 K, TN < 1 K J < 0 (R = Me) 35.4 K, TC < 1 K J < 0 (R = i-Pr) –Mn-NIT-Mn-NIT– −340 to −475 K TC = 7.6 K (R = i-Pr) TC = 8.1 K (R = Et) TC = 8.6 K (R = n-Pr) –Ni-NIT-Ni-NIT– −610 K TC = 5.3 K (Ferrimagnet) –Eu-NIT-Eu-NIT– −23.4 K –Gd-NIT-Gd-NIT– –Zn-NIT-Zn-NIT– −17.6 K Antiferromagnetic chain –Mn–BNOH – TN = 5.5 K –Mn–BNOF – TN = 5.3 K –Mn–BNOCl – TC = 4.8 K –Mn–BNOBr – TC = 5.3 K –Mn–TNOP– TN = 11.0 K

R = Me, i-Pr Mn(hfac)2 (NITR) R = Me, Et, i-Pr, n-Pr, Ph Ni(hfac)2 (NITMe) Eu(hfac)3 (NITEt) Gd(hfac)3 (NITEt) Zn(hfac)2 (NITi-Pr) Mn(hfac)2 (BNOH ) Mn(hfac)2 (BNOF ) Mn(hfac)2 (BNOCl ) Mn(hfac)2 (BNOBr ) Mn(hfac)2 (TNOP )

TC or TN (magnetism) Ref.

–Cu-NIT-Cu-NIT–

39, 46, 68, 77

67, 69 72 73 70 71 71, 76 78 55 84 82 82 58

chain. Depending on the nature of the additional interchain interaction, the chain polymers become an antiferromagnet or a ferri/ferromagnet. In these magnets, the interchain magnetic interactions are much weaker than the intrachain interaction, and therefore the transition temperatures to three-dimensionally ordered states are relatively low.

2.6.1

Structure and Magnetic Properties of Ferrimagnetic 1D Chains Formed by Manganese(II) and Nitronyl Nitroxides

Manganese(II) ions form with various NITR (R = i-Pr, Et, n-Pr) crystalline chain complexes in which Mn(II) is hexacoordinated with four oxygen atoms of two hfac molecules and two oxygen atoms of two different NIT radicals. The other oxygens of the two NITR radicals are coordinated to the adjacent Mn(II) ions (Scheme 8). All the complexes are ferrimagnetic and order at ca. 8 K. Quantitative analyses of the magnetic susceptibility data are performed by using a model of classical-quantum chains [66-78]. The results are summarized in Table 3.

Scheme 8

82

2.6.2

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Structure and Magnetic Properties of Ferrimagnetic 1D Chains Formed by Manganese(II) and Triplet bis-Aminoxyl Radicals

Let us explain the molecular and crystal structures of the metal complexes with BNO derivatives in some detail [55–58]. Their structures are solved in the monoclinic P21/n space group (No. 14) with Z = 4. The crystal structure data are listed in Table 4. The analyses reveal that the manganese(II) ions have octahedral coordination with four oxygen atoms of two hfac anions and two oxygen atoms of two different BNO molecules. The latter is bound to the Mn(II) ion in cis configuration. As a result, the Mn ions and biradical molecules form a helical 1D polymeric chain structure along the crystal b axis. All the hexacoordinated Mn(II) ions have either or configuration along a given chain. Two tert-butylaminoxyl groups are rotated out of the phenylene ring plane in a conrotatory manner but with different angles; each BNO molecule in the crystal has no symmetry element and therefore chiral, i. e., R or S. The 1D polymeric chains are therefore isotactic as all units of the same chirality reside on a given chain (Fig. 8). The crystal lattice is as a whole achiral due to the presence of an enantiomeric chain. The nearest neighbor interchain distances of Mn(II)-Mn(II) and Mn(II)-carbon atoms of the aminoxyl moiety in the three complexes are listed in Table 5. The 1D complexes [Mn(hfac)2 BNOH ] and [Mn(hfac)2 BNOF ] have antiferromagnetic ground states due to a negative intersublattice exchange interaction. Their saturation magnetization values, MS = 3 µB /f.u. agree well with a theoretical limit assuming the antiferromagnetic coupling between the Mn(II) and radical spins. The magnetization of [Mn(hfac)2 BNOH ] was studied using a single crystalline sample. In Fig. 9 the M(H ) curves are given along the three principal axes at 1.8 K. While Mb , the magnetization projection on the b-axis, rises linearly with increasing field,

Fig. 8. A view of a 1D chain formed by [Mn(II)(hfac)2 ] with bis-aminoxyl BNOF .

C24 H24 N2 O6 F12 Mn 719.38 9.212(3) 16.620(3) 20.088(2) 98.46(1) 3042(1) 4 0.30 × 0.15 × 0.95 mm3 Rigaku AFC5R 1.571 Refined 3256 434 0.055 0.058 1.90

C24 H23 N2 O6 F13 Mn 737.37 9.351(4) 16.626(3) 20.167(3) 100.01(2) 3087(1) 4 0.10 × 0.10 × 0.50 mm3 Rigaku AFC7R 1.586 Fixed calc. 3420 415 0.069 0.064 2.96

Mn(II)(hfac)2 (BNOF ) C24 H23 ClN2 O6 F12 Mn 753.83 8.953(4) 17.020(4) 20.094(5) 98.66(2) 3027(1) 4 0.05 × 0.10 × 0.10 mm3 Rigaku Raxis-IV 1.654 Fixed calc. 2844 415 0.106 0.122 2.95

Mn(II)(hfac)2 (BNOCl )

8.499(1) (55402)* 9.11(3) (44402)*

8.5679(6) (65502)* 9.457(8) (65501)*

Mn(II)(hfac)2 (BNOF )

8.953(4) (65501)* 8.86(1) (45502)*

Mn(II)(hfac)2 (BNOCl )

* Symmetry operators: (1) X, Y, Z ; (2) 1/2 − X, 1/2 + Y, 1/2 − Z ; (3) −X, −Y, −Z ; (4) 1/2 + X, 1/2 − Y, 1/2 + Z

¨ Mn(II)-Mn(II) distance (A) ¨ Mn(II)-C distance (A)

Mn(hfac)2 BNO H )

9.244(4) (65502)* 9.01(1)(55602)*

Mn(II)(hfac)2 (BNOBr )

C24 H23 BrN2 O6 F12 Mn 798.28 9.244(4) 17.155(5) 20.431(7) 99.56(3) 3195(1) 4 0.20 × 0.20 × 0.40 mm3 Rigaku AFC7R 1.662 Fixed calc. 1771 415 0.064 0.026 2.51

Mn(II)(hfac)2 (BNOBr )

Table 5. The nearest neighbor interchain Mn(II)-Mn(II) and Mn(II)-carbon atom distances as shown in Fig. 8.

Empirical formula Formula wt. ¨ a (A) ¨ b (A) ¨ c (A) b (◦ ) ¨ 3) V (A Z Crystal dimensions Diffractometer dcalc (g cm−3 ) Hydrogen Observations Variables R Rw GOF

Mn(hfac)2 (BNOH )

Table 4. Crystallographic data for [Mn(hfac)2 BNOR ]n complexes (R = H, F, Cl, or Br).

2.6 One-dimensional Metal-Aminoxyl Systems

83

84

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Fig. 9. (a) The magnetization curves for [Mn(hfac)2 BNOH ] at 1.8 K. (b) Orientation of the magnetization vectors M in the ac-plane in [Mn(hfac)2 BNOH ].

sharp metamagnetic transitions occur at 250 and 450 Oe along the c- and a-axes, respectively. From the M(H ) curves MS was concluded to lie perpendicular to the b-axis. Its orientation in the ac-plane was determined considering the projections Ma and Mc of the magnetization vector on the a- and c-axes extrapolated to zero external field: Ma = 1.2 µB /f.u. and Mc = 2.5 µB /f.u. (it crystallizes in a monoclinic structure with a = 9.212 Å, b = 16.620 Å, c = 20.088 Å and β = 98.46) [85]. These values are very close to the projections of MS = 3.0 µB /f.u. on the crystal axes in case Ms values are oriented along the four {101}-type directions (Fig. 9b). At zero external field, all the orientations [101], [101], [101], and [101] are equivalent, and the total magnetization is compensated. When the magnetic field is applied along the c-axis, [001], the metamagnetic transition corresponds to the re-orientations of the magnetization vector M[101] → M[101] and M[101] → M[101] , thus giving a cprojection Mc = 1.5(cos 25.94◦ + cos 23.02◦ ) µB = 2.73 µB . Similarly, for H//a-axis the metamagnetic transition corresponds to the re-orientations M[101] → M[101] and M[101] → M[101] and Ma = 1.5(cos 72.53◦ + cos 58.51◦ ) µB = 1.23 µB . With further increasing field the two magnetization vectors smoothly rotate to give a saturation at ca. 30 kOe. The low-field susceptibilities of the [Mn(hfac)2 BNOH ], [Mn(hfac)2 BNOF ], [Mn(hfac)2 BNOCl ] and [Mn(hfac)2 BNOBr ] complexes with ferrimagnetic ground state as a function of temperature are shown in Fig. 10. The TC values were determined as 4.8 and 5.3 K for complexes of BNOCl and BNOBr , respectively. Below

2.6 One-dimensional Metal-Aminoxyl Systems

85

Fig. 10. Temperature-dependence of the low-field susceptibility (5 Oe) for the ferrimagnetic complexes: (a) [Mn(hfac)2 BNOH ], [Mn(hfac)2 BNOF ], and (b) [Mn(hfac)2 BNOCl ], [Mn(hfac)2 BNOBr ].

Fig. 11. Magnetization curves for [Mn(hfac)2 BNOCl ] (O) and [Mn(hfac)2 BNOBr ] () at 1.7 K. Inset shows the low-field cycling.

TC the magnetic behavior of both the complexes is very similar to each other. The magnetization at 1.8 K shows saturation at about 30 000 Oe with a maximal value of ca. 3 µB (Fig. 11), which corresponds to the antiparallel alignment of the Mn(II) and NO group spins. The compounds show narrow hysteresis with the coercive force less than 20 Oe (Inset of Fig. 11). Above TC , the product of the molar paramagnetic susceptibility and temperature, χmol T , for all the BNOR complexes increases steadily with decreasing temperature and passes over a maximum at 8–9 K. These dependencies coincide with the paramagnetic range, therefore only the data for [Mn(hfac)2 BNOH ] are displayed in Fig. 12.

86

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Fig. 12. Temperature-dependence of the product χmol T for the [Mn(hfac)2 BNOH ] complex (circle symbols). The solid line is a fit by the use of Eq. (2). The best-fit parameters are listed in the inset table for this compound and for those of the complexes with BNOF , BNOCl and BNOBr .

The [Mn(hfac)2 BNOR ] compounds under consideration have three-spin periodicity along the 1D chains and the extensive analysis of intrachain exchange interactions requires numerical calculations using the spin Hamiltonian: H = −2J1 (s3i−2 s3i−1 + s3i−1 s3i + αs3i S3i+1 ) i

where α = J2 /J1 and s and S are the NO group and Mn(II) spin operators, respectively. However, considering the high temperature value of χmol T and lowtemperature saturation magnetization, some preliminary conclusions can be made about the strength of the intrachain magnetic couplings [86]. As a matter of fact, taking the saturation magnetization value in the ordered state into account, we may consider two configurations: (i) ferrimagnetic chains formed by the 2/2 total spins of biradicals antiferromagnetically coupled with the 5/2 spins of Mn(II) ions, and (ii) ferromagnetic chains formed by the pseudo-spins consisting of two NO groups of different biradicals bridged by Mn(II) ions, i. e., S = 3/2 ferromagnetic chains. The former configuration should yield a characteristic high temperature minimum in plots of χmol T against T with the value approaching to 5.38 emu K mol−1 , while the high temperature value of 1.875 emu K mol−1 is expected for the latter. The observed χmol T values at 300 K vary between 2.2 and 2.4 emu K mol−1 , which is slightly larger than the theoretical limit 1.88 emu K mol−1 expected for S = 3/2 ferromagnetic chain compounds. Hence, the basic intrachain exchange interaction in [Mn(hfac)2 BNOR ] compounds can be determined assuming a ferromagnetic chain structure with S = 3/2. A classical-spin approximation was employed for the analysis and the experimental χmol T dependencies were treated by the expression [87]: χmol T = N

g 2 µ2B 1 + U (T /T0 ) S(S + 1) 3kB 1 − U (T /T0 )

(2)

where U (T /T0 ) = coth(T0 /T ) − T /T0 , T0 = (2J/kB )S(S + 1) and the other symbols have their usual meaning. A comparative analysis of the behavior on the basis of a classical and quantum ferro- or ferrimagnetic chains [88] showed that both the approaches yield similar values for the exchange interaction at elevated temperatures T > 2J/kB . Therefore, Eq. (2) was first fitted to the experimental plot of χmol T

2.6 One-dimensional Metal-Aminoxyl Systems

87

against T in the high-temperature region, then the fits were extended down to ca. 50 K. The effective exchange coupling value 2J/kB between the pseudo-spins was found to be 23 K for all the four [Mn(hfac)2 BNOR ] compounds with R = H, F, Cl and Br. Using this value, the interchain exchange parameters of ferrimagnetic compounds [Mn(hfac)2 BNOCl ] and [Mn(hfac)2 BNOBr ] were then evaluated. Near TC , T0 /T > 10. Thus, taking U (T /T0 ) = 1 − T /T0 , and introducing the interchain exchange interaction by χtot = 1/(1/χmol − λ ), one obtains: T1 =

3TC2 2T0 − TC

(3)

where T1 = (2z J /kB )S(S + 1). This equation reduces to that given by Richards [89] when neglecting TC in the denominator. The evaluated values of 2J /kB are +0.018 K and +0.022 K for [Mn(hfac)2 BNOCl ] and [Mn(hfac)2 BNOBr ], respectively. For the metamagnetic complexes [Mn(hfac)2 BNOH ] and [Mn(hfac)2 BNOF ] these values are −0.018 K and −0.010 K [90], respectively. Thus, the ratios |J/J | are ∼ 10−3 for all the complexes studied. The Mn complexes with BNOCl and BNOBr have topologically the same crystal structure with the complexes of BNOH and BNOF . From the observed intermolecular distance, the strongest interchain interaction in [Mn(hfac)2 BNOR ] is judged to arise from the Mn–Mn and Mn–Cmiddle distance (Table 5). From the dipole-dipole interaction term, magnetic interaction between nearest neighbor magnetic moments is always antiferromagnetic (Fig. 13). This argument can be confirmed by the fact that 2J /kB does not exhibit a regular change throughout the [Mn(hfac)2 BNOR ] series, but changes abruptly in sign when the nearest interacting pair is changed. Polycrystalline samples of [Mn(hfac)2 BNOR ] showed broad singlet EPR spectra at room temperature with H pp = 303 G, g = 2.0055 for R = H, H pp = 315 G, g = 2.0095 for R = Cl and H pp = 158 G, g = 2.018 for R = Br. The temperature dependencies of H pp are shown in Fig. 14. The large H pp values indicate that the low dimensional exchange interaction dominates in these crystals [91–93].

Fig. 13. (a) Schematic drawing of the magnetic structure of [Mn(hfac)2 BNOCl ] and [Mn(hfac)2 BNOBr ], and (b) [Mn(hfac)2 BNOH ] and [Mn(hfac)2 BNOF ]. Broken lines show the antiferromagnetic interaction between the ferro/ferri-magnetic 1D chains. Solid lines show the ferromagnetic interchain interaction.

88

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Fig. 14. Temperaturedependence of H pp for the [Mn(hfac)2 BNOR ] complexes (R = H, Cl, or Br).

2.7

Two-dimensional Metal-Aminoxyl Systems

In principle, there are no difficulties in the design of materials that undergo magnetic phase transitions at higher temperatures: the magnetic network of coupled metal ions and radicals should be extended from one to two or three dimensions, and the strong magnetic coupling between the spin centers would ensure high transition temperatures (see also three conditions given in Introduction) [94]. One of the first two-dimensional structures is a series of copper(II) β-diketonate complexes [95– 101]. The coordination of the oxygen atoms of the aminoxyl groups of neighboring molecules completes the distorted octahedra around the Cu(II) ions. Temperaturedependence of the magnetic susceptibility for these complexes fits a theory of isolated ferromagnetic coupled pairs. Enaminoketones of 3-imidazolin-1-oxyl form layered polymeric structures with Cu(II), Co and Ni [102–105]. However, whereas all these complexes have extended structures, the magnetic coupling is limited to the directly bonded metal-aminoxyl couples. Table 6. Metal-aminoxyl-based two-dimensional complexes. Formula

Crystal or powder

{Mn(pfbz)2 }2 (NITR) R = Me, Et {Mn(pfpr)2 }2 (NITR) R = Me, Et {Mn(hafc)2 }3 (NITBzald)2 · 0.5CHCl3

Powder

{Mn(hafc)2 }3 (TNOPB)2 · n-C7 H16 {Mn(hafc)2 }3 (TNOP)2

J/kB (J /kB )

TC or TN (magnetism)

Ref. 74

Crystal

TC = 24 K TC = 20.5 K TC = 24 K TC = 20.5 K TC = 6.4 K (Ferrimagnet) TC = 3.4 K

Crystal

TC = 9.5 K

Powder Powder

−149.6 K (0.075 K) −367 K

104 105 106 56 57 36

2.7 Two-dimensional Metal-Aminoxyl Systems

89

Thus the problem is not only how to control the construction and the structure of extended systems in a desired fashion in order to extend the properties from the individual building blocks to the whole lattice, but also how to maintain strong magnetic coupling throughout the extended structure. Moreover, there is much difficulty in making single crystal of the complex which have high dimensionality. Therefore, there are still few examples of aminoxyl-metal ion-based two-dimensional magnetic complexes (Table 6) [36, 56, 57, 74, 106–108]. The flexibility of the ligands are important to make good crystals of two- or three-dimensional complexes. In the metal-nitronyl nitroxide systems, there are limitations to make two or three dimensional complexes topologically. The approach by high spin oligo-aminoxyls and transition metal complexes has a greater advantage than metal-nitronyl nitroxide systems Table 6.

2.7.1

Structure and Magnetic Properties of Ferrimagnetic 2D Sheets Formed by Manganese(II) and Nitronyl Nitroxides

Some ferrimagnetic 2D manganese(II)-nitronyl nitroxide systems are synthesized by the complexation of manganese (II) bispentafluorobenzoate (Mn(pfbz)2 ) or manganese (II) bispentafluoropropionate (Mn(pfpr)2 ) with 2-alkyl nitronyl nitroxide (NITR, R = Me, Et) (Fig. 15). None of the complexes gave single crystals amenable for an X-ray crystal structure analysis. It is assumed that 2D sheet structures are formed in these complexes. The complexes of [{Mn(pfbz)2 }2 (NITR)] (R = Me, Et) and [{Mn(pfpr)2 }2 (NITR)] (R = Me, Et) are ferrimagnetic, and the transition into ordered states occurs at ca. 23 K. The quantitative analysis of the magnetic susceptibility of these complexes was performed by using a model of a Heisenberg classical-quantum spin system (see Table 6).

Fig. 15. Scheme of the supposed magnetic interactions in (a) [Mn(pfpr)2 (NITMe)] and (b) [{Mn(pfbz)2 }2 {NITR)].

90

2.7.2 2.7.2.1

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Structure and Magnetic Properties of Ferrimagnetic 2D Sheets Formed by Manganese(II) and High-spin tris-Aminoxyl Radicals Crystal and Molecular Structure of 2D Systems

[{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 The X-ray crystal structure of the complex reveals that the manganese(II) ion has an octahedral coordination with four oxygen atoms of two hfac anions bound to the metal ion in the equatorial plane, while the axial positions are occupied by the two oxygen atoms of the two aminoxyl groups. Six triradical molecules and six Mn ions make a hexagon from which an extended honeycomb network is constructed by sharing its edge (Fig. 16a). A disordered n-heptane molecule is contained in each hexagonal cavity. The 2D network sheets form a layered structure in which the adjacent layers are slide by a radius of the hexagon from the superimposable disposition with a mean inter-plane distance of 3.58 Å (Fig. 16b). On the basis of the spin density and distance, the strongest inter-plane spin–spin interaction is judged to arise from the carbons (3.78 Å apart) of the benzene rings para on the one hand and meta on the other to the aminoxyl groups (Fig. 16b ). This type of interaction is expected to be ferromagnetic as dictated by the McConnell’s theory [109, 110]. [{Mn(hfac)2 }3 (TNOB)2 ] While satisfactory analytical data were obtained [36], the complex does not give a single crystal amenable for an X-ray crystal structure analysis. It is assumed that, while a 2D sheet structure is formed, the reduced symmetry of TNOB relative to TNOPB must be responsible for the difficulty in growing single crystals. It may be assumed that a 2D-heterospin system consisting of a ferromagnetically coupled triradical TNOPB serving as a tris(monodentate) bridging ligand and attached to a paramagnetic transition metal ion Mn(II) in an antiferromagnetic fashion would be realized. The 2D network is schematically given in Fig. 17. As in TNOB, the network may be stacked ferromagnetically across the layer.

2.7.2.2

Magnetic Properties of 2D Systems

Below T C [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 When the measurement was carried out in a low field, the magnetization values showed a sharp rise at TC = 3.4 K (Fig. 18). The low-field susceptibility at 3 K is extremely large, as expected for a ferro/ferrimagnet. The spontaneous magnetization was observed below TC , demonstrating the transition to a bulk magnet. The field dependence of the magnetization for [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 is shown in Fig. 19. When the measurement was carried out at 1.8 K, its magnetization reaches to ca. 9 µB at 30 000 Oe and becomes saturated. If the interaction between the Mn(II) and TNOPB is antiferromagnetic (J2 < 0 in Chart I (b)), the saturated magnetization value is expected to be 9 µB (5/2 × 3 − 3/2 × 2 = 9/2), in good agreement with the observed value. Note that, when a single-domain state is reached, the low tem-

2.7 Two-dimensional Metal-Aminoxyl Systems

91

Fig. 16. (a) View along the c axis of a layer showing the hexagons made of six tris-aminoxyl TNOPB and six manganese complexes Mn(hfac)2 . (b) View of the shortest contact between the layers.

92

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Fig. 17. Schematic drawing of the estimated 2D network structure for [{Mn(hfac)2 }3 (TNOB)2 ].

Fig. 18. Plots of magnetization against T for the complex [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 measured at a field of 1 Oe (◦) and spontaneous magnetization (•).

Fig. 19. Field dependence of the magnetization of [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 measured at 1.8 K (◦) and 10.8 K (+).

2.7 Two-dimensional Metal-Aminoxyl Systems

93

perature (1.78 K) magnetization curve of [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 can satisfactorily be described above 6 kOe by the B3/2 (x) Brillouin function with the molecular field coefficient λ ≈ 0.28 emu mol−1 . The latter mainly includes the inplane interactions. Proceeding from this analysis, we cannot evaluate the weakest inter-plain exchange interaction parameter, which seems mainly responsible for the occurrence of 3D long-range order in this complex. [{Mn(hfac)2 }3 (TNOB)2 ] The low-field magnetization measurements carried out in 5 Oe, showed a sharp rise at TC = 9.2 K (Fig. 20). The low-field susceptibility at 9 K is extremely large, as expected for a ferro/ferrimagnet. The field dependence of the magnetization for [{Mn(hfac)2 }3 (TNOB)2 ] are shown in Fig. 21. The spontaneous magnetization was observed below TC , demonstrating the transition to a bulk magnet. When the measurement was carried out at 1.8 K, the magnetization for [{Mn(hfac)2 }3 (TNOB)2 ] reaches to ca. 9 µB at 30 000 Oe and becomes saturated. If the interaction between the Mn(II) and TNOB is antiferromagnetic (J2 < 0 in Chart I (b)), the saturated magnetization value is expect to be 9 µB (5/2 × 3 − 3/2 × 2 = 9/2) in good agreement with the observed value.

Fig. 20. Plots of magnetization against T for the complex [{Mn(hfac)2 }3 (TNOB)2 ] measured in a field of 5 Oe (◦) and spontaneous magnetization (•).

Fig. 21. Field dependence of the magnetization of [{Mn(hfac)2 }3 (TNOB)2 ] measured at 1.8 K (◦), 5.0 K (♦), and 15.0 K (×).

94

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Above T C [111] [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 The magnetic susceptibility of [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 as, represented by a plot of χmol T against T , is shown in Fig. 22. With increasing temperature, this dependence is first characterized by a sharp maximum at 2.5 K followed by a smooth minimum in the range 80–140 K, which rather resembles a plateau. The value of χmol T = 5.71 emu K mol−1 within this interval is very close to the theoretical limit, 5.625 emu K mol−1 , which is expected for three stable non-interacting S = 3/2 spins per mole. According to Fig. 16a, this spin configuration can be considered as formed by one S = 5/2 spin of the Mn(II) ion coupled antiferromagnetically with two 1/2 spins of two different triradicals TNOPB. Any other stable spin configuration, either two S = 3/2 TNOPB and three S = 5/2 Mn(II) spins or six decoupled S = 1/2 spins of TNOPB and three S = 5/2 spins of Mn(II), yield much higher values of χmol T (>15 emu K mol−1 ). With further increasing temperature, χmol T shows however a substantial increase. This circumstance points to the contribution of the thermal excitations of the 3/2 spin species above about 140 K. Owing to low TC , the in-plane inter-trimer interaction, J1 ,

Fig. 22. The temperature-dependence of χmol T for the layered complex [{Mn(hfac)2 }3 (TNOP)2 ] · n-C7 H16 . Open circles are the experimental data, and the solid and the dashed lines are the least squares fits by the use of Eq. (5) for 2D and 3D lattices, respectively. The fits performed for stable trimers below 120 K coincide with the corresponding 2D and 3D fits for unstable trimers and are not shown in the Figure. The inset shows details of the low temperature behavior as a plot of χ/T against λ /T .

2.7 Two-dimensional Metal-Aminoxyl Systems

95

is suggested to be weak compared to the intra-trimer interaction, JTR , which mainly governs the high temperature properties of the [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 complex. Then, the temperature variation of χmol T can be analyzed by the expression for an isolated ferrimagnetic (1/2, 5/2, 1/2) linear trimer modified accordingly by introducing weak inter-trimer interactions. The energy level scheme of this trimer was calculated using the isotropic spin Hamiltonian H = −2JTR (s1 S2 + S2 s3 ). The eigenvalues E(ST , S13 ) of this Hamiltonian are E(3/2, 1) = 7JTR , E(5/2, 1) = 2JTR , E(5/2, 0) = 0 and E(7/2, 1) = −5JTR (here ST = S2 + S13 and S13 = s1 + s3 ) and give the susceptibility in the form [112]: N g 2 µ2B 15 4 5e5JTR /kB T + 5e7JTR /kB T + 16e12JTR /kB t (0) χmol = 1+ 3kB T 4 5 2 + 3e5JTR /kB T + 3e7JTR /kB T + 4e12JTR /kB T (0)

= χ0 mol Q(JTR /kB T )

(4)

(0)

where χ0 mol = (N g 2 µ2B /3kB T )(15/4) and the other symbols have their usual meaning. For negative JTR , the product STR (STR + 1) is equal to 15/4 and starts to increase near TTR = JTR /kB . Therefore, at low temperatures this magnetic system can be considered as a 2D honeycomb-like network formed by stable S = 3/2 spins. Within the scope of this approximation the paramagnetic susceptibility of the complex [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 was analyzed below 120 K by using the hightemperature power series expansion up to order eight for a 2D honeycomb lattice [113]. A satisfactory fit to the experimental temperature variation of χmol T was possible down to 18 K only, with the in-plane interaction parameter J1 /kB = +0.225 K. From this analysis it was concluded that the contribution from the inter-plane exchange interaction becomes important below ca. 20 K. In the range 4.5–120 K a good fit was possible within(0)the scope of the molecular field approximation, i. e., taking (0) χmol = χ0 mol / 1−λ χ0 mol with λ = +0.333 emu mol−1 . In Fig. 22, where the results of these two analyses are shown as plots of χmol T against T ; the curves corresponding to both fits are indistinguishable above 20 K. This fact can point that in the ratio between the inter-plane and in-plane exchange integrals, J2 /J1 , is not so small as, e. g., for typical 2D systems with J2 /J1 = 10−2 –10−3 [114]. At higher temperatures (0) effect of the strong intra-trimer interaction can be taken into account putting χmol (0) instead of χ0 mol in the expression for the paramagnetic susceptibility χmol . Hence, in the molecular field approximation one obtains: (0)

χmol =

χ0 mol

(0)

Q −1 (JTR /kB T ) − λ χ0 mol

(5)

The dashed line in Fig. 22 is the fit to the experimental data over the whole temperature range 4.5–280 K made by the use of Eq. (5). As seen, the experimental data can well be described within this approximation both at high and low temperatures thus proving the importance of the thermal excitations within the (1/2, 5/2, 1/2) trimers for the complex [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 . The best agreement was obtained at JTR /kB = −176.4 K and λ = +0.333 emu mol−1 . The power series

96

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

expansion for the 2D honeycomb network [113], with STR (STR + 1)/(15/4) replaced by Q(JTR /kB T ), also gives a good fit within the range 18–280 K. The fit parameters for this approximation are JTR /kB = −175.4 K and J1 /kB = +0.226 K. The low temperature behavior of the paramagnetic susceptibility is plotted in the inset in Fig. 22 as C/T against λ /T (C being the Curie constant). As can be seen, the effect of low-dimensionality is not essential in [{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 . [{Mn(hfac)2 }3 (TNOB)2 ] The temperature-dependence of the molar magnetic susceptibility χ for [{Mn(hfac)2 }3 (TNOB)2 ] was investigated at several magnetic fields. When the measurement was carried out in a magnetic field of 5000 Oe, the product of the molar susceptibility and temperature (χmol T ) increased with decreasing temperature, and showed a maximum at 10 K (Fig. 23). The value of χmol T = 7.1 emu K mol−1 is larger than the theoretical limit, 5.625 emu K mol−1 , which is expected for three stable noninteracting S = 3/2 spins per mole. According to Fig. 17, this spin configuration can be considered as formed by one S = 5/2 spin of the Mn(II) ion coupled antiferromagnetically with two 1/2-spins of two different triradicals TNOB. Any other stable spin configuration, either two 3/2 (TNOB) and three 5/2 (Mn(II)) spins or six decoupled 1/2 spins of TNOB and three 5/2 spins of Mn(II), yield much higher values of χmol T (>15 emu K mol−1 ). The ESR spectrum of the complex consisted of a single line (g = 2.0057 and H pp = 392.4 G at room temperature). The temperature-dependence of the g and H pp values is shown in Fig. 24. The g value decreased with decreasing temperature and the decrease leveled off at TC down to 5.9 K. This g value behavior indicates that the internal magnetic field increases with decreasing temperature. The signal narrowed with decreasing temperature, reached a minimum width at ca. 15 K, and then broadened at lower temperature. These temperature dependencies indicate that the effect of the internal magnetic field becomes substantial below 15 K. While clearly separated g and g were not observed in the ESR spectra, the upfield shift appeared to be mostly due to g and the shift of g appeared to lag behind. A 1D

Fig. 23. Plot of χmol T against T for the complex [{Mn(hfac)2 }3 (TNOB)2 ] measured in a field of 5000 Oe.

2.8 Three-dimensional Metal-Aminoxyl Systems

97

Fig. 24. The temperature-dependence of g (◦) and H pp (•) for [{Mn(hfac)2 }3 (TNOB)2 ].

ferrimagnetic chain would have exhibited a much larger g anisotropy. The H pp value is reasonable for a 2D spin structure [91–93].

2.7.3

Conclusion

It is shown that perfect 2D ferrimagnetic sheets with heterospin ferromagnetic (J1 > 0)-antiferromagnetic (J2 < 0 in Chart I (b)) networks together with ferromagnetic stacking of the layers are realized in the metal–aminoxyl radical systems. While the kinetic instability of the triradical TNOB did not allow us to study the magnitude of the intramolecular ferromagnetic exchange coupling, it is expected to be considerably greater in TNOPB. The interaction in TNOB leads by its symmetry to an isosceles triangular relation. One between the two nitroxide radicals at positions 3 and 5 of the same benzene ring is estimated from that of m-phenylenebis(N -tertbutylnitroxide) and the analogs to be J/kB = 200–500 K. The other interaction between the 4 and 3 (and 5) aminoxyl radical centers is estimated to be ca. 67 K. Altogether, the intramolecular ferromagnetic coupling in TNOB is estimated to be stronger than in TNOPB and contributes to the higher TC value of 9.5 K compared to 3.4 K in the latter when self-assembled with the aid of Mn(II) ions.

2.8

Three-dimensional Metal-Aminoxyl Systems

In Sections 6 and 7 we described one- and two-dimensional complexes, respectively. For one-dimensional complexes, the transition temperatures to ferri-ferromagnets are about 5–6 K, while for two-dimensional complexes TC grows up to ca. 9 K. In order to make complexes with higher transition temperature, the dimensions of the spin network must be raised. The flexibility of the ligands is more important to make good crystals of three dimensional complexes than two dimensional systems. Owing to these difficulties, only one well-defined three-dimensional complex has been reported in the metal-aminoxyl systems.

98

2.8.1

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Crystal and Molecular Structure of the 3D System

An X-ray crystal structure analysis of an orthorhombic crystal of the complex reveals the formation of a parallel crosses-shaped 3D polymeric network (Fig. 25). The oxygen atoms of the terminal aminoxyl groups of triradical bis{3-(N -tertbutyloxylamino)-5-tert-butylphenyl}aminoxyl (TNOP) are ligated to two different manganese(II) ions to form a 1D chain in the b/c plane of the crystal. Since any manganese ion in an octahedral position is attached to the two aminoxyl oxygens from two different triradical molecules in a trans disposition, the triradical molecules are in zigzag orientation along the chain. The diarylaminoxyl unit is in a chiral propeller conformation and the R and S forms alternate along the chain. The middle aminoxyl group of the ligand TNOP molecule on one chain is used to link its oxygen with that of the same chirality in the adjacent chains extended in the b/ − c diagonal direction through a third Mn(II) ion in an octahedral position with the intersecting angle of 54.4◦ , establishing a parallel crosses-shaped 3D polymeric network (Fig. 26). The three-connected nets of the intersecting angle of 0 and 90◦ would have produced lattices corresponding to graphite and the silicon sublattice of thorium silicide, respectively.

Fig. 25. Crystal structure of the 3D metal-radical complex [{Mn(hfac)2 }3 (TNOP)2 ]. For clarity the CF3 and (CH3 )C groups are not shown. a, b and c denote the orthorhombic crystal axes. The Mn(1) and Mn(2) ions are shown by filled circles.

2.8 Three-dimensional Metal-Aminoxyl Systems

99

Fig. 26. Schematic drawing of the crystal of the 3D metal-radical complex [{Mn(hfac)2 }3 (TNOP)2 ].

2.8.2 2.8.2.1

Magnetic Properties of the 3D System Below T C

The temperature-dependence of the magnetization M for a single crystalline sample of [{Mn(hfac)2 }3 (TNOP)2 ] was investigated at 5 Oe. When the sample was cooled within the field of 5 Oe, the field-cooled magnetization showed an abrupt rise at TC = 46 K (Fig. 27). The field dependence of the magnetization at 5 K showed two important features. First, the magnetization rose sharply at low field, reached a value of ca. 9 µB (50 000 emu G mol−1 ) at 220 Oe and became saturated. The saturation value is in good agreement with a theoretical one of 9 µB (5/2 × 3 − 3/2 × 2 = 9/2) expected for the antiferromagnetic coupling between the d5 Mn(II) ion and S = 3/2 triradical TNOP. Secondly, a conspicuous magnetocrystalline anisotropy was found in which the easy axis of magnetization lies along the c axis of the crystal lattice and the hard axis lies perpendicular to it (Fig. 28).

2.8.2.2

Above T C

The temperature-dependence of the molar magnetic susceptibility χmol was investigated at a field of 5000 Oe (Fig. 29). The χmol T value of 8.64 emu K mol−1 at 300 K is larger than a theoretical value of 5.63 emu K mol−1 expected for a short-range antiferromagnetic ordering of six 1/2 spins of TNOP and three 5/2 spins of d5 Mn(II) for [{Mn(hfac)2 }3 (TNOP)2 ]. As T is lowered, χmol T value increased monotonically in proportion to the increase in the correlation length within the network. Together with

100

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Fig. 27. The temperature-dependence of the magnetization of the single crystal of [{Mn(hfac)2 }3 (TNOP)2 ]. (a) The field cooled magnetization (FCM, ◦), the zero field cooled magnetization (ZFC, ) along c-axis ; (b) The field cooled magnetization (FCM, ), the zero field cooled magnetization (ZFC, ♦) along the b-axis.

Fig. 28. Magnetization curves for [{Mn(hfac)2 }3 (TNOP)2 ] at 1.8 and 25 K along the three principal crystallographic axes.

2.8 Three-dimensional Metal-Aminoxyl Systems

101

Fig. 29. Temperature-dependence of the product χmol T of the [{Mn(hfac)2 }3 (TNOP)2 ] complex in the paramagnetic temperature range. The open circles are the experimental data, and the solid and dashed lines are calculated for the fixed trimer spin STR = 3/2 in the quantumclassical and classical-classical approximations, respectively.

the lack of a minimum at lower temperature, the room temperature χmol T value also points to the operation of strong (more negative than −300 K) antiferromagnetic coupling between the Mn(II) ion and the nitroxide radical of TNOP in which the onset of the intramolecular ferromagnetic coupling is meaningful. In the low-temperature range, the magnetic behavior is qualitatively equivalent to a 3D ferromagnetically coupled network of S = 3/2 spins consisting of the S(1/2) − S(5/2) − S(1/2) units. The exchange interactions determining the isotropic magnetic properties of the [{Mn(hfac)2 }3 (TNOP)2 ] compound were evaluated from analysis of the temperature-dependence of the paramagnetic susceptibility. All the attempts to describe this dependence within the frame of 3D ferro- or ferrimagnetic models by combining different magnetic sublattices made up either from Mn(II) and TNOP molecules or (1/2, 5/2, 1/2) species formed by Mn(II) and nitroxide groups were unsuccessful (See Fig. 29). This was accounted for the effect of a magnetic low dimensionality the [{Mn(hfac)2 }3 (TNOP)2 ] compound exhibits at least in the paramagnetic region. Although this complex forms a well defined 3D network with respect to the chemical bonding, the spin–spin couplings between Mn(II) and triradical species can be different along different directions, which can in turn modify essentially the paramagnetic behavior of χmol (T ) in the temperature range below 300 K. The paramagnetic susceptibility of [{Mn(hfac)2 }3 (TNOP)2 ] was examined by a model in which the triradicals were assumed to form 1D-ferrimagnetic chains with the Mn(1) ions in positions 1a and 1b (Fig. 25), while the Mn(2) ions in positions 2 link them through the exchange interaction with the middle nitroxide group of (TNOP). This assumption means that the exchange interaction between Mn(1) and the terminal nitroxide group is substantially stronger than the interaction between Mn(2) and the middle nitroxide group of (TNOP).

102

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

As a matter of fact, the [{Mn(hfac)2 }3 (TNOP)2 ] complex is not a true 1Dchain compound because the magnetic contribution of the Mn(2) ions linking the . . .–Mn(1)–(TNOP)–Mn(1)–. . . chains can be neither neglected nor considered as a kind of paramagnetic impurity. Moreover, the 1D-chains themselves have a four-spin periodicity which does not allow to perform any exhaustive analysis by the use of existing analytical expressions derived for ferrimagnetic chains with two-spin periodicity [60, 88, 115–119]. Therefore the approach applied to interpret the paramagnetic susceptibility of [{Mn(hfac)2 }3 (TNOP)2 ] contained some simplifications. This complex on the whole was considered as a two sublattice ferrimagnet formed by isolated Mn(2) ions and . . .–Mn(1)–(TNOP)–Mn(1)–. . . chains with a positive intersublattice exchange interaction. Then, in the molecular-field approximation, the low-field paramagnetic susceptibility of this complex can be written in the conventional form: χmol =

T2

(CMn + Cch )T + CMn Cch (2λ − λMn − λch ) − (CMn λMn + Cch λch )T − CMn Cch [(λ )2 − λMn λch ]

(6)

where λ is the intersublattice molecular-field coefficient, λMn and λch are the intrasublattice molecular-field coefficients for the Mn(2) and 1D-chain sublattices and: CMn = N

g 2 µ2B SMn (SMn + 1) and Cch = χch T 3kB

(7)

are, accordingly, the Curie constants of the Mn(2) and chain sublattices. In Eqs. (6) and (7) the intrachain exchange interaction is included in χch , which is hence a temperature dependent quantity named “constant” for convenience only. To calculate the temperature-dependence of χch T , the . . .–Mn(1)–(TNOP)– Mn(1)–. . . chain was approximated by a model in which molecular species having stable spins in the temperature region up to 300 K were isolated. According to the chain structure, two possible configurations were considered: (i) a ferrimagnetic (5/2 − 3/2) chain formed by nitroxide radicals (TNOP) with SR = 3/2 antiferromagnetically coupled with Mn(1) and (ii) a ferrimagnetic (3/2 − 1/2) chain formed by the trimeric spin species made up of one Mn(1) ion and two terminal nitroxide groups of different triradicals (STR = 3/2) antiferromagnetically coupled with the middle nitroxide group spin (S = 1/2). The (5/2 − 3/2) configuration was analyzed in the Heisenberg classical-classical spin approximation [88]. No satisfactory fit was possible with negative 2Jch /kB and positive λ values. The determination of the exchange interaction parameters for the (3/1 − 1/2) configuration was made in the quantum-classical chain approximation by using the analytical expression for the paramagnetic susceptibility of a ferrimagnetic chain derived by Seiden [60, 115]: (χch T )Qu

N µ2B 2 3 STR (STR + 1) + + = 3kB 4 1 − P(γ )

2 · STR (STR + 1)P(γ ) − STR Q(γ ) + 0.25Q (γ )

(8)

2.8 Three-dimensional Metal-Aminoxyl Systems

103

Table 7. Exchange parameters of [{Mn(hfac)2 }3 (TNOP)2 ]. The number of nearest neighbors for the intersublattice exchange interaction 2J /kB is taken as 2 and 6, respectively. Model

2Jch /kB [K]

λ [emu mol−1 ]

2J /kB [K]

JTR /kB [K]

Class.-Class. Quant.-Class. Quant.-Class.

−900.30 −520.20 −520.20

+5.90 +5.20 +5.10

+4.4 +3.9 +3.8

STR = 3/2 STR = 3/2 −350

where γ = −

and

2Jch STR , kB T

P(γ ) =

(1 + 12γ −2 ) sinh γ − (5γ −1 + 12γ −3 ) cosh γ − γ −1 + 12γ −3 , sinh γ − γ −1 cosh γ + γ −1

Q(γ ) =

(1 + 2γ −2 ) cosh γ − 2γ −1 sinh γ − 2γ −2 sinh γ − γ −1 cosh γ + γ −1

The best fits were found near the zero values for χMn and χch and the final fit was made with two variables, 2Jch /kB and λ . In Fig. 29 the calculated and experimental temperature-dependencies of χmol T for [{Mn(hfac)2 }3 (TNOP)2 ] are compared. They are in good agreement in a wide temperature range down to about 55 K. The corresponding parameters are given in Table 7. The results obtained show that the [{Mn(hfac)2 }3 (TNOP)2 ] complex is characterized by very strong intrachain interactions. The possibility to isolate trimeric spin species (1/2, 5/2, 1/2) in the . . .–Mn(1)–(TNOP)–Mn(1)–. . . chain indicates that the exchange interaction between Mn(1) and terminal nitroxide group exceeds essentially the interaction energy between the NO groups of the TNOP radical, which is characterized by the exchange integral 2J/kB = +480 K [33]. 2.8.2.3

Effect of Intra-trimer Interaction

Because of the strong intra-trimer exchange interaction the model with fixed STR value seems to be applicable in the temperature range up to 300 K. However, this approach will fail at temperatures higher than the intra-trimer interaction parameter JTR /k. To reveal the role of the intra-trimer interaction in the temperaturedependence of χmol T , a fitting procedure was made with STR replaced by the effec3kB tive moment of the (1/2, 5/2, 1/2) trimer µ2TR = (χTR T ) in Eq. (8). Here N g 2 µ2B χTR T for an isolated trimer is given by Eq. (4). The temperature variation of χmol T of [{Mn(hfac)2 }3 (TNOP)2 ] appeared to be unaffected by the intra-trimer exchange interaction. In fact, a change of the fit parameters lies within the accuracy of the procedure (see Table 7). Hence |JTr |/kB was estimated to be larger than 350 K.

104

2.8.3

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

Conclusion

It is concluded that the perfect 3D ferro/ferrimagnet with heterospin ferromagnetic (J1 > 0)-antiferromagnetic (J2 < 0) network is realized in [{Mn(hfac)2 }3 (TNOP)2 ]. Magnetic properties of the ferrimagnetic 3D metal-radical complex [{Mn(hfac)2 }3 (TNOP)2 ] can adequately be described in the Heisenberg exchange approximation assuming that two magnetic sublattices, 1D ferrimagnetic . . .–Mn(1)–(TNOP)–Mn(1)–. . . chains with four-spin periodicity and Mn(2) ions, form a collinear ferrimagnetic structure with a positive exchange coupling. Due to a strong exchange interaction between Mn(1) and terminal NO groups within the chain, it can be considered approximately as a two-spin ferrimagnetic chain made up of middle nitroxide groups (S = 1/2) antiferromagnetically coupled with the trimer spin species having spins S = 3/2.

2.9

Summary and Prognosis

The supramolecular approaches have been successfully applied to the construction of high-dimensional network structures by using magnetic metal ions with organic free radicals serving as bridging ligands. These extended structures are difficult to construct by using only covalent bonds. The method employed takes advantage of the intermolecular bonding interactions considerably weaker than conventional covalent bonds, mostly coordination bonds plus hydrogen bonds, hydrophobic interactions, van der Waals forces, etc. The products are formed under thermodynamic control rather than kinetic control. The magnetic structures correspond nicely to the crystal structure and network structures have been successfully used for making molecular-based magnets in which the spins order at finite temperature. Elsewhere such network structures are intended for making switches, host molecular cages like “organic zeolites” and enzymatic pockets, electric conductors, non-linear optical materials, etc. As far as the magnetic properties are concerned, the exchange coupling through these weaker bonds are generally rather weak. For the coupling to be strong, the unpaired electrons have to be bound to each other through one or two σ -bonds or several π-bonds. Thus the supramolecular approaches have been limited to metal coordination compounds. Design of appropriate free radical ligands, i. e., high-spin oligo-aminoxyl radicals as bridging ligands is a key to our strategy. The number and the configuration of the coordination sites in the free radical ligands control the dimensions of the magnetic structures of these metal–radical ligand complexes. A sophistication of the design of new high-spin bridging ligands should lead to magnetic materials having higher TC . Furthermore, a clear-cut one-to-one correspondence has been found between tacticity of the extended molecular structures and dimension of the crystal structures; while the isotactic polymeric chains remain one-dimensional and are difficult to have strong interchain interactions, as in [Mn(hfac)2 BNO], [Mn(hfac)2 TNOP] · nC6 H14 ]and [Mn(hfac)2 TNOPB], the syndiotactic chains have a tendency to grow

References

105

into 2D ([{Mn(hfac)2 }3 (TNOPB)2 ] · n-C7 H16 ) and 3D ([{Mn(hfac)2 }3 (TNOP)2 ]) networks by extending interchain connectivity. The ligands employed so far are conformationally labile and chiral only in the crystals. The right chirality of each ligand and the consequent tacticity have been selected during the self-assembling and crystallization processes. Furthermore, isotactic chains of opposite chirality cancel each other out and there is no net chirality exhibited by the bulk crystals. Ongoing studies should employ ligands stable with respect to chirality to dictate the dimension of the resulting metal complexes. Once such crystals are obtained, they might become chiral magnets that would show interesting photophysical behavior. The usefulness of the heterospin systems as a versatile design strategy for high TC molecule-based magnetic materials would be increased by incorporation of the chiral aspect into the complex structures.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]

H. Iwamura, K. Inoue, T. Hayamizu, Pure Appl. Chem. 1996, 68, 243–252. H. Iwamura, J. Phys. Org. Chem. 1998, 15, 299–304. H. Iwamura, Adv. Phys. Org. Chem. 1990, 26, 179–253. R. Chiarelli, M. A. Novak, A. Rassat, J. L. Tholence, Nature 1993, 363, 145. M. Kinoshita, P. Turek, M. Tamura, K. Nozawa, D. Shiomi, Y. Nakazawa, M. Ishikawa, M. Takahashi, K. Awaga, T. Inabe, Y. Maruyama, Chem. Lett. 1991, 1225. J. Schweizer, B. Gillon, in P. M. Lahti (Ed.): Magnetic Properties of Organic Materials, Marcel Dekker, Basel, Switzerland 1999, p. 449–473. Y. Y. Lim, R. S. Drago, Inorg. Chem. 1972, 11, 1334–1338. P. M. Boymel, G. R. Eaton, S. S. Eaton, Inorg. Chem. 1980, 19, 727–735. P. M. Boymel, G. A. Barden, G. R. Eaton, S. S. Eaton, Inorg. Chem. 1980, 19, 735–739. P. M. Bilgren, G. A. Barden, G. R. Eaton, S. S. Eaton, Inorg. Chem. 1985, 24, 4268–4272. R. S. Drago, T. C. Kuechler, M. Kroeger, Inorg. Chem. 1979, 18, 2337–2342. M. H. Dickman, R. J. Doedens, Inorg. Chem. 1981, 20, 2677–2681. M. H. Dickman, L. C. Porter, R. J. Doedens, Inorg. Chem. 1986, 25, 2595. S. S. Eaton, G. R. Eaton, Coord. Chem. Rev. 1978, 26, 207–262. S. S. Eaton, G. R. Eaton, Coord. Chem. Rev. 1988, 83, 29–72. O. P. Anderson, T. C. Kuechler, Inorg. Chem. 1980, 19, 1417–1422. A. Bencini, C. Benelli, D. Gatteschi, D. Zanchini, D. Carnegie, J. Am. Chem. Soc. 1984, 106, 5813–5818. A. Bencini, C. Benelli, D. Gatteschi, C. J. D. W., R. L. Carlin, J. Am. Chem. Soc. 1985, 107, 2560–2561. C. T. Cazianis, E. R. Eaton, Can, J. Chem. 1974, 52, 2454–2462. E. F. Ullman, J. H. Osiecki, D. G. B. Boocock, R. Darcy, J. Am. Chem. Soc. 1972, 94, 7049. V. I. Ovcharenko, A. B. Gel’man, V. N. Ikorskii, Zh. Strukt. Khim. 1989, 30, 142. J. F. W. Keana, Chem. Rev. 1978, 78, 37. L. B. Volodarsky, V. A. Reznikov, V. I. Ovcharenko, Synthetic Chemistry of Stable Nitroxides, CRC Press, Coca Raton 1994.

106

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals

[24] (a) D. Luneau, J. Laugier, P. Rey, G. Ulrich, R. Ziiessel, P. Legoll, M. Drillon, J. Chem. Soc., Chem. Commun. 1994, 741–742; (b) A. Caneschi, F. Ferraro, D. Gatteschi, P. Rey, R. Sessoli, Inorg. Chem. 1988, 29, 4217. [25] P. F. Richardson, R. W. Kreilick, J. Am. Chem. Soc. 1977, 99, 8183. [26] H. Oshio, J. Chem. Soc., Dalton Trans. 1999, 2641–2643. [27] P. T. Matsunaga, d. T. McCall, M. D. Carducci, R. J. Doedens, Inorg. Chem. 1990, 29, 1655. [28] A. Rassat, Pure & Appl. Chem. 1990, 62, 223–227. [29] H. C. Longuet-Higgins, J. Chem. Phys. 1950, 18, 265–274. [30] M. S. Platz, in W.T. Borden (Ed.): Diradicals, Wiley, New York 1982, p. 195–258. [31] S. Nakazono, S. Karasawa, N. Koga, H. Iwamura, Angew. Chem., Int. Ed. Engl. 1998, 37, 1550–1552. [32] A. Calder, A. R. Forrester, P. G. James, G. R. Luckhurst, J. Am. Chem. Soc. 1969, 91, 3724–3727. [33] T. Ishida, H. Iwamura, J. Am. Chem. Soc. 1991, 113, 4238–4241. [34] F. Kanno, K. Inoue, N. Koga, H. Iwamura, J. Phys. Chem. 1993, 97, 13267–13272. [35] T. Mitsumori, K. Inoue, N. Koga, H. Iwamura, J. Am. Chem. Soc. 1995, 117, 2467–2478. [36] K. Inoue, H. Iwamura, Adv. Mater. 1996, 8, 73–76. [37] K. Oniciu, K. Matsuda, H. Iwamura, J. Chem. Soc. Perkin II 1996, 907–913. [38] H. Iwamura, N. Koga, Pure Appl. Chem. 1999, 71, 231–238. [39] A. Caneschi, D. Gatteschi, R. Sessoli, P. Rey, Acc. Chem. Res. 1989, 22, 392–398. [40] A. Caneschi, D. Gatteschi, J. Laugier, L. Pardi, R. P., Inorg. Chem. 1988, 27, 1031. [41] D. Gatteschi, P. Rey, in P. M. Lahti (Ed.): Magnetic Properties of Organic Materials, Marcel Dekker, Basel, Switzerland 1999, p. 601–627. [42] R. N. Mushin, P. V. Schastnev, S. A. Malinovskaya, Trends in Applied Theoretical Chemistry, Kluwer, Dordtrecht 1992. [43] R. N. Mushin, P. V. Schastnev, S. A. Malinovskaya, Inorg. Chem 1992, 31, 4118. [44] L. C. Porter, R. J. Doedens, Inorg. Chem. 1985, 24, 1006–1010. [45] L. C. Porter, M. H. Dickman, R. J. Doedens, Inorg. Chem. 1986, 25, 678–684. [46] D. Gatteschi, J. Laugier, P. Rey, C. Zanchini, Inorg. Chem. 1987, 26, 938. [47] A. Caneschi, A. Grand, L. Laugier, P. Rey, R. Subra, J. Am. Chem. Soc. 1988, 110, 2307. [48] M. M. Rohmer, A. Grand, M. Bernard, J. Am. Chem. Soc. 1990, 112, 2875. [49] N. V. Pervukhina, N. V. Podlberezskaya, V. I. Ovcharenko, S. V. Larionov, Zh. Strukt, Khim. 1991, 32, 123. [50] A. Caneschi, D. Gatteschi, R. Sessoli, S. K. Hoffman, J. Laugier, P. Rey, R. Sessoli, C. Zanchini, J. Am. Chem. Soc. 1988, 110, 2795. [51] A. Caneschi, D. Gatteschi, J. Laugier, L. Pardi, R. P., Inorg. Chem. 1988, 27, 2390. [52] P. Rabu, M. Drillon, H. Iwamura, G. Goerlitz, T. Itoh, K. Matsuda, N. Koga, K. Inoue, Eur. J. Inorg. Chem. 2000, 211–216. [53] G. Goerlitz, T. Hayamizu, T. Itoh, K. Matuda, H. Iwamura, Inorg. Chem. 1998, 37, 2083– 2085. [54] M. Kitano, Y. Ishimaru, K. Inoue, N. Koga, H. Iwamura, Inorg. Chem. 1994, 33, 6012– 6019. [55] K. Inoue, H. Iwamura, J. Chem. Soc., Chem. Commun. 1994, 2273–2274. [56] K. Inoue, H. Iwamura, Synth. Met. 1995, 71, 1791–1794. [57] K. Inoue, T. Hayamizu, H. Iwamura, Mol. Cryst. Liq. Cryst. 1995, 273, 67–80. [58] K. Inoue, T. Hayamizu, H. Iwamura, Chem. Lett. 1995, 745–746. [59] J. C. Bonner, M. E. Fisher, Phys. Rev. A 1964, 135, 640–658. [60] E. Coronado, M. Drillon, R. Georges, in C. J. O’Connor (Ed.): World Scientific Publishing Co. Ptc. Ltd 1993, p. 27–66.

References

107

[61] O. Kahn, Molecular Magnetism, VCH publishers, Inc., New York 1993. [62] Scaling, Scaling factors employed here contains all the errors due to the sample purity, weighing the samples, calibration of SQUID susceptometer, etc. in addition to the deviation of the g factor from 2.00 as assumed in this study. If the last term is predomiant, f = g2 and, therefore, f = 0.954 corresponds to a g factor of 1.95 (= 2 × 0.954). [63] Y. Ishimaru, Chem. Lett. 1994, 1693–1696. [64] H. Iwamura, New J. Chem. 1998, 201–210. [65] F. Iwahori, K. Inoue, H. Iwamura, J. Am. Chem. Soc. 1999, 121, 7264–7265. [66] J. Laugier, P. Rey, C. Benelli, D. Gatteschi, C. Zanchini, J. Am. Chem. Soc. 1986, 108, 6931. [67] C. Benelli, A. Caneschi, D. Gatteschi, J. Laugier, P. Rey, in P. Delhaes, M. Drillon (Eds.): Organic and Inorganic Low Dimensional Crystalline Materials, Plenum, New York 1987. [68] A. Caneschi, D. Gatteschi, J. Laugier, P. Rey, J. Am. Chem. Soc. 1987, 109, 2191. [69] A. Caneschi, D. Gatteschi, P. Rey, R. Sessoli, Inorg. Chem. 1988, 27, 1756. [70] A. Caneschi, D. Gatteschi, J. P. Renard, P. Rey, R. Sessoli, Inorg. Chem. 1989, 28, 2940. [71] C. Benelli, A. Caneschi, D. Gatteschi, L. Pardi, P. Rey, Inorg. Chem. 1989, 28, 275. [72] A. Caneschi, D. Gatteschi, J. P. Renard, P. Rey, R. Sessoli, Inorg. Chem. 1989, 28, 2940. [73] A. Caneschi, D. Gatteschi, J. P. Renard, P. Rey, R. Sessoli, Inorg. Chem. 1989, 28, 3314. [74] A. Caneschi, D. Gatteschi, J. P. Renard, P. Rey, R. Sessoli, J. Am. Chem. Soc. 1989, 111, 785. [75] A. Caneschi, D. Gatteschi, J. P. Renard, P. Rey, R. Sessoli, Inorg. Chem. 1989, 28, 1976. [76] C. Benelli, A. Caneschi, D. Gatteschi, L. Pardi, P. Rey, Inorg. Chem. 1990, 29, 4223. [77] C. I. Cabello, A. Caneschi, R. K. Carlin, D. Gatteschi, P. Rey, R. Sessoli, Inorg. Chem. 1990, 29, 2582. [78] A. Caneschi, D. Gatteschi, R. Sessoli, C. I. Cabello, P. Rey, A. L. Barra, L. C. Brunel, Inorg. Chem. 1991, 30, 1882. [79] H. Iwamura, K. Inoue, N. Koga, T. Hayamizu, in O. Kahn (Ed.): Magnetism: A Supramolecular Function, Vol. C484 1996, p. 157–179. [80] K. Inoue, H. Iwamura, Mol. Cryst. Liq. Cryst. 1996, 286, 133. [81] K. Inoue, H. Iwamura, Proc., Mater. Res. Soc. 1996, 413, 313. [82] K. Inoue, F. Iwahori, A. S. Markosyan, Y. Hosokoshi, H. Iwamura, Coord. Chem. Rev. 2000, 198, 219. [83] H. Kumagai, K. Inoue, Angew. Chem. Int. Ed. 1999, 38, 1601. [84] F. Iwahori, K. Inoue, (to be published) . [85] A. S. Markosyan, H. Iwamura, K. Inoue, Mol. Cryst., Liq. Cryst. 1999, 334, 549. [86] A. S. Markosyan, H. Iwamura, K. Inoue, Mol. Cryst. Liq. Cryst. 1999, 334, 549–568. [87] R.L. Carlin, Magnetochemistry, Springer, Berlin and Heidelberg, Germany 1986. [88] E. Coronado, M. Drillon, P. R. Nugteren, L. J. de Longh, D. Beltran, R. Georges, J. Am. Chem. Soc. 1989, 111, 3874. [89] P. M. Richards, Phys. Rev. 1989, B 10, 4687. [90] F. Iwahori, A. S. Markosyan, K. Inoue, (unpublished results) . [91] R. Hoogerbeets, A. J. van Duyneveldt, Physica 1989, B 121, 233. [92] K. Nagata, I. Yamamoto, H. Takano, Y. Yokozawa, J. Phys. Soc. Jpn. 1977, 43, 857. [93] K. Nagata, Y. Tazuke, J. Phys. Soc. Jpn. 1977, 43, 337. [94] H. E. Stanley, Introduction to Phase Transitions and Critical Phenomena, Clarendon Press, Oxford, England 1971. [95] C. Espie, J. Laugier, R. Ramasseul, A. Rassat, P. Rey, Nouv. J. Chim. 1980, 4, 205. [96] J. C. Espie, R. Ramasseul, A. Rassat, P. Rey, Bull. Soc. Chim. Fr. 1983, 2, 33. [97] J. Laugier, R. Ramasseul, P. Rey, J. C. Espie, A. Rassat, Nouv. J. Chim. 1983, 7, 11. [98] C. Benelli, D. Gatteschi, C. Zanchini, J. M. Latour, P. Rey, Inorg. Chem. 1986, 25, 4242.

108 [99] [100] [101] [102] [103] [104]

[105]

[106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119]

2 Magnetic Ordering in Metal Coordination Complexes with Aminoxyl Radicals R. Briere, A. M. Giroud, A. Rassat, P. Rey, Bull. Soc. Chim. Fr. 1980, 2, 147. M. Drillon, A. Grand, P. Rey, Inorg. Chem. 1990, 29, 771. A. Grand, P. Rey, R. Subra, Inorg. Chem. 1983, 22, 391. V. N. Ikorskii, V. I. Ovcharenko, Zh. Neorg. Khim. 1990, 35, 2093. V. I. Ovcharenko, S. V. Larionov, V. K. Mohosoeva, L. B. Volodarsky, Zh. Neorg. Khim. 1983, 28, 151; Chem. Abstr. 1983, 98, 118506w. V. I. Ovcharenko, V. N. Ikorskii, K. E. Vostrikova, A. B. Burdukov, G. V. Romanenko, N. V. Pervukhina, N. V. Podberezskaya, Izv. Sib. Otd. Akad. Nauk S. S. S. R., Ser. Khim. Nauk 1990, 5, 100. N. V. Pervukhina, V. N. Ikorskii, N. V. Podberezskaya, P. S. Nikitin, A. B. Gel’man, V. I. Ovcharenko, S. V. Larionov, V. V. Bakakin, Zh. Strukt. Khim. 1986, 27, 61; Chem. Abstr. 1987, 106, 11395r. A. Caneschi, D. Gatteschi, M. C. Melandri, P. Rey, R. Sessoli, Inorg. Chem. 1990, 29, 4228. A. Caneschi, D. Gatteschi, R. Sessoli, Inorg. Chem. 1993, 32, 4612. K. Inoue, H. Iwamura, J. Am. Chem. Soc. 1994, 116, 3137. H. M. McConnell, J. Chem. Phys. 1963, 39, 1910. A. Izuoka, S. Murata, T. Sugawara, H. Iwamura, J. Am. Chem. Soc. 1987, 109, 2631. A. S. Markosyan, Y. Hosokoshi, K. Inoue, Phys. Lett. 1999, A 261, 212–216. A. S. Markosyan, T. Hayamizu, H. Iwamura, K. Inoue, J. Phys. Condens. Matter 1998, 10, 2323. K. Yamaji, J. Kondo, J. Phys. Soc. Jpn. 1967, 23, 516. R. Navarro, in L. J. de Jongh (Ed.): Magnetic Properties of Layered Compounds, Kluwer Acad. Publ. 1990. J. Seiden, J. Phys. (Paris) 1983, 44, L-947. M. Drillon, E. Coronado, D. Beltran, R. Georges, Chem. Phys. 1983, 79, 449. M. Verdaguer, A. Gleizes, J. P. Renard, J. Seiden, Phys. Rev. 1984, B 29, 5144. Y. Pei, O. Kahn, J. Sletten, J. P. Renard, R. Georges, J. C. Gianduzzo, J. Curely, Q. Xu, Inorg. Chem. 1988, 27, 47. X. Qiang, J. Darriet, J. L. Soubeyroux, R. J. Georges, J. Magn. Magn. Mater. 1988, 74, 219.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

3

Organic Kagome Antiferromagnets Kunio Awaga, Nobuo Wada, Isao Watanabe, and Tamotsu Inabe

3.1

Introduction

Geometrical frustration in antiferromagnetic systems with triangular coordination symmetry has recently been of interest in physics. In such a triangle the two nearest neighbors to a given spin are themselves nearest neighbors and antiferromagnetic coupling among them cannot be completely satisfied. This frustration prevents long-range magnetic order from being established and enables novel kinds of lowtemperature magnetic state to develop [1–3]. The Heisenberg Kagome (named after a form of Japanese basket weaving [4]) antiferromagnet, whose lattice is shown in Scheme 1, is one of the most interesting of these frustrated systems.

Scheme 1

Whereas the number of the nearest neighbors is six in the simple triangular lattice, it is reduced to be four in the Kagome lattice. This enables more freedom for the alignment of the magnetic moments on the Kagome lattice. A possible ground state for the antiferromagnetic classical spins on a regular triangular lattice is the so-called 120◦ structure (Scheme 2a), in which the magnetic moments are all parallel to the triangular plane and the neighboring moments make an angle of 120◦ with each other. √ √ The magnetic moments in the same direction form a 3 × 3 superlattice whose unit is indicated by the broken lines. The spin alignment between the two neighbors is energetically disadvantageous to the simple antiparallel one, but the net moment becomes zero. The 120◦ structure can result in long-range magnetic order with the periodicity in the simple triangular Heisenberg antiferromagnets, indicating that it is a unique solution to spin frustration. Although it is possible to write down a coplanar

110

3 Organic Kagome Antiferromagnets

Scheme 2

120◦ structure on the Kagome lattice, as shown in Scheme 2b, the reduced number of nearest neighbors in this lattice enables non-coplanar 120◦ spin orientation, or a so-called paper Origami structure [5, 6]. Scheme 2c shows the internal spin space folded in two along the broken line. On each plane the magnetic moments have the 120◦ structure, but the moments on the different planes are not coplanar. Because continuous spin folding takes place with no energy cost, it is predicted that the classical Kagome antiferromagnet has rich, non-trivial ground-state degeneracy. The actual ground state might be governed by subtle effects, such as a quantum effect, a single-ion magnetic anisotropy, next-nearest-neighbor interactions, and so on. For these, the Kagome antiferromagnet is theoretically expected to have characteristic features – non-Neel ´ states, residual entropy at absolute zero temperature, and so on.

3.2

Inorganic Kagome Antiferromagnets

Despite the interest in the Kagome antiferromagnets, few antiferromagnetic Kagome systems have yet been studied, owing to the lack of suitable model compounds. They had, in addition, been limited to inorganic materials, before we discovered the organic Kagome antiferromagnet described in Section 3. In this section we will briefly review inorganic Kagome antiferromagnets.

3.2.1

SrGa12-x Crx O19 (SCGO(x))

This is a quasi-two-dimensional oxide containing antiferromagnetically interacting S = 3/2 Cr3+ ions in Kagome bilayers [7]. Although the antiferromagnetic Weiss constant is between −200 and −500 K, depending on the Cr concentration, there is no long-range magnetic order down to at least 1.5 K [8]. Instead hysteretic behavior, indicative of a spin-glass transition, is observed at 3.5 K [9]. Neutron scattering [8]

3.3 Organic Kagome Antiferromagnet, m-MPYNN·X

111

and muon spin rotation (µSR) [10] measurements indicate that the spins are not strictly frozen at these low temperatures with short-range antiferromagnetic correlation. Single-crystal magnetic susceptibility measurements reveal strong magnetic anisotropy at the spin-glass-like transition, suggesting that the magnetization component normal to the Kagome planes freezes completely, whereas the component parallel to the planes does not [11]. SCGO(x) suffers from a severe disadvantage as a model material, however, in that the coverage of the Kagome lattice sites is significantly less than 100% and, depending on the method of preparation, ranges from 88–95%. A significant proportion of moments also reside on a triangular lattice.

3.2.2

+ Jarosite, AM3 (OH)6 (SO4 )2 (A = Na+ , K+ , Rb+ , Ag+ , NH+ 4 , H3 O , 3+ 3+ etc., and M = Fe or Cr )

The structures of jarosite materials belong to the rhombohedral space group R3m, and are identified as a Kagome lattice in which MO6 octahedra are linked through their vertices in layers that are well-separated by hydroxide, sulphate, and hydronium groups [12, 13]. Fe-jarosite, in which the Fe3+ ions are in the ground state of 6 A1g and nearest-neighbor exchanges can approximate well to Heisenberg symmetry, has been studied extensively by magnetic susceptibility [13, 14], Mossbauer ¨ [13, 14], neutron-scattering [15] and µSR [16] measurements. Long-range magnetic order is observed below TN ≈ 50 K for the compounds in which A = K+ , Na+ , etc., with the 120◦ structure [13–16], whereas the compound with A = H3 O+ has an anomaly, characteristic of spin-glass freezing, in the magnetic susceptibility at 17 K [17]. Besides these two examples, the spin lattices in La4 Cu3 MoO12 [18] and RCuO2.66 [19] are reported to identical with the Kagome lattice. The 2D solid 3 He adsorbed on graphite has also been studied as an antiferromagnetic system [20].

3.3 3.3.1

Organic Kagome Antiferromagnet, m-MPYNN·X Crystal Structure

We recently discovered that crystals of m-N -methylpyridinium nitronylnitroxide (abbreviated m-MPYNN) have an antiferromagnetic Kagome lattice [21]. m-MPYNN · I was obtained by reaction of m-pyridyl nitronylnitroxide and methyl iodide. Recrystallization of m-MPYNN·I with the presence of excess TBA·A (TBA = tetrabutylammonium and A = BF4 , ClO4 ) gave a crystalline solid solution, m-MPYNN·Ax ·I1−x . The reaction of equivalent amounts of m-MPYNN·I and Ag·A resulted in immediate precipitation of Ag·I, leaving iodide-free m-MPYNN·A in the solution. Recrystallization of m-MPYNN·I, m-MPYNN·Ax ·I1−x , and m-MPYNN·A from their acetone solutions resulted in hexagonal single crystals containing one acetone molecule per three m-MPYNN. The crystals of the simple iodide salt, m-MPYNN·I·(1/3)(acetone),

112

3 Organic Kagome Antiferromagnets

Fig. 1. (a) Organic 2D layer of m-MPYNN projected on to the ab plane. (b) Bond-alternated hexagonal lattice; J1 and J2 are the intradimer and interdimer magnetic interactions, respectively. (c) Kagome lattice.

were not stable – in air they immediately turned into powder as a result of evaporation of the solvent of crystallization – whereas the crystals containing BF4 or ClO4 were stable. The structure of m-MPYNN·X·(1/3)(acetone) (X = I, BF4 , ClO4 , etc.) belongs to a trigonal space group. The m-MPYNN molecules exist as a dimer, and the dimer units form a 2D triangular lattice parallel to the ab plane. Fig. 1a shows a projection of the organic layer of m-MPYNN on to the ab plane. The radical dimer is located on each side of the triangles – in the other words the m-MPYNN molecules form a bondalternated hexagonal lattice, as schematically shown in Fig. 1b. In the intradimer arrangement there is a very short intermolecular, interatomic distance of less than 3 Å between the NO group and the pyridinium ring. This short contact is probably caused by an electrostatic interaction between the positive charge on the pyridinium ring and the negative charge polarized on the oxygen atom. In the interdimer arrangement, on the other hand, there is weak contact between the NO groups. The NO–NO contact means overlap between the singly-occupied molecular orbitals (SOMOs), which always contributes to antiferromagnetic coupling.

3.3 Organic Kagome Antiferromagnet, m-MPYNN·X

113

Fig. 2. Side view of the organic 2D layers. Nine m-MPYNN dimers are drawn on the surface of the hexagonal prism.

Figure 2 shows a side view of the trigonal lattice, where nine units of the mMPYNN dimers on the surface of the hexagonal prism are drawn. The unit cell includes two organic layers at the height z = 0 and z = 1/2, between which there is a big separation. One-third of the anions are in the organic layer, joining the m-MPYNN molecules; those remaining are between the layers, compensating the excess positive charge in the organic layers. The crystal solvent, acetone, is located between the organic layers at the center of the triangle.

3.3.2

Magnetic Susceptibility

The temperature-dependence of the paramagnetic susceptibility, χp , for mMPYNN·BF4 ·(1/3)(acetone) is shown in Fig. 3, in which χp T is plotted as a function of temperature.

Fig. 3. Temperature-dependence of the paramagnetic susceptibilities χp for mMPYNN · BF4 · (1/3)(acetone).

114

3 Organic Kagome Antiferromagnets

Fig. 4. Temperature-dependence of the ac magnetic susceptibilities, χac , for oriented single crystals of m-MPYNN · BF4 ·(1/3)(acetone).

The value of χp T increases as the temperature is reduced from room temperature to ca. 10 K, indicating that the intradimer coupling, J1 , is ferromagnetic. After passing through a maximum near 10 K, χp T decreases rapidly, suggesting that the interdimer magnetic interaction, J2 , is antiferromagnetic. The observed temperaturedependence can be readily interpreted in terms of the ferromagnetic J1 and the antiferromagnetic J2 by use of Eq. (1): χ=

4C T {3 + exp(−2J1 /kB T )} − 4J2 /kB

(1)

where C is the Curie constant and kB the Boltzmann constant. The derivation of Eq. (1) is described elsewhere [21]. The solid curve in Fig. 3 is the theoretical best fit to the data, obtained with J1 /kB = 11.6 K and J2 /kB = −1.6 K. Below |J1 |/kB K the radical dimer can be regarded as a spin-1 Heisenberg spin located at the mid point of each side of the triangle, as shown in Fig. 1c. It is expected that the interdimer antiferromagnetic coupling J2 gives rise to spin frustration among the triplet spins. The spin lattice in Fig. 1c is exactly coincident with the Kagome lattice. Therefore, the magnetic system in this organic material will he characterized as a spin-1 Kagome antiferromagnet in the temperature range below |J2 |/kB K. Figure 4 shows the temperature-dependence of the ac magnetic susceptibilities, χac , for oriented single crystals of m-MPYNN·BF4 ·(1/3)(acetone) below 0.8 K [22]. The magnetic field was parallel to the c axis. The value increases with decreasing temperature down to 0.24 K, and, after passing through a maximum approaches zero at absolute zero. We have confirmed no magnetic anisotropy in this behavior. This clearly indicates that the ground state is not an antiferromagnetic ordered state but a spin-gap state. The low-temperature data is, in fact, a good fit to the gap equation:

χ=A kB T

f

exp − kB T

(2)

3.3 Organic Kagome Antiferromagnet, m-MPYNN·X

115

where A is a constant and is the magnetic gap. The parameter f depends on the density of the excited states against the excitation energy, but is fixed at 1 in this analysis [23]. The solid curve in Fig. 4 is the theoretical curve obtained with A = 0.52 and /kB = 0.25 K. Spin or gap states have been observed in spin Peierls systems [24] and Haldene gap systems [25], although these precedents were 1D magnetic systems. The ground states of the spin-frustrated inorganic systems described in Section 2 are either the 3D ordered state or the spin-glass state. As far as we are aware this is the first example of a spin-gap state resulting from spin frustration. It is worth noting here that Anderson predicted the so-called resonating valence bond (RVB) state on a triangular antiferromagnetic lattice, which brought about a spin gap [26]. It is possible that the ground state of this material can be characterized in terms of the RVB state.

3.3.3

Heat Capacity

The temperature-dependence of the heat capacity, cp , of m-MPYNN · BF4 · (1/3) (acetone) has been examined down to 0.12 K [22]. The results below 3 K are shown in Fig. 5. The value of cp gradually increases with decreasing temperature. After a broad maximum at 1.4 K, cp decreases. Below 0.24 K, where χac shows the spin-gap ground state, cp increases again. The temperature of maximum cp , 1.4 K, almost agrees with |J2 |/kB . Monte Carlo calculation indicated that the heat capacity of the spin-1/2 Kagome Heisenberg antiferromagnet has maximum short-range magnetic ordering at 1.4|J |/kB [27]. The anomaly observed at 1.4 K is probably because of the short-range ordering which results from J2 . The reason of the increase in cp below 0.24 K is not clear, but it suggests another anomaly below 0.1 K. It is notable that there is no signal indicative of long-range ordering in the temperature range down to 0.1 K, which is 8% of |J2 |/kB . This is indicative of the presence of spin frustration in this magnetic system. Plots of cp /T increase gradually as the temperature is reduced to 0.12 K (not shown). We calculated the entropy change accompanying the anomaly at 1.4 K to be (S = 4.7 J K−1 mol−1 , by using the data above 0.12 K and subtracting the contributions of the lattice and the excited state resulting from J1 . Because the triplet spins on the m-MPYNN dimers lose magnetic freedom as a result of

Fig. 5. Temperature-dependence of the heat capacities cp for m-MPYNN·BF4 ·(1/3)(acetone).

116

3 Organic Kagome Antiferromagnets

the short-range ordering, the magnetic entropy is theoretically calculated to be Sm = (R/2) ln 3 = 4.567 J K−1 mol−1 . The observed value of S is already larger than Sm , despite S being obtained by use of data acquired above 0.12 K. This means that there is an unknown amount of freedom besides the magnetic freedom; this cooperates with the short-range magnetic ordering in this ultra-low temperature range.

3.3.4

Positive Muon Spin Rotation

We performed µSR measurements on m-MPYNN·BF4 ·(1/3)(acetone) in the temperature range down to 30 mK, to clarify whether or not magnetic transitions occur, and to confirm the non-magnetic ground state [28]. Positive muon spin rotation (µ+ SR) is a good microscopic probe for monitoring such a magnetic state of the system. A muon spin is completely polarized along a beam direction, even in the zero-field (ZF) condition, and depolarized after the stop at a potential-minimum position in the samples, interacting with a local field at a muon site [29]. Long-range or shortrange ordering of the dimer spins can be recognized as a change in the depolarization behavior of the muon spin, because a static or a dynamically fluctuating component of the internal field which is accompanied by the magnetic transition strongly affects the muon spin polarization. Figure 6 shows ZF-µ+ SR time spectra obtained at 265, 100, and 2.9 K, and at 30 mK. In this figure the asymmetry parameter of the muon spin at time t, A(t), is defined as [F(t)− B(t)]/[F(t)+ B(t)], l, where F(t) and B(t) are muon events counted by the forward and backward counters, respectively. The asymmetry at each temperature was normalized to unity t = 0, to enable comparison of differences between depolarization behavior. The depolarization behavior cannot be described either by a simple Gaussian function or a Lorentzian function. For convenience the ZF-µ+ SR spectra obtained were analyzed by means of a power function, A0 exp(−λt)β , where A0 is the initial asymmetry at t = 0 and λ is the depolarization rate. The solid curves in Fig. 6 are the best fit results obtained by use of this power function. The temperature-dependence of A0 , λ, and β was obtained from the best fit analysis. All are independent of temperature below ca 100 K, showing that the static and

Fig. 6. ZF-µ+ SR time spectra obtained for m-MPYNN·BF4 ·(1/3)(acetone) at 265 K, 100 K, 2.9 K, and 30 mK.

3.3 Organic Kagome Antiferromagnet, m-MPYNN·X

117

dynamic properties of the local field at the muon site are temperature-independent. This is different from other types of Kagome magnet in which strong enhancement of the rate of polarization, by the critical slowing-down behavior of magnetic moments, is observed in the region of a magnetic transition temperature [10, 16, 30, 31]. The value of λ decreases slightly above 150 K, probably because of the motional narrowing effect indicating that the muon starts to diffuse through the crystal. The half-width of the distribution of the static internal field at the muon site, H , was estimated, from the longitudinal field-dependence of the µ+ SR time spectra (not shown), to be 10 ± 2 G. It is known that a muon implanted into a crystal which contains F− ions forms a strong FµF state by hydrogen-bonding. In this case, the distance between the F- ion and the muon is similar to a nominal F− ionic radius of 1.16 Å. Assuming the distance between the stopped muon and the 19 F-nucleus in m-MPYNN-BF4 is the nominal F− ionic radius, the dipole field of the 19 F-nucleus at the muon site is estimated to be ca 8.5 G, a value comparable with the H value obtained. Although the reason for the absence of muon spin precession observed in other fluorides [32] is still unclear, it can be concluded that the implanted muon is expected to stop near the F− ion and enter into hydrogen-bonding, and that the static internal field at the muon site originates from the 19 F-nuclear dipole field. In conclusion, the temperature-independent depolarization behavior resulting from the distributed static internal field induced by the 19 F-nuclear dipoles at the muon site was observed down to 30 mK. The width of the field distribution was 10 ± 2 G. No clear long-range magnetic ordering of the dimer spins was observed. Taking into account results from magnetic susceptibility measurements, it is concluded that the ground state of m-MPYNN·BF4 is non-magnetic.

3.3.5

Distorted Kagome Lattices

In this section, we will describe the structures and magnetic properties of mEPYNN·I and m-PPYNN·I, where m-EPYNN and m-PPYNN are m-N-ethyl- and m-N -propylpyridinium nitronylnitroxides, respectively, and compare them with those of m-MPYNN·I. The materials were obtained by the same procedure as for m-MPYNN·I. Recrystallization of m-EPYNN·I and m-PPYNN·I from acetone or acetone-benzene resulted in hexagonal single crystals. Results from elemental analysis were indicative of the chemical formulas m-EPYNN·I·1.5H2 O and mPPYNN·I·0.5H2 O. Table 1 shows the unit cell dimensions of the m-R-PYNN·I series, determined by use of X-ray diffraction data in the range 20◦ < 2θ < 25◦ . Whereas the crystal of m-MPYNN · I has a trigonal structure, m-EPYNN · I and m-PPYNN · I crystallize as monoclinic systems. It was, however, found that lattice transformations for m-EPYNN · I and m-PPYNN · I led to cell dimensions quite similar to those of m-MPYNN · I. A schematic comparison between the trigonal and monoclinic cell is shown in Scheme 3. The transformed lattice constants are a = 16.14(6) Å, b = 16.19(4) Å, c = 24.02(6) Å, α = 89.5(2)◦ , β = 90.1(2)◦ , γ = 119.9(1)◦ , and V = 5440(26) Å3 for m-EPYNN·I and a = 16.30(1) Å, b = 16.30(1) Å, c = 24.71(1) Å, α = 89.7(4)◦ ,

118

3 Organic Kagome Antiferromagnets

Table 1. Cell and magnetic properties of m-R-PYNN·I.

Structure a (Å) b (Å) c (Å) β (◦ ) V (Å3 ) Z J1 /kB J2 /kB

m-MPYNN·I

m-EPYNN·I

m-PPYNN·I

Trigonal 15.876(5) – 23.583(6) – 5147(3) 12

Monoclinic 16.14(6) 28.07(8) 24.02(5) 91.1(2) 10876(53) 24

Monoclinic 28.153(9) 16.407(14) 24.705(11) 90.25(3) 11411(11) 24

10.2 −1.6

9.6 −1.7

6.6 −1.0

Scheme 3

β = 90.1(4)◦ , γ = 119.5(3)◦ , and V = 5709(5) Å3 for m-PPYNN·I. The lattice parameters obtained, a, b, c, and V , are slightly larger than the corresponding ones for m-MPYNN·I, but the differences between them are very small. Although we could not complete full structural analyses of m-EPYNN·I and m-PPYNN·I, probably because of positional disorders of the iodide ion and the crystal solvent, it is expected that they have a slightly distorted Kagome lattice. The decrease in crystal symmetry means distortion of the equilateral triangle to an isosceles triangle. This will significantly affect the low-temperature magnetic properties, as described later. Figure 7 shows the temperature-dependence of χp T for the m-R-PYNN·I series. The plots show quite similar temperature-dependence – the value of χp T increases as temperature is reduced from room temperature to ca 10 K. After passing through a maximum χp T decreases rapidly. This temperature-dependence is well explained by Eq. (1). Strictly speaking, m-EPYNN·I and m-PPYNN·I should include two kinds of J2 , because of the distortion of the Kagome lattice in these materials. The distortion is, however, so small we can ignore the difference. The solid curves going through the plots for the three compounds in Fig. 7 are the theoretical best fits to the data, obtained with the parameters listed in Table 1. The values of J1 and J2 decrease systematically with extension of the N -alkyl chain, presumably because of expansion of the 2D lattice. The temperature-dependence of the ac susceptibility, χac , for the three compounds are shown in Fig. 8. Although m-MPYNN·I has the regular antiferromagnetic Kagome lattice, as has m-MPYNN·BF4 , the value of χac continues to increase down to 0.05 K with no evidence of the spin-gap state. This is probably because

3.3 Organic Kagome Antiferromagnet, m-MPYNN·X

119

Fig. 7. Temperature-dependence of the paramagnetic susceptibilities, χp , for m-MPYNN·I, mEPYNN·I and m-PPYNN·I.

Fig. 8. Temperature-dependence of the ac magnetic susceptibilities, χac , for m-MPYNN·I, mEPYNN·I, and m-PPYNN·I.

m-MPYNN·I is chemically unstable; evaporation of the solvent gradually takes place. The dependence can be explained by the Eq. (3):

χ=A kB T

f

exp − kB T

+

Cdef T

(3)

where the first term is the same as in Eq. (2) and the second term is for the Curie contribution of lattice defects. When the data are fit to Eq. (3), the values obtained are: A = 0.64, /kB = 0.25 K, and Cdef = 0.025 emu K mol−1 (6.6%). The values of A and are very close to the corresponding values for m-MPYNN·BF4 . The temperature-dependence of the distorted Kagome materials m-EPYNN·I and mPPYNN·I is similar, but their values of χac below 0.2 K are ca. five times larger than that for m-MPYNN·I. The values are too large to be explained by the contribution of lattice defects. It is considered that their behavior is intrinsic and the spin-gap state is readily collapsed by the small distortion of the Kagome lattice.

120

3.3.6

3 Organic Kagome Antiferromagnets

Summary

The crystal structures and magnetic properties of the m-R-PYNN-X series have been studied. In the crystal of m-MPYNN·X the ferromagnetic dimers formed the triangular lattice with weak interdimer antiferromagnetic coupling. The magnetic system can be regarded as a spin-1 Kagome antiferromagnet at low temperatures. Single-crystal EPR revealed the 2D Heisenberg character of the spin system The low-temperature magnetic behavior was indicative of the spin-gap ground state; this was possibly identical with the RVB state. The temperature-dependence of the heat capacity showed that short-range magnetic ordering resulted from interdimer antiferromagnetic interaction, but that there was no long-range ordering down to 0.1 K. It also suggested an unknown amount of freedom which cooperated with the short-range magnetic ordering. The detection of µ+ SR revealed the temperatureindependent depolarization behavior down to 30 mK. In the other words, this study was strongly indicative of the absence of long-range magnetic ordering of the dimer spins and the non-magnetic ground state. Extension of the N -alkyl chain in the m-RPYNN·I series resulted in distortion of the Kagome lattice, and consequent collapse of the spin-gap ground state.

Acknowledgment The authors would like to thank their co-workers (Masao Ogata, Tsunehisa Okuno, Akira Yamaguchi, Morikuni Hasegawa, Masahiro Yoshimaru, Wataru Fujita, Takeo Otsuka, Hideo Yano, Tatsuya Kobayashi, Seiko Ohira, and Hiroyuki Imai) for their important contributions to the work reported herein.

References [1] [2] [3] [4] [5] [6]

P. Fazekas and P.W. Anderson, Philos. Mag. 1974, 30, 423. X.G. Wen, F. Wilczek, and A. Zee, Phys. Rev. B 1989, 39, 11413. P. Chandra and P. Coleman, Phys. Rev. Lett. 1991, 66, 100. I. Syozi, Progr. Theor. Phys. 1951, 6, 306. I. Ritchey, P. Chandra, and P. Coleman, Phys. Rev. B 1993, 47, 15342. E.F. Shender, V.B. Cherepanov, P.C. Holdsworth, and A.J. Berlinsky, Phys. Rev. Lett. 1991, 70, 3812. [7] X. Obradors, A. Labarta, A. Isalgue, J. Tejada, J. Rodriguez, and M. Pernet, Solid State Commun. 1988, 65, 189. [8] C. Broholm, G. Aeppli, G. Espinosa, and A.S. Cooper, Phys. Rev. Lett. 1990, 65, 3173; S.-H. Lee, C. Broholm, G. Aeppli, T. G. Perring, B. Hessen, and A. Taylor, Phys. Rev. Lett. 1996, 76, 4424. [9] A.P. Ramirez, G.P. Espinosa, and A.S. Cooper, Phys. Rev. Lett. 1990, 64, 2070; Phys. Rev. B 1992, 45, 2505.

References

121

[10] Y.J. Uemura, A. Keren, K. Kojima, L.P. Le, G.M. Luke, W.D. Wu, Y. Ajiro, T. Asano, Y. Kiruyama, M. Mekata, H. Kikuchi, and K. Kakurai, Phys. Rev. Lett. 1994, 73, 3306. [11] P. Schiffer, A.P. Ramirez, K.N. Franklin, and S-W. Cheong, Phys. Rev. Lett. 1996, 77, 2085. [12] R. Wang, W.F. Brandley, and H. Steinfink, Acta Crystallogr. 1965, 18, 249. [13] M. Takano, T. Shinjo, and T. Takada, J. Phys. Soc. Jpn 1971, 30, 1049. [14] M. Takano, T. Shinjo, M. Kiyama, and T. Takada, J. Phys. Soc. Jpn 1968, 25, 902. [15] M.G. Townsent, G. Longworth and E. Roudaut, Phys. Rev. B 1986, 33, 4919. [16] A. Keren, K. Kojima, L.P. Le, G.M. Luke, W.D. Wu, Y.J. Uemura, M. Takano, H. Dabkowska, and M.J.P. Gingras, Phys. Rev. B 1996, 53, 6451. [17] A.S. Wills and A. Harrison, J. Chem. Soc., Faraday Trans. 1996, 92, 2161. [18] D.A.V. Griend, S. Boudin, V. Caignaert, K. Poeppelmeier, Y. Wang, V.P. Dravid, M. Azuma, M. Takano, Z. Ho, and J.D. Jorgensen, J. Am. Chem. Soc. 1999, 121, 4787. [19] M.D. Nunez-Regueiro, C. Lacroix and B. Canals, Phys. Rev. B, 1996, 54, 736. [20] Y.R. Wang, Phys. Rev. B 1992, 45, 12608, and references cited therein. [21] K. Awaga, T. Okuno, A. Yamaguchi, M. Hasegawa, T. Inabe, Y. Maruyama, and N. Wada, Phys. Rev. B, 1994, 49, 3975. [22] N. Wada, T. Kobayashi, H. Yano, T. Okuno, A. Yamaguchi, and K. Awaga, J. Phys. Soc. Jpn. 1997, 66, 961. [23] L.N. Bulaevskii, Soviet Phys. Solid State 1969, 11, 921. [24] I.S. Jacobs, J.W. Bray, H.R. Hart Jr, L.V. Interrante, J.S. Kasper, G.D. Watkins, Phys. Rev. B 1976, 14, 3036. [25] F.D.M. Haldene, Phys. Lett. 1983, 93A, 464. [26] P.W. Anderson, Mater. Res. Bull. 1973, 8, 153. [27] T. Nakamura and S. Miyashita, Phys. Rev. B 1995, 52, 9174. [28] I. Watanabe, N. Wada, H. Yano, T. Okuno, K. Awaga, S. Ohira, K. Nishiyama, and K. Nagamine, Phys. Rev. B, 1998, 58, 2438. [29] Y.J. Uemura, T. Yamazaki. D.R. Harshman, M. Senba, and E.J. Ansaldo, Phys. Rev. B 1985, 31546. [30] A. Keren, L.P. Le, G.M. Luke, W.D. Wu, Y.J. Uemura, Y. Ajiro, T. Asano, H. Huriyama, M. Mekata, and H. Kikuchi, Hyperfine Interactions 1994, 85, 181. [31] S.R. Dunsiger, R.F. Kiefl, K.H. Chow, B.D. Gaulin, M.J.P. Gingras, J.E. Greedan, A. Keren, K. Kojima, G.M. Luke, W.A. MacFarlane, N.P. Raju, J.E. Sonier, Y.J. Uemura, and W.D. Wu, Phys. Rev. B 1996, 54, 9091. [32] J.H. Brewer, S.R. Kreitzman, D.R. Noakes, E.J. Ansaldo, D.R. Harshman, and R. Keitel, Phys. Rev. B 1986, 33, 7813.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

4

Magnetism in TDAE-C60 Ales Omerzu, Denis Arcon, Robert Blinc and Dragan Mihailovic

4.1

Introduction

The purpose of this paper is to review recent work on one of the most interesting new magnetic organic compounds this decade, namely the ferromagnetic fullerene compound tetrakis-dimethylaminoethylene-C60 (TDAE-C60 ) (Fig. 1) first discovered in 1991 by Fred Wudl and collaborators at the University of California in Santa Barbara [1]. At the time of their discovery, the Curie temperature of TC = 16 K was an order of magnitude higher than the previous record [2] and brought the research field of p-electron ferromagnetism from the realms of the esoteric into the mainstream. Very soon after the discovery of an efficient method for producing useful quantities of fullerenes using a carbon arc by Huffman and Kraetschmer [3] attempts to dope them with various dopants yielded immediate and spectacular results. Whereas a Bell Laboratories group [4] were rewarded with the discovery of fullerene superconductivity by doping with alkali metals, the efforts of the Santa Barbara group [1] in doping with the strong organic electron donor TDAE yielded a material with very unusual low-temperature magnetic properties. When TDAE (a liquid at room temperature) was mixed with C60 in solution it crystallized into small particles with could be removed as a precipitate. The result was TDAE+ C− 60 , a charge-transfer salt with a monoclinic structure quite different to the high-symmetry cubic or tetragonal ones of other doped fullerene compounds (Fig. 3). The structure contains unusually short distances between adjacent C60 buckyballs of 9.99 Å [5] which is believed to

Fig. 1. Schematic diagrams of C60 (left) and TDAE (right) molecules. Carbon atoms are shown as dark spheres, nitrogen atoms as light spheres, and hydrogen atoms as small spheres.

124

4 Magnetism in TDAE-C60

lead to coupling between spins localized on C− 60 ions and the formation of a ferromagnetically correlated spin state. Since TDAE-C60 was discovered, there has been significant experimental progress in elucidating the magnetic and electronic properties of this material. However, very few new compounds have been found to show similar behavior, and until very recently [6] TDAE-C60 remained the compound with the highest TC . For example TDAE-C70 [7] and many different TDAE-doped C60 derivatives show no evidence for a ferromagnetic (FM) state down to 4 K – which is true also when TDAE is substituted with other electron donors [8] – and although there have been some reports of possible ferromagnetism at higher temperatures [9], up until very recently such unidentified ferromagnetic organic compounds (UFOs) have so far failed the reproducibility test. The current record thus stands at 19 K, for cobaltocene-3aminophenylmethano[60]fullerene [6], which has a TC approximately ∼3.5 K higher than TDAE-C60 .

4.2 4.2.1

Synthesis and Structure Synthesis

TDAE-C60 was first synthesized in powder form by reacting either powder C60 with liquid TDAE at room temperature or C60 in toluene solution with TDAE. The synthesis was performed in oxygen-free surroundings because TDAE is highly sensitive to oxygen immediately forming diethyl urea. TDAE-C60 precipitate was then thoroughly washed and dried. A method for growing single crystals of organic charge-transfer (CT) salts has been developed in the seventies for demands of a new emerging field of low-dimensional organic conductors [10]. The method is applicable to two-component systems of donors and acceptors, both soluble in an organic solvent (usually acetonitrile or some other solvent with low dielectric constant). The resulting CT salt is no longer soluble in non-polar solvent and precipitates out from solution. For growing a single crystal of significant size, the rate of charge-transfer reaction must be very slow. For this purpose crystal growing cells are used. In these cells reactants are dissolved in two vessels separated by a suitable filter (fritted glass), or a larger space filled with a pure solvent. The separation enables slow diffusion of one reactant into the solution of another and consequently, slow growth of crystals. Soon after its discovery in 1991 a similar method has been employed for growing single crystals of TDAE-C60 but for unknown reasons these method originally failed. Later, in 1994 an adapted procedure was tried with large surplus of TDAE, so shifting the balance of the diffusion process strongly in the direction of TDAE’s diffusion into the C60 compartment. The first crystals where not of good quality, but during the next few years the method has improved and today it is possible to grow crystals of high quality for both α and α modifications (see later for explanation of the two modifications of TDAE-C60 ). A typical size of such crystals is around 1 mm.

4.2 Synthesis and Structure

125

Fig. 2. A crystal growing cell for TDAE-C60 .

This enables all important physical experiments to be performed except neutron scattering (the size of TDAE-C60 crystal for the neutron scattering should be at least 2 × 2 × 2 mm3 primarily because of diffuse scattering on protons). Sizes of crystal growing cells can be different: from 10 mL to several hundreds of milliliters. The usual size is 2 × 30 mL with additional free space of approximate 20 mL which is initially filled with pure solvent. The photograph (Fig. 2) shows the shape of a typical cell. The two compartments are connected with a narrow tube, which is divided in the middle by an additional fritted glass filter which slows diffusion. Before filling, the tube must be carefully cleaned in chromic acid, washed with deionized water and dried. The solvent (toluene) must be freshly distilled under an inert atmosphere and transferred into a glove-box with an oxygen concentration less then 1 ppm. C60 and TDAE are used as commercially shipped (C60 Hoechst gold grade 95% pure, and pure TDAE from Fluka or Aldrich). Two solutions are prepared separately in an inert atmosphere with concentrations 2 mg mL−1 and 0.3 mg mL−1 for C60 and TDAE, respectively. The solutions are filtered and then poured into two compartments of the tube. Finally, the free space between two compartments is filled with pure toluene. The tube is carefully sealed with vacuum grease and transferred to a thermostatted bath for a period from three to six months, depending on the thermostatting temperature. When crystals of the α modification are needed the growing cell should be thermostatted at temperatures between 8 and 10◦ C. For the ferromagnetic α modification, the thermostatting temperatures are between 20 and 25◦ C. When the crystal growth is completed, the tube is transferred back to a glove-box and crystals are extracted from it, washed with hexane and dried. For all experimental purposes, the crystals must be kept under vacuum or an inert gas.

126

4.2.2

4 Magnetism in TDAE-C60

The Lattice Structure

The lattice structure of powder specimens was first determined by Stevens et al. [5] to be monoclinic C2/m with one formula unit per unit cell. Later, a structural analysis performed on single crystals [11] showed the room temperature structure to be monoclinic with unit cell dimensions a = 15.858(2) Å, b = 12.998(2) Å, c = 19.987(2) Å, β = 93.37◦ and four formula units per unit cell. The space group was found to be C2/c and not C2/m as originally reported from powder data. The unit cell in fact consists of two subcells (Fig. 3) which are stacked along the c-direction so that the unit cell size in the c-direction is doubled. In one of the subcells the TDAE ions are shifted along the b-axis for about 0.02 Å and in the other by the same distance in the opposite direction. The TDAE coordinates are (0.5, 0.502, 0.75), (0.5, 0.498, 0.25), (0, 0.002, 0.75) and (0, −0.002, 0.25) with the C = C bond parallel to the c-axis, whereas the C60 coordinates are (0, 0.5, 0), (0, 0.5, 0.5), (0.5, 0, 0) and (0.5, 0, 0.5). The C60 –C60 center-to-center distance is shortest along the c axis and amounts to 9.99 Å at room temperature. The atomic coordinates are collected in Table 1, together with the isotropic displacement parameters. The C60 molecules were found to be executing large amplitude re-orientations at room temperature, so that large anisotropic thermal displacements factors of the C60 carbon atoms were found. The thermal displacement parameters for some of the C60 carbon atoms at room temperature are in fact so large that the C60 atomic coordinates may well represent only an average over one or more disordered structures involving fractional atomic occupancy. On the other hand, the TDAE N and C atomic coordinates are well defined already at room temperature. At 80 K the C60 center to center distance along the c-axis decreases to 9.87 Å and the C60 atomic coordinates are much better defined than at room temperature. At the time of writing no systematic structural analysis has yet been performed on the different modifications (ferromagnetic α or non-ferromagnetic α ) of the material, nor has there been any extensive temperature dependence structural study either at low temperatures or high temperatures.

Fig. 3. Schematic unit cell of TDAE-C60 consisting of two subcells which are stacked along the c-direction. The arrows indicate the small shifts of the TDAE molecules in the unit cell along the b-direction.

4.2 Synthesis and Structure

127

Table 1. Atomic coordinates of TDAE-C60 and isotropic displacement factors. The atoms assigned with a ‘prime’ belong to the TDAE molecule while the others form the C60 molecule. x/a C(1) C(2) C(3) C(4) C(5) C(6) C(7) C(8) C(9) C(10) C(11) C(12) C(13) C(14) C(15) C(16) C(17) C(18) C(19) C(20) C(21) C(22) C(23) C(24) C(25) C(26) C(27) C(28) C(29) C(30) N(1() N(2() C(1() C(2() C(3() C(4() C(5()

0.5596 0.5578 0.4752 0.4149 0.4827 0.6258 0.6136 0.4355 0.3441 0.4657 0.5326 0.6210 0.6909 0.6793 0.5851 0.4993 0.3524 0.3310 0.3297 0.3846 0.4048 0.4912 0.6444 0.7103 0.6918 0.6332 0.4613 0.3756 0.2863 0.2953 0.4373 0.5761 0.5033 0.3764 0.4164 0.6092 0.6292

y/b

z/c

0.0996 −0.0432 −0.0589 0.0620 0.104 0.1123 −0.0852 −0.1312 0.0309 0.1992 0.2466 0.1919 0.042 −0.0362 −0.1849 −0.2101 −0.1402 −0.0606 0.1245 0.2018 0.2467 0.2714 0.2029 0.0767 −0.1221 −0.2021 −0.2515 −0.2149 −0.0439 0.0929 0.4688 0.5340 0.5022 0.3927 0.5174 0.4815 0.6145

0.1572 0.1689 0.1684 0.1520 0.1598 0.1310 0.1456 0.1404 0.1169 0.1132 0.0673 0.0778 0.0946 0.1037 0.1139 0.1119 0.0861 0.0979 0.0693 0.0669 −0.000 0.0007 0.0287 0.0331 0.0488 0.0584 0.0563 0.0379 0.0253 0.0155 0.6727 0.6869 0.7148 0.6970 0.6067 0.6286 0.7211

U 0.166 0.171 0.148 0.150 0.197 0.118 0.145 0.130 0.123 0.127 0.106 0.266 0.229 0.199 0.088 0.092 0.175 0.217 0.194 0.109 0.070 0.062 0.166 0.124 0.098 0.087 0.075 0.092 0.138 0.142 0.0535 0.0584 0.0452 0.058 0.068 0.076 0.077

128

4 Magnetism in TDAE-C60

4.3

The Electronic Structure

Ferromagnetism in insulators is very rare, and there are not very many examples known. One would therefore expect that TDAE-C60 may also be an itinerant ferromagnet, as the discoverers originally thought [1]. However, very soon after the discovery of the material it was found from the powder infrared absorption spectra that the TDAE-C60 shows no Drude tail and exhibits no absorption at low frequencies [12]. Subsequent careful infrared work by Degiorgi et al. [13] confirmed this finding and confirmed that material is most probably an insulator. Microwave absorption measurements [14] also appeared to confirm the insulating behavior of the powder and the issue was finally unambiguously settled by DC and AC conductivity measurements on single crystals of TDAE-C60 as a function of temperature [15] , when direct electrical contacts on TDAE-C60 single crystals became possible. For the measurements the two-contact method was used, as the millimeter-size samples have resistances of ∼100 k at room temperature, much higher than the gold contacts used to attach wires on the samples. The DC conductivity as a function of temperature is shown in Fig. 4. It has an activated temperature dependence: σDC = σ0 exp(−E a /kB T )

(1)

with two activation energies E a = 0.34 for T > T0 , and E a = 0.14 for T < T0 , where T0 = 150 K is the temperature below which rotations of the C60 molecules start to slow down [36]. Below 100 K the conductance become unmeasurable and there are no evidence of a re-entrant rise of conductivity down to 4.2 K.

Fig. 4. The activated temperature dependence of DC conductivity of TDAE-C60 plotted as log σ against 1/T .

4.3 The Electronic Structure

129

Fig. 5. (a) The conductance as a function of frequency in TDAE-C60 for a number of different temperatures. (b) The frequency-dependent part of the conductance for different temperatures as a function of frequency (the DC part G 0 has been subtracted).

The frequency dependence of conductivity at several different temperatures is show in Fig. 5a. It shows a clear crossover from the frequency-independent behavior at low frequencies to a power-law dependence at higher frequencies. When the frequency-independent DC part of the conductivity is subtracted from the total conductivity only the frequency-dependent part remains with a power-law frequency dependence (Fig. 3b) σAC = Bωs

(2)

with an exponent s ≈ 1. The total conductivity can be expressed as a sum of two components: a temperature-dependent and frequency independent DC conductivity and a temperature-independent and frequency-dependent AC conductivity σ (T, ω) = σDC (T ) + σAC (ω)

(3)

where σDC and σAC are given by eq. (1) and eq. (2), respectively. The model proposed for the mechanism of electrical transport in TDAE-C60 is the following: since no real energy gap of order 0.2 eV was observed in the IR measurements, the Mott–Hubbard insulating state is ruled out and there must be another reason for the activated behavior of the conductivity. Because of the strong coupling of electrons to phonons on the C60 molecule, an electron, when added to the neutral molecule, gives rise to a relaxation of the equatorial bond conjugation

130

4 Magnetism in TDAE-C60

and reduction in energy by E b ≈ 0.1 eV [16]. Thus each electron on C60 should be considered as a small polaron confined to the C60 molecule, and we should expect classic thermally activated phonon-assisted conductivity: σ ≈ ω exp(−E b /kB T )

(4)

where E b is the polaron binding energy and ω should signify the rate of rotation of the molecule. The AC part of the conductivity can be described by intermolecular tunneling which is temperature independent. The orientational disorder in the system causes a power-law frequency dependence for the tunneling (the behavior usually observed in random systems). In summary, the macroscopic conductivity of TDAEC60 is determined by the intermolecular hopping where there is a crossover between two conductivity regimes: a low-frequency temperature-dependent phonon-assisted hopping and a high-frequency temperature-independent tunneling.

4.4 4.4.1

The Magnetic Properties The Bulk Magnetic Properties

By measuring magnetic susceptibility χ and magnetization M we get information about the nature of macroscopic magnetic state of a sample. The methods for measuring the bulk magnetic properties are roughly divided into static and dynamic ones. With static methods we measure the bulk magnetic moment of a sample in some static external field. Dynamic methods, such as AC susceptibility measurements, give information about how a sample’s magnetic moment responds to a time-varying field and is the ideal choice for measuring magnetic susceptibility. With these methods the ferromagnetic transition in α-TDAE-C60 has been directly and reliably proven [17]. In Fig. 6 we see the temperature-dependence of the real and the imaginary part of the AC susceptibility for two α-TDAE-C60 samples: a powder (left) and a single crystal (right). Considering single crystals first, a steep rise of the real part of the susceptibility and a peak in the imaginary part at 16 K indicate a ferromagnetic transition. Below TC = 16 K, the real part saturates because the measured susceptibility χ0 is not equal to intrinsic susceptibility of the material (0. They are connected by relationship: χ = χ0 /(1 + Dχ0 )

(5)

where D is the demagnetization factor, which depends only on a geometry of the sample. When χ0 diverges at ferromagnetic transition, χ saturates at value 1/D and remains at that value as long as Dχ0 1. A slight decrease in the real part of the susceptibility below 10 K is related to the hump in the imaginary part, and is due to additional phenomena observed in single crystals of α-TDAE-C60 , namely a reentrant spin glass transition [18].

4.4 The Magnetic Properties

131

Fig. 6. The real and the imaginary parts of the AC susceptibility as a function of temperature for a powder (left) and a single crystal (right) of α-TDAE-C60 .

All the features that are sharply visible in the AC susceptibility of the single crystal are smeared for the powder sample. This is mainly because of very poor homogeneity and large surface area of the nano-size crystallites in the powder sample. The temperature and the field dependence of magnetization of α-TDAE-C60 are shown in Fig. 7. The behavior of the magnetization measured in a low magnetic field are characteristic for the order parameter of the second-order phase transition. It is equal to zero above the transition temperature TC , and continuously rises below TC . The field dependence of the magnetization in the ferromagnetic phase is also characteristic of a ferromagnet: a very sharp increase with a saturation in quite low field H < 50 Oe. Although strongly non-linear, the magnetization curve shows no significant hysteresis (although some authors claim to have observed a hysteresis loop a few Oersteds wide [19]). The bulk magnetic properties of the metastable α -TDAE-C60 phase are quite different from those for the α modification. There is no sign of any ferromagnetic transition down to 2 K. The system behaves as a paramagnet. The AC susceptibility of α -TDAE-C60 crystals is shown on the left of Fig. 8. χAC is very small, corresponding to the Curie susceptibility χ = C/T of non-interacting spins (the solid line in Fig. 8). The hump below 16 K is due to a small part of the sample which has already transformed into the α modification. The paramagnetic nature of α -TDAE-C60 becomes more obvious in the static magnetization measurements. The magnetization

132

4 Magnetism in TDAE-C60

Fig. 7. The temperature-dependence of the magnetization of α-TDAE-C60 measured in a magnetic field 10 Oe (left) and the magnetization curve of α-TDAE-C60 measured in fields −100 Oe < H < 100 Oe at 2 K.

is linearly related to the magnetic field M = χ H and also obeys the Curie law (right panel of Fig. 8). When M −1 is plotted against temperature it clearly extrapolates to zero (insert). The field-dependence of magnetization is also typically paramagnetic (left of Fig. 9). It fits extremely well to the theoretical Brillouin formula for a system of non-interacting magnetic moments: M = N µ tanh(µH /kB T )

(6)

where N is the number of spins in the system, µ the magnetic moment, H is the magnetic field, kB the Boltzmann constant, and T the temperature. It is surprising to find that N corresponds to one spin per pair of ions TDAE+ -C− 60 . One would expect that both ions posses one unpaired spin but this is not the case. Following the effective number of spins up to 100 K reveals a gradual increase of the effective spin number. The same happens also in the ferromagnetic α modification (Fig. 9. (right)). As mentioned before, the α -TDAE-C60 is the metastable modification of TDAEC60 . It irreversibly transforms into the α -TDAE-C60 . These transformation can be followed by measuring the magnetization of α -TDAE-C60 after several cycles of annealing at temperatures slightly above room temperature which speeds up the transformation [20]. The transformation is not gradual as can be seen in Fig. 10. The first annealing cycle doesn’t increase the magnetization much, but the second cycle causes an abrupt change for more then two orders of magnitude. After the third

4.4 The Magnetic Properties

133

Fig. 8. The real (black) and the imaginary (white) part of the AC susceptibility (left) and the magnetization in a static magnetic field of 100 Oe (right) of α -TDAE-C60 as a function of temperature. The insert shows the inverse magnetization which extrapolates to zero as T → 0.

Fig. 9. The field dependence of magnetization of α -TDAE-C60 at 2 K. The solid line is the theoretical Brillouin function (left). The effective number of spins per formula unit as a function of temperature (right).

134

4 Magnetism in TDAE-C60

Fig. 10. The temperature dependence of magnetization after different numbers of annealing cycles of ((-TDAE-C60 .

cycle the magnetization is already saturated. Further annealing slowly damages the sample and it’s magnetization gradually falls again. It is important to note that the ferromagnetic transition temperature always stays at 16 K in all stages between α TDAE-C60 and α-TDAE-C60 .

4.4.2

The Spin-glass Behavior of α -TDAE-C60

In the early stage of research on the nature of magnetic ordering in TDAE-C60 , some experimental results appeared to be in contradiction with the hypothesis of a long-range ferromagnetic ordering in TDAE-C60 . In the first place, the temperature dependence of the ESR linewidth [21] showed a relatively small line broadening and no frequency shift with a non-exponential and very slow decay of the magnetization [22]. Both of these features are characteristic of random magnetic systems without a long range ordering, and naturally, it was suggested that TDAE-C60 could be a spin-glass. After a firm experimental establishment of a direct connection between orientational degrees of freedom of C60 molecules and magnetic interactions in the system [23], this hypothesis seemed to be even more plausible (these connection was later demonstrated also by theoretical calculations [24]). By freezing C60 molecules in random orientations one can obtain a distribution of exchange interactions in the system, and consequently, magnetic disorder and frustration – two essential conditions for a spin-glass. Later measurements of linear and non-linear susceptibility partly confirmed the spin-glass hypothesis [25]. The linear susceptibility χ1 exhibit a broad peak centered at 10 K and the non-linear susceptibilities χ3 and χ5 diverge at

4.4 The Magnetic Properties

135

the same temperature. The only feature which deviated from the spin-glass behavior was the absence of any shift of the peak position at the temperature axis with frequency, which is so characteristic for spin glasses. Obviously, TDAE-C60 possesses some characteristics of spin-glasses and some of ferromagnets and it is not impossible that both phases coexist together. This idea was shown to be essentially correct by latest measurements on single crystals of α-TDAE-C60 , where it was conclusively shown that the ferromagnetic transition at 16 K is followed by a reentrant spin-glass transition at 7 K.

4.4.3

Electron-spin Resonance

As already indicated in the previous section, electron spin resonance (ESR) is a very valuable experimental tool for the investigation of magnetic properties of materials. In the first place, with ESR we can determine the static spin susceptibility by integrating the ESR absorption spectra (the static spin susceptibility χs is directly proportional to the absorption integral). Also, with ESR we can detect very weak magnetic signals, undetectable with macroscopic techniques. From the shape and the width of the ESR absorption spectra we can get additional information about spin dynamics and eventual presence of internal fields. In Fig. 11 we can see a shape of an ESR spectrum of α-TDAE-C60 in the ferromagnetic phase which develops at temperatures below the ferromagnetic transition temperature TC = 16 K. The overall width of 80 Gauss and a complicated structure indicate the presence of internal magnetic fields due to a complex distribution of magnetic domains and shape anisotropy.

Fig. 11. The shape of ESR spectrum of α-TDAE-C60 in the ferromagnetic phase.

136

4 Magnetism in TDAE-C60

Fig. 12. Temperature-dependence of the low-temperature spin susceptibility (ESR integrated intensity) for different annealing temperatures.

The ESR linewidth can be correlated with a µSR measurement which shows the existence of internal magnetic fields of similar magnitude [26]. In the powder samples we cannot observe sharp features in the spectra, and only inhomogeneous line broadening is observed below TC . The ESR technique was used for tracing the transformation from the metastable, non-ferromagnetic α modification to the stable, ferromagnetic α modification of TDAE-C60 [20]. The low-temperature spin susceptibility χs rises with the annealing temperature until 110◦ C, when the sample starts to polymerise. From the plots in Fig. 12, we can see that ferromagnetic correlations start to become significant even at temperatures much higher than TC . This effect can be attributed to the applied magnetic field of 3.4 kGauss used in the X-band ESR measurement. The ESR line-width in the paramagnetic phase gives us information about relaxation processes in the system. For TDAE-C60 , an anomaly in the temperature dependence of the linewidth centered at T0 = 160 K is characteristic (Fig. 13). The sharp decrease of the linewidth is related to slowing of rotational motion of C60 molecules which is a gradual process, but appears sharp when viewed at the ESR time-scale (10−10 s). It is interesting to note that T0 slightly shifts towards lower temperatures when a sample transforms from α to α modification. It seems that C60 molecules have more room to rotate in the ferromagnetic α modification.

4.4 The Magnetic Properties

137

Fig. 13. The ESR linewidth as a function of temperature for different annealing temperatures.

4.4.4

Ferromagnetic Resonance

Magnetic resonance in the magnetically ordered phase is different with respect to the ordinary electron spin resonance in the sense that one cannot deal with only one electron separate from the all the rest due to the presence of the large exchange fields between electronic spins. Instead of the usual ESR signal in the magnetically ordered phase, a coherent precession of all the electronic spins, i. e. the precession of the entire magnetization of the sample, can be excited. The relation between the resonance frequency and resonance field becomes strongly non-linear [27a] and it strongly depends on the type of the ordering [27b]. While in simple uniaxial ferromagnets only one resonance mode is predicted, in antiferromagnets and weak ferromagnets two resonance modes should be found. Here it is interesting to compare uniaxial ferromagnets and weak ferromagnets. In both when the external magnetic field is perpendicular to the easy axis one finds a resonance mode with a dip in the resonance field-resonance frequency relation at a resonance field equal to the anisotropy field. But in weak ferromagnets a high frequency antiferromagnetic type mode is present as well. The magnetic resonance in spin-glasses is very complicated and strongly depends on the thermal history like for instance that the line position for field-cooled samples when the measured field is applied parallel with respect to the case when the measured field is applied anti-parallel to the cooling field is different due to the hysteresis effects. In many ways superparamagnets behave similar to normal paramagnets and the resonance position of the observed lines depends linearly on the resonance frequency. So the (anti)ferromagnetic resonance technique is extremely sensitive technique for the determination of the ground state and microscopic parameters.

138

4 Magnetism in TDAE-C60

4.4.4.1

Ferromagnetic Resonance in TDAE-C60

Magnetic resonance experiments on TDAE-C60 single crystals were performed in a wide frequency range between 30 MHz and 245 GHz. High-frequency experiments in the microwave region (1.2 GHz, L-band; 9.6 GHz, X -band; 94 GHz, W -band; and 245 GHz) at T = 5 K < TC and a||H showed a single resonance line with some inner structure [28, 29]. It should be noted in the 245 GHz experiment no additional lines in were observed anywhere between 0 and 10 T. The same observation holds also for the X -band and L-band experiments where the largest sweep field was between 0 and 10 kGauss. The resonant frequency is linearly dependent on the resonant field in the microwave region, so no definite conclusion can be made on the basis of these experiments. Much more interesting behavior is found in the radio-frequency region. Above 110 MHz there is still only one resonant line. Below 110 MHz a new line emerges at zero field which shifts to higher resonant fields with decreasing resonant frequency. The two resonant lines merge together below 50 MHz. The dependence of the resonant frequency on the resonant field is strongly non-linear in the low frequency region (Fig. 14). A dip in the resonance frequency-resonance field relation is predicted for uniaxial ferromagnets as well as for weak ferromagnets. Since we have not observed any other resonant modes – which should be found in weak ferromagnet – at higher frequencies as well as in the experiments where the high frequency magnetic field was parallel to the external magnetic field we have to rule out the possibility of weak ferromagnetism. The strongly non-linear behavior (Fig. 14) also eliminates any possibility of a superparamagnetic state in TDAE-C60 single crystals and we conclude from the ferromagnetic resonance that the magnetic state is that of a normal Heisenberg ferromagnet. Agreement between theory and experiment becomes quantitative, if one takes into account also the demagnetizing field effects. For the anisotropy field the obtained values are HK = 29 Gauss and for the demagnetizing field Hdem = −39 Gauss. A non-linear dependence of the resonance field-resonance frequency relation disappears above the transition temperature in the paramagnetic phase (Fig. 14b). To conclude this section we would like to stress again that the resonant frequencyresonant field relation of the magnetic resonance signal observed below TC confirms

Fig. 14. Resonant frequency-resonant field-dependence in TDAE-C60 single crystal at (a) T = 5 K and a||H , and (b) T = 20 K and a||H .

4.4 The Magnetic Properties

139

the presence of long range magnetically ordered state below TC = 16 K. The presence of a dip in the resonant frequency-resonant field relation at H = 31 Gauss at T = 5 K and a||H , as well as the existence of a zero-field gap at 110 MHz, seem to show that TDAE-C60 single crystals below TC = 16 K described in this work were essentially Heisenberg ferromagnets. The anisotropy field appears to be rather small, which is not so unusual for a purely organic compound. Nevertheless the fact that the linewidth of the observed resonances is comparable to the shifts suggests that the distribution of the local fields is rather large and the possibility of the coexistence of a long range ferromagnetic order and short range spin-glass effects can therefore not be excluded. 4.4.4.2

Nuclear Magnetic Resonance of Powdered and Single-crystal TDAE-C60

NMR in TDAE-C60 In TDAE-C60 there are two suitable NMR probes: 1 H from the methyl groups of the TDAE molecule and 13 C mainly from the C60 and partially from the TDAE molecule. The problem with 13 C NMR is, that 13 C isotope occurs in a very small natural abundance, r = 1.108%, which makes experiments in TDAE-C60 single crystals difficult. On the other hand, the 1 H NMR probe is extremely sensitive and even experiments on 1–2 mg samples can be performed. Further, the 13 C and 1 H probes also differ in one very important aspect. While 13 C atoms are mainly on the C molecule so that their 2 p orbital contribute to the z 60 molecular t1u orbital filled with one accepted electronic spin, 1 H NMR probes come exclusively from the TDAE methyl groups, and we shall see, their different position in the crystal has some important consequences. 4.4.4.3

1 H NMR

in Powdered TDAE-C60

In powdered samples the transformation from non-magnetic α modification to ferromagnetic α modification occurs rapidly [30] even at room temperature and by the time one prepares the experiment the sample is already transformed. That is the reason why we focus in this subsection only to a well annealed powdered TDAE-C60 samples. In powdered TDAE-C60 at room temperature at a Larmor frequency ωL /2π = 270 MHz two proton NMR lines are observed [31], designated as A and B lines (Fig. 15). They are of nearly equal intensity (Fig. 15) and at room temperature they are separated by 42 kHz. The temperature dependence of the two lines is very different. The position of the A-line is nearly T -independent whereas the position of the Bline (Fig. 16) follows a Curie–Weiss law with a negative Curie temperature down to around 50 K. This means that those protons, which contribute to the B-line are close to the C60 spins and due to the hyperfine contact interaction B-line exhibit a paramagnetic shift δν. On the other hand the protons which contribute to the A-line, seem to be in a position where some sort of spin cancellation mechanism causes zero electron spin density. A trivial explanation could be that roughly half of the sample was destroyed during the sample preparation due to the high air sensitivity of the powdered TDAE-C60 samples, but additional ESR measurements disagreed with

140

4 Magnetism in TDAE-C60

Fig. 15. The temperature evolution of the 1 H NMR lines in a well annealed powdered TDAEC60 sample.

Fig. 16. The temperature-dependence of the paramagnetic shifts of the 1 H NMR A and B lines in well annealed powdered TDAE-C60 sample.

4.4 The Magnetic Properties

141

such a large proportion of the decomposition of the sample. Thus we believe that both line are intrinsic. They could reflect the incomplete transformation from α to α modification or two different 1 H protons of the same TDAE molecule. 4.4.4.4

1 H NMR

in TDAE-C60 Single Crystals

The transformation between the two modifications is in TDAE-C60 single crystals is much slower than in powdered samples and this fact has enabled the study of the individual properties of the ferromagnetic [32] and non-magnetic modifications respectively as well as the difference between the two modifications [33]. It was shown in the previous subsection that in powder samples the intensities of the “non-magnetic” and “magnetic” methyl proton A and B lines are nearly equal at room temperature. Approximately the same is true for well-annealed single crystal samples, which show a ferromagnetic transition. On the other hand in the nonmagnetic α samples, at room temperature the intensity of the “magnetic” line B (i. e. the line which shows a paramagnetic frequency shift following a Curie–Weiss law) is much weaker than the intensity of the “non-magnetic” A line. With temperature cycling between 200 K and 330 K one can change the relative intensities of the two lines [34]. This most probably reflects the annealing process that changes single crystals from the non-magnetic to ferromagnetic type. With decreasing temperature both the A and B lines broaden and the linewidth of the B line increases to 25 kHz at 170 K. This is the temperature range where the 13 C NMR rotational motion of the C− 60 ions freezes out as we shall see from the experiment. The protons are thus also sensitive to orientational ordering of the C− 60 ions through the coupling of the methyl protons with the unpaired electron at the C60 site. At 50 K the half-width of the B line is already about 250 kHz. Below TC , in ferromagnetic α type samples, the proton NMR spectra are several MHz broad (Fig. 17), and the ferromagnetic B-line is shifted with respect to the

Fig. 17. 1 H NMR spectra of TDAE-C60 single crystal at some representative temperatures.

142

4 Magnetism in TDAE-C60

non-magnetic A line. However as already noted in the powdered samples, the ferromagnetic B-line slowly disappeared and the intensity of the non-magnetic A-line increases. The disappearance of the B-line indicates that at low temperatures the responsible mechanism for spin-cancellation is most probably spin-pairing in agreement with the magnetization measurements and AC susceptibility. The non-magnetic A-line also approaches a double peaked line-shape at T = 5 K which is in agreement with the spin-pairing mechanism into singlet-like state. If the ground state of the non-magnetic A-line is of a density wave (either spin or charge density wave) the amplitude of the density wave should be very small, since the shifts of the two peaks of the A-line are rather small as compared with the shift of the magnetic B-line. In the non-magnetic α modification the TDAE proton NMR spectra, on the other hand, exhibit just one proton line. The observed 1 H NMR lines are narrower [34] and the position of this line is close to the proton Larmor frequency and thus coincides with the position of the A line of the modification α. Its position does not change with decreasing temperature even down to 4 K. In spite of an extensive search with a field swept (± 1000 G) superconducting magnet we were unable to find any other, more shifted B-type proton line in the modification α . More information about the dynamics of the local magnetic fields in the α modification of TDAE-C60 below 10 K can be obtained from the proton spin-lattice relaxation time T1 . The temperature dependence of the proton spin-lattice relaxation time T1 of the methyl protons of TDAE-C60 crystals of modification α is shown in Fig. 18. The proton spin–lattice relaxation rate is practically temperature independent between room temperature and 20 K. Between 20 K and 10 K the relaxation rate slowly increases with decreasing temperature whereas a dramatic decrease of the relaxation rate occurs below 10 K and the behavior is of the activated type. It was noticed that the temperature dependence of the proton T1 in modification α is somewhat similar to the temperature dependence of the T1 in mesoscopic size magnetic systems like iron clusters [35] Fe8 with a S = 0 ground state which is separated with a gap from the lowest excited S = 1 triplet state. In an attempt to describe the observed proton spin-lattice relaxation rate T1−1 quantitatively it was assumed that the ground state is a singlet state with S = 0 and that T1−1 has contributions

Fig. 18. The temperature-dependence of the 1 H spin-lattice relaxation rate of the α modification methyl proton line.

4.4 The Magnetic Properties

143

proportional to the probability of occupation of the excited spin energy levels; the dominant contribution being from the first excited triplet state (total spin S = 1). Using such a model a value for the singlet–triplet gap was found to be E T = 19.1 K [33]. The large methyl proton shifts of the A-line in the ferromagnetic α modification + are evidently due to hyperfine contact interactions. The C− 60 ions and the TDAE ions are thus indeed strongly exchange coupled. A part of the exchange coupling between the C− 60 ions is thus of indirect nature and takes place via the intermediate TDAE+ group. However the presence of non-shifted B-line and its temperature dependence suggests that the mechanism responsible for a spin-cancellation is some kind of pairing of spins into singlet-like ground state. On the basis of 1 H NMR one cannot exclude the possibility that we have a dimerized spin-Peierls or some sort of density wave (spin or charge) ground state. 4.4.4.5

13 C NMR

of Powdered TDAE-C60

As already mentioned, the small size of the TDAE-C60 single crystals and low natural abundance of the 13 C isotope has prevented measurements of 13 C NMR of single crystals. We thus present the results obtained on the well annealed powdered samples. The main difference between the 1 H and 13 C NMR in TDAE-C60 is the much smaller contact hyperfine interaction in the later case. This is most clearly seen when one compares the 1 H and 13 C shifts. The extremely large values of the TDAE methyl proton NMR shifts of the A-line which amount to 9000 ppm are much bigger than 13 C NMR shift of 188 ppm with respect to TMS [36]. The the relatively small C− 60 obvious explanation is that the unpaired electron spin density is non-zero at the methyl proton sites whereas at the 13 C sites the electron spin density is strongly reduced due to the fact that the unpaired electron orbital at the C− 60 ions is mainly of π-character. This has dramatic consequences in the second moment of the 13 C NMR line as well as in the spin-lattice relaxation time which are both determined by the dipolar interaction with the unpaired electron. In view of the low natural abundance of the 13 C nuclei and the fact that there are nearly ten times less TDAE than C60 carbons, the observed 13 C spectra can be 13 safely assigned to the C− 60 ions. This is also confirmed by the observed C NMR lineshift with respect to TMS which amounts at room temperature to 188 ppm. It agrees rather well with the value of 185 ppm observed for the electrochemically prepared 13 C− 60 in solution. Comparing the shifts of the C NMR lines with respect to TMS we notice that the shifts of pure C60 , C70 as well as K6 C− 60 which are all insulating with a completely filled highest occupied molecular orbital-are in the same range between 132 and 155 ppm. However, the shift of the 13 C NMR line in Rb3 C60 is very close (182 ppm) to the one observed in TDAE-C60 (188 ppm). Both compounds have in common that they have only a partially filled lowest unoccupied molecular orbital. Since in K6 C60 the t1u orbital is completely filled, the shift (12 ppm) of the line with respect to the C60 is entirely due to the chemical shift of the additional 6 electrons. We can thus estimate the contribution of each added electron in the t1u orbital to the chemical shift to be ∼2 ppm. The rest of the observed shift 43 ppm in TDAE-C60 with respect to the pure C60 is therefore due to the Knight shift – i. e. due to a non-zero

144

4 Magnetism in TDAE-C60

Fig. 19. Temperature-dependence of the second moment of the 13 C NMR line in powdered TDAE-C60 .

spin-density at the 13 C site. The non-zero spin-density at the carbon sites is a result of the curved surface of the C60 molecule. The temperature dependence of the second moment of the 13 C spectra is shown in Fig. 19. The 13 C spectra are clearly motionally narrowed by the nearly isotropic rotation of the C− 60 ion above 170 K. The spectral shape, which is Lorentzian, does not change between room temperature and 170 K. The second moment M2 amounts here to less than 1 kHz2 . Between 150 K and 70 K a 13 C linewidth transition takes place due to a freezing out of the C− 60 “isotropic” rotation on the NMR time scale. The second moment M2 increases to 160 kHz2 . The analysis [36] of the temperature dependence of the 13 C NMR linewidth transition led us to the activation energy for the C− 60 reorientation to be E a ≈ 130 meV. The 13 C NMR data summarized above demonstrate the nearly isotropic rotation of the C− 60 ions above 150 K and the existence of the orientational ordering transition between 150 K and 100 K. The question whether the resulting orientational order of the C− 60 ions is perfect, or whether there is some residual static orientational disorder at low temperatures cannot be answered by the 13 C powder line-shape data. 13 C spin-lattice relaxation data on the other hand seem to demonstrate the existence of residual disorder. The recovery of the 13 C spin magnetization after 180◦ –t–90◦ pulse sequence is definitely non-exponential due to a distribution of spin-lattice relaxation times. The stretch exponential behavior reflects either the presence of the orientational disorder of C60 molecules or the distribution of local magnetic fields which is characteristic for inhomogeneous ferromagnets or spin-glasses.

4.5

Conclusion

The macroscopic measurements presented thus far give a self-consistent picture of the electronic and magnetic properties of TDAE-C60 . All the data point towards a low-temperature insulating Heisenberg-like ferromagnetic state with low anisotropy, low magnetization (∼10 Gauss) and very small hysteresis (if any).

References

145

On the other hand, relatively little is known about the microscopic nature of the magnetic interactions leading to its rather remarkable properties. For example, the magnetic measurements appear to show that there is an interplay between the ferromagnetic long-range order and short-range glass-like behavior. Although it appears to be reasonably clear that this somehow arises from orientational disorder of the C60 molecules, until a full structural study is completed at low temperatures in both the α and α phases in correlation with the magnetic properties, the microscopic details will not be understood. It is also appears to clear by now that the ferromagnetic interactions arise between the fullerenes, and that the spins on the donor TDAE appear to be less important in relation to the low-temperature ferromagnetic state than originally thought. Since non-TDAE fullerene ferromagnets have been discovered, whose properties are virtually identical [6], we can conclude that TDAE is not essential for the positive effective exchange interaction between the fullerenes in this material. However, there are some indications that there might be structural effects due to the peculiar molecular structure of the TDAE donor, which might play a role in determining the rotational degrees of freedom of the C60 molecules. It was shown by molecular orbital calculations that the TDAE in the +1 state appears to have a nearly degenerate configurational ground state [37], and infrared measurements [37] seem to show strong non-harmonic behavior which could be considered as good evidence for the existence of such a near-degenerate ground state of the donor. Curiously, the temperature ∼130 K where C60 molecules start to slow down their rotation and where the TDAE spins start to pair up [33] appears to be close to the temperature where the non-harmonic effects start to be visible in the optical spectra [37]. It remains to be shown how much of this is coincidental, or in fact these observations are connected in some way. As already mentioned, apart from TDAE-C60 there exist a number of similar compounds with different donors exhibiting a low-temperature ferromagnetic ground state [6]. The fact that the Curie temperatures of these new compounds is very similar suggests that the fullerene ferromagnetic materials may be limited to temperatures below 20 K. However, until we understand the microscopic origin of the ferromagnetic exchange interaction in these materials, it is perilous to make predictions about the TC values of future fullerene charge-transfer compounds.

References [1] P.-M. Allemand. K. C. Khemani, A. Koch, F. Wudl, K. Holczer, S. Donovan, G. Gruner, J. D. Thompson, Science 1991, 253, 301–303. [2] M. Takahashi, P. Turek, Y. Nakazawa, M. Tamura, K. Nozawa, D. Shiomi, M. Ishikawa, M. Kinoshita, Phys. Rev. Lett. 1991, 67, 746–748. [3] W. Kratschmer, ¨ K. Fostiropoulos, D. R. Huffman, Chem. Phys. Lett. 1990, 170, 167–170. [4] R. C. Haddon, A. F. Hebard, M. J. Rosseinsky, D. W. Murphy, S. J. Duclos, K. B. Lyons, B. Miller, J. M. Rosamilia, R. M. Fleming, A. R. Kortan, S. H. Glarum, A. V. Makhija, A. J.

146

[5] [6] [7] [8]

[9] [10] [11]

[12]

[13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

4 Magnetism in TDAE-C60 Muller, ¨ R. K. Eick, S. M. Zahurak, R. Tycko, G. Dabbagh, F. A. Thiel, Nature (London) 1991, 350, 320–322. P. W. Stephens, D. Cox, J. W. Lauher, L. Mihaly, J. B. Wiley, P. M. Allemand, A. Hirsch, K. Holczer, Q. Li, J. D. Thompson, F. Wudl, Nature (London) 1991, 355, 331–332. A. Mrzel, A. Omerzu, P. Umek, D. Mihailovic, Z. Jaglicic, Z. Trontelj, Chem. Phys. Lett., 1998, 298, 329–333. K. Tanaka, A. A. Zakhidov, K. Yoshizawa K. Okahara, T. Yamabe, K. Yakushi, K. Kikuchi, S. Suzuki, I. Ikemoto, Y. Achiba, Phys. Rev. B 1993, 47, 7554–7559. (a) H. Klos, I. Rystau, W. Schutz, ¨ B. Gotschy, A. Skiebe, A. Hirsch, Chem. Phys. Lett. 1994, 224, 333–336; (b) A. Otsuka, T. Teramoto, Y. Sugita, T. Ban, G. Saito Synth. Met., 1995, 70, 1423–1424. Y. Li, D. Zhang, F. Bai, D. Zhu, B. Yin, J. W. Li, Z, Zhao, Solid State Commun. 1993, 86, 475–457. (a) H. Anzai, J. Cryst. Growth 1975, 33, 185–187; (b) M. L. Kaplan, ibid., 161–164. L. Golic, R. Blinc, P. Cevc, D. Arcon, D. Mihailovic, A. Omerzu, P. Venturini in Fullurenes and Fullerene Nanostructures (Eds. H. Kuzmany, J. Fink, M. Mehring, S. Roth), World Scientific, Singapore, 1996, p.p. 531–534. D. Mihailovic, P. Venturini, A. Hassanien, J. Gasperic, K. Lutar, S. Milicev in Progress in Fullerene Research (Eds. H. Kuzmany, J. Fink, M. Mehring, S. Roth), World Scientific, Singapore, 1994, p.p. 275–278. F. Bommeli, L. Degiorgi, P. Wachter, D. Mihailovic, A. Hassanien, P. Venturini, M. Schreiber, F. Diedrich, Phys. Rev B 1995, 51, 1366–1369. A. Schilder, H. Klos, I. Rystau, W. Schutz, ¨ B. Gotschy, Phys. Rev. Lett. 1994, 73, 1299– 1302. A. Omerzu, D. Mihailovic, N. Biskup, O. Milat in S. Tomic, Phys. Rev Lett. 1996, 77, 2045–2048. K. Harrigaya, J. Phys. Soc. Jpn. 1991, 60, 4001–4004. A. Suzuki, T. Suzuki, R. J. Whitehead, Y. Maruyama, Chem. Phys. Lett. 1994, 223, 517– 520. A. Omerzu, D. Mijatovic, D. Mihailovic, to be published. L. Dunsch, D. Eckert, J. Frohner, A. Bartel, K.-H. Muller, J. Appl. Phys. 1997, 81, 4611– 4613. A. Mrzel, P. Cevc, A. Omerzu, D. Mihailovic, Phys. Rev B 1996, 53, R2922–2925. P. Venturini, D. Mihailovic, R. Blinc, P. Cevc, J. Dolinsek, D. Abramic, B. Zalar, H. Oshio, P.-M. Allemand, A. Hirsch, F. Wudl, Int. J. Mod. Phys. B 1992, 6, 3947–3951. R. Blinc, P. Cevc, D. Arcon, D. Mihailovic, P. Venturini, Phys. Rev. B 1994, 50, 1–3. D. Mihailovic, D. Arcon, P. Venturini, R. Blinc, A. Omerzu, P. Cevc, Science 1995, 268, 400–402. T. Sato, T. Saito, T. Yamabe, K. Tanaka, Phys. Rev. B 1997, 55, 11052–11055. K. Tanaka, T. Sato, K. Yoshizawa, K. Okahara, T. Yamabe, M. Tokumoto, Chem. Phys. Lett. 1955, 237, 123–126. A. Lappas, K. Prassides, K. Vavekis, D. Arcon, R. Blinc, P. Cevc, A. Amato, R. Feyerhen, F. N. Gygax, A. Schenck, Science 1995, 267, 1799–1802. (a) S.V. Vonsovskii, Ferromagnetic Resonance, Pergamon Press, Oxford, 1966. (b) D. Arcon, R. Blinc, A. Omerzu, Molecular Physics Reports 1997, 18/19, 89–97. R. Blinc, K. Pokhodnia, P. Cevc, D. Arcon, A. Omerzu, D. Mihailovic, P. Venturini, L. Golic, Z. Trontelj, J. Luunik, ˆ J. Pirnat, Phys. Rev. Lett. 1996, 76, 523–527. R. Blinc, P. Cevc, D. Arcon, A. Omerzu, M. Mehring, S. Knorr, A. Grupp, A.-L. Barra, G. Chouteau, Phys. Rev. B 1998, 58, 14416–14423. D. Arcon, R. Blinc, P. Cevc, T. Jesenko, Europhys. Lett. 1996, 35, 469–472.

References

147

[31] R. Blinc, J. Dolinsek, D. Arcon, D. Mihailovic, P. Venturini, Solid State Commun. 1994, 89, 487–491. [32] D. Arcon, J. Dolinsek, R. Blinc, K. Pokhodnia, A. Omerzu, D. Mihailovic, P. Venturini, Phys. Rev. B 1996, 53, 14028–14031. [33] D. Arcon, R. Blinc, P. Cevc, A. Omerzu, Phys. Rev. B 1999, 59, 5247–5251. [34] R. Blinc, D. Arcon, P. Cevc, D. Mihailovic, A. Omerzu, Appl. Magn. Reson. 1996, 11, 203–228. [35] A. Lascialfari, D. Gatteschi, F. Borsa, A. Cornia, Phys. Rev. B 1997, 55, 14341–14349. [36] D. Arcon, J. Dolinsek, R. Blinc, Phys. Rev. B 1996, 53, 9137–9143. [37] K. I. Pokhodnia, J. Papavassiliou, P. Umek, A. Omerzu, D. Mihailovic, J. Chem. Phys. 1999, 110, 3606–3611.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

5

Triarylmethyl and Amine Radicals R.-J. Bushby

5.1

Introduction

Recent work on high-spin species based on triarylmethyl radical and triarylamine radical cation building blocks has helped us to understand the rules which may ultimately lead to a proper ferromagnetic polymer [1]. Following a short discussion of the chemistry of triarylmethyl radicals and triarylamine radical cations this chapter describes their elaboration into triplet biradical, quartet tetraradical, high-spin oligomeric and ultimately polymeric systems.

5.2

Monoradicals (S = 1/2)

The study of triarylmethyl radicals goes back over 100 years to the very origins of organic free radical chemistry [2]. In 1900, Gomberg reported that, when a solution of triphenylmethyl chloride 1 (Fig. 1) in benzene was reduced with finely divided silver, an intensely yellow solution was obtained which behaved as if it contained the triphenylmethyl radical 2 [3].

Fig. 1. Gomberg’s preparation of triphenylmethyl 2 [3]. Reagents (i) Ag/toluene; (ii) air.

150

5 Triarylmethyl and Amine Radicals

For example, when this solution was exposed to the air it decolorized and when the resultant solution was evaporated the peroxide 4 was isolated. Alternatively, when the yellow solution was evaporated in the absence of air, a radical dimer was obtained. Gomberg incorrectly identified this dimer as hexaphenylethane (arising by coupling of two radicals 2 through their α-positions) and only much later was it shown to have the structure shown in formula 3 and to arise by coupling of the α-position of one radical to the para position of another [4]. The interesting story of how this mistake arose, how it was propagated for so long and how a clear understanding of Gomberg’s work eventually emerged has been reviewed in detail by McBride [5]. We now know that Gomberg’s yellow benzene solution was a mixture containing mainly the dimer 3, which is in dynamic equilibrium with a small percentage of the free radical 2. Simple resonance theory suggests that the free spin in the triphenylmethyl radical 2 will be distributed between the α, ortho and para positions and this is reinforced by HMO calculations which yield spin densities of 4/13 on the α-carbon and 1/13 on each ortho and para site. Higher levels of theory give similar positive spin densities for these sites but also predict a small negative spin density on each alternate carbon. They show that the central carbon is close to being sp2 trigonal/planar and that the benzene rings are twisted out of the plane giving a propeller-like conformation. This twisting out of the plane substantially reduces the repulsion between the phenyl ortho hydrogens whilst only slightly reducing the conjugation between the radical center and the π-orbitals of the benzene rings. For triphenylmethyl in the gas phase the tortional angle is 40–50◦ [6], for tris(para-nitrophenyl)methyl 5 in the crystal it is 30◦ [7] and for the crystalline 1:1 complex of tris(pentachlorophenyl)methyl and benzene the three rings are twisted by 46, 53 and 54◦ (Fig. 2, inset) [8]. In general one expects the introduction of bulky ortho substituents into the radical 2 to increase the tortional angle of the aryl ring, to decrease the degree of conjugation and to enhance the free spin density on the α-carbon. Substituents also profoundly affect the monomerdimer equilibrium and data for relevant compounds is collected in Table 1 [9–14]. Other factors being equal, the two main factors that influence the equilibrium are resonance stabilization of the radical and steric hindrance to dimerization. Hence, increasing the extent of the conjugated system in 2 by introducing para-phenyl substituents increases the degree of dissociation (Table 1, entries 1-4) and para nitro and methoxy groups markedly increase the degree of dissociation (entries 5 and 7). The importance of steric factors is most clearly seen in comparing compounds 6 and 7 (Fig. 3). In tris(2,6-dimethoxyphenyl)methyl 6 the ortho methoxy groups shield the radical from attack both above and below the plane and prevent dimerization. As a result the system is completely dissociated even in the solid state. The sesquixanthydryl 7 is similarly stabilized by ortho oxygens but is forced to be planar. It is much less stable and in this case the equilibrium strongly favors the dimer [13]. The most remarkable example of a stabilized triarylmethyl is that of the deep red perchlorinated radical 5 (Fig. 2). This system is completely dissociated. Also, whereas almost all other triarylmethyl radicals react rapidly with oxygen, the perchlorinated radical 5 is stable in air up to 300◦ C! The large dihedral twist of the phenyl substituents in this system means that the spin is largely localized on the α-carbon but this carbon is completely shielded by three benzenes and six ortho chlorines. Effectively the radical is in a cage [14]. Although 5 is heat and air stable, like most triarylmethyl radicals, it

5.2 Monoradicals (S = 1/2)

151

Fig. 2. Synthesis of tris(pentachlorophenyl)methyl 5 [14]. Reagents (i) BCH; (ii) Bu4 NOH; (iii) para-chloranil. Inset: Geometry of the radical taken from the X-Ray crystal structure of the 1:1 complex with benzene [8]. (Reproduced by permission of the Royal Society of Chemistry).

Fig. 3. Similarly substituted triarylmethyl derivatives but with very different geometries and degrees of steric hindrance [13].

is sensitive to light undergoing a photocyclization to give a 9-phenylfluorenyl derivative [15]. Although many triarylmethyl radicals have been made by reduction of the corresponding halide, the preparation of the perchlorinated radical 5 provides an example of the other main synthetic route: oxidation of the corresponding carbanion.

152

5 Triarylmethyl and Amine Radicals

Table 1. Degree of dissociation of triarylmethyl dimers. The degree of dissociation of some of these systems has proved controversial with quite widely divergent figures being quoted in the literature according to which experimental method was employed and even sometimes when two laboratories have used the same experimental method. However, within the present context, the general trends are more important than the absolute values. Except as noted the figures are based on cryoscopic measurements.

a b

Entry

Radical formed

Dissociation of the dimer (%)

Conditions

Ref.

1 2 3 4 5 6

(C6 H4 )3 C• p-C6 H5 C6 H4 (C6 H5 )2 C• ( p-C6 H5 C6 H4 )2 C6 H5 C• ( p-C6 H5 C6 H4 )3 C• ( p-O2 NC6 H4 )3 C• ( p-ButC6 H4 )3 C•

2–3 15 79 100 100 57–79a ca. 100b

[9] [9] [9] [9] [9] [10] [11]

7

( p-CH3 OC6 H4 )3 C•

ca. 100b

8 9 10 11

(o-CH3 OC6 H4 )3 C• Compound 6 (Fig. 3) Compound 7 (Fig. 3) Compound 5 (Fig. 2)

ca. 100b ca. 100 ca. 0 ca. 100

5◦ C, benzene 5◦ C, benzene 5◦ C, benzene 5◦ C, benzene 5◦ C, benzene 0◦ C, toluene −30 to 100◦ C, benzene or toluene −30 to 100◦ C, benzene or toluene 5◦ C, benzene 25◦ C, benzene or ether 150◦ C, methyl benzoate Up to 300◦ C, neat

[11] [12] [13] [13] [14]

Based on magnetic susceptibility. Based on EPR measurements.

The triphenylamine radical cation 9 is also made by a one-electron oxidation; in this case of the corresponding amine 8 (Fig. 4) [16]. It is unstable, irreversibly dimerizing with the loss of two protons to give the benzidine 10. The dimerization process is second order in 9 and so involves a radical cation/radical cation coupling reaction and not the addition of a radical cation to a neutral amine molecule. Rates of dimerization give an indication of the effect of substituents on the stability of triarylamino radical cations and relevant data is presented in Table 2 [17, 18]. As shown, substituent effects mostly parallel those in the triarylmethyl series. Like the triarylmethyls, triarylamine radical cations are stabilized by extending the conjuga-

Fig. 4. Synthesis of the radical cation of triphenylamine 9 [16]. Reagents: (i) – e− ; (ii) – 2H+ .

5.3 Diradicals (S = 1)

153

Table 2. Rates of dimerization of triarylamine radical cations in acetonitrile at room temperature. Entry 1 2 3 4 5 6 7 8

System N+ .

( p-NO2 C6 H4 )(C6 H5 )2 ( p-Et2 NSO2 C6 H4 )3 N+ . (C6 H5 )3 N•+ ( p-But C6 H4 )(C6 H5 )2 N+ . ( p-C6 H5 C6 H4 )(C6 H5 )2 N+ . ( p-C6 H5 C6 H4 )3 N•+ ( p-CH3 OC6 H4 )(C6 H5 )2 N+ . ( p-CH3 CH2 OC6 H4 )3 N+ .

Rate of dimerization (M−1 s−1 )

Ref.

1.4 × 104

[16] [17] [17] [16] [16] [17] [16] [17]

4.4 × 103 1.1 × 103 1 × 102 6 × 10 2.4 × 10 6 × 10−1 8.8 × 10−2

tion with a para phenyl substituent, (entries 3, 5 and 6) or by a para alkoxy substituent (entries 7 and 8). However, unlike the neutral radicals, these radical cations are destabilized by electron withdrawing para substituents (entries 1 and 2). Even systems such as tris(para-biphenylyl)aminium, which slowly dimerize in solution, can be isolated as stable solids [18]. In general the amine radical cations are much more air-stable and thermally stable than their all-carbon counterparts and far more have been isolated. Some have found a use as “easy-to-handle chemical oxidants” [19] and tris(4-bromophenyl)aminium hexachloroantimonate is commercially available. The radical 2 and the radical cation 9 are isoelectronic and X-ray crystallography shows that they have very similar propeller-like geometries. Hence, in tris(parabiphenylyl)aminium perchlorate the dihedral angles about the central nitrogen are 43, 46, and 27◦ [20]. EPR studies show that the introduction of the nitrogen only perturbs the free spin distribution a little. Hence the EPR spectrum of the free radical 2 shows a(6H), 2.55 G, ortho-H; a(6H), 1.11 G, meta-H; a(3H), 2.78 G, para-H [21] whereas the radical cation 9, generated by treating triphenylamine with BF3 in SO2 shows a(6H), 2.28 G, ortho-H; a(6H), 1.22 G, meta-H; a(3H), 3.32 G, para-H [22]. In the light of knowledge gleaned from studies of these monoradicals, workers who have tried to construct high-spin systems based on triarylmethyl and triarylamine building blocks have generally opted for systems in which the stability of the spincarrier is enhanced by perchloro, para-phenyl, para-tert-butyl and/or ortho-alkoxy substituents.

5.3

Diradicals (S = 1)

Just as triphenylmethyl 2 was the first “stable” organic monoradical to be observed experimentally, so the closely related “Schlenk hydrocarbon” 13 (Fig. 5) was the first “stable” organic triplet diradical. Following Gomberg’s initial success [3] Stark and coworkers [23] attempted to make a metaquinonoid diradical by reduction of the dichloride 11 (Fig. 5).

154

5 Triarylmethyl and Amine Radicals

Fig. 5. Preparation of the Schlenk hydrocarbon 13 [24, 25]. Reagents (i) Ag/benzene. Inset: EPR spectrum, frozen toluene matrix, −180◦ C [27]. (Reproduced by permission of WileyVCH.)

However, it was only when the experiments were repeated by Schlenk and Brauns [24, 25] using rigorously air-free experimental conditions that it became clear that reduction of the dichloride 11 gave first the monochloro-monoradical 12 and finally the diradical 13. Like the corresponding monoradical 2, the diradical 13 tends to selfassociate but, since 13 is “bifunctional”, this results in a complex mixture of oligomers rather than a simple dimer. These oligomers, the monoradical 12 and the diradical 13 are all extremely sensitive to oxygen which instantly discharges their color [24] giving peroxide products [26]. In solution, where the molecules are rapidly tumbling and spin-relaxation is also very rapid, EPR spectra for organic triplet species are usually too broad to be observed. However, when a solution containing the diradical 13 in toluene is frozen at −180◦ C a strong triplet EPR spectrum [27] is observed which is superimposed on a broad singlet arising from the monoradicals present (Fig. 5, inset). The triplet component of the spectrum has been fitted to a single species, with zero field splittings |D/ hc| = 0.0079 cm−1 , |E/ hc| = 0.0005 cm−1 . A study of the temperature dependence of the intensity of the EPR signal shows that the Curie Law is obeyed and this is consistent with a triplet ground state [28]. The size of the energy difference between the triplet and first excited singlet spin state is unknown and estimates have varied from as high as 1 eV (23 kcal mol−1 ) [27] down to as low as 2.6 kcal mol−1 [29]. The lower number is probably closer to the truth

5.3 Diradicals (S = 1)

155

but even this is so large that there will not be a significant population of the excited singlet state at normal temperatures. The only experimentally measured value for a compound in this family is that for metaquinodimethane for which the triplet state is favored by 9.6 kcal mol−1 [30]. The geometry of the Schlenk hydrocarbon 11 is also unknown but it seems probable that there is a mixture of diastereoisomers since the two stereogenic α-carbon centers can either be right hand helical [usually designated plus (P)] or left-hand helical [usually designated minus (M)] [31]. This has been clearly demonstrated by Vecciana for the perchlorinated Schlenk hydrocarbon 14 [32]. The synthesis of this is shown in Fig. 6.

Fig. 6. Synthesis of the perchlorinated Schlenk hydrocarbon 14 [32]. Reagents (i) CHCl3 /AlCl3 ; (ii) C6 Cl5 H; (iii) Bu4 NOH; (iv) para-chloranil; (v) chromatographic separation of meso and rac isomers.

156

5 Triarylmethyl and Amine Radicals

Like the corresponding perchlorinated monoradical 5, the diradical 14 is thermally stable and it can be stored for months at room temperature. Indeed, it is sufficiently stable to be purified by chromatography. In this way it has been possible to separate the two diastereoisomers. In the meso isomer the helicity of the two stereogenic centers is opposed and in the rac isomer both are the same. The diastereoisomers can be interconverted by heating in MeCN/THF (G ‡ = 23.4 kcal mol−1 ). As expected they show significantly different physical and spectroscopic properties. Hence, in a frozen THF matrix at 143 K, the meso isomer shows |D/ hc| = 0.0152 cm−1 , |E/ hc| = 0.0051 cm−1 and the rac isomer |D/ hc| = 0.0085 cm−1 , |E/ hc| < 0.0003 cm−1 . Both have triplet ground states. This separation of the diastereoisomers is important result since it underlines the fact that all of the high-spin systems based on triarylmethyl radical and triarylamine radical cation building blocks are complex mixtures of diastereoisomers (up to 2(N −1) for a system comprised of N centers) [33]. Hence, the common practice of fitting spin-splitting, zero-field splitting, etc., data using a single parameter set is always an approximation. Whereas the Schlenk hydrocarbon 13 is largely oligomerized in solution this tendency to self-associate can be suppressed by suitable substitution. Derivatives 14–19 (Figs. 6 and 7) [29, 32–36] of the Schlenk hydrocarbon are all monomeric in solution and 14–17 and 19 all have triplet ground states with large singlet-triplet energy gaps. Studies by EPR spectroscopy 4–80 K and by magnetometry 2–80 K of frozen THF and Me-THF matrices of the highly substituted and presumably highly nonplanar diradical 18 show the presence of two isomers, which are probably diastereoisomers, both of which have singlet ground states. For the major isomer E ST = −0.2 kcal mol−1 and for the minor isomer −0.02 kcal mol−1 [33]. The solid diradicals 16–19 can all be stored at room temperature under vacuum or in an argon atmosphere. Solid samples of the diradicals 14 and 19 give an effective moment µeff. close to that of 2.83 µB expected for a triplet molecule [29, 35, 36] but those for 16–18 are lower (2.2–2.5 µB ) and analysis of these systems is complicated by the presence of S = 1/2 impurities [29]. In comparing the diradicals shown in Fig. 7 with the bisradical-cations shown in Fig. 8 it is reasonable to assume that the replacement of the α-carbons with αnitrogens reduces the energy difference between the singlet and triplet states [37] but unfortunately the magnitude of this reduction is not known. Certainly these amminium cations, like the corresponding hydrocarbons, have triplet ground states and the magnitudes of the zero field splittings are similar confirming that the heteroatoms only perturb the spin-distribution in quite a minor way. Most are more stable than the equivalent all-carbon diradicals. Their stability has usually been probed using cyclic voltammetry [38–41]. The simple diradical 21a [38– 40] is not very stable but these radical cations are greatly stabilized by ortho or para methoxy [41–43], phenylamino or diphenylamino [44] substituents. Diphenylamino and phenylamino substituents prove to be particularly good at stabilizing these radical cations but they also have the undesirable effect that E ST is reduced to the extent that it becomes of the order of kT or even that the ordering of the states is reversed. A potential problem in building high-spin systems based on charged triarylamine radical cation blocks rather than neutral triarylmethyl radical building blocks is

5.3 Diradicals (S = 1)

157

Fig. 7. Triplet diradical derivatives of the Schlenk hydrocarbon 15 [34], 16–18 [29, 35], and 19 [36].

the coulombic effect. A priori, it seemed possible that that cation-cation repulsion would make the polyradical polycation species either difficult to prepare or at highly unstable. However, in one system where this effect has been measured it was shown that cation–cation repulsion is moderated by the shielding from the counterions. Whilst the cation-cation repulsion can be measured it is scarcely significant [41]. It is important to note that, in all of the triplet ground-state biradicals discussed so far, the topology is that of a metaphenylene. Equivalent orthophenylene and paraphenylene species always have singlet ground states. Because metaquinodimethane hydrocarbons cannot be represented by a classical valence bond formula in which each π -electron is formally paired with one on a neighboring carbon they are called “non-Kekule´ hydrocarbons”. As in other non-Kekule´ hydrocarbons [45] the ferromagnetic spin coupling in metaquinodimethane derivatives is best understood in terms of Hund’s Rule. [46, 47]. The application of Hund’s Rule to systems such as atomic carbon (Fig. 9) is well known but, whereas Hund’s rule applies to all atomic systems, it does not apply in the same universal way to every molecular system

158

5 Triarylmethyl and Amine Radicals

Fig. 8. Triarylamine analogs of the Schlenk hydrocarbon 20 [42], 21a [38–40], 21b [42, 43], and 22 [44].

containing degenerate or near-degenerate SOMOs. In atomic systems, Hund’s Rule depends on the fact that the co-centered singly occupied atomic orbitals are always strictly degenerate, orthogonal ( ψ1 ψ2 dτ = 0) and coextensive ( ψ12 ψ22 dτ = 0). In molecular systems the SOMOs are sometimes not quite degenerate and they may or may not be coextensive. It is this second point which is most often crucial to determining the energy difference between the spin states. In the Schlenk hydrocarbon 13 (Fig. 9) or the 3,4 -dimethylenebiphenyl derivatives shown in Fig. 10 it is clear that the SOMOs are properly coextensive (they share atoms in common) and in situations such as this the triplet state is always strongly preferred. However, in a long chain α,ω-polymethylene [• CH2 (CH2 )n CH•2 ] or the 3,3 -dimethylenebiphenyl derivatives shown in Fig. 10 the SOMOs are “disjoint”. They no longer overlap in their spatial distribution and they no longer share atoms in common. In situations such as this there is little or no interaction between the spins and the singlet and triplet states are close to being degenerate. An additional complication arises in molecular systems when changes in geometry, environment or the introduction of substituents or heteroatoms lifts the degener-

5.3 Diradicals (S = 1)

159

Fig. 9. Hund’s Rule illustrated for the case of atomic carbon and the Schlenk hydrocarbon. Note that, since the singly occupied orbitals py and ψ16 are symmetric with respect to a vertical plane and px and ψ17 symmetric we have orthogonality in both cases (ψ1 ψ2 dτ = 0) but that in each case they are also coextensive ( ψ12 ψ22 dτ = 0).

acy of the SOMOs. What is found is that non-Kekule´ α-diradicals with coextensive SOMOs are remarkably tolerant to such perturbations and the triplet state remains the ground state even when the perturbation is quite substantial. These triplets are known as “robust” triplets [48] and all attempts to build high-spin polyradicals have relied on “robust” triplet building blocks. The most frequently exploited alternative to ferromagnetic coupling 1,3 through benzene is 3,4 -coupling through biphenyl (Fig. 10). As already pointed out, this gives the required pair of degenerate coextensive orbitals and it is instructive to compare this with isomers in which there is 3,3 -coupling through biphenyl where the singlyoccupied orbitals are disjoint and there is expected to be a negligible exchange

160

5 Triarylmethyl and Amine Radicals

Fig. 10. Ferromagnetic coupling 3,4 and 3,3 through biphenyl [41, 49]. In the 3,4 case the relevant singly occupied orbitals are both orthogonal and coextensive leading to a significant exchange interaction. In the 3,3 case they are disjoint (they occupy separate regions of space) and the exchange interaction is negligible.

5.4 Triradicals (S = 3/2)

161

interaction. The coextensive diradical 23 has a triplet ground state. However, in the isomeric disjoint diradical 24, the singlet state lies 0.1 kcal below the triplet state [49]. The situation for the corresponding biphenyldiamine diradical dications 25 and 26 appears to be similar with only the 3,4 -coupled diradical having a triplet ground state. In this series the temperature dependence of the EPR spectrum for the 3,3 coupled system 26 is complex and it has been interpreted on the assumption that there is a mixture of rotamers about the central biphenyl linkage [41]. In other diradicals coupled 3,3 through biphenyl the splitting between the singlet and triplet states is very close to zero [50]. Other alternatives to 1,3-coupling through benzene that have been explored are 1,3-coupling through triazene [51], 1,6-coupling through naphthalene [52] and 4,4 -coupling through metaterphenyl [53]. In principle, many other non-Kekule´ quinodimethane nuclei could be exploited [54].

5.4

Triradicals (S = 3/2)

The two most obvious ways of extending the triplet diradical 13 to give a quartet triradical are through the central ring as in the “Leo” triradical 27 (Fig. 11) or through one of the peripheral benzene rings as in the “linear” system 31 (Fig. 12). Although the Leo triradical 27 has been known for many years [55] and it is known to give a quartet EPR spectrum there appears to be no experimental evidence that the quartet is actually the ground state. However, the corresponding perchlorinated triradical 28 has been shown to possess a quartet ground state. Like the corresponding perchlorinated diradical 14, it is remarkably stable and different diastereoisomers have been isolated [56]. The tris-amino radical cations 29a and 29b [38, 39, 43] also have quartet ground states. There is conflicting information regarding their stability [39, 43, 57] but the tris-diphenylamine-substituted radical cation 30 is a stable, isolable solid [58]. In this case, the quartet and doublet states are comparable in energy. As usual, although “amino” substituents are very good at stabilizing these radical cations they are sufficiently strongly perturbing that they annul the splitting between the high-spin and low-spin states. All of the linear triradicals 31 (Fig. 12) have quartet ground states with negligible population of the first excited doublet state <100 K. [59]. However, although the triradical 32, in which the coupling is 3,4 through biphenyl, also has a quartet ground state the ferromagnetic coupling is relatively weak (J/k ca. 90 K) leading to a significant temperature dependence of χ T [60]. Addition of one electron to the triradicals 31a–31c gives anion-diradicals with triplet ground states (lithium salt, frozen matrices) and this has been interpreted in terms of a species in which the excess charge is essentially localized at on of the terminal α-carbons. A similar one-electron reduction of the triradical 31d gives a singlet anion-diradical. In this case the excess charge is thought to be localized on the central α-carbon next to the carbanion-stabilizing biphenyl substituent effectively insulating the two unpaired electrons from each other [61]

162

5 Triarylmethyl and Amine Radicals

Fig. 11. Quartet tetraradicals based on the Leo triradical 27.

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

163

Fig. 12. Quartet triradicals with a linear architecture [59, 60].

5.5

Monodisperse High-spin Oligomers (S = 2 – ca. 10)

In principle it is possible to extend the simple motif shown in Fig. 12 and by linking many α-carbon radical centers (or α-nitrogen radical cation centers) 1,3 through phenylene to build up a linear high-spin oligomer or polymer. Indeed, this sort of approach was successfully exploited by Itoh and Iwamura in their pioneering work on high-spin polycarbenes [62] which eventually led to an pentacarbene S = 5 oligomer [63]. However, for building high-spin polymers and oligomers, linear architectures are not ideal and most workers setting out along this path have incorporated elements of branching, rings or 3D networking into their designs. The first problem with linear architectures is that this type of design can lead to an unfavorable scaling of the energy difference between the ground and first excited spin states and hence an increased probability of thermal population of a low-spin state. The simplest cases are illustrated in Fig. 13. If we assume a Heisenberg model and that the coupling constant J associated with each 1,3-phenylene pathway in 33– 36 is the same this leads to a singlet-triplet splitting of 2J for metaquinodimethane

164

5 Triarylmethyl and Amine Radicals

Fig. 13. Scaling of the energy separation between high- and low-spin states as a function of the architecture of the oligomer.

33, a doublet-quartet splitting of J for the (hypothetical) triradical 34 and a doubletquartet splitting of 3J for the (hypothetical) triradicals 35 and 36 [64]. Clearly the last situation is to be preferred. The second argument for branched/cyclic/networked architectures rather than linear architectures is that in building organic polyradicals it is impossible to achieve absolute chemical purity and homogeneity. The effect of a site defect, where the synthesis fails to deliver the desired spin at one of the sites, is catastrophic for a linear chain since it decouples the ferromagnetically coupled spins on one side from those on the other side. However, cyclic and networked structures in which spins are linked through many independent pathways can sustain site defects. The percentage

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

165

of defect sites that can be sustained within a infinite spin lattice is usually calculated using percolation theory [47, 65]. The limit is 0% for a one-dimensional linear array (each spin with two nearest neighbors), ca. 30% for a two-dimensional hexagonal lattice (each spin with three nearest neighbors), ca. 43% for a two-dimensional square lattice (each spin with four nearest neighbors), ca. 54% for two such lattices stacked one above the other (each spin with five nearest neighbors) and ca. 69% for a threedimensional cubic lattice (each spin with six nearest neighbors. Clearly, the greater the number of connective pathways the better. An equivalent conclusion was reached using the analysis of monodisperse oligomeric structures introduced by Rajca [29]. This is shown in Fig. 14 in which the filled circles represent spin-half sites, the lines represent ferromagnetic coupling pathways and the open circles represent spinless or spin defect sites. In the linear and branched/dendritic architectures defects in or near the periphery hardly affect the total observable “spin value” whereas non-peripheral defects are very damaging. The advantage of the branched/dendritic structures over linear structures is that there is a higher proportion of peripheral sites and hence the probability of defects occurring at a damaging site is that much less. Monocyclic structures are better still since it takes two defect sites to significantly reduce the total “spin value” Rajca terms the linear and branched structures “zero-proof” and the cyclic structures “one-proof”. Clearly the situation becomes better still as further connectivity is introduced into the design. The third fundamental difficulty with linear architectures only really becomes apparent for very large arrays. This is the “dimensionality” problem. Just as it is impossible to form a one-dimensional crystal so it is impossible to envisage other disorder-order transitions such as the paramagnet-ferromagnet transition in a onedimensional system [66]. For a linear chain of N spins in which each center is ferromagnetically coupled only to its neighbor on either side, the energy lost in introducing a spin-coupling defect is independent of the length of the chain but the entropy gain scales as T ln N . However strong the local coupling, at real temperatures and with a sufficiently long chain there is bound to be “entropy-driven” formation of spin-coupling defects. For a combination of these reasons the preferred design strategies are networked > cyclic > branched/dendritic > linear. Both branched and cyclic tetraradical derivatives of the Schlenk hydrocarbon have been investigated [34, 59, 67–69]. Synthetic routes to these are shown in Figs. 15 and 16 [70]. The tetraradical 37a was obtained by oxidizing a 0.05 M solution of the tetraanion in tetrahydrofuran with two molar equivalents of iodine at −78◦ C [34, 67]. When the solution was frozen, the m = 1 region of the EPR spectrum showed the symmetrical eight peak pattern expected for a pentuplet tetraradical superimposed on a signal ascribed to a monoradical impurity. NMR measurement of the magnetic moment in the range 133–163 K gave µeff. = 4.3 µB which approaches the spin-only value of 4.9 µB . Even tetraradical 37c, the most sterically hindered of the tetraradicals shown in Fig. 15, has a pentuplet ground state with negligible thermal population of lower spin states [59] despite the fact that it must be significantly nonplanar. Attempts to prepare an equivalent perchlorinated tetraradical by treating the chlorocarbon 38 with tetrabutylammonium hydroxide and then chloranil unfortunately failed even when using a large excess of reagent and very long reaction

166

5 Triarylmethyl and Amine Radicals

Fig. 14. The effect of the number and position of “spin defects” on the “spin value” for linear, dendritic, and cyclic arrays of ferromagnetically coupled S = 1/2 sites.

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

167

Fig. 15. Syntheses of branched pentet tetraradicals [34, 67, 68]. Reagents (i) 1 mol. BuLi, −30◦ C; (ii) 4,4 -di-tert-butylbenzophenone; (iii) EtOCOCl; (iv) BuLi, −25◦ C, (MeO)2 CO; (v) EtOCOCl; (vi) Li, THF; (vii) I2 , −78◦ C. Ar = para-tert-butylphenyl.

168

5 Triarylmethyl and Amine Radicals

Fig. 16. Syntheses of a cyclic pentet tetraradical 39 [69]. Reagents (i) 1 mol. BuLi, <0◦ C; (ii) 3-bromo-4-tert-butylbenzophenone; (iii) NaH, THF, MeI; (iv) t-BuLi, −78◦ C; (v) 1-(4-tertbutylbenzoyl)pyrrolidine; (vi) t-BuLi, −78◦ C; (vii) acid work-up; (viii) NaH, THF, MeI; (ix) Na/K alloy, THF; (x) I2 , −95◦ C.

times [68]. Only triradical species could be obtained. The difficulty of obtaining the desired product was ascribed to the very high degree of steric hindrance. The cyclic tetraradical 39 was obtained in a similar manner to the branched tetraradicals 37 (Fig. 16) [69]. The m = 1 region of the EPR spectrum showed the expected eight-line pattern plus peaks due to about 10% doublet and quartet

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

169

Fig. 17. Syntheses of a sextet pentaradical 40 [60]. Reagents (i) 1 mol. BuLi; (ii) 4,4 -di-tertbutylbenzophenone; (iii) NaH, THF, MeI; (iv) tert-BuLi, −78◦ C; (v) B(OMe)3 ; (vi) Pd(PPh3 )4 , BaOH, toluene, MeOH; (vii) Na/K alloy, THF; (viii) I2 , 95◦ C.

impurities. A Brillouin function fit to the field dependence of the magnetization of a frozen 10−3 M solution at 2 K fitted a theoretical S = ca. 1.9 curve. The magnetization and EPR data for the pentaradical 40 (Fig. 17) can be fitted on the assumption of a sextet ground state and a spin system in which, the couplings 1,3 through benzene are strong (J1 /k 200 K) but the couplings 3,4 through biphenyl are weak (J2 /k ca. 90 K). Fits to the field dependence of the magnetization at 1.8–5 K give S = ca. 2.4 compared with an expected value of 2.5 [60].

170

5 Triarylmethyl and Amine Radicals

The heptaradicals 41–43 (Fig. 18) all have octet ground states and were prepared through simple variations of the routes already described [60, 71]. In the case of the heptaradical 41, susceptibility studies give a constant moment in the range 5– 80 K indicating strong ferromagnetic coupling. Although the dominant species is S = ca. 3.5 the data had to be fitted to a model in which only 93% of the α-carbons carry spin (7% are “spin-defect” sites). Magnetization data for a dilute solution in 2-methyltetrahydrofuran at >5 K fits Brillouin curves 3.5 > S > 2.5 but at lower temperatures significant intermolecular antiferromagnetic interactions lead to lower values [71]. The data obtained for the dendritic heptaradical 42 is very similar. The heptaradical 43 shows a temperature dependent moment arising from the fact that it is built upon 3,4 -coupling through the biphenyl unit which is relatively weak. The field-dependence of the magnetization at 5 K fits S = ca. 3.3 compared with the expected value of 3.5 [60]. As pointed out earlier, cyclic polyradical architectures are particularly desirable. The cyclic octaradical 44 (Fig. 19) was made by a variant on the route used to prepare the tetraradical 39 [72]. A 10−3 –10−2 M solution showed no significant intermolecular interactions and the field dependent behavior <10 K corresponded to S = ca. 3.8 [70, 72] which is close to the expected value of S = 4.0. The branched decaradical 45 (Fig. 19) gave a constant moment over the temperature range 20–100 K indicating strong local ferromagnetic coupling. Although the dominant species is S = 5.0 the data had to be fitted to a model in which only 93% of the α-carbons carry spin (7% are spin-defect sites). Magnetization data for a dilute solution in tetrahydrofuran at low temperatures indicated a predominant undecet ground state [29, 71]. Synthesis of the tricyclic tetradecaradical 47 (Fig. 20) is somewhat complicated by the multiplicity of diastereoisomers obtained for the precursor 46 [73]. However, three of these were isolated in pure form. The field dependence of the magnetization for a dilute matrix in deuterated THF/Me-THF at 1.8–5 K corresponded to an S = ca. 6.2 rather than S = 7.0 system. Somewhat surprisingly for a system in which all of the couplings are 1,3 through benzene, this system showed a strongly temperature dependent moment. Even at low temperatures there is significant thermal population of low-spin states. Furthermore, this polyradical proved particularly thermally labile. When a solution was left at room temperature for 24 h it degenerated giving an S = ca. 1 product. Quenching studies showed that the product had incorporated deuterium from the solvent. The unexpectedly weaker-than-usual spin-coupling and the poorer-than-usual stability both probably arise from increased non-planarity in the π-system which, in this case, is particularly sterically crowded. Attempts to build the very high-spin structures using the branched/dendritic architectures shown in Fig. 21 were rather unsuccessful. Attempts to prepare the pentadecaradical radical 48 gave a product for which the field-dependent magnetization data could not be fitted to a single Brillouin function and rather than being close to the expected value of S = 7.5 it fell in the range S = 2.5–3.5. Attempts to prepare the hentriaconaradical 49 were even less successful. In both cases the failure is most probably attributable to the presence of spin-defects at key non-peripheral sites. If this is the problem then it is likely to prove a fatal weakness in any attempt to build high-spin dendrimers [74].

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

Fig. 18. Octet ground state heptaradicals [60, 71]. Ar = para-tert-butylphenyl.

171

172

5 Triarylmethyl and Amine Radicals

Fig. 19. Nonet ground state octaradical 44 [72] and undecet ground state decaradical 45 [29, 71]. Ar = para-tert-butylphenyl.

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

173

Fig. 20. Synthesis of a tricyclic tetradecaradical 47 [73] (i) tert-BuLi 195–223 K; (ii) 1-(4-tertbutylbenzoyl)pyrrolidine; (iii) NaH, THF, MeI; (iv) Na/K alloy, THF; (v) I2 < 168 K. Ar = para-tert-butylphenyl.

174

5 Triarylmethyl and Amine Radicals

Fig. 21. Target very-high-spin dendrimer polyradicals [74]. Ar = para-tert-butylphenyl.

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

175

This failure of the “dendrimer” strategy together with the poor spin coupling and instability problems encountered with the tetradecaradical 47 seem like a dead-end so far as making higher spin systems based on the Schlenk hydrocarbon buildingblocks is concerned. However, an ingenious solution was found through a design that uses the strong one-proof cyclic motif found in 39 and 44 linked to other rings and branches, not 1,3 through benzene, which introduces steric hindrance, but 3,4

through biphenyl. The synthesis of the hexadecaradical 51, which is designed on this basis, is shown in Fig. 22. The precursor 50 was obtained as a single cis/trans isomer of low symmetry, the 1 H NMR in C6 D6 showing two non-equivalent tert-Bu resonances for each of the four diether branches and eight other non-equivalent tert-Bu resonances. Fits to the magnetization of 51 against field at low temperatures indicates values of S in the range 6.6–7.2 compared to an expected value of 8.0. The penalty for employing the 3,4 biphenyl coupling unit is that the ferromagnetic coupling through this particular unit is weak and the moment strongly temperature dependent [60]. The current limit of this line of development is reached with the tetracosaradical 52 whose synthesis is shown in Fig. 23. This should yield S = 12.0. The magnetization data in the temperature range 1.8–5 K corresponds quite closely to S = 10.0 (Fig. 24). The shortfall is ascribed to a small population of defect sites which are bound to arise if there is anything short of 100% step-yield at each α-carbon site. The data obtained can be fitted on the assumption of a step-yield of 98% [75]. The importance of these studies of monodisperse oligomers is that they have established both the design rules for “high-spin” organics and the protocols necessary for their characterization. In terms of design the preference must be network > cyclic > branched/dendritic > linear and clearly it is going to be very difficult to make further progress with anything other than polycyclic or networked architectures. In all cases the presence of defects means that the observed total spin values are less than the theoretical maximum because of spin imperfections inherent in the synthetic methods employed but this is only disastrous for linear and branched architectures. The requirement for strong local ferromagnetic coupling can be achieved by building on the Schlenk hydrocarbon motif and exploiting 1,3-coupling through benzene but this can conflict with the equally important need for a reasonable degree of planarity in the π system and for an acceptable level of thermal stability. Hence it has proved necessary to exploit the relatively weak 3,4 -coupling through biphenyl (<100 K). All of the Schlenk hydrocarbon derivatives described in this section are highly sensitive to the air necessitating vacuum-line-handling techniques. Some are also close to the limit of thermal stability requiring manipulations to be carried out using chilled solutions. In all cases intermolecular anti-ferromagnetic interactions complicate the analysis of the data [76]. In these polyradicals there is quite strong ferromagnetic through-bond coupling but, in the condensed state, there are many through-space contacts most of which are only weakly antiferromagnetic. None-the-less their sum is significant and the full intramolecular coupling is only observed when the polyradicals are dispersed in a dilute frozen matrix (typically 10−2 –10−3 M). If the sample cannot be diluted to a level at which these intermolecular effects are negligible, as with the data shown in Fig. 24, a mean field correction sometimes has to be made, in-

176

5 Triarylmethyl and Amine Radicals

Fig. 22. Synthesis of a hexadecyl radical 51 [60] Reagents (i) t-BuLi; (ii) B(OMe)3 ; (iii) Pd(PPh3 )4 , Ba(OH)2 , toluene-methanol; (iv) Na/K alloy; (v) I2 . Ar = para-tert-butylphenyl.

5.5 Monodisperse High-spin Oligomers (S = 2 – ca. 10)

177

Fig. 23. Synthesis of a tetracosyl radical [75]. Reagents (i) tert-BuLi, 195–253 K; (ii) ZnCl2 , Et2 O, 195–253 K; (iii) Pd(PPh3 )4 , 100◦ C; (iv) Na/K alloy, THF; (v) I2 . Ar = para-tertbutylphenyl.

178

5 Triarylmethyl and Amine Radicals

Fig. 24. SQUID magnetometry (H = 0–5.0 T) for the polyradical 52 in d8 -THF, θ = −0.03 K. The experimental data are represented by the symbols and the theoretical fits by the continuous lines. The fitting parameters for the field-dependence plots at T = 1.8, 3, 5, and 10 K are S = 10.0, 10.0,10.0, 9.8 and Msat × 102 = 2.01, 2.01, 2.00, and 1.96 emu. The fitting parameters for the temperature dependence plot are J/kB = 6.6 K, N = 1.33 × 10−7 mol and Mdia = 1.3 × 10−5 emu.

troducing an additional disposable parameter. The intrinsically dilute nature of these systems (the spin-bearing units have a RMM of the order of 400) and the need to use dilute solutions places an absolute premium on the sensitivity of the magnetometer employed [77].

5.6 High-spin Polymers (up to Sn = ca. 48)

5.6

179

High-spin Polymers (up to Sn = ca. 48)

In most of the oligomer systems described in the previous section it is possible to achieve the desired high dilution and to virtually eliminate intermolecular antiferromagnetic interactions. The paramagnetic molecules are then isolated from each other and because the spin systems are also close to being monodisperse the magnetization versus field plots can be analyzed by fitting to the appropriate Brillouin function. In polymer systems, although it is normal practice to measure their properties in as dilute a matrix as is practicable, it can be difficult to obtain a sufficiently dilute solution to be sure that complications arising from intermolecular and intramolecular through-space antiferromagnetic interactions have been eliminated. Furthermore, the spin systems in polymers are always polydisperse and, even if the problem of intermolecular interactions is overcome, the magnetization versus field plots cannot be fitted to a single Brillouin function {In comparison to plots for single-spin species, the magnetization for a disperse spin system apparently increases too rapidly at low fields (where the behavior is dominated by the larger moments) and too slowly at high fields (where the behavior is dominated by the smaller moments) [78]}. This makes a rigorous analysis of the magnetometer data for the polymers difficult and so, under these circumstances, a “by eye” comparison of the data with Brillouin plots for single spin species has often been used by workers to give a rough estimate of the magnitude of the “effective spin”. The synthesis of the simplest type of high-spin polymer based on a linear repeating Schlenk hydrocarbon motif 53 is shown in Fig. 25 [79]. The resultant polyradical is unstable and the intensity of the EPR signal decreases rapidly at room temperature. The results of magnetization studies at 10 K fitted “by eye” to a Brillouin functions for S = 2. Although the desired limiting elimination of intermolecular antiferromagnetic interactions was not achieved in this case, the data indicates an average chain length of at least four ferromagnetically coupled spins.

Fig. 25. Synthesis of a simple linear oligomer of the Schlenk hydrocarbon [79]. Reagents (i) BuLi; (ii) methyl benzoate; (iii) P, HI; (iv) KH; (v) I2 , 180 K.

A much more elaborate but also much more successful synthesis is shown in Fig. 26 This was based on that of the high spin oligomer 52 (Compare Fig. 26 with Fig. 23) and contains a randomly networked array of cyclic Schlenk hydrocarbon tetramers liked to each other 3,4 through biphenyl [80]. Magnetization of the polymer 54 as a function of temperature and field is shown in Fig. 27. The data was fitted to a model in which it was assumed that 97% of the α-carbons bear spin and that there

180

5 Triarylmethyl and Amine Radicals

Fig. 26. Synthesis of a high-spin polymer 54 based on networked rings of triarylmethyl radicals [80]. Reagents (i) tert-BuLi, 195–253 K; (ii) ZnCl2 , 195–293 K; (iii) Pd(PPh3 )4 ; (iv) Na/K alloy, 283 K; (v) I2 , 170 K. Ar = para-tert-butylphenyl.

5.6 High-spin Polymers (up to Sn = ca. 48)

181

Fig. 27. SQUID magnetometry for high-spin polymer 54 in THF-d8 . Plot A: T against T . Plot B: M/Msat against H/T [80]. (Reproduced by permission of the American Chemical Society.) Representative parameters (with standard errors) for the number average, Sn = 48 ± 2, (Ss 66) at 1.8 K: p = 0.968 ± 0.003 and Msat 102 = 1.072 ± 0.004 emu (in the units of magnetic moment) with the parameter dependence of 0.094; S0 , n, and w12 are set to 6.0, 13, and 0.2, respectively. At T = 3, 5, 10, and 20 K, the Sn are 45, 40, 30, and 20, respectively.

are relatively low-lying easily thermally populated low-spin states. Defining number average S, Sn as i xi Si /σi xi (xi = fraction of systems with spin Si ) and a spinaverage S, Ss as i xi Si2 /i xi Si and fitting the field dependence data for a polymer Mn ca. 105 reproducibly gave Sn > 40 at 1.8 K and at best Sn = 48, Ss ca. 66. Sn decreases steeply with increasing temperature. Samples left at room temperature for several weeks decomposed and on cooling to low temperatures <1% of the original magnetism remained showing that it is indeed due to the organic component and not adventitious metal or metal oxide impurities. One of the first attempts to produce a ferromagnetic polymer was due to Torrance et al who reported that when sym-triaminobenzene was oxidized with iodine [81] it produced a polymer showing weak ferromagnetism (hysteresis behavior). It was admitted that the results obtained were irreproducible and although attempts were made to exclude the possibility that this was due to adventitious metal or metal oxide

182

5 Triarylmethyl and Amine Radicals

Fig. 28. High-spin polymer 55 based on repeating metaphenylenediamine radical cation spinbearing units [82]. Reagents (i) CuI/K2 CO3 ; (ii) NOBF4 .

impurities this still seems the most likely explanation. No other groups have reported successfully repeating these experiments. The structure of the polymer obtained by Torrance et al. is not at all clear. Structurally better defined are the linear metaphenylenediamine polymers obtained by Tanaka and co-workers the synthesis of which is shown in Fig. 28 [82]. The doped polymer 55 shows an “effective S” of 1.0 at 2 K. A synthesis of networked polymers 56 in which the coupling is 4,4 through metaterphenyl is shown in Fig. 29 [47, 53, 83]. Variants using alternative “ferromagnetic coupling units” [84] and “spin-bearing units” [85] have also been reported. These polymers proved difficult to “dope” in a clean and quantitative manner [53, 84]. The neutral polymers have limited solubility in chlorocarbon solvents but are even less soluble when oxidized to the radical cation. As a result they tend to precipitate from solution in an incompletely oxidized form. Even partially “doped” materials show ferromagnetic spin-coupling as evidenced by magnetometer and EPR nutation resonance studies [86]. When “doped” in very dilute solution essentially quantitative formation of the amine radical cation is achieved [85, 87]. When analyzed in an equivalent manner to that employed for polymer 54, the magnetization data for polymer 56 can be fitted to a distribution Sn = ca. 7.3, Ss = ca. 10.7 (Fig. 30). These doped amines are much more thermally stable than the all-carbon systems. Samples of the neat polymer 56 can be stored indefinitely at room temperature under vacuum. When air is let into the sample the spin concentration only declines by ca. 10% per day.

5.7

Conclusions and Prospects (Beyond S = ca. 48?)

Even although none of these polymers show hysteresis or remnant magnetization behavior, there has been significant progress in the last few years. Methods for the synthesis and characterization high-spin organics based on triarylmethyl radical and triarylamine radical cation building blocks have been established and the design rules for high-spin organics have been clarified. For the monodisperse oligomers cited in this chapter, we have seen S = 0.5, 2 (1900) [3]; S = 1.0, 13 (1915) [24]; S = 1.5, 27 (1937) [55]; S = 2.0, 37 (1990) [34]; S = 6.6–7.2, 51 (1997) [58]; S = ca. 10,

5.7 Conclusions and Prospects (Beyond S = ca. 48?)

183

Fig. 29. Synthesis of high-spin polymers 56 in which the amine radical cation centers are coupled 4,4 through metaterphenyl [83]. Reagents. (i) K2 CO3 /Cu; (ii) BBr3 ; (iii) RBr/K2 CO3 ; (iv) Br2 ; (v) BuLi/−78◦ C; (vi) (iPrO)3 B/−78◦ C; (vii) HCl/H2 O; (viii) Pd(PPh3 )4 ; (ix) NOBF4 .

52 (1998) [75]. For the polymers progress has been even more rapid: S = ca. 1.0, 55 (1992) [82]; S = ca. 2.0, 53 (1993) [79]; S = ca. 2.5, 56 (1996) [83]; S = ca. 3.5, 56 (1998) [84]; Sn = ca. 7.5, 56 (1999) [85]; Sn = ca. 48, 54 (1999) [80]. On this basis it is safe to predict that we will shortly see ferromagnetically coupled spin-clusters well beyond these limits. However this is only a part of the problem [29, 37, 47, 48, 88]. The ultimate ferromagnetic polymer would combine room temperature magnetism and good processability with heat and air stability. These are not easy to obtain! One promised advantage of these polymers [1] is that they exploit a through-bond spin-coupling mechanism and so it should be possible to obtain stronger exchange interactions than in the nitronylnitroxides [89] and other molecular solids where the

184

5 Triarylmethyl and Amine Radicals

Fig. 30. Field dependence at 2 K of the magnetization of a dilute matrix of the polymer 56 R = C4 H9 , Mw 8500, fitted to a theoretical line calculated for Sn = 7.3, Ss = 10.7 by summation of Brillouin functions [85].

spin-coupling is through-space [90]. Unfortunately, in polymers such as 54 and 56, it has proved necessary to introduce the weak 3,4 -biphenyl and 4,4

-metaterphenyl spin-coupling units and so this promised advantage has been lost. As a result, very high-spin behavior is limited to very low temperatures. Systems that exploit 1,3phenyl and other strong coupling units will need to be reinvestigated. The polymers 54 and 56, rely on solution chemistry for their synthesis and for the step through which the spin is introduced. This is important since they rely on the π bond connectivity to mediate the spin-spin interaction so that the upper limit to the ferromagnetically coupled spin-clusters is ultimately determined by their molecular weight and hence by their solubility! Hence, another practical challenge is to develop non-solution methods of polymer synthesis and for the “doping” step {a solution to this may already have been found [91]}. Finally, perhaps the most difficult problem to solve is that of stability. Compared to inorganic materials, the structures of organic compounds are easier to optimize but, from the standpoint of magnetic materials they only provide a limited palate of stable spin-bearing building blocks and those which are available seem very difficult to reconcile with the competing needs for processability and strong spin-coupling. Among these the triarylmethyl radical and triarylamine radical cation building blocks remain some of the most promising. The further exploration of these will undoubtedly provide an interesting challenge to chemists for many years to come.

References

185

References [1] (a) N. Mataga, Theoret. Chim. Acta 1968, 10, 372–376; (b) A. A. Ovchinnikov, ibid. 1978, 47, 297–304. [2] V. D. Scholle, E. G. Rozantsev, Russ. Chem. Rev. 1973, 42, 1011–1019. [3] M. Gomberg, Chem. Ber. 1900, 33, 3150–3163. [4] H. Lankamp, W. T. Nauta, C. MacLean, Tetrahedron Lett. 1968, 249–254. [5] J. M. McBride, Tetrahedron 1974, 30, 2009–2022. [6] P. Anderson, Acta Chim. Scand. 1965, 19, 622–628, 629–637. [7] P. Anderson, B. Klewe, Acta Chim. Scand. 1962, 16, 1817–1818, 1967, 21, 2599–2607. [8] J. Veciana, J. Carilla, C. Miravitlles, E. Molins, J. Chem. Soc., Chem. Commun. 1987, 812–814. [9] W. A. Waters, Chemistry of Free Radicals, Clarendon Press, Oxford, 1948. [10] P. W. Selwood, R. M. Dobres, J. Amer. Chem. Soc. 1950, 72, 3860–3863, [11] D. Dunnebacke, W. P. Neumann, A. Penenory, U. Stewen, Chem. Ber. 1989, 122, 533–535. [12] H. Lund, J. Amer. Chem. Soc. 1927, 1346–1360. [13] M. J. Sabacky, C. S. Johnson, R. G. Smith, H. S. Gutowsky, J. C. Martin, J. Amer. Chem. Soc. 1967, 89, 2054–2058. [14] M. Ballester, Acc. Chem. Res.1985, 18, 380–387. [15] (a) S. T. Bowden, W. J. Jones, J. Chem. Soc. 1928, 1149–1158; (b) G. R. Luckhurst, J. N. Ockwell, Tetrahedron Lett. 1968, 4123–4126. [16] S. C. Creason, J. Wheeler, R. F. Nelson, J. Org. Chem. 1972, 37, 4440–4446. [17] R. F. Nelson, R. H. Philip, J. Phys. Chem. 1979, 83, 713–716. [18] G. M. Brown, G. R. Freeman, R. I. Walter J. Amer. Chem. Soc. 1977, 99, 6910–6915. [19] D. H. R. Barton, R. K. Haynes, G. Leclerc, P. D. Magnus, I. D. Menzies, J. Chem. Soc., Perkin Trans. 1 1975, 2055–2065. [20] R. I. Walter, J. Amer. Chem. Soc. 1955, 77, 5999–6002. [21] P. B. Ayscough, A. P. McCann, R. Wilson, Proc. Chem. Soc. 1961, 16–17. [22] H. van Willigen, J. Amer. Chem. Soc. 1967, 89, 2229–2230. [23] (a) O. Stark, O. Garben, Chem. Ber. 1913, 46, 659–666; (b) O. Stark, O. Garben, ibid. 1913, 46, 2252–2259; (c) O. Stark, O. Garben, L. Klebahn ibid. 1913, 46, 2542–2544; (d) O. Stark, L. Klebahn ibid. 1914, 47, 125–130. [24] W. Schlenk, M. Brauns Chem. Ber. 1915, 48, 661–669. [25] (a) W. Schlenk, M. Brauns Chem. Ber. 1915, 48, 716–728; (b) W.Schlenk, M Brauns, ibid. 1915, 48, 661–669. [26] A. Rajca, S. Rajca, S. R. Desai, V. W. Day J. Org. Chem. 1997, 62, 6524–6528. [27] G. Kothe, K.-H. Denkel, W. Summermann, Angew. Chem., Int. Edn. English 1970, 9, 906–907. [28] G. R. Luckhurst, G. F. Pedulli, M. Tiecco J. Chem. Soc. B 1971, 329–334. [29] A. Rajca, Chem. Rev. 1994, 94, 871–893. [30] P. G. Wenthold, J. B. Kim. W. C. Lineberger, J. Amer. Chem. Soc. 1997, 119, 1354–1359. [31] K. P. Meurer, F. Vogtle, Topics Curr. Chem. 1985, 127, 1–76. [32] J. Veciana, C. Rovira, M. I. Crespo, O. Armet, V. M. Domingo, F. Palacio, J. Amer. Chem. Soc. 1991, 113¸ 2552–2561. [33] A. Rajca, S. Rajca, J. Chem. Soc., Perkin Trans. 2 1998, 1077–1082. [34] A. Rajca, J. Amer. Chem. Soc. 1990, 112, 5890–5892. [35] A. Rajca, S. Utamaanya, J. T. Xu, J. Amer. Chem. Soc. 1991, 113, 9235–9241. [36] A. Rajca, S. Utamapanya, J. Org. Chem. 1992, 57, 1760–1767. [37] R. J. Bushby in Magnetic Properties of Organic Materials, (Ed. P. M. Lathi), M Dekker, NY, 1999, pp. 179–185.

186

5 Triarylmethyl and Amine Radicals

[38] K. Yoshizawa, A. Chano, A. Ito, K. Tanaka, T. Yamabe, H. Fujita, J. Yamauchi, Chem. Lett. 1992, 369–372. [39] K. Yoshizawa, A. Chano, A. Ito, K. Tanaka, T. Yamabe, H. Fujita, J. Yamauchi, M. Shiro, J. Amer. Chem. Soc. 1992, 114, 5994–5998. [40] K. R. Stickley, S. C. Blackstock, Tetrahedron Lett. 1995, 36, 1585–1588. [41] (a) R. J. Bushby, D. R. McGill, K. M. Ng, N. Taylor, J. Chem. Soc. Chem. Commun. 1996, 2641–2642; (b) R. J. Bushby, D. R. McGill, K. M. Ng, N. Taylor, J. Chem. Soc. Perkin Trans. 2 1997, 1405–1414. [42] K. Sato, M. Yano, M. Furuichi, D, Shiomi, T. Takui, K. Abe, K. Itoh, A. Higuchi, K. Katsuma, Y. Shirota, J. Amer. Chem. Soc. 1997, 119, 6607–6613 [43] K.R. Stickley, S.C. Blackstock, J. Amer. Chem. Soc. 1994, 116, 11576–11577. [44] (a) M.M. Wienk, R.A.J. Janssen, J. Chem. Soc., Chem. Commun. 1996, 267–268; (b) M.M. Wienk, R.A.J. Janssen, J. Amer. Chem. Soc.1996, 118, 10626–10628; (c) K. R. Stickley, T. D. Selby, S. C. Blackstock, J. Org. Chem. 1997, 62, 448–449. [45] G. Allinson, R. J. Bushby, J.-L. Paillaud, J. Mater. Sci.: Mater. Electronics 1994, 5, 83–88. [46] W. T. Borden, E. R. Davidson, J. Amer. Chem. Soc. 1977, 99, 4587–4594. [47] R. J. Bushby, D. R. McGill, K. M. Ng in Magnetism a Supramolecular Function, (Ed.: O. Kahn), Kluwer, Dordrecht, 1996, pp. 181–204. [48] D. A. Dougherty, Acc. Chem. Res. 1991, 24, 88–94. [49] A. Rajca, S. Rajca, J. Amer. Chem. Soc. 1996, 118, 8121–8126. [50] A. Rajca, S. Utamapanya, D. J. Smithhisler, J. Org. Chem. 1993, 58, 5650–5652. [51] T. D. Selby, K. R. Stickley, S.C. Blackstock, Org. Lett. 1999, 2, 171–174. [52] T. D. Selby, S.C. Blackstock, Org. Lett. 1999, 1, 2053–2055. [53] R. J. Bushby, D. R. McGill, K. M. Ng, N. Taylor, J. Mater. Chem. 1997, 7, 2343–2354. [54] (a) M. S. Platz in Diradicals, (Ed: W. T. Borden), Wiley, New York, 1982, pp. 195–258; (b) J. A. Berson in The Chemistry of the Quinonoid Compounds Vol. 2 Part 1, (Eds.: S. Patai, Z. Rappoport), Wiley, New York, 1988, pp. 455–536. [55] (a) M. Leo, Chem. Ber. 1937, 70, 1691–1694; (b) G. Schmauss, H. Baumgartel, H. Zimmerman, Angew. Chem. Internat. Edn. Engl. 1965, 4, 594–564; (c) W. Wilker, G. Kothe, H. Zimmerman, Chem. Ber. 1975, 108, 2124–2136. [56] J. Veciana, C. Rovira, N. Ventosa, M. I. Crespo, F. Palacio, J. Amer. Chem. Soc. 1993, 115, 57–64. [57] K. Yoshizawa, M. Hatanaka, H. Ago, K.Tanaka, T. Yamabe, Bull. Chem. Soc. Japan 1996, 69, 1417–1422. [58] K. R. Stickley, T. D. Selby, S. C. Blackstock, J. Org. Chem. 1997, 62, 448–449. [59] A. Rajca, S. J. Utamapanya, J. Amer. Chem. Soc. 1993, 115, 2396–2401. [60] A. Rajca, J. Wongsriratanakul, S. Rajca, J. Amer. Chem. Soc. 1997, 119, 11674–11686. [61] S. Rajca, A. Rajca, J. Amer. Chem. Soc. 1995, 117, 9172–9179. [62] (a) H. Iwamura, Pur. Appl. Chem. 1986, 58, 187–196; (b) Y. Teki, K. Itoh in Magnetic Properties of Organic Materials, (Ed. P. Lathi), 1999, Marcel Dekker, N.Y. [63] I Fujita, Y. Teki, T. Takui, T. Kinoshita, K. Itoh, F. Miko, Y. Sawaki, H. Iwamura, A. Izuoka, T. Sugawara, J. Amer. Chem. Soc. 1990, 112, 4074–4075. [64] E. Sin, Coord. Chem. Rev. 1970, 5, 313–347. [65] D. Stauffer, A. Aharony, Introduction to Percolation Theory, Taylor and Frances, London 1994. [66] P.M. Chaitkin in Review of the Physics of Low–Dimensional Systems, (Eds.: J. S. Miler, A. J. Epstein), NY, Acad. Sci., 1977, pp. 128–144. [67] A. Rajca, J. Amer. Chem. Soc. 1990, 112, 5889–5890. [68] D. Ruiz–Molina, J. Veciana, F. Palacio, C. Rovira, J. Org. Chem. 1997, 62, 9009–9017. [69] A. Rajca, S. Rajca, S. R. Desai, J. Amer. Chem. Soc.1995, 117, 806–816.

References

187

[70] (a) A. Rajca, S. Janici, J. Org. Chem. 1994, 59, 7099–7107; (b) A. Rajca, J. Org. Chem. 1991, 56, 2557–2563. [71] (a) A. Rajca, S. Utamapanya, S. Thayumanavan, J. Amer. Chem. Soc. 1992, 114, 1884– 1885; A. Rajca, S. Utamapanya, Mol. Cryst. Liq. Cryst. 1993, 232, 305–312. [72] A. Rajca, S. Rajca, R. Padmakumar, Angew. Chem., Int. Edn. Engl. 1994, 33, 2091–2093. [73] A. Rajca, K. Lu, S. Rajca, J. Amer. Chem. Soc. 1997, 119, 10335–10345. [74] A. Rajca, S. Utamapanya, J. Amer. Chem. Soc. 1993, 115, 10688–10694. [75] A. Rajca, J. Wongsriratanakul, S. Rajca, R. Cerny Angew. Chem, Inter. Edn. Engl. 1998, 37, 1229–1232. [76] K. K. Anderson, D. A. Doughery, Adv. Mater. 1998, 10, 668. [77] A. Rajca, Mol. Cryst Liq. Cryst. 1997, 305, 567–577. [78] M. M. Murray, P. Kaszynski, D. A. Kaisaki, W. Chang, D. A. Dougherty, J. Amer. Chem. Soc. 1994, 116, 8125–8161. [79] S. Utamapanya, H. Kakegawa, L. Bryant, A. Rajca, Chem. Mater. 1993, 5, 1053–1055. [80] A. Rajca, S. Rajca, J. Wongsriratanakul, J. Amer. Chem. Soc. 1999, 121, 6308–6309. [81] J. B.Torrance, S. Oostra, A. Nazzal, Syn. Metals 1987, 19, 709–714. [82] (a) K. Yoshizawa, K. Tanaka, T. Yamabe, J. Yamauchi, J. Chem. Phys. 1992, 96, 5516– 5522; (b) A. Ito, K. Ota, K. Yoshizawa, K. Tanaka, T. Yamabe, J. Yamauchi, Chem. Phys. Lett. 1994, 223, 27–34; (c) A. Ito, K. Ota, K. Tanaka, T. Yamabe, T. Yamauchi, K. Yoshizawa, Macromol.1995, 28, 5618–5625. [83] R. J. Bushby, K. M. Ng, J. Chem. Soc., Chem. Commun. 1996, 659–670; R. J. Bushby, D. R. McGill, K. M. Ng, N.Taylor, ibid. 1996, 2641–2642. [84] R. J. Bushby, D. Gooding, J. Chem. Soc., Perkin Trans. 2 1998, 1069–1075. [85] R. J. Bushby, D. Gooding, M. E. Vale, Phil. Trans. Roy. Soc. 1999, 357, 2939–2957. [86] D. Shiomi, K. Sato, T. Takui, K. Itoh, D. R. McGill, K. M. Ng, R. J. Bushby, Syn. Met. 1997, 85, 1721–1722; D. Shiomi, K. Sato, T.Takui, K. Itoh, D. R. McGill, K. M. Ng, R.J.Bushby, Mol. Cryst. Liq. Cryst. 1997, 306, 513–520. [87] R. J. Bushby, D. Gooding, M. Thornton–Pett, M. E. Vale, Mol. Cryst. Liq. Cryst. 1999, 334, 167–176. [88] Magnetic Properties of Organic Materials, (Ed. P. M. Lathi), M. Dekker, NY, 1999 [89] H. Tamura, Y. Nakazawa, D. Shiomi, K. Nozawa, Y. Hosokoshi, M. Ishikawa, M. Takahashi, M. Kinoshita, Chem. Phys. Lett. 1991, 186, 401–404. [90] J. Veciana, J. Cirujeda, J. J. Novoa, M. Deumai in Magnetic Properties of Organic Materials, (Ed. P. M. Lathi), M Dekker, NY, 1999, pp. 573–600. [91] Professor Rajca, who kindly read a draft of this chapter comments that “the problem of generating triarylmethyl radicals from insoluble polyether precursors corresponding to network 52 (Fig. 26) was solved; the number average Sn > 300 is reproducibly obtained. This and the AC susceptibility studies will be described in the upcoming VIIth ICMM meeting in San Antonio in the September of 2000”.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

6

High-spin Metal-ion-containing Molecules Talal Mallah and Arnaud Marvilliers

6.1

Introduction

The discovery of magnetic bistability in Mn12 where the magnetization stays blocked after removing an external applied magnetic field [1] has prompted several research groups to design new polynuclear complexes bearing a low-lying high spin ground state [2]. The origin of this behavior is due to the axial magnetic anisotropy and the relatively large magnetic moment (S = 10) of the molecule [3] as for metallic or metal-oxide superparamagnetic particles [4]. The originality of the molecular systems like Mn12 is the strict monodispersion of the sample which allow to ascribe the observed phenomenon to a single molecule. One-molecule based devices for information storage may now become a reality. However, if Mn12 were to be used, the device should be refrigerated to below liquid helium temperature since above 4 K relaxation effects will destroy the induced magnetic moment. Thus, one of the challenges in this area is the synthesis of new molecules possessing very high spin ground state well separated from the first excited states and having a large magnetic anisotropy so that the blocking of the magnetic moment may be observed at high temperature. For example, Mn12 has an S = 10 spin ground state with a zero field splitting parameter due to an Ising type anisotropy D = −0.5 cm−1 [5]. This leads to an anisotropy energy barrier of 50 cm−1 (DS2z ) and hysteresis loops are observed only at temperatures lower than 3 K. If bistability is to be present at 30 K for example, the anisotropy energy barrier has to be as high as 500 cm−1 assuming everything else being equal. A high spin state is a necessary but not a sufficient condition to observe the blocking of the magnetization. Powell’s [Fe17 + Fe19 ] compound has the highest spin ground state S = 33/2 reported to date in a polynuclear system [6]. But, probably because of the very small magnetic anisotropy of the molecules, no blocking temperature had been observed down to T = 2 K. Several research programs throughout the world are devoted to the preparation of polynuclear complexes containing a large number of paramagnetic metal ions. Two approaches are mainly used leading in several cases to high spin molecules. The first approach is a “one-pot”, the second is a rational one. The paper is organized in five parts, after this introduction the three following sections are devoted to the self-assembly approach, then to the host-guest approach and finally to the rationale strategy. Within each part, the magnetic properties of the chemical systems will be presented and discussed. The last section consists of a general conclusion.

190

6 High-spin Metal-ion-containing Molecules

6.2

Self-assembly of Molecular Clusters

To date it seems difficult to rationalize the synthetic approaches that have been used to prepare the polynuclear clusters by a one-pot approach. However, many beautiful reactions were performed leading to interesting new systems. The common feature of almost all the reactions and the complexes obtained is the presence of particular chelating ligands. These ligands possess the general following characteristics: (i) they have more than one coordination mode, (ii) may be terminal and bridging simultaneously, (iii) can be bidentate or tridentate in some cases without stabilizing mononuclear complexes. We focus on two families of compounds where two archetypal ligands play the key role: carboxylate and 2-hydroxypyridine derivatives. A common structural feature to most of the polynuclear complexes of the carboxylate and the 2-hydroxypyridine families is the presence of metal ions sitting at the corners of a triangle leading to competition between the antiferromagnetic interactions between the metal ions. This makes very difficult the prediction of the nature of the ground state of such molecules.

6.2.1

Competing Interactions and Spin Frustration

Let us take a simple example of three S = 2 spins sitting on the corners of an isosceles triangle as shown in Scheme 1 and see how the nature of the ground state varies as a function of the relative amplitude of the exchange interactions between the local spins. The spin Hamiltonian (Eq. 1) used to calculate the energy levels due to the interaction of the three local spins S A , S B and SC is given by:

Scheme 1

H = −J (S A · S B + S A · SC + S B · SC ) + (J − J )(S B · SC )

(1)

The expression for the energy levels (Eq. 2) is: J − J ∗ ∗ J E(S, S ∗ ) = − S(S + 1) + S (S + 1) 2 2

(2)

where S ∗ = S B + SC to |S B − SC | and S = S ∗ + S A to |S ∗ − S A |. Assuming only antiferromagnetic interactions, the ground state may be S = 2, 1 or 0 depending on the J /J ratio (if we restrict J /J in the range 0–2) as shown in Fig. 1. Figure 1 reveals other important features. For J /J = 1/2, 2/3, and 3/2, the ground state is accidentally degenerate and the system is said to be frustrated (Toulouse original

6.2 Self-assembly of Molecular Clusters

191

Fig. 1. Plot of −E(S, S ∗ ) against A for an isosceles triangle of S = 2 spins.

work referred to a square Ising-type spin topology with three ferromagnetic and one antiferromagnetic interactions) [7]. A small perturbation in the chemical surroundings of the metal ions may slightly change the J /J ratio so that degeneracy is lifted leading to a well defined ground state. A system with a non-degenerate ground state is in a situation of stable equilibrium as a result of the competition between the exchange coupling interactions. We must stress that in the particular example of three S = 2 local spins, no spin frustration occurs when J /J is equal to one (the ground state is not degenerate); the magnetic properties of the system are those of a S = 1 ground state. Many molecules prepared by the one-pot approach have spin topologies that may lead to competing interactions so that the spin ground state is difficult to predict. A very small perturbation in the chemical surroundings of some of the metal ions may slightly change the amplitude of the coupling and thus leads to a completely different spin ground state. For example, II the complexes of general formula [MnIII 2 Mn O(O2 CR)6 L3 ] where R is CH3 or C6 H5 and L is C5 H5 N or H2 O (Fig. 2), spin ground states ranging from 1/2 to 13/2 have been observed depending on the nature of R and L which induces different exchange interactions between the metal ions within the triangle [8].

Fig. 2. PLATON projection of [Mn3 O(2Fbenzoato)6 (Pyr)3 ].2MeCN, which has crystallographic C2 symmetry. Thermal ellipsoids at the 50% probability level. The F-benzene moieties and the hydrogen atoms have been omitted for clarity (reproduced with permission).

192

6.2.2

6 High-spin Metal-ion-containing Molecules

The Carboxylate Family

The triangle arrangement of the trinuclear complex shown in Fig. 2 is found in many of the FeIII , MnIII and MnIV high nuclearity clusters belonging to the carboxylate family. This is a very stable framework that can be obtained with the metal ions FeIII , CrIII , MnII , MnIII or MnIV with pyridine or/and water to complete the coordination sphere of the metal ions. Depending on the nature of the metal ion, its oxidation degree and the terminal ligands different spin ground states may be stabilized [9]. The importance of these trinuclear complexes lies in their use as starting point for the synthesis of higher nuclearity complexes. A hexanuclear complex has been prepared by the association of two fragments of the trinuclear FeIII one [10]. The addition of three molar equivalents of the chelating ligand 1,1-bis(N -methylimidazol-2yl)ethanol (Scheme 2) to one molar equivalent of [FeIII 3 O(O2 CCH3 )(C5 H5 N)3 ]ClO4 leads to the formation of a hexanuclear complex (Scheme 3) made of two fragments of the original one by displacement of some of the carboxylate ligands. For six interacting FeIII ions (S = 5/2) the spin of the ground state can take a value between 0 and 15 depending on the relative amplitude and the nature of the exchange interaction between neighboring metal ions. Competing spin interactions lead in this case to a spin ground state S = 5.

Scheme 2

Scheme 3

Starting from a similar but slightly different trinuclear complex [FeIII 3 O(O2 CCH2 Cl)(H2 O)3 ]NO3 in the presence of Fe(NO3 )3 lead to a cyclic decanuclear FeIII complex (Fig. 3) [11]. The nature of the ground state in such molecule is easy to predict; depending on the nature of the exchange coupling interaction, it may be S = 0 (antiferromagnetic) or S = 25 (ferromagnetic). Magnetization measurements show clearly a S = 0 ground state [12].

6.2 Self-assembly of Molecular Clusters

193

Fig. 3. ORTEP drawing of the planar projection of [Fe(OMe)2 (O2 CCH2 Cl)]10 with 50% probability thermal ellipsoids and atom labels; prime and unprimed atoms are related by the center of inversion. Chlorine and hydrogen atoms are omitted for clarity (reproduced with permission).

Tetranuclear manganese complexes possessing a butterfly-like core have been used as starting point to prepare high nuclearity clusters (Scheme 4); Mn(4) and Mn(3) form the body of the butterfly and the two fragments Mn(1)–O and Mn(2)–O the wings [13].

Scheme 4

Many compounds having the same core with FeIII , MnIII , and VIII exist [14]. Christou and coworkers were able to merge two Mn tetranuclear [Mn4 O2 (O2 CCH3 )7 (pic)2 ]− (pic− is the deprotonated picolinic acid) molecules into a new octanuclear one [Mn8 O4 (O2 CCH3 )12 (pic)4 ] (Fig. 4). The reaction is based on the use of the Jahn-Teller effect that permits the abstraction by (CH3 )3 SiCl of one acetate group belonging to a manganese ion of the butterfly body. Actually the structure of the tetranuclear complex shows that among the seven acetate groups only one has a long Mn-O bond distance thus

194

6 High-spin Metal-ion-containing Molecules

Fig. 4. Labeled structure of complex Mn8 O4 (OAc)12 (pic)4 . To avoid congestion, not all symmetry-equivalent atoms have been labeled. The µ3 -O atoms are O5, O5 , O6, and O6 (reproduced with permission).

Fig. 5. ORTEP representation of the anion complex Mn8 O6 C16 (O2 CPh)7 .(H2 O)− 2 at the 50% probability level. For clarity, atoms not bound to Mn are de-emphasized (reproduced with permission).

enough labile to be attacked by (CH3 )3 SiCl [15]. Replacing the acetate groups by benzoate and performing the same reaction lead to a new octanuclear compound [Mn8 O6 Cl6 (O2 CC6 H5 )7 (H2 O)2 ]− with a different structure (Fig. 5) [16].

6.2 Self-assembly of Molecular Clusters

195

The general reaction can be written: 2[Mn4 O2 (O2 CC6 H5 )9 (H2 O)]− + 8(CH3 )3 SiCl → [Mn8 O6 Cl6 (O2 CC6 H5 )7 (H2 O)2 ]− + 8(CH3 )3 SiO2 CC6 H5 + 3C6 H5 CO2 H + H+ + 2Cl− The spin ground state of this octanuclear complex has been found to be equal to S = 11. The zero field splitting parameter D was estimated to be −0.04 cm−1 . Despite the high value of the spin ground state, no blocking of the magnetization was observed above T = 2 K. Chemical oxidation in acetonitrile of the benzoate tetranuclear complex using dibenzoyl peroxide leads to the formation of a new MnIII nonanuclear complex [Mn9 Na2 O7 (O2 CC6 H5 )15 (CH3 CN)2 ] possessing a spin ground state S = 4 [16]. The elegant reactions performed by Christou and coworkers that led to these complexes by coupling two tetranuclear ones are rather rare examples of “rationale” design. Most polynuclear complexes are to date obtained by a self-assembly process where prediction is still not reliable. However, as we will see, given a stable metal-ion framework as for the Mn12 derivatives it is possible to introduce small changes that keep the overall structure but may lead to new complexes with different spin ground states.

6.2.2.1

Mn12 Derivatives

The most studied compound in the Mn-carboxylate compounds is the so called Mn12 . It was prepared in 1980 by Lis by mixing KMnO4 and Mn(CH3 CO2 )2 .4H2 O in the molar ratio 1/2.5 in a 60% solution of acetic acid. The compound has the chemical formula [Mn12 O12 (O2 CCH3 )16 (H2 O)4 ].2CH3 CO2 H.4H2 O. The structure consists III by eight µ -O2− of a central MnIV 3 4 O4 cubane core connected to a ring of eight Mn anions (Fig. 6). The 16 acetate ligands and the four water molecules complete the coordination sphere of the metal ions. The four water molecules occupy the axial positions of two MnIII metal ions [17]. Using a.c. susceptibility measurements in zero applied static field in order to avoid saturation effect shows that the spin ground state of Mn12 is S = 10 [5]. High-field EPR spectra using different frequencies lead to the conclusion that the M S = −10 component of the S = 10 ground manifold has the lowest energy. EPR provided as well a good estimation of the zero field splitting parameter (D = −0.5 cm−1 ). The imaginary component of the susceptibility (χ ) was found to be different from zero below T = 9 K with maximum values between 7 and 5 K depending on the frequency of the a. c. magnetic field. This behavior is similar to what is observed in superparamagnetic particles but with a very important difference, that is the sample of Mn12 is made of identical molecules while a sample of metallic or metal-oxide particles has a size distribution. The most spectacular behavior is the magnetic bistability observed when a field is applied to an oriented crystal of Mn12 (Fig. 7) [1]. The origin of the hysteresis cycle is purely molecular since no 3D order was observed above T = 2 K. Christou succeeded to prepare a whole new family related to the original Mn12 acetate. Replacing acetate by benzoate affords a new compound with the same

196

6 High-spin Metal-ion-containing Molecules

Fig. 6. ORTEP representation of [Mn12 O12 (O2 CPh)16 (H2 O)4 ] at the 50% probability level. For clarity, the peripheral carboxylates are de-emphasized and only one phenyl carbon atom is included (reproduced with permission).

Fig. 7. Hysteresis loops of Mn12 recorded parallel to the c axis with a SQUID magnetometer at 2.2 K (outer loop) and 2.8 K (inner loop). A whole loop was recorded in ca. 8 h. The dotted lines are a guide for the eye only (reproduced with permission).

6.2 Self-assembly of Molecular Clusters

197

molecular structure but with a different ground state. The spin ground state for [Mn12 O12 (O2 CC6 H5 )16 (H2 O)4 ] is S = 9 in zero field [3]. The substitution of four MnIII ions by four FeIII is possible when performing the reaction using FeII instead of MnII . The structure of the new Mn8 Fe4 compound is similar to that of the parent one but the substitution process induces a drastic change in the magnetic properties. The new compound has a S = 2 ground state instead of S = 10 for Mn12 [18]. Starting with a propionate-Mn12 and adding tetraphenylphosphonium iodide leads to a new monoanion Mn12 : PPh4 [Mn12 O12 (O2 CC2 H5 )16 (H2 O)4 ] [19]. Electrochemical studies show that the extra electron is located on one of the manganese atoms belonging to the external ring. The ground state has been found to be S = 19/2. Thus, keeping the same overall metal-oxide framework it was possible to prepare molecules with different spin ground states: S = 10 for Mn12 acetate, S = 9 for Mn12 propionate, S = 19/2 for one-electron reduced Mn12 propionate and S = 2 for Mn8 Fe4 acetate. The few examples given above show how the chemistry of these systems is rich but how the prediction of the structure and the nature of the ground state is still out of reach.

6.2.3

The Hydroxypyridonate Family

The richness of the chemistry of the hydroxypyridonate family lies mainly in the flexibility of the ligands [20]. These ligands are able to simultaneously act as chelating and bridging ligands. As stated by Winpenny, the 2-hyroxypyridone derivatives are a much less well-behaved ligands than the carboxylate derivatives because once deprotonated they may show six different coordination modes within the same complex; this is illustrated in Scheme 5. Few examples of polynuclear complexes involving 3d metals and a hydroxypyridonate ligand had been reported before 1987 [21]. Winpenny et al. extended the chemistry of mixed carboxylate/pyridonate complexes and prepared a new series involving almost all the 3d metal ions [20].

Scheme 5

198

6 High-spin Metal-ion-containing Molecules

Fig. 8. Structure of Co24 (µ3 -0H)14 (µ2 -OH)4 (µ3 -OMe)2 (µ3 -Cl)2 (Cl4 )(mph)22 (reproduced with permission).

One of the largest polynuclear molecule containing paramagnetic ions is the tetraicosanuclear cobalt(II) complex obtained from the reaction of CoCl2 with two molar equivalent of Namph in methanol. Recrystallization from ethyl acetate containing 0.1% water by weight leads to crystals of [Co24 (µ3 -OH)14 (µ2 -OH)4 (µ3 OMe)2 (µ3 -Cl)2 (Cl4 )(mph)22 ] (Fig. 8) [22]. The small amount of water in ethyl acetate is probably the source of hydroxides necessary to grow the core of the complex. The magnetic properties of Co24 indicate that the spin ground state in not less than S = 9 which makes this complex a good candidate to observe magnetic bistability. The structure of the core is reminiscent to that of [Co(OH)4 ]2+ cubes with one corner missing and for some metals a chloride or a methoxide replacing the OH− groups. The periphery of the molecule is made of a layer containing twenty-two 2methylhydroxypyridonate ligands that adopt three different coordination modes as those shown in Schemes 5a, 5b and 5c. The global structure of Co24 is similar to that of [Fe17 + Fe19 ] mentioned in the introduction even though the synthetic route adopted by Powell et al. is different. According to Powell, the formation of such oligomeric

6.2 Self-assembly of Molecular Clusters

199

species which are at midway between mononuclear (or binuclear) complexes and infinite 2D array of (Fe(OH)2 )+ may be controlled by the iron/polydentate ligand ratio and the pH of the solution [6].

6.2.3.1

Role of the Solvent

The solvent can occasionally play an important role in stabilizing complexes with different nuclearities; a nice example can be taken from the chemistry of NiII pyridonate/acetate. The reaction of hydrated NiII (O2 CCH3 )2 with excess Hchp at 130◦ C for 1 h, followed by removal of unreacted Hchp and acetic acid, gives a bright green solid. Extraction of the residue with methanol gives after slow evaporation of the solution green needles in 40% yield. The structure reveals a linear trimeric NiII complex: [Ni3 (chp)4 (O2 CCH3 )2 (CH3 OH)6 ] [23]. The central Ni atom is surrounded by for oxygen atoms coming from four chp ligands; the two other oxygen atoms are located in axial positions and belong to two different acetate groups (Fig. 9). Only the oxygen atom of the chp ligand is linked to the metal ions while the nitrogen atom of chp is involved in hydrogen-bonding with the methanol groups. When extracting the green solid obtained from the above reaction by THF instead of methanol, a new dodecanuclear complex of formula [Ni12 (O2 CCH3 )12 (chp)12 (H2 O)(THF)6 ] with a completely different structure has been isolated [24].

Fig. 9. The structure of the trinuclear complex [Ni3 (chp)4(O2 -CMe)2 (MeOH)6 ]. Hydrogen bonds with O· · ·N distances in the range 2.654–2.698 Å shown as dotted lines (reproduced with permission).

200

6 High-spin Metal-ion-containing Molecules

Fig. 10. The structure of the dodecanuclear complex Ni12 (O2 CCH3 )12 (chp)12 (H2 O)(THF) (reproduced with permission).

The twelve Ni atoms form a ring held by bridging acetate and chp ligands (Fig. 10). The chp ligands are all bridging by their oxygen atom while the acetate groups have two bonding modes; the external bridge two Ni atoms while the internal are trinucleating. The core structure of this compound is similar to that of Fe10 the so-called “ferric wheel” mentioned above [11]. However, the nature of the exchange coupling interaction between adjacent NiII ions is ferromagnetic while it was found to be antiferromagnetic in Fe10 . The origin of the ferromagnetic interaction can be rationalized by the small Ni– O–Ni angle of the Ni2 O2 core. The mean value of this angle found equal to 96.4 ± 1.5◦ is consistent with a ferromagnetic interaction on the basis of the experimental studies carried out by Hatfield on binuclear hydroxo-bridged CuII complexes [25]. Simulating the experimental data leads to a JNiNi value of 9.4 cm−1 based on the following spin Hamiltonian (Eq. 3): H = −JNiNi

12

Si · Si+1

(3)

i=1

The spin ground state corresponds to the sum of the local S = 1 spins leading to a value of S = 12.

6.3 Host-Guest Approach

6.2.3.2

201

Role of the Substituents of Hydroxypyridine

Winpenny and coworkers performed again the same reaction by replacing chp by mhp and recrystallized the green solid from CH2 Cl2 -diethyl ether. A new undecanuclear [Ni11 (µ3 -OH)6 (µ-O2 CCH3 )6 (mph)9 (H2 O)3 ][CO3 ] compound is formed (Fig. 11) [24]. The structure of this compound is fundamentally different from that of the other two. The dramatic change in the structure is probably the result of the presence of mhp instead of chp. The mhp ligand in Ni11 is bonded to Ni by its nitrogen and oxygen atoms whereas only the oxygen atom of chp is involved in metal ligation in Ni3 and Ni12 . The chemistry of the hydroxypyridonate derivatives is vast. We have just given few examples to show how using the same synthetic route but introducing slight changes leads to the formation of different structures. Again here, prediction is not an easy task.

Fig. 11. The structure of the undecanuclear cation of [Ni11 (µ3 -OH)6 (O2 -CMe)6 (mhp)9 (H2 O)3 ]+ . Atoms not involved in metal atom bridging are excluded for clarity (reproduced with permission).

6.3

Host-Guest Approach

Serendipity is the rule for the synthesis of most of the polynuclear complexes prepared by self-assembly. Up to now, few synthetic methods were conceived to design polynuclear molecules in a rationale way. Lippard and coworkers showed that host– guest interaction is an original and interesting route to obtain a new kind of discrete species that may possess a high spin ground state and presents the properties ascribed to a single molecule magnet.

202

6.3.1

6 High-spin Metal-ion-containing Molecules

Hexanuclear Iron(III) Rings

The structure of the first synthesized compound of formula [NaFe6 (OCH3 )12 (dbm)6 ] Cl.12CH3 OH.CHCl3 belonging to a new series that emerged only few years ago consists of a ring of six FeIII metal ions bridged by methoxides; the coordination sphere of the metal ions is completed by the bidentate dibenzoylmethane ligand (Fig. 12) [26]. The structure of this compound qualifies as an example of [12] metallacrown-6 type [27]. The important feature is the presence of a sodium cation in the octahedral cavity formed by six oxygen atoms coming from six bridging methoxides. Using 23 Na NMR, Gatteschi and coworkers gave evidence that the molecular structure is retained in solution [28]. The spin ground state of this Fe6 Na cyclic complex and that of a related one with ClO− 4 as counter anion was found to be S = 0 as expected from the antiferromagnetic interaction (JFeFe = −20.4 and −19.9 cm−1 for [Fe6 Na]Cl and [Fe6 Na]ClO4 respectively, the interaction between non-adjacent FeIII

Fig. 12. ORTEP representation of the cation [NaFe6 (OCH3 )12 (dbm)6 ]+ with atom labels. Hydrogen atoms have been omitted for clarity. An inversion center relates primed to unprimed atoms. Thermal ellipsoids enclose 50% probability (reproduced with permission).

6.3 Host-Guest Approach

203

ions is neglected) between adjacent metal ions as for the Fe10 ferric wheel. Gatteschi and coworkers succeeded for the first time to estimate the single ion anisotropy of FeIII (D = −0.2 cm−1 ) within these Fe6 Na complexes that possess a non-magnetic ground state. The origin of the compound magnetic anisotropy revealed by susceptibility measurements on single crystals of the perchlorate derivative can be explained on the basis of single ion contributions mainly. This is an important step to the understanding of relaxation effects of iron(III) oxo clusters that have a superparamagnetictype behavior.

6.3.1.1

Role of the Template

The amplitude of the antiferromagnetic interaction between two adjacent iron(III) ions depends on the Fe–O–Fe angle; in order to modulate the amplitude of the interaction Gatteschi and coworkers tried to prepare new compounds similar to Fe6 Na using Li+ and K+ as templates. The reaction with K+ does not proceed to the formation of a new Fe6 K complex. While with Li+ a new Fe6 Li complex which has the same overall structure as Fe6 Na is obtained [29]. Because of the smaller ionic radius of Li+ (0.68 Å) in comparison to that of Na+ (0.97 Å), the size of the ring given by the Fe1–Fe1 separation is smaller for Fe6 Li than for Fe6 Na: 6.272(3) and 6.425(1) Å respectively. Furthermore, the nearest-neighbor Fe-Fe separation decreases considerably from 3.2152(5) to 3.140(1) Å; the Fe–O–Fe angles are reduced by more than 2.5◦ . The investigation of the magnetic properties of Fe6 Li shows that the amplitude of the antiferromagnetic interaction is reduced by about 30% (JFeFe = 14.30 and 14.68 cm−1 for [Fe6 Li]ClO4 and [Fe6 Li]PF6 respectively) in comparison to Fe6 Na, however the spin ground state is still S = 0.

6.3.2

Hexanuclear Manganese Rings

As has already been shown in cyanide-bridged three dimensional molecular-based magnets, where changing the nature of the metal ions and keeping the overall general structure enable tuning of the exchange coupling interaction and lead to the formation of a room temperature magnet [30], for some discrete species this may represent a powerful tool to modulate their magnetic properties. Christou and coworkers showed that substituting 4 MnIII by 4 FeIII in the external ring of Mn12 acetate (S = 10) leads to the formation of a new Mn8 Fe4 acetate with a S = 2 ground state (see Section 2). A preparation method similar to that of Fe6 Na affords a new Mn6 Na complex when manganese is used instead of iron. The overall structure of the hexanuclear Mn6 Na species [NaMn6 (OCH3 )12 (dbm)6 ]+ is similar to that of Fe6 Na; the main difference comes from the presence of the Jahn–Teller d4 MnIII metal ion that leads to a tetragonal elongation along three perpendicular directions of the three crystallographically independent MnIII ions within the cyclic core [31]. The ground state was found to be equal to S = 12 due to the ferromagnetic interaction between adjacent MnIII metal ions. In a d4 tetragonally elongated octahedron, the dx 2 −y 2 orbital is

204

6 High-spin Metal-ion-containing Molecules

empty and due to the particular orientation of the axes’ distortion on the metal ions within the ring, the singly occupied dz 2 orbitals of nearest-neighbor MnIII do not overlap. The ferromagnetic interaction responsible of the stabilization of the largest spin ground state has its origin in the overlap between the semi-occupied dz 2 orbital on one center and the empty dx 2 −y 2 orbital on an adjacent one through the methoxy ligands. This mechanism was first invoked by Goodenough [32] to explain the ferromagnetic interaction in some metal-oxide three dimensional networks and later by Girerd [33] to rationalize the magnetic properties of a binuclear µ-oxo, µ-acetato MnIII complex. Another complex belonging to the same family has been obtained by the same synthetic procedure as for Mn6 Na. A MnII metal ion plays the role of the template leading to a heptanuclear Mn7 complex. Conductivity measurements confirmed the non-ionic character of the compound leading to the following mixed-valence formuIII lation: MnII 3 Mn4 [34]. The analysis of the magnetic properties is quite difficult in such complex since even though the central ion is a MnII , the position of the other two MnII is difficult to locate within the ring. On the other hand, the presence of triangular motives leads to competition between the different exchange coupling interactions making, as pointed out in Section 2, prediction of the nature of the ground state very difficult. However, the χM T value at low temperature indicates a S = 17/2 ground state. Least-squares fits of the experimental magnetization data with field data at two different temperatures leads to the conclusion that spin ground states S = 17/2 or S = 19/2 are possible and that the zero-field splitting parameter is ca. 0.25 cm−1 . As a conclusion to this part, it is important to note that the prediction of the structure and thus of the nature of the ground state is not as difficult as for the complexes reviewed in the preceding section so that the template approach may be a valuable and powerful tool to prepare new molecules with expected electronic structure. Pecoraro and coworkers designed polynuclear complexes based on the same concept using different metal ions of the first series as well as lanthanides as templates in order to tune the nuclearity of the clusters [27b].

6.4

Step-by-step Rationale Approach

The synthesis of high spin molecules has been since the middle of the eighties considered as one possible route to the preparation of genuine molecular-based ferromagnets. Kahn stated that a strategy used to prepare molecular-based ferromagnets “consists of synthesizing high-spin molecules or chains and then assembling them in a ferromagnetic fashion within the crystal lattice” [35]. To date, there are very few reports in the literature of high spin discrete species (molecules) interacting together and leading to molecular ferromagnets (see below). However, these ideas contributed in a constructive manner to the preparation of polynuclear complexes possessing a high spin ground state using a rationale approach. The important feature of the rationale approach is the possibility that the chemist has to predict the

6.4 Step-by-step Rationale Approach

205

nature of the ground state, the size of the molecule, its shape and eventually the order of magnitude of the magnetic anisotropy that plays an crucial role in determining the blocking temperature of the magnetic moment. Gatteschi and coworkers has, for instance, predicted that an oxo-bridged tetranuclear FeIII complex possessing the structure depicted in Scheme 6 would have a S = 5 ground state due to the expected antiferromagnetic interaction between the central and the three peripheral S = 5/2 FeIII metal ions. Furthermore, assuming a local zero-field splitting parameter D(FeIII ) = −0.2 cm−1 , Gatteschi predicted that the S = 5 ground state would have a D(5) parameter equal to −0.17 cm−1 . The first results on the tetranuclear [Fe4 (OCH3 )6 (dbm)6 ] complex that possess the postulated structure reveal by low temperature HF-EPR studies that D(5) is equal to −0.2 in excellent agreement with the predicted value. Gatteschi and coworkers went a step further and estimated that a decanuclear FeIII complex (Scheme 7) would have a S = 10 ground state and a D(10) zero-field splitting parameter equal to −0.12 cm−1 [28]. This compound has not been reported yet but its topology is reminiscent of dendritic-like molecules of first generation [36]. A step-by-step convergent method developed for dendrimers may be useful to realize the synthesis of this decanuclear complex by first preparing the peripheral trinuclear dendron and then assembling three such complexes around a central FeIII metal ion. This stepwise approach enables the chemist to have a good idea about the magnetic anisotropy. But, since the magnetic anisotropy of a compound depends as well on the relative orientation of the molecules within the crystal lattice, prediction may become an arduous task to fulfil.

Scheme 6

Scheme 7

To overcome this difficulty Langmuir–Blodgett technique [37] may be a useful tool to organize the molecules in mono- or multilayer films so that the anisotropy axes (assuming an axial anisotropy) of a collection of molecules may be oriented in the same direction. The first example is given by Coronado and Mingotaud who succeeded to prepare Langmuir-Blodgett films of the benzoate derivative of Mn12 [38]. Magnetic studies on one monolayer reveals that a high degree of organization

206

6 High-spin Metal-ion-containing Molecules

has been achieved with the anisotropy axes of the molecules lying perpendicular to the plane of the monolayer; when the applied magnetic field is parallel to the monolayer, the observed hysteresis loop (at T < 4 K) is softer than when the field is perpendicular to the layer.

6.4.1

Complex as Ligand and Complex as Metal

In the following text, we focus on the tactics that enabled chemists to prepare high spin molecules where it is possible to predict in most cases the nature of the spin ground state as well as the overall molecular structure of the target molecules. Two prerequisite conditions have to be fulfilled in order to put in practice the stepwise approach. Firstly, a central molecular complex (referred to as the ‘core’ in the following) that plays the role of a ligand (complex as ligand) i. e. possessing on its periphery atoms bearing lone electron pairs able to coordinate to a metal ion, should be conceived and then synthesized. This complex must be stable and inert since all reactions will be performed in solution. The second condition is the preparation of the peripheral complex which must be able to behave as a Lewis acid with respect to the central molecule (complex as metal). This may be achieved when such complexes bear ligands that can easily be substituted by the core. Furthermore, since the objective is the synthesis of a discrete molecule, the peripheral complex should possess chelating ligands that prevent the formation of extended lattices. And finally, in order to minimize intermolecular interactions at low temperature and to insure that the properties of a single molecule can be studied, bulky peripheral ligands and charged species are preferred to neutral ones so that the counter-ions may dilute the magnetic species preventing dipole–dipole interactions, for example. Within this approach, the nuclearity of the target molecule will depend on the connectivity of the core. By connectivity, we mean the number of sites that can receive metal ions; for example, trisoxalatochromate(III) has a connectivity 3 (Scheme 8), since it can coordinate to three metal ions by its six oxygen atoms. Another complex that can play the role of a core is hexacyanochromate(III) which possess a connectivity 6 due to the six nitrogen atoms present at the vertices of the octahedron (Scheme 9). The connectivity given by the core is an upper value to the number of peripheral complexes that may be attached. The number of coordination sites of the peripheral complex depends on the nature of the core. For trisoxalatochromate(III), the oxalate oxygen atoms are bidentate so that two coordination sites in cis position

Scheme 8

6.4 Step-by-step Rationale Approach

207

(Scheme 10) should be available within the peripheral complex in order to construct a polynuclear species; while for hexacyanochromate(III) the nitrogen atoms are monodentate and the peripheral complex must possess one available coordination site as depicted in Scheme 11. Other parameters like the nature of the solvent, the nature of the counter-ions, the relative charge of the core and the peripheral molecules or the size of the ligands may be decisive to stabilize complexes with different nuclearities other than those predicted on the simple connectivity properties of the core.

Scheme 9

Scheme 10

6.4.2

Scheme 11

Predicting the Spin Ground State

The value of the spin ground state will depend on the nature of the exchange coupling interaction between the core and the peripheral metal ions through the bridging ligands. For a given bridging ligand, the nature of the exchange interaction is greatly influenced by the nature of the two interacting metal ions and to be more precise by the number of their d electrons. It is not the object of this section to go deep in the mechanism of exchange coupling interactions, we would just like to specify the model that we will use to make prediction. The phenomenon of electron exchange interaction is expressed by the Heisenberg Hamiltonian H = −J S A · S B where S is the local spin of a metal ion taken in its ground state and J is the interaction energy. The above Hamiltonian relies on the fact that the interaction between the metal ions is weak so that the spin keeps its local properties. In the case of interacting metal ions bearing more than one electron, the semi-occupied orbitals on each metal ion are considered and J can be expressed (Eq. 4) as the sum of the individual jij interactions between the semi-occupied orbitals ai of metal ion A and bj of B taken two by two: J=

1 jij n A n B i, j

(4)

208

6 High-spin Metal-ion-containing Molecules

where n A and n B are the number of unpaired electrons on A and B respectively. The semi-occupied orbitals considered are not pure metallic orbitals but they contain a contribution from the bridging ligands as it was proposed by Anderson, Hoffmann and Kahn [39]. The parameter jij is the sum of two contributions of opposite signs: one positive jijF contributes to ferromagnetism (parallel alignment of the interacting spins) and is proportional the bielectronic exchange integral and the other is negative jijAF contributes to antiferromagnetism (antiparallel alignment of the interacting spins). Because most of the complexes obtained by the step-by-step approach are bimetallic and since the orbital analysis of the interaction is made by considering a binuclear low symmetrical unit comprising the central metal, the bridging ligand and one peripheral metal ion, we will adopt here the orbitals proposed by Kahn [39c, 40]. These orbitals are strictly localized and are built so that they may not necessarily be orthogonal to each other. In this case, the antiferromagnetic contribution is proportional to the resonance integral βij and to the overlap integral Si j between orbitals ai localized on metal A and bj localized on metal B. These orbitals are well-adapted to our systems which are generally bimetallic and to our chemical approach. Thus, the amplitude of the antiferromagnetic interaction is proportional to the degree of overlap between the semi-occupied localized orbitals; the higher the degree of overlap, the strongest the interaction is. Now, it is clear that two orthogonal ai and bj orbitals will have zero overlap so that only the ferromagnetic contribution will remain leading to the highest possible spin for the ground state.

6.4.3

Antiferromagnetic Approach

One of the first high-spin molecules designed by a step-by-step strategy was reported by Kahn and coworkers in 1986. Using the Cu(pba)2− (connectivity 2) as the core (Scheme 12) and [Mn(Me6 -[14]ane-N4 )]2+ (Scheme 10, M = Mn) as the peripheral complex, it was possible to obtain a trinuclear linear CuMn2 species [41]. The magnetic properties show clearly that the spin ground state S = 9/2 arises from the antiferromagnetic interaction between the central S = 1/2 and the two peripheral S = 5/2 local spins (Fig. 13).

Scheme 12

To enhance the value of the magnetic moment, one possibility is to use a central complex with higher connectivity. Trisoxalatochromate(III) which has a connectivity 3 is a good candidate to do so. Unfortunately, the reaction of K3 [Cr(C2 O4 )3 ] with the

6.4 Step-by-step Rationale Approach

209

Fig. 13. Experimental ( ) and calculated (—) plots of χM T against T for CuMn2 . The inset shows an expansion of the χM T axis in the 100–250 K temperature range, as evidence of the minimum in χM T (reproduced with permission).

above mentioned MnII complex did not lead to any well defined compound. While using a different CrIII complex possessing a tris bidentate ligand (Scheme 13, same connectivity as Cr[C2 O4 ]3− 3 ) as the core leads to a tetranuclear CrMn3 complex with a low-lying S = 6 ground state. Fitting experimental χM T data leads to JCrMn = −3.1 cm−1 (Fig. 14) [42].

Scheme 13

To increase the nuclearity of this kind of complexes and to stabilize a higher spin ground state, hexacyanochromate(III) was used as the assembling core. The peripheral MnII metal ion is chelated by a pentadentate ligand with the sixth coordination

210

6 High-spin Metal-ion-containing Molecules

Fig. 14. Experimental (◦) and calculated (—) plots of χM T against T for CrMn3 .

Fig. 15. Experimental (◦) and calculated (—) plots of χM T against T for CrMn6 .

site occupied by a water molecule which can be substituted by the nitrogen end of [Cr(CN)6 ]3− . With a connectivity 6, a heptanuclear CrMn6 complex was expected and obtained [43]. The antiferromagnetic interaction between the central S = 3/2 CrIII and the six surroundings S = 5/2 MnII leads to a S = 27/2 ground state (Fig. 15).

6.4 Step-by-step Rationale Approach

6.4.4

211

Ferromagnetic Approach

In 1971, Ginsberg predicted that for the CrIII –O–NiII linear sequence, the interaction between CrIII and NiII might be ferromagnetic [32b]. If we assume that this sequence is part of an extended cubic 3D network (this is what Ginsberg was referring to in his paper) where the local symmetry around the metal ions is octahedral, the CrIII has three unpaired electron in t2g orbitals and NiII has two unpaired electrons in eg orbitals. In term of the orbital model developed by Kahn, we are in the situation of strict orthogonality, the overlap integral between the CrIII t2g and the NiII eg semi-occupied orbitals is zero and the ferromagnetic contribution dominates leading to an alignment of the local spins within the ground state. Kahn et al. were the first to test this prediction by preparing a tetranuclear [Cr(oxNi(Me6 3− as the core and [14]ane-N4 ))3 ](ClO4 )3 (ox = C2 O2− 4 ) complex using [Cr(ox)3 ] 2+ [Ni(Me6 -[14]ane-N4 )] (Scheme 10, M = Ni) at the periphery [35]. The magnetic data shows that the interaction is ferromagnetic with a spin ground state S = 9/2 and an exchange coupling parameter JCrNi through the oxalate bridge equal to 5.3 cm−1 . Unfortunately, the structure of this compound has not been reported. Later, Okawa and coworkers succeeded to solve the structure of a similar complex using dithiooxalate instead of oxalate as bridging ligand (Fig. 16) [44]. The spin ground state was found, as expected, to be S = 9/2.

Fig. 16. An ORTEP view of the tetranuclear cation [Cr((C2 O2 S2 )(Ni(Me6 -[14]ane-N4 ))3 ]3+ .

212

6 High-spin Metal-ion-containing Molecules

Fig. 17. Experimental (◦) and calculated (—) plots of χM T against T for CrNi6 .

When [Cr(CN)6 ]3− is mixed with six molar equivalents of the mononuclear [Ni(tetren)H2 O]2+ (tetren = tetraethylenepentamine, M = Ni, Scheme 11), a heptanuclear CrNi6 complex with an octahedral symmetry is isolated [45]. Again in this case, as expected, the strict orthogonality of the semi-occupied orbitals on the central CrIII on one hand and on the peripheral six NiII on the other, leads to a ferromagnetic interaction stabilizing the S = 15/2 state as the ground state. The best fit parameters of the χM T = T experimental data leads to: J = 15.6 cm−1 , |D| = 0.008 cm−1 and g = 2.04 (Fig. 17). The structure of this compound was partly solved, only the position of the heavy atoms and their surroundings could be determined (Fig. 18). Shortly after, Spiccia and Murray reported the crystal structure of a very similar complex with [Fe(CN)6 ]4− as the core surrounded by six CuII ions chelated by a tetradentate ligand (Fig. 19) [46].

6.4.4.1

Dynamic Magnetic Properties of CrNi6

The dynamic magnetic properties of CrNi6 was investigated down to 200 mK using a. c. susceptibility measurements at different frequencies of the oscillating magnetic field (0.7 Oe). A maximum of the real component (χ ) of the magnetic susceptibility is observed at T = 0.38 K at a frequency of 1000 Hz (Fig. 20). Associated with the maximum, the out of phase component (χ ) increases and, on cooling, reaches a maximum at 0.24 K. This maximum may be due either to three-dimensional ordering or to a blocking of the magnetic S = 15/2 moment due to the decrease of its relaxation time at low temperature. Measurements were thus carried out at different frequencies: 1000, 300, and 30 Hz and using a different apparatus at 680 MHz [47]. The maximum of χ is found to be frequency-dependent. It shifts to low temperature when the frequency is decreased. These observations are in line with the occurrence

6.4 Step-by-step Rationale Approach

213

Fig. 18. Structure of the cation [Cr(CNNi(tetren))6 ]9+ . Only the first coordination sphere of the metal ions is represented (reproduced with permission).

of blocking of the magnetization as a result of superparamagnetic behavior and not because of a 3D magnetic order. The magnetic moment relaxation time (τ ) is proportional to the inverse of the frequency (f) of the applied magnetic field: τ = 1/2π f . The blocking temperature (TB ) at a given frequency corresponds to the temperature of the maximum of the χ signal. Plotting ln(τ ) against 1/TB gives a straight line (Fig. 21). The blocking phenomenon is thermally activated (Arrhenius type), as already observed in magnetic particles and in some high spin molecules [1, 3, 48]. The relaxation time can be expressed as τ = τ0 exp(E A /kTB ), where τ0 is the relaxation time at infinite temperature and E A /k is the activation energy. τ0 and E A are found equal to 1.1×10−11 s and 3.98 cm−1 respectively. Assuming an axial anisotropy, the energy barrier is given by DSz2 where Sz = 15/2. It is possible to compute the value of the zero field splitting parameter D to be 0.07 cm−1 . This value is an order of magnitude higher than the value extracted from fitting the χM T = f (T ) data (|D| = 0.008 cm−1 ). One reasonable explanation to this discrepancy is that our assumption of an axial anisotropy as responsible of the blocking of the moment at low temperature is wrong. The CrNi6 heptanuclear complex has an octahedral symmetry and the local magnetic axial anisotropy of the six NiII ions will cancel out. Nevertheless, the experimentally observed blocking of the magnetic moment is due to the presence of a kind of magnetic anisotropy which is probably cubic and not axial in this case.

214

6 High-spin Metal-ion-containing Molecules

Fig. 19. ORTEP diagram of [Fe(CN)Cu(tpa)6 ][ClO4 ]8 .3H2 O. For clarity only the atoms in Fe(CN)Cu(N4 )6 are labelled (reproduced with permission).

6.4.5

Role of the Organic Ligand

One crucial point of the approach developed above is the presence of a chelating ligand that cannot be substituted by the core. We, and others, have already shown that by changing the number of available coordination sites around the ’complex as metal’ by changing the nature of the organic ligand, it is possible to built systems with different dimensionality (structurally) and original architecture [49–51]. Murray et al. reacted ferricyanide with [Ni(bpm)2 ]2+ and obtained a pentanuclear complex of formula [Fe(CN)6 ]2 [Ni(bpm)2 ]3 .7H2 O (bpm = bis (1-pyrazolyl) methane) where three Ni(bpm)2 units bridge two ferricyanide molecules (Fig. 22)

6.4 Step-by-step Rationale Approach

215

Fig. 20. Plots of real (χ ) and imaginary (χ ) susceptibility against temperature at a frequency of 1000 Hz for CrNi6 .

Fig. 21. ln(τ ) = f (1/TB ) where τ = 1/2π f and f = 1000, 300, 30 Hz and 685 MHz for CrNi6 .

[51e]. The magnetic properties of the pentanuclear complex reveal at TC = 23 K a long-range ferromagnetic order due the ferromagnetic intermolecular interaction mediated by a three dimensional network of hydrogen bonds. It is worth noting here that this is one of the few examples of polynuclear complexes where the interaction between the molecules leads to a ferromagnetic long-range order. Since, in the framework of the mean field approximation, the ordering temperature of three dimensional networks is proportional to S(S + 1), a compound containing molecules with a spin state S = 16, for example, and interacting ferromagnetically in the same

216

6 High-spin Metal-ion-containing Molecules

Fig. 22. Structure of the pentanuclear complex [Fe(CN)6 ]2 [Ni(bpm)2 ]3 showing the atomlabeling scheme and 20% probability ellipsoids (reproduced with permission).

manner as in Murray’s compound would be ferromagnetic a room temperature. In order to investigate the intramolecular interaction, Murray and coworkers dehydrated the compound by heating it at 200◦ C for 36 h. Dehydration leads to the suppression of the long-range magnetic order; the magnetic properties of the dehydrated sample indicate that the ground state of the pentanuclear complex is either a S = 4 or S = 3. The structure of this compound merits some comments. It is surprising that a discrete µ-cyano assembly is obtained instead of an extended one. Actually, the few examples reported in the literature of compounds obtained from the reaction of hexacyanometallates and a mononuclear complex possessing two available sites in cis position are polymeric [51d, 52]. On the other hand, Murray mentioned in 2+ his paper that when Ni(bpm)2+ 2 is replaced by Ni(bipy)2 , a pentanuclear complex with the same structure is obtained. In order to understand the particular role of these two mononuclear NiII complexes, we attempted to built a structural model similar to that reported by Gatteschi (three dimensional) [51d] and Morpugo (one dimensional) [52] by using Corey–Pauling–Kotlun (CPK) molecular models. It was clear that an extended structure could not be built because the steric hindrance induced by the bulky bpm or bipy ligands. In order to achieve an extended structure, 2+ the hexacyanometallate must bridge more than three Ni(bpm)2+ 2 (or Ni(bipy)2 ) molecules; this turned out not to be possible when the chelating ligand has bulky groups attached on two well identified atoms coordinated to NiII and not anywhere else. This leads to the conclusion that it is possible to design the organic ligand in order to prepare complexes of expected nuclearities or extended systems with a particular architecture.

6.4 Step-by-step Rationale Approach

6.4.6

217

Molecules with Two Shells of Paramagnetic Species

One of the advantages of the step-by-step approach is the possibility to use the strategy developed for the synthesis of dendritic molecules in order to design polynuclear complexes containing at each generation an increased number of paramagnetic species. Let us concentrate on the method that may be used to design molecules with a central metallic core surrounded by two shells of metal ions. The same method may then be extended to molecules containing more than two shells. In the case of a core of connectivity 3, the topology of a target two-shell molecule is represented in Scheme 14. In order to prepare such molecules, one should design three kinds of complexes: the first kind is the core that as stated above should be able to play the role of a ligand towards the complexes of the first shell, the second kind is the periphery molecule, it must play the role of a metal towards the complexes of the second shell and the third kind is the complex of the first shell that should play simultaneously the role of a metal and a ligand towards the core and the peripheral complexes respectively. The core and the peripheral complexes are easy to design (see above), the difficulty resides in the design of the first shell complex which in addition to the requirements mentioned above must be able to transmit efficiently the magnetic interaction so that the low-lying spin ground state of the polynuclear complex will be well separated from the excited states.

Scheme 14

Schemes 15 and 16 show two possible examples of such molecules; Scheme 15 represents a mononuclear complex adapted to a core possessing a monodentate Lewis base like the hexacyanometallates while Scheme 16 represents a complex with a tetradentate ligand leaving two coordination sites in cis position allowing the reaction with the trisoxalatochromate(III) for example. These complexes bear two imidazol groups that once deprotonated may serve to coordinate two peripheral molecules achieving thus the synthesis of two-shell complexes.

Scheme 15

Scheme 16

218

6 High-spin Metal-ion-containing Molecules

6.4.6.1

Synthesis and Magnetic Properties of a CrNi3 Complex

The synthesis of two-shell complexes may be carried out by two different methods: the first method is convergent, it consists of preparing a trinuclear complex and then as a second step assembling six trinuclear complexes around the hexacyanochromate(III) core (or three trinuclear complexes if the trisoxalatochromate(III) is used as the core), the second method is divergent and consists of building up around the core the two shells step by step. We have used this latter approach to prepare a tetranuclear complex that can be considered as the first step towards the synthesis of a two-shell complex. The reaction between ((C4 H9 )4 N)3 [Cr(CN)6 ] and [Ni(imdipa)Cl]PF6 in acetonitrile leads to the formation of a tetranuclear complex of formula [(NC)3 Cr(CNNi(imdipa))3 ]Cl2 PF6 . The investigation of the magnetic properties shows that the spin ground state is S = 9/2 as expected from the ferromagnetic interaction between the central CrIII and NiII metal ion through the cyanide bridge. The experimental data showing the dependence of χM T on T (Fig. 23) were fitted by use of the spin Hamiltonian (Eq. 5): ∗

H = −JCrNi SCr · S + D

Sz2

S(S + 1) − + β [gCr SCr + gNi S ∗ ] H 3

(5)

where S ∗ = S N i1 +S N i2 +S N i3 and S = SCr +S ∗ , D is the zero-field splitting parameter within the ground state S = 9/2. The fit results are: JCrNi = 12.4 cm−1 , gCr = 1.98, gNi = 2.27 and |D| = 0.7 cm−1 ; the JCrNi value is in the same range as that of the heptanuclear CrNi6 complex [45] (15.6 cm−1 ). An important feature revealed by the magnetic properties measurements is the rather relatively large zero-field splitting parameter (0.7 cm−1 ) associated to S = 9/2 ground state. It is clear that this value is only a rough estimate, HF-EPR studies are needed to get a much better idea on

Fig. 23. Experimental (◦) and calculated (—) plots of χM T [(NC)3 Cr(CNNi(imdipa))3 ]Cl2 PF6 .

against

T

for

6.4 Step-by-step Rationale Approach

219

the amplitude of the magnetic anisotropy. However, the zero-field splitting in the tetranuclear complex is two order of magnitude larger than that found for CrNi6 (0.008 cm−1 ). This is in line with the lower symmetry of the tetranuclear complex in comparison to the octahedral heptanuclear one. We are currently working on the synthesis of the second step that will enable, by deprotonating the six imidazol functions present on the three NiII complexes, coordination of six peripheral metal complexes and thus achieve the synthesis of a decanuclear two-shell complex.

6.4.6.2

A Pentanuclear Complex with Three Different Paramagnetic Species

Another elegant way to design a molecule with two shells of paramagnetic species is to use organic radicals as chelates for the peripheral metallic complexes. The first idea was to prepare a tetranuclear complex similar to that reported by Kahn and coworkers [Cr(oxNi(Me6 -[14]ane-N4 ))3 ](ClO4 )3 [35] (see above) but by substituting the tetradentate ligand of the peripheral complex by two bidentate organic radicals. The synthesis of the mononuclear complex [Ni(IM2-py)2 (NO3 )]NO3 (IM2-py = 2-(2-pyridyl)-4,4,5,5-tetramethyl-4,5-dihydro-1H-imidazolyl-1-oxy, Scheme 17) was achieved. Unfortunately, the crystal structure was not solved but that of the mononuclear NiII complex with the hydroxylamine derivative of the radical was (Fig. 24). It shows that the two bidentate ligands are coordinated in cis position as required for the trisoxalatochromate(III) core; the remaining coordination sites are occupied by a chelated nitrate anion. The reaction between [Ni(IM2-py)2 (NO3 )]NO3 and the tetrabutylammonium or the potassium salt of [Cr(C2 O4 )3 ]3− leads to non-pure compounds that were identified to be a mixture of the postulated tetranuclear complex contaminated with variable amount of the trinuclear complex CrNi2 depending on the preparation procedure. Trying the perchlorate salt of the NiII complex seems to give better results but no well identified pure compound could be obtained yet.

Scheme 17

The structure of the NiII complex (Fig. 24) shows that the ligands are bulky so that the reaction with hexacyanochromate(III) may lead to a pentanuclear complex similar to that described by Murray and coworkers (Fig. 22) [51e]. Let us first see what will be the ground state of such pentanuclear complex that will contain two S = 3/2 CrIII , three S = 1 NiII and six S = 1/2 organic radicals. We have already shown that the exchange interaction between CrIII and NiII is ferromagnetic due to the orthogonality of their semi-occupied orbitals. On the other hand, Rey studied the magnetic properties of the mononuclear Ni(hfac)2 (IM2-py) (hfac = hexafluo-

220

6 High-spin Metal-ion-containing Molecules

Fig. 24. Structure of the monocation [Ni(HIM2-py)2 (NO3 )]+ .

roacetylacetonate) complex and observed a ferromagnetic interaction between NiII and the radical ligand [53]. The exchange coupling parameter JNiRad was found to be equal to 128 cm−1 . Having these results in mind, we expect that within the pentanuclear complex the interaction between the different magnetic species will be ferromagnetic leading to a S = 9 ground state. The reaction between [Ni(IM2-py)2 (NO3 )]NO3 and K3 [Cr(CN)6 ] leads, when performed in water, to the formation of a precipitate. The characterization of the compound leads to the following formula: [Cr(CN)6 ]2 [Ni(IM2-py)2 ]3 .7H2 O [54]. The new compound is soluble in most common solvents but is insoluble in water as expected for a neutral species. Unfortunately, crystals suitable for X-rays analysis have not been obtained. To give some evidence supporting our assumption of the presence of a discrete species within the compound, we recorded the UV-visible electronic spectrum and compared it to that of the mononuclear NiII complex. The UV-visible spectrum (MeOH, c = 3×10−2 M) presents a band assigned to the 3 T2g ← 3 A transition of the NiII chromophores (E = 11 363 cm−1 , ε = 35.9 L mol−1 cm−1 ) 2g which is shifted to higher energy in comparison to that of the mononuclear [Ni(IM2py)2 (H2 O)2 ]2+ complex (E = 10 416 cm−1 , ε = 45 L mol−1 cm−1 ). It has already been shown that the nitrogen end of the cyanide induces a stronger crystal field than water [55], thus the solubility of the compound is not because of decomposition of an extended network in solution but because our compound is made of discrete polynuclear species keeping the same structure (presence of bridging cyanides) in solution as found in the solid state by infrared studies. To check our assumption that the molecular species is a pentanuclear complex with a structure similar to that reported by Murray and in order to investigate the nature of the ground state, we carried out magnetization measurements on a powder

6.4 Step-by-step Rationale Approach

221

Fig. 25. Experimental (◦) and calculated (—) plots of χM T against T for [Cr(CN)6 ]2 [Ni(IM2py)2 ]3 .7H2 O.

sample. On cooling down χM T increases and reaches a maximum at T = 6.5 K, then decreases (Fig. 25). The value at the maximum (42 cm3 mol−1 K) is very close to the expected one for a S = 9 ground state (45 cm3 mol−1 K for an average g value of 2) which corresponds to the parallel alignment of the local spins of the eleven paramagnetic species. A calculation of χM T as a function of temperature for different values of JCrNi and JNiRad by setting the local g values to gCr = 1.98, gNi = 2.07 and gRad = 2.00 was performed. The calculated curves for JNiRad = 150 K (105 cm−1 ) and JCrNi = 9, 11, 13, 15 and 17 K are shown in Fig. 26. The best agreement between the experimental and the calculated data is obtained in the temperature range 300– 8 K, for JCrNi = 13 K (9 cm−1 ) and JNiRad = 150 K (105 cm−1 ). It is possible at this level to introduce a parameter θ that takes into account the decrease of χM T at low temperature as due to antiferromagnetic intermolecular interactions, a very good agreement is obtained with θ = −0.45 K (Fig. 25). The sign and the values of the exchange coupling parameters are in the same range as what has already been found for CrNi6 [45] (JCrNi = 15.6 cm−1 ) and Ni(hfac)2 (IM2-py) [53] (JNiRad = 128 cm−1 ). The decrease of χM T at low temperature may be due to zero-field splitting effect and not to intermolecular antiferromagnetic interaction. Unfortunately, zerofield splitting could not be included because of prohibitive calculation time. The magnetization vs. field plot (Fig. 27) (T = 2.2 K, H = 0–140 kOe) shows that the value at saturation (18.1 Bohr Magneton) corresponds to the expected S = 9 ground state. AC susceptibility measurements performed down to 100 mK show the absence of long-range magnetic order, a behavior different from the ferromagnetic order found for Murray’s pentanuclear complex [Fe(CN)6 ]2 [Ni(bpm)2 ]3 .7H2 O. The origin of the difference in behavior at low temperature is probably due to the absence of a network of hydrogen bonds in our compound linking the pentanuclear species together; this is only an assumption since the crystal structure of our compound

222

6 High-spin Metal-ion-containing Molecules

Fig. 26. Value of χM T calculated by fixing JNiRad to 150 K and varying the value of JCrNi for [Cr(CN)6 ]2 [Ni(IM2-py)2 ]3 .7H2 O.

Fig. 27. Plots of magnetization against field for [Cr(CN)6 ]2 [Ni(IM2-py)2 ]3 .7H2 O, (◦) experimental, (—) Brillouin function for S = 9 (g = 2), (- - -) sum of the Brillouin functions for six S = 1/2 (g = 2), three S = 1 (g = 2.07), and two S = 3/2 (g = 1.98).

has not been solved yet. The solubility of our compound and the presence of only two bands in the 2000–2200 cm−1 region of the infrared spectrum assigned to two kind of cyanide bonds: bridging and non-bridging is in favor of the absence of a hydrogen-bond network involving the pentanuclear species. Actually, the infrared spectrum of Murray’s complex reveals, in addition of bands assigned to bridging and terminal cyanides, the presence of bands assigned to terminal cyanides which form hydrogen-bonds with the oxygen atoms of water molecules present in the structure.

6.5 Conclusion

223

The use of organic radicals to stabilize large spin ground states may be particularly interesting within a rationale approach. We have described an example of a complex where the organic radicals are located on the periphery. Rey and coworkers showed that it is possible to use bis-bidentate radicals to bridge two metal ions and prepare an extended 2D network by a two-step reaction [56]. The organic radicals within the network are bridging the metal ions. Using the same approach, it may be possible to design discrete species where the organic radicals are located not only at the periphery but within the body of the polynuclear complex. A judicious choice of the different paramagnetic species may lead to the stabilization of a very high spin ground state.

6.5

Conclusion

The self-assembly strategy has produced a great number of new complexes, some of them possess a high spin ground state. The possibilities in this area of research are open and the results obtained in the last ten years proof that the efforts made are rewarding. On the other hand, the multistep approach that we have developed is laborious and arduous. Preparing a perfectly structured two- or three-shell high spin molecules is a challenge. The question is: is it worth the effort to try to design and prepare such systems? The answer is probably: yes. The study of the properties of a molecule made of well identified fragments (the constituent of each shell) gives valuable information on many physical parameters: magnetic anisotropy, amplitude and nature of the exchange coupling interaction, optical transitions. . . This information is crucial to our understanding of the properties of large polynuclear complexes and are of great value to perform magnetostructural correlation of such complex molecules.

References [1] R. Sessoli, D. Gatteschi, A. Caneschi, M. A. Novak, Nature 1993, 365, 141. [2] (a) D. Gatteschi, A. Caneschi, L. Pardi, R. Sessoli, Science 1994, 265, 1054; (b) D. Gatteschi, A. Caneschi, R. Sessoli, A. Cornia, Chem. Soc. Rev. 1996, 101. [3] R. Sessoli, H. L. Tsai, A. R. Schake, S. Wang, J. B. Vincent, K. Folting, D. Gatteschi, G. Christou, D. N. Hendrickson, J. Am. Chem. Soc. 1993, 115, 1804. [4] C. P. Bean, J. D. Livinsgston, J. App. Phys. 1959, 30 , 1205. [5] A. Caneschi, D. Gatteschi, R. Sessoli, A.-L. Barra, L. C. Brunel, M. Guillot, J. Am. Chem. Soc. 1991, 113, 5873. [6] (a) A. K. Powell, S. L. Heath, D. Gatteschi, L. Pardi, R. Sessoli, G. Spina, F. Giallo, F. Pieralli, J. Am. Chem. Soc. 1995, 117, 2491; (b) S. L. Heath, A. K. Powell, Angew. Chem. Int. Ed. Engl. 1992, 31, 191.

224

6 High-spin Metal-ion-containing Molecules

[7] (a) G. Toulouse, Comm. Phys. 1977, 2, 115; (b) O. Kahn, Chem. Phys. Lett. 1997, 265, 109. [8] (a) J. K. McCuster, H. G. Jang, S. Wang, G. Christou, D. N. Hendrickson, Inorg. Chem., 1992, 31, 1874; (b) J. Ribas, B. Albela, H. Stoeckli–Evans, G. Christou, Inorg. Chem. 1997, 36, 2352; and references therein. [9] (a) R. D. Cannon, R. P. White, Prog. Inorg. Chem. 1988, 36, 195; (b) A. B. Blake, A. Yavari, W. E. Hatfield, C. N. Sethulekshmi, J. Chem. Soc., Dalton Trans. 1985, 2509; (c) A. R. E. Baikie, M. B. Hursthouse, D. B. New, P. Thornton, J. Chem. Soc., Chem. Commun. 1978, 62; (d) J. B. Vincent, H. R. Chang, K. Folting, J. C. Huffman, G. Christou, D. N. Hendrickson, J. Am. Chem. Soc. 1987, 109, 5703; (e) S. E. Wohler, r. J. Wittebort, S. M. Oh, T. Kambara, D. N. Hendrickson, D. Inniss, C. E. Strouse, J. Am. Chem. Soc. 1987, 109, 1063. [10] (a) D. F. Harvey, C. A. Christmas, J. K. McCusker, P. M. Hagen, R. K. Chadha, D. N. Hendrickson, Angew. Chem. Int. Ed. Engl. 1991, 30, 598; (b) J. McCuster, C. A. Christmas, P. M. Hagen, R. K. Chadha, D. F. Harvey, D. N. Hendrickson, J. Am. Chem. Soc. 1991, 113, 6114. [11] K. L. Taft, S. J. Lippard, J. Am. Chem. Soc. 1990, 112, 9629. [12] K. L. Taft, C. D. Delfs, G. C. Papaefthymiou, S. Foner, D. Gatteschi, S. J. Lippard, J. Am. Chem. Soc. 1994, 116, 823. [13] (a) E. Libby, J. K. McCuster, E. A. Schmitt, K. Folting, D. N. Hendrickson, G. Christou, Inorg. Chem. 1991, 30, 3486; (b) S. Wang, J. C. Huffman, K. Folting, W. E. Streib, E. B. Lobkovsky, G. Christou, Angew. Chem. Int. Ed. Engl. 1991, 30, 1672. [14] (a) M. W. Wemple, D. K. Coggin, J. B. Vincent, J. K. McCuster, W. E. Streib, J. C. Huffman, D. N. Hendrickson, G. Christou, J. Chem. Soc., Dalton Trans. 1998, 719; (b) S. L. Castro, Z. Sun, C. M. Grant, J. C. Bollinger, D. N. Hendrickson, G. Christou, J. Am. Chem. Soc. 1998, 120, 2365; (c) Z. Sun, C. M. Grant, S. L. Castro, D. Hendrickson, G. Christou, Chem. Commun. 1998, 721. [15] E. Libby, K. Folting, C. J. Huffman, J. C. Huffman, G. Christou, Inorg. Chem. 1993, 32, 2549. [16] H. L. Tsai, S. Wang, K. Folting, W. E. Streib, D. N. Hendrickson, G. Christou, J. Am. Chem. Soc. 1995, 117, 2503. [17] T. Lis, Acta Cryst. 1980, B 36, 2042. [18] A. R. Schake H. L. Tsai, R. J. Webb, K. Folting, G. Christou, D. N. Hendrickson, Inorg. Chem. 1994, 33, 6020. [19] H. J. Eppley, H. L. Tsai, N. Vries, K. Folting, G. Christou, D. N. Hendrickson, J. Am. Chem. Soc. 1995, 117, 301. [20] S. Parsons, R. E. P. Winpenny, Acc. Chem. Res. 1997, 30, 89 and references therein. [21] (a) F. A. Cotton, P. E. Fanwick, R. H. Niswander, J. C. Sekutowski, J. Am. Chem. Soc. 1978, 100, 4725; (b) M. Berry, W. Clegg, C. D. Garner, I. H. Hillier, Inorg. Chem. 1982, 21, 1342; (c) F. A. Cotton, L. R. Falvello, S. Han, W. Wang, Inorg. Chem. 1983, 22, 4106; (d) W. Clegg, C. D. Garner, M. H. Al–Samman, Inorg. Chem. 1983, 22, 1534; (e) Y. Nishida, S. Kida, Bull. Chem. Soc. Jpn. 1985, 58, 383. [22] E. K. Brechin, S. G. Harris, A. Harrison, S. Parsons, A. G. Whittaker, R. E. P. Winpenny, Chem. Commun. 1997, 653. [23] A. J. Blake, E. K. Brechin, A. Codron, R. O. Gould, C. M. Grant, S. Parsons, J. M. Rawson, R. E. P. Winpenny, J. Chem. Soc., Chem. Commun. 1995, 1983. [24] A. J. Blake, C. M. Grant, S. Parsons, J. M. Rawson, R. E. P. Winpenny, J. Chem. Soc., Chem. Commun. 1994, 2363. [25] W. E. Hatfield, in Magneto–Structrual Correlations in Exchange Coupled Systems, NATO ASI Series C (Eds.: R. D. Willet, D. Gatteschi, O. Kahn), 1985, pp. 555.

References

225

[26] A. Caneschi, A. Cornia, S. J. Lippard, Angew. Chem. Int. Ed. Engl. 1995, 34, 467. [27] (a) M. Lah, V. L. Pecoraro, Comments Inorg. Chem. 1984, 3, 205; (b) V. L. Pecoraro, A. J. Stemmler, B. J. Gibney, J. J. Bodwin, H. Wang, J. W. Kampf, A. Barwinski, Prog. Inorg. Chem. 1997, 45, 83. [28] A. Caneschi, A. Cornia, A. C. Fabretti, S. Foner, D. Gatteschi, R. Grandi, L. Schenetti, Chem. Eur. J. 1996, 2, 1379. [29] G. L. Abbati, A. Cornia, A. C. Fabretti, W. Malavasi, L. Schenetti, A. Caneschi, D. Gatteschi, Inorg. Chem. 1997, 36, 6443. [30] (a) V. Gadet, T. Mallah, I. Castro, M. Verdaguer, J. Am. Chem. Soc. 1992, 114, 9213; (b) T. Mallah, S. Thiebaut, ´ M. Verdaguer, P. Veillet, Science 1993, 262, 1554; (c) W. R. Entley, G. S. Girolami, Science 1995, 268, 397; (d) S. Ferlay, T. Mallah, R. Ouahes, ` P. Veillet, M. Verdaguer, Nature 1995, 378, 701; (e) S. Ferlay, T. Mallah, R. Ouahes, ` P. Veillet, M. Verdaguer, Inorg. Chem. 1999, 38, 229. [31] G. L. Abbati, A. Cornia, A. C. Fabretti, A. Caneschi, D. Gatteschi, Inorg. Chem. 1998, 37, 1430. [32] (a) J. B. Goodenough, Magnetism and the Chemical bond, Interscience, New York, 1963; (b) A. P. Ginsberg, Inorg. Chim. Acta 1971, 45. [33] R. Hotzelmann, K. Wieghardt, U. Florke, ¨ H.-J. Haupt, D. C. Weatherburn, J. Bonvoisin, G. Blondin, J.-J. Girerd, J. Am. Chem. Soc. 1992, 114, 1681. [34] G. L. Abbati, A. Cornia, A. C. Fabretti, A. Caneschi, D. Gatteschi, Inorg. Chem. 1998, 37, 3759. [35] Y. Pei, Y. Journaux, O. Kahn, Inorg. Chem. 1989, 28, 100. [36] (a) V. Balzani, S. Campagna, G. Denti, A. Juris, S. Serroni, M. Venturi, Acc. Chem. Res. 1998, 31, 26; (b) M. Fisher, F. Voglte, ¨ Angew. Chem. Int. Ed. Engl. 1999, 38, 884 and references therein. [37] A. Ulman, An Introduction to Ultrathin Organic Films: From Langmuir–Blodgett to Self–Assembly, Academic Press, Boston, 1991. [38] M. Clemente–Leon, H. Soyer, E. Coronado, C. Mingotaud, C. J. Gomez–Garcia, P. Delhaes, ` Angew. Chem. Int. Ed. Engl. 1999, 37, 2842. [39] (a) P. W. Anderson in Magnetism, Vol. 1 (Eds.: T. G. Rado, H. Suhl), Academic Press, New York, 1963; (b) P. J. Hay, J. C. Thiebault, R. Hoffmann, J. Am. Chem. Soc. 1975, 97, 4884; (c) J. J. Girerd, Y. Journaux, O. Kahn, Chem. Phys. Lett. 1981, 82, 534. [40] O. Kahn, Molecular Magnetism, VCH, Weinheim, Germany, 1993. [41] Y. Pei, Y. Journaux, O. Kahn, A. Dei, D. Gatteschi, J. Chem. Soc., Chem. Commun. 1986, 1300. [42] Y. Journaux, R. Ruiz, A. Aukauloo, Y. Pei, Mol. Cryst. Liq. Cryst. 1997, 305, 193. [43] A. Scuiller, T. Mallah, A. Nivorozkin, J. L. Tholence, M. Verdaguer, P. Veillet, New J. Chem. 1996, 20, 1. [44] M. Mitsumi, H. Okawa, H. Sakiyama, M. Ohba, N. Matsumoto, T. Kurisaki, H. Wakita, J. Chem. Soc., Dalton Trans. 1993, 2991. [45] C. Auberger, M. Verdaguer, P. Veillet, T. Mallah, J. Chem. Soc., Chem. Commun. 1995, 61. [46] R. J. Parker, D. C. R. Hockless, B. Moubaraki, K. S. Murray, L. Spiccia, Chem. Commun. 1996, 2789. [47] (a) T. Mallah, S. Ferlay, A. Scuiller, M. Verdaguer, in Magnetism : A Supramolecular Function, NATO ASI Series Vol. 484, (Eds. O. Kahn), Kluwer Academic, 1996, pp. 597; (b) N. Vernier, G. Bellessa, T. Mallah, M. Verdaguer, Phys. Rev. B 1997, 56, 75. [48] (a) S. M. J. Aubin, M. W. Wemple, D. M. Adams, H. L. Tsai, G. Christou, D. Hendrickson, J. Am. Chem. Soc. 1996, 118, 7746; (b) S. M. J. Aubin, S. Spagna, H. J. Eppley, R. E. Sager, G. Christou, D. N. Hendrickson, Chem. Commun. 1998, 803;

226

[49]

[50]

[51]

[52]

[53]

[54] [55] [56]

6 High-spin Metal-ion-containing Molecules (c) S. M. J. Aubin, N. R. Dilley, M. W. Wemple, M. B. Maple, G. Christou, D. N. Hendrickson, J. Am. Chem. Soc. 1998, 120, 839. (a) M. Ohba, N. Usuki, N. Fukita, H. Okawa, Angew. Chem. Int. Ed. Engl. 1999, 38, 1795; (b) M. Ohba, N. Usuki, N. Fukita, H. Okawa, Inorg. Chem. 1998, 37, 3349; (c) M. Ohba, H. Okawa, N. Fukita, Y. Hashimoto, J. Am. Chem. Soc. 1997, 119, 1011; (d) M. Ohba, H. Okawa, T. Ito, A. Ohto, J. Chem. Soc., Chem. Commun. 1995, 1545; (e) M. Ohba, N. Maruono, H. Okawa, T. Enoki, J.-M. Latour, J. Am. Chem. Soc. 1994, 116, 11566. (a) N. Re, R. Crescenzi, C. Floriani, H. Miyasaka, N. Matsumoto, Inorg. Chem. 1998, 37, 2717; (b) H. Miyasaka, N. Matsumoto, N. Re, E. Gallo, C. Floriani, Inorg. Chem. 1997, 36, 670; (c) N. Re, E. Gallo, C. Floriani, H. Miyasaka, N. Matsumoto, Inorg. Chem. 1996, 35, 6004. (a) J. L. Heinrich, P. A. Berseth, J. R. Long, J. Chem. Soc., Chem. Commun. 1998, 1231; (b) K. Van Langenberg, S. R. Batten, K. J. Berry, D. C. R. Hockless, B. Moubaraki, K. S. Murray, Inorg. Chem. 1997, 36, 5006; (c) S. Ferlay, T. Mallah, J. Vaissermann, F. Bartolome, ´ P. Veillet, M. Verdaguer, Chem. Commun. 1996, 2481; (d) S. El Fallah, E. Rentshler, A. Caneschi, R. Sessoli and D. Gatteschi, Angew. Chem. Int. Ed. Engl. 1996, 35, 1947; (e) K. Van Langenberg, S. R. Batten, K. J. Berry, D. C. R. Hockless, B. Moubaraki, K. S. Murray, Inorg. Chem. 1997, 36, 5006. (a) G. O. Morpugo, V. Mosini, P. Porta, G. Dessy, V. Fares, J. Chem. Soc., Dalton Trans. 1981, 111; (b) H. Z. Kou, D. Z. Liao, P. Cheng, Z. H. Jiang, S. P. Yan, G. L. Wang, X. K. Yao, H. G. Wang, J. Chem. Soc., Dalton Trans. 1997, 1503. P. Rey, D. Luneau, A. Cogne, in Magnetic Molecular Materials, NATO ASI Series Vol. 198 (Eds.: D. Gatteschi, O. Kahn, J. S. Miller, F. Palacio), Kluwer Academic, Dordrecht/Boston/London, 1990, pp. 203. A. Marvilliers, Y. Pei, J. Cano Boquera, K. E. Vostrikova, C. Paulsen, E. Riviere, ` J.-P. Audiere, ` T. Mallah, Chem. Commun. 1999, 1951. D. F. Shriver, S. H. Shriver, S. E. Anderson, Inorg. Chem. 1965, 5, 725. K. Fegy, D. Luneau, T. Ohm, C. Paulsen, P. Rey, Angew. Chem. Int. Ed. Engl. 1998, 37, 1270.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

7

Electronic Structure and Magnetic Behavior in Polynuclear Transition-metal Compounds Eliseo Ruiz, Santiago Alvarez, Antonio Rodr´ıguez-Fortea, Pere Alemany, Yann Pouillon, and Carlo Massobrio

7.1

Introduction

The observed magnetic behavior of transition-metal complexes containing more than one paramagnetic metal atom often differs from that predicted by the sum of the magnetic properties of each individual unit bearing unpaired electrons [1]. This phenomenon is due to a coupling of the electron spins and is termed intramolecular antiferromagnetism or ferromagnetism, depending upon whether antiparallel or parallel spin alignment is found in the ground state, respectively. Since the discovery of intramolecular magnetic coupling in 1951 by Guha [2] and then by Bleaney and Bowers [3] in a compound known at that time as copper(II) acetate monohydrate, the mechanism of exchange coupling and its relation with the electronic structure have been the subject of a large number of experimental and theoretical studies [4, 5]. From the theoretical point of view, it was soon realized that the study of the electronic structure of magnetically coupled systems is more challenging that the problem of chemical bonding in closed-shell molecules. While simple reasoning based on molecular topology, overlap between atomic orbitals and electronegativity often allow qualitative interpretations in closed-shell systems [6], no single qualitative model is able to explain satisfactorily all features of exchange coupled systems and there are still a number of controversies about the advantages and limits of the various approaches that have been devised. The direct calculation by means of sophisticated ab initio methods of the energy differences between the ground and low lying excited spin states has been also hindered by serious computational problems: energy splittings of the order of 100 cm−1 (∼0.3 kcal mol−1 ) or even smaller must be obtained as differences between total energies which are up to seven orders of magnitude larger [7]. If one adds to this situation that the compounds of experimental interest are formed by at least 40 atoms (including two or more transition metal ones) one can understand the relative scarcity of attempts to use the ab initio approach until the last few years. The rapid development of hardware and software technologies, together with the emergence and consolidation of new ab initio methods applied to the electronic structure of molecular compounds (mainly based on density functional theory, DFT) [8, 9] in the last decade has dramatically changed the situation. Thus, the simplest cases of exchange coupled systems, i. e. dimers of transition-metal ions with a single unpaired

228

7 Electronic Structure and Magnetic Behavior

electron on each metal atom, can be treated in an almost routine way using standard quantum chemical software packages on desktop computers. The study of more complex systems, involving two atoms with more than one unpaired electron per center or compounds with more than two paramagnetic centers is nowadays the subject of an intense research and their study will be possibly affordable in a routine way soon. The availability of quantitative or semiquantitative approaches to the electronic structure of magnetically coupled systems has, on the other hand, renewed the interest on the earlier qualitative models. The exploration of their strengths and weaknesses, together with the establishment of their limits of applicability has allowed researchers in this field to use these simple methods as valuable complementary tools in the interpretation of the results obtained with the more sophisticated ab initio calculations. The conceptual simplicity of these qualitative models often provides a much deeper insight into the physical origins of exchange coupling than that obtained by using the more accurate, but also more obscure, first principles techniques. In the first part of this chapter (Sections 2–4) we will present the key concepts involved in the most widely used qualitative models and ab initio approaches to the electronic structure of magnetically coupled systems, discussing their merits and limits. In the second part (Section 5) we will show how some of these theoretical approaches have been applied to selected examples.

7.2

Phenomenological Description of Exchange Coupling: the Heisenberg Hamiltonian

The study of intramolecular exchange interactions started, as mentioned above, with the analysis of the magnetic properties of copper(II) acetate. This compound is actually dimeric, with four acetate ligands bridging two copper(II) ions (1), each bearing one unpaired electron. The magnetic susceptibility of this compound exhibits a broad maximum as a function of temperature and becomes negligible below 100 K. This behavior can be rationalized through the phenomenological Heisenberg Hamiltonian that describes the exchange interaction between the two paramagnetic centers: Hˆ = −J SˆA · SˆB

(1)

where SA and SB are the total spins on each metal ion (SA = SB = 1/2 in this case) and J is known as the exchange coupling constant [1]. With the above definition, positive values of J indicate a ground state with parallel spins, that is, a ferromagnetic interaction, while negative values correspond to an antiferromagnetic coupling. (Alternative definitions for the exchange coupling constants are also common in the literature. In some cases, the Heisenberg Hamiltonian is expressed as Hˆ = +J SˆA · SˆB , whereupon positive values of J indicate antiferromagnetic coupling, and negative ones correspond to the ferromagnetic situation. In other instances, the Hamiltonian used is Hˆ = −2J SˆA · SˆB and the reported coupling constants must be multiplied

7.2 Phenomenological Description of Exchange Coupling: the Heisenberg Hamiltonian

229

by 2 to make them comparable with those obtained using Eq. 1.) The experimental magnetic susceptibility indicates that the electron spins in copper(II) acetate are antiferromagnetically coupled with J = −296 cm−1 . If the two spins of the metal ions interact, the local spins SA and SB are not good quantum numbers and we must use the total spin: Sˆ = SˆA + SˆB

(2)

to characterize the pair states. It is relatively easy to show that the eigenvalues of the Heisenberg Hamiltonian can be expressed as a function of the total spin quantum number S: J E(S) = − S(S + 1) 2

(3)

which, for two local doublet states found in copper(II) acetate leads to singlet (S = 0) and triplet (S = 1) states separated by an energy gap of magnitude J : E S − ET = J

(4)

It is important to note here that exchange coupling constants are not determined directly from experiment. The usual procedure is to use the model Hamiltonian to derive an expression for the temperature dependence of the magnetic susceptibility and to fit the experimental susceptibility to this expression, treating the coupling constant as an adjustable parameter [4]. This procedure, which is straightforward for exchange coupled dimers is, however, difficult for many polynuclear or extended systems, for which an analytical expression for the temperature dependence of the magnetic susceptibility is still lacking. In such cases, approximate models or simulation procedures are often used to obtain the temperature dependence of the magnetic

230

7 Electronic Structure and Magnetic Behavior

susceptibility and to fit it to the experimental data [5]. The use of different models may, however, lead to significantly different “experimental” coupling constants; this makes comparison with calculated constants difficult.

7.3

Qualitative Models of the Exchange Coupling Mechanism

The Heisenberg Hamiltonian is purely phenomenological and it does not provide any information on the real mechanism of the interaction between the two unpaired electrons. The theoretical interpretation of exchange interactions has traditionally been based on ideas developed for infinite solid lattices [10–13]. Since it has been realized empirically that the bridging atoms between the metal ions bearing the unpaired electrons determine the sign and magnitude of the exchange interaction, such qualitative treatments focus on the various types of overlap between ligandcentered and metal d orbitals. Extension of these ideas to cases involving molecular, rather than atomic bridging species has lead to the two most widely used qualitative models for intramolecular exchange interactions that will be briefly reviewed here.

7.3.1

Orthogonal Magnetic Orbitals

To relate the experimentally available quantity, the coupling constant J , to the electronic structure of the compound, let us consider a centrosymmetric model system Ma –X–Mb , where Ma and Mb are two paramagnetic centers (transition metal atoms in our case) with one unpaired electron each and X is a closed-shell diamagnetic bridge (or set of bridges). The unpaired electron on each paramagnetic center occupies one of the d orbitals of the transition metal atoms, da and db . In copper(II) acetate, for example, the unpaired electrons can be found on x2 –y2 -like copper orbitals oriented toward four oxygen atoms of the acetato bridges (2).

The combination of the da and db orbitals with ligand-centered φx orbitals leads to two molecular orbitals ϕ1 and ϕ2 (see 3, where only the contributions of one of the acetato bridges are depicted for simplicity) that play a key role in the qualitative orbital models proposed to rationalize magnetic behavior of dinuclear complexes. If we restrict our analysis to the two unpaired electrons occupying molecular orbitals ϕ1

7.3 Qualitative Models of the Exchange Coupling Mechanism

231

and ϕ2 (active-electron approximation) the following many-electron configurations arise (4) [14]: S1 = |ϕ1 αϕ1 β| S2 = |ϕ2 αϕ2 β| 1 S3 = √ |ϕ1 αϕ2 β| − |ϕ1 βϕ2 α| 2 T,+1 = |ϕ1 αϕ2 α|; T,−1 = |ϕ1 βϕ2 β|; 1 T,0 = √ |ϕ1 αϕ2 β| + |ϕ1 βϕ2 α| 2

(5)

Although the energy of the lowest triplet can be evaluated for a single determinant, e.g. T,+1 , the lowest singlet state will be represented by a linear combination of S1 and S2 : S = λ1 S1 + λ2 S2

(6)

In the limit of non-interacting metal ions we have |λ1 | = |λ2 |, while in the opposite extreme, for strong metal-metal bonding, |λ1 | |λ2 |. In a centrosymmetric dimer, ϕ1 and ϕ2 belong to different symmetry species and there is no contribution of S3 to the lowest singlet state because it belongs to a different symmetry species than S1 and S2 . In what follows it will be assumed that the MOs themselves correspond to an SCF solution for the triplet state [15]. After introducing some plausible approximations, Hay, Thibeault and Hoffmann [16] proposed the following expression for the singlet–triplet splitting: 1 (ε1 − ε2 )2 E S − E T = −J12 + (J11 + J22 ) − 2 2K 12

(7)

where ε1 and ε2 are the energies of molecular orbitals ϕ1 and ϕ2 , and Ji j and K i j are the Coulomb and exchange integrals, respectively, expressed in the molecular orbital basis: −1 Ji j = ϕi (1)ϕ j (2)|r12 |ϕi (1)ϕ j (2) (8) −1 K i j = ϕi (1)ϕ j (2)|r12 |ϕ j (1)ϕi (2) (9)

232

7 Electronic Structure and Magnetic Behavior

The interpretation of experimental data is simpler if Eq. (7) is expressed in terms of two-electron integrals involving localized molecular orbitals ϕa and ϕb defined as follows: 1 ϕa = √ (ϕ1 + ϕ2 ) 2

and

1 ϕb = √ (ϕ1 − ϕ2 ) 2

(10)

Because of their orthogonality, ϕa |ϕb = 0, ϕa and ϕb are sometimes called orthogonal magnetic orbitals. The singlet-triplet splitting, Eq. (7), can be now rewritten as: E S − E T = J = 2K ab −

(ε1 − ε2 )2 Jaa − Jab

(11)

which is the key equation of the Hay, Thibeault, and Hoffmann (HTH) model relating the exchange coupling constant to the electronic structure of the compound. Because K ab , Jaa , and Jab are all positive and Jaa > Jab , both terms in Eq. (11) must be positive. The first term in this equation represents, therefore, a positive, ferromagnetic, contribution to the overall coupling constant, whereas the minus sign preceding the second term in Eq. (11) indicates an antiferromagnetic contribution to the exchange coupling constant: J = JF + JAF

(12)

JF = 2K ab

(13)

with

and JAF = −

(ε1 − ε2 )2 Jaa − Jab

(14)

If ϕ1 and ϕ2 are degenerate or nearly degenerate, the antiferromagnetic contribution vanishes and a triplet ground state results. On the other hand, a significant splitting between these two molecular orbitals will yield a singlet ground state. Eq. (11) thus suggests that we can focus on the orbital energy difference ε1 − ε2 as a measure of the singlet-triplet energy splitting for these compounds [16]. The HTH model has been mainly applied within the extended Huckel ¨ framework, [17–20] the simplest all-valence-electron model in quantum chemistry. Since twoelectron interactions are not included in this method, actual singlet-triplet energy differences cannot be calculated with such approach. One can nevertheless assume that extended Huckel ¨ calculations reproduce correctly the qualitative changes in the orbital energies as a function of the structure and the nature of the substituents [6]. Thus, the HTH model allows the study of the variation of J within a family of dinuclear compounds if one assumes that all two-electron integrals in Eq. (11) are practically insensitive to structural changes (or to changes in the non-bridging ligands) [21]. Within this approximation, that has been used to explain a large number

7.3 Qualitative Models of the Exchange Coupling Mechanism

233

of experimentally observed magneto-structural correlations [22–30], the analysis is restricted to the variation of the one electron term (ε1 − ε2 )2 in Eq. (11). Extension of the HTH model to more complex systems, such as dinuclear compounds with more than one unpaired electron on each metal atom or compounds with more than two metal atoms, is not straightforward [31]. The number of configurations that must be considered to describe the spin states of the system grows rapidly with the number of unpaired electrons and the analysis becomes more and more complex. Explicit expressions similar to Eq. (11) have been derived only for dimers with the same number of unpaired electrons on each metal atom. In these expressions, the antiferromagnetic term can be analyzed in terms of separate contributions from pairs of orbitals. In the case of a dimer with two d8 ions in local octahedral environments, for example, one must consider four molecular orbitals: ϕ1 and ϕ2 arising from the in-phase and out-of-phase combinations of the x2 − y2 orbitals of the two metals, and ϕ3 and ϕ4 involving the z2 orbitals. Two orthogonalized magnetic orbitals are now constructed from each of these pairs (ϕ1 and ϕ2 give raise to ϕa and ϕb , while ϕc and ϕd are constructed from ϕ3 and ϕ4 using expressions analogous to Eq. 10). The total coupling constant in terms of orbital energies and two-electron integrals involving the orthogonalized magnetic orbitals is, within this approximation: J = E S − ET = =

1 (E T − E Q ) 2

1 1 (ε1 − ε2 )2 1 (ε3 − ε4 )2 − (K ab + K ad + K bc + K cd ) − 2 4 Jaa − Jab 4 Jcc − Jcd

(15)

which shows that the antiferromagnetic term can be traced to the separate contributions from the x2 − y2 and z2 orbitals [16]. Here, the energy values E S , E T and E Q correspond to the spin states with S = 0, S = 1 and S = 2, respectively. Despite its success in explaining a large amount of experimental data, the shortcomings of the HTH model are evident. The model allows only to study trends within a family of compounds with the same Ma –X–Mb core. Since the actual values of the coupling constant are not directly evaluated, it is impossible to predict one of the most important properties in intramolecular magnetism, that is, the relative coupling ability of different bridging ligands. The model fails also in the case of the most interesting compounds, those exhibiting ferromagnetic coupling. Since the term (ε1 − ε2 )2 is supposed to be negligible for these compounds, changes in K i j , even if small in absolute terms, can be crucial in the study of magneto-structural correlations. An additional problem that limits the applicability of the HTH model is that the localization criterion, see Eq. (10), is only valid when the Ma and Mb centers are symmetry-related, either through an inversion center, or through a twofold axis. This fact precludes the application of the HTH model to heterodinuclear compounds which are of great interest as potential building blocks for molecule-based ferrimagnetic systems [32].

234

7.3.2

7 Electronic Structure and Magnetic Behavior

Natural Magnetic Orbitals

To remedy, at least partially, some of the deficiencies of the HTH model, an alternative approach to analyze the relationship between electronic structure and the exchange coupling mechanism was devised [4, 33]. The exchange interaction in the Ma –X–Mb system may be viewed as the borderline case of a very weak chemical bond. In this case it is appropriate to use a Heitler-London type of wavefunction: =

1 2 ) 2(1 ± Sab

χa (1)χb (2) ± χa (2)χb (1)]

(16)

where the positive sign holds for the singlet and the negative sign for the triplet state. In this approximation, χa is defined as the singly occupied molecular orbital for the Ma –X fragment in its local ground state and χb is defined in the same way with respect to the X–Mb fragment. This type of orbitals has been called natural magnetic orbitals by Kahn and coworkers [4, 33–35]. The cutting of the Ma –X–Mb system into two Ma –X and X–Mb fragments with a common bridging region X is not rigorous and constitutes certainly the weak point of this approach since too much weight of the bridge’s orbitals is introduced in this way in the wavefunctions describing each state. One way suggested to determine the natural magnetic orbital χa consists in contracting the atomic orbitals of Mb until all orbital interactions between the magnetic center Mb and its surroundings become negligible. Natural magnetic orbitals are by construction non-orthogonal and their overlap integral Sab = χa | χb

(17)

plays a key role in the qualitative orbital model devised by Kahn and coworkers to predict the magnitude of exchange interactions in magnetically coupled systems [36–45]. Considering the two states described by the Heitler-London wavefunctions (Eq. 16) the singlet-triplet splitting can be expressed as: 2 E S − E T = J = 2K + 4h ab Sab − 2Sab (2h aa + j)

(18)

where ˆ a h aa = χa |h|χ ˆ b h ab = χa |h|χ −1 k = χa (1)χb (2)|r12 |χa (2)χb (1) −1 j = χa (1)χb (2)|r12 |χa (1)χb (2)

(19)

In these expressions hˆ is the one-electron Hamiltonian that takes into account the kinetic energy of the electron and its interactions with the nuclei and with all the passive electrons. For a centrosymmetric Ma –X–Mb system, h aa = h bb .

7.4 Quantitative Evaluation of Exchange Coupling Constants

235

The 2k term in Eq. (18) is always positive, and therefore represents a ferromagnetic contribution. Assuming Sab to be small enough, Sab and h ab are of opposite sign and the second term in Eq. (18) represents an antiferromagnetic contribution to the overall coupling constant. The sign of the last term in Eq. (18) cannot be easily determined because h aa and j have opposite signs. For small enough of Sab values one can assume that the contribution of this third term to the coupling constant will be much smaller than the rest and it is often neglected in a qualitative analysis, resulting in: J = JF + JAF ≈ 2k + 4h ab Sab

(20)

In a more elaborate model one can include the contribution of the metal–metal charge-transfer configurations χa (1)χa (2) and χb (1)χb (2) to the singlet state. This treatment gives rise to an additional antiferromagnetic term stabilizing the singlet state. There is no definitive answer yet as to which of the two antiferromagnetic contributions dominates the value of J . It has been however suggested that in transition metal dinuclear compounds the metal-metal charge-transfer configurations are in general too high in energy to couple significantly with the low lying singlet state [4]. The qualitative interpretation of the exchange coupling is now based on the overlap Sab between natural magnetic orbitals. If Sab vanishes, according to Eq. (20) we should expect a triplet ground state with J = 2k. The strength of the antiferromagnetic coupling provided by different bridging ligands (or sets of bridging ligands) X can now be compared, provided that one is able to calculate Sab . This is, however, a difficult task since, as mentioned above, the definition of natural magnetic orbitals is not rigorous. This difficulty has relegated the application of this qualitative model to the cases in which Sab can be predicted to vanish for symmetry reasons. The search for compounds with orthogonal natural magnetic orbitals has lead to some success in the synthesis of molecules with “designed” ferromagnetic coupling, as in Cu(II)Cr(III)[37] or Cu(II)V(IV) complexes [39]. It is also worth to mention in this section the work of Gudel ¨ et al., who explored the exchange interaction, for instance in several transition metal oxo complexes [46, 47], using a perturbative approach to obtain the different contributions to the exchange interaction. This approach has also been employed by von Seggern et al. [48]. to study the ferromagnetic coupling in end-on azido Cu(II) complexes. The parameters needed in these calculations, such as the h ab transfer integrals are obtained from simple angular overlap approximations or extended Huckel ¨ calculations, while the orbital energy differences U are usually estimated from experimental spectroscopic data [49].

7.4

Quantitative Evaluation of Exchange Coupling Constants

Although successful qualitative predictions have been derived from the models discussed above, non-empirical calculations including electron correlation are needed to reach quantitative estimates of exchange coupling constants. As can be deduced

236

7 Electronic Structure and Magnetic Behavior

from the preceding discussion, the consideration of configuration interaction is an unavoidable step in any attempt to calculate the spectrum of the low lying states of a magnetic polynuclear compound. The precise calculation of the small energy gaps needed for the evaluation of coupling constants is a great computational challenge. In this section we will briefly review the fundamental aspects of the different approaches that have been proposed to reach this goal.

7.4.1

Perturbative and Variational Calculations of State Energy Differences

In one of the pioneering works in theoretical molecular magnetism, de Loth et al. solved this dilemma by calculating the singlet–triplet energy difference in copper(II) acetate directly [50]. For this purpose these authors devised a perturbative development up to second order of the configuration interaction problem which allowed them not only a first semiquantitative computation of the coupling constant in this compound, but also a decomposition of such parameter in different contributions that permitted a qualitative understanding of the physical mechanisms responsible for the exchange coupling phenomenon. The calculation of the coupling constant within this approach consists of two separate steps. The first is the construction of magnetic orbitals performing a restricted open-shell SCF calculation. This calculation yields two singly occupied molecular orbitals ϕ1 and ϕ2 that are then transformed into orthogonal localized magnetic orbitals ϕa and ϕb defined by Eq. (10). The M S = 0 components of the singlet and the triplet states are defined by: 1 1 0 = √ | . . . ϕi αϕi β . . . ϕa αϕb β| + | . . . ϕi αϕi β . . . ϕa βϕb α| 2 1 = √ (1 + 2 ) 2 (21) 1 3 0 = √ | . . . ϕi αϕi β . . . ϕa αϕb β| − | . . . ϕi αϕi β . . . ϕa βϕb α| 2 1 = √ (1 − 2 ) 2 where ϕi stands for one of the doubly occupied molecular orbitals and 1 , 2 are shorthand notations for the two Slater determinants involved in the above expressions. The zeroth-order singlet–triplet energy splitting is given by the well-known exchange integral: (0)

E ST = 2K ab

(22)

that represents a ferromagnetic contribution, called potential exchange, favoring a triplet ground state. The second order contribution to the singlet–triplet splitting is given by: 2 1 | Hˆ |i i | Hˆ |2 i (2) E ST = (23) E0 − Ei

7.4 Quantitative Evaluation of Exchange Coupling Constants

237

where E 0 is the energy corresponding to 1 or 2 and E i are the energies of the determinants i which describe excited states. It is interesting to note that the number of determinants i which contribute to the second-order perturbation term of the singlet–triplet separation is reduced to those which interact with both 1 and 2 . For this reason, these determinants are much less numerous than those involved in the second order CI corrections to the energy of the singlet and triplet states. The determinants i which contribute to the second-order correction must differ, at most, by two spin orbitals from 1 and 2 to give a non-zero numerator. Fig. 1 shows a schematic representation of all types of i determinants along with the name of their contribution to the second order correction to the singlet-triplet energy splitting.

Fig. 1. Schematic representation of the determinants that interact with both 1 and 2 in the second-order correction to the singlet-triplet energy splitting.

238

7 Electronic Structure and Magnetic Behavior

Table 1. Numerical values (cm−1 ) of various contributions to the singlet-triplet energy splitting, J , for Cu(II) acetate after de Loth et al. [50]. Values in the fourth column correspond to the accumulated sum of the different contributions to J . The corresponding experimental value is −297 cm−1 [4, 51, 52]. Perturbation

Contribution

Zeroth order Second order

Potential exchange Kinetic exchange Double-spin polarization M → L charge transfer L → M charge transfer Kinetic exchange + polarization Kinetic exchange Kinetic exchange + polarization

Higher order

Ji +233.6 −204.3 −52.0 0 −5.9 −2.3

Ji +233.6 +29.3 −22.7 −22.7 −28.6 −30.9

−89.3 or −213.5 −120.2 or −244.4

As can be deduced from Table 1, the leading contributions for copper acetate are those of potential exchange Eq. (22) and kinetic exchange (second-order term, see Fig. 1). These two contributions have been shown to correspond, respectively, to J F and J AF in the qualitative HTH model Eq. (11) which therefore corresponds to the neglect of all other second and higher order terms in the perturbative expansion of the singlet-triplet gap. The actual figures in Table 1 show that the kinetic exchange barely compensates the direct exchange, bringing the singlet-triplet splitting almost to zero. This result, which has been found in all cases studied with this methodology, is rather disappointing, since earlier qualitative explanations of exchange coupling such as the HTH model had been based on just these two terms. Addition of other contributions to the second order expression (see Table 1) results in a progressive stabilization of the singlet state relative to the triplet. The value given by de Loth et al. for J at the second order level is −30.9 cm−1 , which has already the correct sign, but is still far from the experimental value of approximately −297 cm−1 [4, 51, 52]. Such disagreement led the authors to consider higher order corrections, although these are so numerous that cannot be calculated in a complete way. Considering two different ways to estimate higher order corrections, de Loth et al. arrive at much better values for J : −120.2 and −244.4 cm−1 . These two values show, however, the weak point of this perturbational approach, that is, convergence. In many cases stopping at the second order level gives just the correct sign of the coupling constant and higher order terms are needed to obtain a quantitative estimate of its value. The treatment of these higher order terms is far from trivial and different treatments can lead to quite different values for such terms. An interesting question that arises at this point is why is the HTH model able to qualitatively predict the changes in J for small structural distortions or changes in the peripheral ligands if the two terms that it retains practically cancel each other? An answer to this question can be found in the work of Daudey et al. that attempted to justify the experimentally observed correlation between the value of J and the Cu—O—Cu angle in planar hydroxo-bridged Cu(II) compounds [53]. The calculated values for J , together with its decomposition in the different zero and second-order terms, are shown in Table 2 for [Cu(tmen)OH]2 Br2 for its experimental geome-

7.4 Quantitative Evaluation of Exchange Coupling Constants

239

Table 2. Numerical values (cm−1 ) of various contributions to the singlet-triplet energy splitting, J , for [Cu(tmen)OH]2 Br2 with its experimental geometry (left column) and with the O–Cu–O angle set to 95◦ , after Daudey et al. [53]. Contribution

Cu–O–Cu 104◦

Potential exchange Kinetic exchange Double-spin polarization L → M charge transfer M → L charge transfer Kinetic exchange + polarization Calculated value Experimental value

+1380 −1306 +11 −477 −35 −236 −393 −509

95◦ +1275 −431 +40 −492 −35 −260 +97 +161

try (Cu—O—Cu = 104◦ ) and for a hypothetical structure having Cu–O–Cu = 95◦ , for which a value of J = 161 cm−1 is predicted from the experimentally-derived magneto-structural correlation. It can be clearly deduced from the values in Table 2 that the variation of the kinetic exchange term is actually the main cause for the variation in J . All other second-order terms, although far from negligible, are much less sensitive to the structural variation analyzed. The success of the HTH model is thus related to the fact that it reduces the antiferromagnetic contribution to the kinetic exchange, which is the only second-order contribution that is strongly affected by the molecular geometry. The perturbational approach of de Loth et al. has been applied with some success to Cu(II) dinuclear complexes with acetato [53], hydroxo [53], alkoxo [54] and oxalato bridges [55], as well as to a heterodinuclear Cu(II) -V(IV) complex, in which the kinetic exchange term is found to be zero as predicted by the qualitative model of Kahn and coworkers [56]. This approach was subsequently employed by Haase and coworkers to study N -oxide [57], hydroxo [58], alkoxo/acetato [59] and terephthalato-bridged [60] Cu(II) dinuclear complexes and hemocyanin models [61]. The difficulties related to the convergence of the perturbational corrections that arise within this approach lead Miralles et al. [62–64] to propose a different strategy which tried to avoid the aforementioned problems. In the so-called differencededicated configuration interaction (DDCI) method low-order perturbation criteria are applied to select the i determinants which contribute directly to the considered energy difference. After this subspace is rationally selected, the calculation of the transition energies is treated variationally via a CI calculation. For two electrons in two active orbitals (ϕa and ϕb ), as present in exchange-coupled Cu(II) dinuclear compounds, the model space that is used consists of the two neutral valence-bond structures |ϕa αϕb β| and |ϕa βϕb α|. The corresponding list of interacting determinants (the DDCI-2 list) involves only those presenting up to two inactive orbitals (i. e. one hole and one particle, or two holes and two particles). Diagonalization of this subspace of single and double excitations permits to obtain the singlet-triplet gap as

240

7 Electronic Structure and Magnetic Behavior

the difference between the first two roots of this variational CI. It has been shown that, since the DDCI space is invariant under a unitary transformation of the active orbitals, it is also possible to generate the DDCI space using the delocalized orbitals ϕ1 and ϕ2 . The DDCI-2 method has been applied with success to different magnetically coupled dinuclear compounds involving one or more unpaired electrons per center, chloro and hydroxo doubly-bridged Cu(II) complexes [65], chloro and azido doublybridged Ni(II) compounds [66], and hydroxo-bridged Cr(III) binuclear complexes [67]. For copper(II) acetate the calculated value at this level of approximation is −77 cm−1 (see Table 3 for a comparison of calculated J values for copper acetate using various quantitative approaches) which correctly reproduces the antiferromagnetic character of the interaction, but strongly underestimates its actual magnitude. This discrepancy might come from the neglect of important contributions to the singlet state, notably the ionic valence-bond structures, |ϕa αϕa β| and |ϕb αϕb β|. If these two determinants are included in the model space, the difference dedicated CI space has to be enlarged up to a DDCI-3 list, involving configurations with two inactive holes and one inactive particle or one inactive hole and two inactive particles [68]. This list increases the size of the CI space by a factor of 50–60, but gives a much better theoretical estimate of J for copper acetate: −224 cm−1 . The DDCI-3 method, even if appealing because of the quality of results it delivers, suffers from its computational complexity, which limits its use to heavily simplified models of the actually synthesized compounds. As will be seen later in the section dedicated to some representative examples, modeling terminal or bridging ligands by substituting bulky groups by simpler ones might induce changes of up to 50–100 cm−1 in the calculated coupling constants. In such cases it is impossible to tell whether the discrepancy between theoretical and experimental values comes from the computational methodology or from the modeling of the real structure.

7.4.2

Ab initio Calculations of State Energies

The “brute force” approach, i. e. the calculation of exchange coupling constants from the energy differences of the thermally populated spin states, has also been employed by some authors. As mentioned above, an essential requirement for an estimation of J using this procedure is the availability of highly accurate energies for all the states involved in the magnetic behavior of the compound under scrutiny. Representative calculations of this type have been performed by Staemmler and coworkers employing their own methodology [69–73], and Pierloot and Ceulemans using the standard complete active space SCF method (CASSCF) with inclusion of the dynamic correlation through second-order perturbation theory (CASPT2) [74, 75]. The quantum chemical strategy proposed by Staemmler and coworkers starts with a first calculation aimed at generating molecular orbitals equally appropriate for all low-lying electronic states. These can be obtained usually from a ROHF calculation for the highest multiplicity state, although a more involved calculation may be necessary [69]. Once these orbitals have been obtained, the energy of each state can

7.4 Quantitative Evaluation of Exchange Coupling Constants

241

be calculated by means of either a full valence configuration (VCI) treatment or a CASSCF calculation. The results show that there is virtually no difference between these two treatments since all electronic states involved are very close in energy. Dynamic correlation effects must be included in a third step on top of the VCI or CASSCF calculation. The method of choice of Staemmler and coworkers is the multiconfiguration coupled electron pair approach (MC-CEPA) [76]. This theoretical approach has been applied to different exchange coupled dimers with more than one unpaired electron per center, oxo-bridged Ti(III), V(III) and Cr(III) complexes [69– 71], and oxo- and sulfur-bridged Ni(II) complexes [72, 73]. It is interesting to remark that these authors have also been the first ones to include explicitly the important effects of spin–orbit coupling and Zeeman splitting in their study of chlorine-bridged dinuclear cobalt(II) complexes [77]. Pierloot and Ceulemans have focused on the study of the exchange interaction in [Ti2 Cl9 ]3− , including parametric expressions of metal-centered spin-orbit coupling and Zeeman splitting due to the d1 configuration of the metal atom [74, 75]. Recently, these authors have published a review of their work reporting also results for [Cr2 Cl9 ]3− [78]. The VCI or the CASSCF calculations correctly reproduce the ordering of the low lying energy states, but the inclusion of dynamic correlation effects is needed if one wants to calculate reliable values of coupling constants. The main reason is that the antiferromagnetic contribution to J is determined by the extent to which “ionic” (charge transfer) configurations are mixed into the dominant “neutral” (covalent) ones in the VCI. As long as the ionic configurations are built up from orbitals which have been optimized for covalent states and no relaxation effects are accounted for, their energies are too high and consequently their contribution to J AF is too small. The quality of results that can be obtained by such an approach can be deduced from the values in Table 3. The CASSCF calculation of the second step yields a value of −24 cm−1 for J in copper acetate, which correctly indicates a singlet ground state, Table 3. Numerical values (cm−1 ) of the singlet-triplet energy splitting, J , for copper(II) acetate calculated using some of the methods discussed in the text. For calculations performed using the broken-symmetry approach, unprojected and projected (in parenthesis, this issue will be discussed in the next section) values are indicated in the table. All data included in this table have been calculated for this review [79–81]. Method DDCI-2 DDCI-3 CASSCF CASPT2 UHF-bs Xα-bs SVWN-bs BLYP-bs B3LYP-bs Experimentala a

Ref. [51].

J −77 −224 −24 −117 −27 −848 −1057 −779 −299 −297

(−54) (−1696) (−2114) (−1558) (−598)

242

7 Electronic Structure and Magnetic Behavior

but strongly underestimates the magnitude of J , accounting for less than 10% of its experimental value (−296 cm−1 ). Consideration of the dynamic electron correlation by means of perturbation theory, as in the CASPT2 method, recovers a substantial part of the antiferromagnetic contribution to J yielding a value of −117 cm−1 that is, however, still far from the experimental one. This value is substantially improved when some bridging ligand-centered molecular orbitals are included in the active space. Despite the success of the ab initio approach in the determination of coupling constants, its weak point is doubtlessly its computational complexity. The use of the state-of-the-art programs needed for this type of calculations is not only out of reach for most experimental chemists interested in molecular magnetism, but is also severely limited by the size of the molecule that one wants to study. The ab initio approach has been only applied so far to strongly idealized models of exchange coupled systems in which bulky terminal ligands have been substituted by smaller ones and in which small structural distortions have been disregarded in order to build a model with as much symmetry elements as possible to reduce the computational burden.

7.4.3

Calculations using Broken-symmetry Functions

A possible solution to the problems outlined above was advanced by Noodleman et al. in 1981 in one of the key papers on the theoretical treatment of magnetically coupled systems [82]. Noodleman’s suggestion in this work was the use of a single configuration model containing non-orthogonal orbitals for the calculation of the coupling constant using either unrestricted Hartree–Fock theory or spin-polarized density functional theory, e. g., the Xα method [83, 84]. The crucial point of this proposal was the use of a state of mixed spin symmetry and lowered space symmetry (a broken-symmetry wavefunction). Let us briefly review the more important features of this method for a system having two paramagnetic centers with one unpaired electron each occupying orthogonal magnetic orbitals ϕa and ϕb [85–87]. Proper spin eigenfunctions for this system are given by: 1 S,0 = √ |ϕa αϕb β| + |ϕa βϕb α| 2 1 T,0 = √ |ϕa αϕb β| − |ϕa βϕb α| 2 T,+1 = |ϕa αϕb α| T,−1 = |ϕa βϕb β|

(24)

The direct calculation of the coupling constant as the energy difference between the singlet and triplet states involves therefore at least one wavefunction, S,0 that cannot be expressed as a single configuration. Noodleman’s suggestion was to use instead a broken-symmetry solution: B S = |ϕa αϕb β| (or its degenerate counterpart B S = |ϕa βϕb α|)

(25)

7.4 Quantitative Evaluation of Exchange Coupling Constants

243

which has M S = 0 but is a state of mixed spin because it can be expressed as a combination of S,0 and T,0 : 1 B S = √ [ S,0 + T,0 ] 2

(26)

and the energy of which is given by: EBS =

1 (E S + E T ) 2

(27)

from which the following expression for the coupling constant can be deduced: J = 2(E B S − E T )

(28)

For the general case, the overlap Sab between non-orthogonal orbitals must be taken into account, and Eq. (28) rewritten as: J=

2(E B S − E T ) 2 1 + Sab

(29)

For small values of Sab (as in the case of exchange coupled dinuclear compounds) this expression reduces to Eq. (28). The use of the broken-symmetry approach in combination with the UHF method yields values for J that are in qualitative agreement with experimental data. For example, the calculated value for J in copper acetate is −54 cm−1 , indicating correctly that the two electrons are antiferromagnetically coupled. The value of J , though, is severely underestimated by this approach. Noodleman’s method can be applied also within the density functional formalism. The J values obtained for copper acetate using different functionals are summarized in Table 3 (values calculated using Eq. (28) are those given in parenthesis). All functionals tested predict an antiferromagnetic coupling in good agreement with experimental observations, even if the value of the coupling constant is strongly overestimated. The local density approximation (SVWN functional) [15, 88] gives the worst results, with a magnitude of J which is almost 10 times larger than the experimental value. Gradient corrected functionals [89, 90] and hybrid functionals strongly reduce the calculated value of J , which is still strongly overestimated. Even the best results obtained with the B3LYP functional [91], yield calculated coupling constants about twice the experimental value. This observation, which is general for all systems studied so far, has led us to propose the use of a modified broken-symmetry approach in which Eq. (28) is simply replaced by: J = E B S − ET

(30)

The use of this equation implies that, when using density functional theory to evaluate the energy of the states involved in the magnetic behavior, the energy of the

244

7 Electronic Structure and Magnetic Behavior

Fig. 2. Experimental exchange coupling constants for different families of binuclear compounds with different bridges, represented as a function of the calculated value with the B3LYP-bs method using the complete, unmodeled structure for each compound.

singlet state E S can be effectively estimated from the energy of the broken-symmetry solution E B S [92]. The calculated value J in copper acetate are substantially better when Eq. (30) is used (values without parenthesis in Table 3) in combination with density functional calculations. Our experience in this field is that the combination of Eq. (30) with the B3LYP functional gives values for the coupling constant which are in excellent agreement with the experimental data for all the compounds studied so far (Fig. 2). The use of Eq. (30) instead of Eq. (28) has lead to some controversy in the recent literature [92–97]. For wavefunction-based methods, such as UHF, it is clear that the spin-projection procedure that leads to Eq. (28) is indeed the right way to tackle the problem. When dealing with density functional calculations the playground is however somewhat different. In density functional theory, the Kohn–Sham wavefunction is only a tool used to obtain the ground state electron density from which the energy is calculated. The use of spin-projection techniques applied to the wavefunction constructed from the Kohn–Sham orbitals has been questioned recently [9]. Wittbrodt and Schlegel [98] have discussed the influence of spin projection on potential energy surfaces finding that these certainly improve UHF and UMP2 results whereas the best results from DFT methods are obtained with the energy values of the brokensymmetry state without projection. In this context it is also interesting to point to the work of Perdew et al. [99, 100] in which it is observed that the broken-symmetry function describes the electron density and the on-top electron pair density with remarkable accuracy even if it gives an unrealistic spin density distribution. These authors conclude that the broken-symmetry function is indeed the correct singledeterminant solution of the Kohn–Sham equations for these systems. For another interesting work related to the adequacy of using broken-symmetry solutions to estimate the energy of the singlet state in organic biradicals, the reader is referred to a recent article by Grafenstein ¨ et al. [101]. The use of broken-symmetry wavefunctions (either with or without the use of spin-projection) can be easily extended to dinuclear compounds with more than one unpaired electron per center [92]. For a compound with the same spin on the two centers the following equation can be used to estimate J in conjunction with DFT

7.4 Quantitative Evaluation of Exchange Coupling Constants

245

calculations using the B3LYP method: J=

2(E B S − E H S ) S H S (S H S + 1)

(31)

where HS stands for the high-spin configuration. A similar expression applies when spin-projection is considered: J=

2(E B S − E H S ) 2 SH S

(32)

Several authors have applied the broken-symmetry approach, introducing in some cases particularities for the estimation of the energy differences between ground and excited states. We can mention as one of the first works in this field the paper of Fukutome in 1981 [102]. Hart and Rappe have applied the Hartree–Fock brokensymmetry approach to the study of oxo-bridged Ti(III) and Fe(III) complexes [103, 104]. A particular feature of these studies is the use of the ROHF method instead the usual UHF one for the calculation of the high-spin state. Ovchinnikov et al. [105] have also explored the applicability of the broken-symmetry method to the estimation of the splitting energies for simple diatomic molecules as previously done by Goursot and coworkers [106]. It is worth to note that there are significant differences between the symmetry of the SOMOs in simple diatomic molecules and in the transition-metal complexes discussed in this review. In a molecule such as O2 , its two orthogonal π ∗ SOMOs cannot be represented as a pair of localized orbitals as in the case of exchange coupled paramagnetic centers. For this type of situations DFT calculations with spin projection techniques seem to yield good results for the calculation of the singlet–triplet splitting [105]. Daul has proposed an alternative to the broken-symmetry method within the density functional theory, providing an expression to calculate the energy difference from single-determinant energies [107]. A similar approach has been developed by Filatov and Shaik, called spinrestricted ensemble-referenced Kohn-Sham method (REKS). The electron densities and energies for the selected states are represented as weighted sums of energies and densities of symmetry-adapted Kohn–Sham determinants [108–110]. An alternative procedure using the scheme based on the Lowdin ¨ annihilator has been used by Cory and Zerner to project the unrestricted Hartree–Fock wavefunction using an INDO model Hamiltonian in their study of ferredoxin models [111]. Finally, we must mention the work of Yamaguchi and coworkers who have also been using broken-symmetry functions in the context of unrestricted Hartree–Fock and MP2 calculations [112]. These authors use also spin-projection techniques to extract J through the expression: J=

2(E B S − E H S ) S 2 H S − S 2 B S

(33)

where S 2 = S(S + 1) is the expectation value for S 2 in each configuration. This expression is similar to that proposed earlier by Ginsberg [113].

246

7 Electronic Structure and Magnetic Behavior

Given the simplicity of the broken-symmetry approach compared to other methods described above, several groups have employed such methodology to study the exchange coupling and the electronic structure in polynuclear transition metal complexes. Noodleman and coworkers have carried out extensive work on complexes with biological interest, mainly in iron–sulfur clusters [85, 86, 114–127] and oxobridged manganese complexes [128–132]. They have employed local or gradient corrected density functionals that provide usually a correct qualitative energy order of the spin states but non-accurate quantitative estimations of exchange coupling constant values. It is also worth mentioning the contribution of Solomon and coworkers in the field of bioinorganic chemistry. Their work in this field is more focused on the analysis of the electronic structure and its relation with a variety of spectroscopic data rather than on the magnetic behavior. As representative examples we can mention their work on Cu(II) model complexes of active centers in blue copper proteins [133– 140] and hemocyanins [141–147], model iron complexes of hemerythrins [148–153], and manganese model complexes of catalases [154, 155]. Yamaguchi and coworkers have performed an extensive study of the magnetic properties of molecular systems, focusing on both transition-metal complexes and organic radicals [156]. They employed several methods, from Hartree–Fock and post-Hartree–Fock to those based on density functional theory. Among the systems with transition metals studied by these authors we can mention the [M2 Cl8 ]2− (M = Cr and Mo) complexes containing metal-metal multiple bonds [157–159], transition metal tetrathiolates [160, 161], iron-sulfur clusters [162–164], Prussian blue analogs [165–167], tetranuclear linear Ni–Cr–Cr–Ni complexes [168, 169], fluorobridged Cu(II), Ni(II) and Mn(II) dinuclear complexes [170], copper clusters [171] and also metalloporphyrins [172]. They have also developed a procedure to calculate the magnetization, combining path-integral and Monte Carlo methods [173–175], applying it to complex systems such as Mn12 compounds [176, 177] or the Fe12 ferric wheel [178]. Bencini and coworkers started in the eighties to apply the Xα method and the broken-symmetry approach [179] to the study of several Cu(II) dinuclear compounds with hydroxo [180], halo [181-183], oxalato-type [184], or carbonato bridges [185], and more recently azido-bridged Cu(II) complexes [186] using new functionals. These authors have also explored the electronic structure of some Fe and Co hexanuclear compounds in search for explanations of spectroscopic data reported for these systems [187–190]. They have also analyzed the influence of substitutions in dioxobenzene-bridged Mo(V) complexes on the exchange coupling constant [191]. Recently, they have extended the use of the broken-symmetry approach to mixed valence compounds by studying Fe(II)Fe(III) complexes [192], oxo-bridged Mn(III)Mn(IV) complexes [193] and the Creutz–Taube ion [194], that has also been studied by Chen et al. [195]. Caneschi et al. have deduced the structural dependence of the exchange coupling constant for hydroxo- and alkoxo-bridged complexes [196] using DFT methods. Density functional calculations were performed by Belanzoni et al. to investigate the electronic structure and exchange coupling in organometallic Fe(II) dinuclear complexes with metal-metal interactions and different types of bridging ligands [197]. McGrady and Stranger have applied DFT based calculations to perform exhaustive studies of the d3 –d3 [M2 Cl9 ]3− and [M2 Cl10 ]4− systems (M =

7.4 Quantitative Evaluation of Exchange Coupling Constants

247

Cr(III), Mo(III) and W(III)) [198–207] and oxo-bridged Mn dinuclear compounds [208–210]. The magneto-structural correlation in dinuclear alkoxo-bridged V(IV) complexes has been explored by Plass by using the broken-symmetry approach combined with a non-local functional where a self-interaction correction has been included [211]. Blanchet-Boiteux and Mouesca have analyzed the exchange coupling in end-on azido-bridged Cu(II) dinuclear complexes and oxo-bridged Cu(II) dimers using DFT methods. Their work is centered on the relation between the exchange coupling constant and the overlap of the magnetic orbitals or the atomic spin populations [212–214]. Recently, Boca et al. have employed the Hartree–Fock method to study the electronic structure in trinuclear cobalt complexes [215]. We have studied the exchange coupling and magneto-structural correlations in several transition metal dinuclear complexes using the broken-symmetry approach and the B3LYP hybrid functional [92]. Considering a classification according to the bridging ligands, we can mention hydroxo- and alkoxo Cu(II) dinuclear complexes [216, 217], end-on azido-bridged Cu(II), Ni(II) and Mn(II) complexes [218], end-to-end azido-bridged Cu(II) and Ni(II) compounds [219], Cu(II) complexes with oxalato-type bridging ligands [220, 221], carboxylato-bridged Cu(II) dinuclear complexes [222], Cu(II) and Ni(II) cyano-bridged binuclear complexes, chlorobridged Cu(II) dimers and bis(oximato)-bridged heterobimetallic compounds [223]. Using the same approach, Castro et al. have studied the exchange coupling in dithiosquarate-bridged Cu(II) complexes [224] and the problem of orbital countercomplementarity in mixed µ-acetato and µ-hydroxo Cu(II) trimers has been explored by Gutierrez et al. [225]. In the next section we will provide a more detailed description of our work in this field. We would like to remark here that the broken-symmetry DFT approach is able to provide quantitative estimates of J that are of a similar quality to those obtained from highly sophisticated multireference ab initio calculations. The power of the brokensymmetry approach resides in its computational simplicity, which can be applied to compounds with more than 100 atoms using relatively modest computational resources and standard quantum chemistry software packages, easily available to a broad community of researchers interested in this field. The broken-symmetry method is also highly appealing because it is relatively easy to adapt for the calculation of coupling constants in higher nuclearity compounds or even to solid state materials with periodically repeating unit cells containing an arbitrary number of paramagnetic centers. In the latter case, construction of the broken-symmetry solution often requires the use of supercells to deal with the symmetry lowering with respect to the original space group of the crystal. A few considerations are in order to properly extend the ideas developed so far to the case of periodic systems, for which coupling constants are typically obtained via a procedure involving two main steps. First, total energies corresponding to distinct values of the total spin are obtained and compared, special emphasis being put on the consistency with experimental results and on the relative stability of the magnetic states under consideration. It should be remarked that, in addition to the ferromagnetic configuration, several spin configurations corresponding to the same total spin (as in the antiferromagnetic or ferrimagnetic situations) need to be taken into account. The second step consists in mapping to appropriate spin Hamiltonians the collection of

248

7 Electronic Structure and Magnetic Behavior

total energy data resulting from the spin configurations thereby obtained. So far most attempts in this direction have been based on the Unrestricted Hartree–Fock approach, implemented within the CRYSTAL code [226], refined in some cases by including the contribution of electron correlation via an a posteriori scheme based on correlation-only density functional formulas. This recipe has allowed the successful investigation of the electronic and magnetic properties of a large amount of transition metal oxides and halides [227–240]. From a methodological point of view, and focusing on the extraction of the superexchange coupling constants from a parametrization of total energy data, the case of the non-cubic Mn3 O4 spinel is particularly instructive. Apart from the ferromagnetic structure, six ferrimagnetic configurations were considered [241]. This gives rise to a system of equations relating the ferromagnetic–ferrimagnetic energy differences to four exchange coupling constants via a spin Hamiltonian which includes firstand second-neighbor interactions. The calculated J parameters were found to be about 40–50% of those derived from experimental data, mostly due to the neglect of electronic correlation in the UHF method. Nevertheless, the sign and the strength of the magnetic interactions turned out to be in agreement with experimental evidence. Solid-state compounds exhibiting interesting and mostly unexplored correlations between structural and magnetic properties may encompass systems obtained by the combined synthesis of inorganic layered networks and molecular species, as in the organic-inorganic hybrid multilayer materials [242]. In these cases, theoretical models may be used to predict structures which are not experimentally available, while ensuring a reliable description of the electronic structure. The possibility of simultaneously optimizing structural parameters and the electronic structure in the search for a configurational ground state accurately described at the density-functional level relies on the first-principle molecular dynamics technique proposed in 1985 by R. Car and M. Parrinello [243]. Within this scheme, the most important idea to be retained is that the total energy can be taken as a function of all the wavefunction coefficients of the occupied states and the positions of the atoms. Given this assumption, one is faced to the global minimization problem of finding the electronic ground state for the relaxed atomic configuration. The novelty of the approach rests on the fact that when we seek the relaxed ground state, the positions and the coefficients of the wavefunctions are varied at the same time, thereby ensuring that at the global minimum of energy the self-consistent solution will be automatically obtained. The application of this method to polynuclear transition metal compounds relevant in the area of molecular magnetism is still at its infancy [244]. However, its flexibility and the extended record of reliability achieved for other systems makes its application to molecular based magnetic materials highly promising.

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

7.5

249

Exchange Coupling in Polynuclear Transition-metal Complexes

After this short review of the computational techniques available for the study of exchange coupled systems, we discuss now in more detail the results of our own research applying one of these methods, the broken-symmetry approach in combination with B3LYP calculations, to the study of different aspects related to the magnetic behavior of magnetically coupled compounds containing transition metal atoms. We start our discussion with the study of the simplest case, i. e. dinuclear Cu(II) compounds, with only two exchange coupled electrons, and increase gradually the complexity of the system considering homodinuclear compounds with more than one electron per paramagnetic center, heterodinuclear compounds and polynuclear clusters.

7.5.1 7.5.1.1

Homodinuclear Compounds Hydroxo- and Alkoxo-bridged Cu(II) Compounds

One of the most extensively studied families of exchange coupled dinuclear compounds is that of the hydroxo- and alkoxo-bridged Cu(II) complexes. Although they provide examples of the simplest case of magnetic interaction, involving only two unpaired electrons, their magnetic behavior is quite varied, exhibiting ferro- or antiferromagnetic character depending on their molecular geometry. Hatfield and Hodgson [245] found a linear correlation between the experimentally determined exchange coupling constant and the Cu–O–Cu bond angle θ (see 5 for the definition of the most relevant geometrical parameters in these compounds). Antiferromagnetic coupling is found for complexes with θ larger than a critical value, θc = 98◦ , whereas ferromagnetism appears at smaller angles. A plot of the calculated J as a function of θ for the model compound yields also a linear correlation (dotted line in Fig. 3). The calculated critical angle θc is, however, slightly smaller (92◦ ) and antiferromagnetic coupling is predicted for the range comprising all experimental angles (96–105◦ ). A more detailed analysis of the structural data shows that a second parameter, the out-of-plane displacement of the hydrogen atom of the hydroxo group (measured by τ as indicated in 5), also affects the value of the exchange coupling constant. The computational results show indeed that the two structural parameters are correlated: if the value of θ is optimized keeping τ fixed, one finds that the

250

7 Electronic Structure and Magnetic Behavior

Fig. 3. Magnetic coupling constants, J , for Cu(II) hydroxo-bridged complexes calculated with the B3LYP-bs method using a double zeta basis set for model 5 as a function of bridging angle θ. The black circles correspond to the experimental values, the dashed line gives the calculated values (triangles) for the planar model (τ = 0◦ ). The squares represent the values calculated for the optimized θ at a fixed value of τ (circled numbers). The uncircled labels indicate the value of τ for each experimental structure [216].

out-of-plane displacement of the hydrogen atoms favors smaller values of θ (Fig. 3, squares, with the fixed τ values given in circles). The most interesting result is that in a plot of J as a function of θ for each (θ, τ ) pair (Fig. 3, solid line) a linear correlation emerges that predicts ferromagnetic coupling for large values of τ (and, hence, small values of θ). This trend is in excellent agreement with the experimental data (Fig. 3, black circles with the corresponding τ values given besides). A similar trend is also found for the related alkoxo-bridged complexes, with the main difference that in this case all compounds are predicted to be antiferromagnetic for the range of (θ, τ ) values experimentally found, although ferromagnetic behavior cannot be ruled out for the so far unknown compounds with τ values larger than 50◦ . The effect of different counterions on the exchange coupling constant can thus be rationalized considering that the position of the hydrogen atoms of the hydroxo bridges is dictated by hydrogen bonding with the counterions. The assumption of the HTH model that small changes in structural parameters affect only the one-electron contribution to the coupling constant can be checked by plotting the value of J against (ε1 − ε2 )2 when θ is varied for a fixed value of τ = 0◦ (Fig. 4). The orbital energies ε1 and ε2 in this plot are those of the Kohn– Sham SOMOs in the triplet state for the model compound shown in 5. The linear dependence expected from Eq. (11) is found for both the hydroxo- and the alkoxobridged (not shown in the figure) compounds. A similar behavior is found for the

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

251

Fig. 4. Singlet–triplet energy separation, J , calculated for the hydroxo-bridged model compound 5 with τ = 0◦ (B3LYPbs method) as a function of the square of the energy gap between the two SOMOs in the triplet state [216]. Table 4. Exchange coupling constants J (cm−1 ) calculated for two model compounds with different terminal ligands. The experimental J and pKb values are also included for comparison [217]. L

Calc. J

Expt. J

pKb

[CuL(µ-OH)2 CuL](NO3 )2 bipy NH3 en

+107 +81 +60

+172

8.77 4.75 4.07

[CuL(µ-OH)2 CuL]Br2 NH3 en tmeen

−354 −402 −502

−509

4.75 4.07 3.29

out-of-plane hydrogen shift at a fixed θ value. These calculations suggest, thus, that there is a sound theoretical basis for the analysis of magneto-structural correlations based on the study of changes in the HOMO–LUMO gap with geometry. Another interesting point that has been analyzed for this family of compounds is the dependence of J on the nature of the terminal ligands. The donor atoms in these ligands are usually nitrogen atoms, belonging sometimes to aromatic, sometimes to aliphatic N-donors. The conclusions regarding a series of calculations for model compounds with different terminal ligands (Table 4) shows that the strength of the antiferromagnetic coupling follows the same trend as the basicity of these ligands: tmeen > en > NH3 > bipy. Considering the J values experimentally found for this family of complexes, a sharp contrast is found between the large number of known antiferromagnetic complexes and the paucity of ferromagnetic systems. A possible strategy for the design of new ferromagnetic compounds is to search for the factors that lead to practically degenerate SOMOs, a search for which theoretical studies may be of great help. One possibility consists in introducing modifications in the bridging ligand. For this purpose, the hydroxo- and alkoxo-bridged Cu(II) compounds offer an excellent op-

252

7 Electronic Structure and Magnetic Behavior

portunity since changes in the nature of the X group bonded to the bridging oxygen atom may influence the magnetic behavior, as evidenced by the fact that alkoxobridged complexes give stronger antiferromagnetic coupling than hydroxo-bridged ones with similar composition and structure [246]. Table 5 shows the calculated exchange coupling constants for a variety of OXbridged dinuclear Cu(II) compounds with ammonia as terminal ligands. From the analysis of these values some conclusions can be drawn: (a) all the alkoxo- and phenoxo-bridged complexes show stronger antiferromagnetic coupling than the hydroxo-bridged ones, in good agreement with the experimental results; (b) substitution of the hydrogen atoms in methoxo-bridged compounds by alkyl groups weakens the antiferromagnetic coupling; (c) in general, the larger the electronegativity of the substituent, the stronger the antiferromagnetic coupling is, although there are a few exceptions to this rule. A particularly surprising result is the exceptionally strong ferromagnetic coupling constant predicted for the oxo-bridged complex of copper(II), as compared to the largest values reported in the literature for azido- and hydroxo-bridged complexes (+170 and +172 cm−1 ). A more realistic model for the oxo-bridged complex, with pentacoordinate Cu(II) atoms having an extra NH3 molecule in the fifth coordination position is also predicted to show a very strong ferromagnetic coupling, with a J value of +685 cm−1 calculated for the optimized structure. Table 5. Exchange coupling constants J (cm−1 ) calculated for [(NH3 )2 Cu(µ-OX)2 Cu(NH3 )2 ]n+ (n = 0 for OX = O, OSO3 , OBR3 and OAlR3 ; n = 4 for OX = OPy; n = 2 for all other OX). The same geometry with θ = 101◦ and τ = 0.0◦ has been taken for all model compounds. The range of experimental values of J , when known, is also included for comparison [246]. X F Me BH3 Et But Ph py H GeH3 SiH3 Ge(OH) 3 COMe NO2 Si(OH) 3 Al(OH) 3 SO3 SOMe Li –

Calc. J −1855 −778 −687 −669 −617 −587 −675 −493 −331 −278 −259 −230 −221 −202 −134 −108 +8 +100 +989

Expt. J

−1064/−65 −852/−166 −855/−242 −509/+172

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

7.5.1.2

253

Cu(II) Compounds with Oxalato, Oxamidato or Related Polyatomic Bridging Ligands

One of the most striking observations in intramolecular magnetism is the ability of some polyatomic bridging ligands to provide a pathway for strong exchange coupling between two paramagnetic centers that are far apart. In this regard, one of the most studied ligands is the oxalate dianion and a series of other groups that can be formally derived from it by replacing the oxygen atoms by either S or NR. Given the variety of structures found for oxalato bridged Cu(II) complexes, this family allows us to analyze the influence of the coordination environment around the metal atoms on the exchange coupling [220, 221]. In recent years, several complexes of general formula [(AA)Cu(µC2 O4 )Cu(AA)]Xn have been structurally and magnetically characterized. Owing to the Jahn–Teller plasticity of the coordination sphere around Cu(II), this ion may appear as four-coordinate, four-coordinate with a fifth ligand at a larger distance (4 + 1 coordination), or four-coordinate with two weakly bound ligands (4 + 2 coordination). In all three cases, the four short ligand–metal bonds can be found approximately in a common plane giving square-planar, square-pyramidal, or square-bipyramidal coordination environments, respectively. In all cases, the unpaired electron is located in a dx 2 −y 2 -type orbital pointing to the four atoms with short metal–ligand distances. Two different orientations of the basal plane with respect to the plane of the bridging oxalate have been identified experimentally, depending on whether they are coplanar or perpendicular. This gives rise to the three different relative dispositions of the two SOMOs indicated in 6–8. Alternatively, five-coordinate Cu(II) ions may present a trigonal bipyramidal environment. In this case, the unpaired electron is located in a dz 2 -type orbital along the pseudotrigonal axis (9). The calculated J for model compounds with these four topologies are given in Table 6. The coplanar disposition of the SOMOs provides the most effective exchange pathway for antiferromagnetic coupling, in good agreement with the available experimental data. On the other hand, the parallel disposition of the SOMOs leads to weak coupling, which can be either ferro- or antiferromagnetic depending on the nature of the terminal ligands and on the detailed geometrical features of the coordination environment around the copper atoms. For the other two topologies, moderate Table 6. Exchange coupling constants J (cm−1 ) calculated for [(NH3 )2 Cu(µ-C2 O4 ) Cu(NH3 )2 ]2+ with different topologies of the two SOMOs. The column labeled Jest corresponds to the empirical estimation of J from Eq. (34), taking the average experimental value (−370 cm−1 ) for the coplanar case as a reference value [220]. Orbital topology Coplanar (6) Perpendicular (7) Parallel (8) Trigonal-bipyramidal (9) a

See text

Jcalc −293 −86 +10 −185

Jest −370 −93 0 −165

Jexp −300 to −400 −75 +1.2 to −37 −75a

254

7 Electronic Structure and Magnetic Behavior

antiferromagnetic coupling is expected. The influence of the orbital topology on the coupling constant has also been analyzed by Julve et al. [247] using empirical rules based solely on the overlap between natural magnetic orbitals (Eq. 11). Taking Ja , the coupling constant for coplanar compounds, as a reference value, these authors suggest the following relationships between the J values in the different compounds (the subscripts correspond to the different orbital topologies in 6–9): Jb = Ja /4 Jc = 0 Jd = 4Ja /9

(34)

The values obtained using these relationships (Table 6) are in a fair agreement with both our calculated data and the experimentally determined coupling constants, showing the excellent predictive power of these empirical rules. The disagreement between the calculated coupling constant and the experimental data for the trigonal-bipyramidal topology can be attributed to the departure of the experimental coordination environment from an ideal trigonal bipyramid. A calculation for the unmodeled structure of this compound yields a coupling constant of −82 cm−1 , in excellent agreement with the experimental value. Chemical substitution at the bridging ligand is expected to have an important influence on the exchange coupling and has therefore been analyzed by performing a series of calculations for model compounds with the general formula [(NH3 )2 Cu(µC2 WXYZ)Cu(NH3 )2 ]2+ where W, X, Y, Z = NH, O, or S. The calculated exchange

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

255

Table 7. Exchange coupling constants J (cm−1 ) calculated for [(NH3 )2 Cu(µC2 WXYZ)Cu(NH3 )2 ]2+ with different bridging ligands and ranges of experimental values for complexes with the analogous bridges [221]. Bridge

W

X

Y

Z

−Jcalc

−Jexp

Oxalate Oxamate cis-Oxamidate

O NR NR NR NR NR S S S NR NR NR S NR NR

O O NR O O NR S O O NR S S S NR NR

O O O NR O NR O S O S NR S S NR NR

O O O O NR NR O O S S S NR S NR NR

293 312 360 347 356 358 485 465 391 504 553 473 829 98 44

284-402 400-425 242–453

trans-Oxamidate Ethylenetetraamidate cis-Dithiooxalate trans-Dithiooxalate cis-Dithiooxamidate trans-Dithiooxamidate Tetrathiooxalate Bipyrimidine Bisimidazole

305–591

523–730 >800 139–236

coupling constants, Table 7 and Fig. 5, show that antiferromagnetic coupling is predicted for all the combinations of W, X, Y, and Z. The magnitude of such coupling is strongly affected by the nature of the bridging ligand, ranging from J = −829 cm−1 for tetrathiooxalate to J = −44 cm−1 for bisimidazole. These calculations confirm the qualitative trend established by Verdaguer et al. using the HTH model [248], that progressive substitution of oxygen by less electronegative donor atoms such as nitrogen or sulfur results in increasingly stronger antiferromagnetic coupling. The exception to this rule comes from the aromatic bridging ligands bipyrimidine and bisimidazole, which show a much weaker coupling than the analogous non-aromatic ligand, ethylenetetraamidate. This fact can be explained by the delocalization of the lone pair orbitals throughout the aromatic system that results in a poorer overlap with the metal d orbitals.

Fig. 5. Ranges of experimental exchange coupling constants for different families of binuclear Cu(II) compounds with bridges of oxalato-type, represented as a function of the calculated value for the corresponding model compound (Table 7) [221].

256

7 Electronic Structure and Magnetic Behavior

Fig. 6. Dependence of the calculated J value on the electronegativity of the terminal ligands X in compounds of type 10 [220].

In contrast with the effect of substitution of the donor atoms, which accounts for changes in the coupling constant of up to 600 cm−1 , different substitution patterns for the same set of donor atoms affects the values of J by less than 90 cm−1 . Although a large part of the variation in the values of J can be associated with the electronegativity of the donor atoms, the wide range of values experimentally found for the same bridging ligands clearly indicates that other factors, such as structural distortions or changes in the nature of the terminal ligands affect the exchange coupling to a significant extent. Once the main factors concerning the role of the bridge are understood, we analyze the effect of the nature of the terminal ligands on the magnitude of the exchange coupling when the orbital topology is kept constant. For this purpose, calculations for a series of compounds in which only a terminal ligand is changed (10 with X = F, Cl, Br or I) were performed. The results (Fig. 6) indicate that, other things being equal, the less electronegative donors induce a stronger antiferromagnetic coupling. This is due to a greater hybridization of the dx 2 −y 2 -type SOMOs towards the bridge induced by better σ -donor terminal ligands, as predicted previously using the qualitative HTH model [30].

7.5.1.3

Carboxylato-bridged Cu(II) Compounds

Copper(II) carboxylates form a large family with many structurally characterized compounds for which magnetic properties have been measured. This wealth of information permits the detailed study of the influence of different factors on the exchange interaction between the two unpaired electrons. The different coordination modes of the carboxylato group (11–14) together with the choice of the bridging ligand substituent R, the terminal ligand L, and the number of bridging ligands, give rise to a large number of possibilities to obtain new compounds with tailored magnetic properties.

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

257

Although in most carboxylato-bridged dinuclear Cu(II) complexes the bridging ligands appear coordinated in a syn–syn fashion (11) [249, 250], some compounds with other bridging modes (12–14) have been prepared and their magnetic properties measured. In order to compare the exchange coupling mediated by a formiato bridge in syn–syn (11), syn–anti (12) [251–256], and anti–anti (13) [257, 258] coordination modes, the corresponding models were used, keeping the geometry of the formiato bridge and the copper sphere fixed. Water molecules were included in all cases as terminal ligands. The Cu · · · Cu distances in these model structures are 2.82, 5.15, and 5.77 Å for the syn–syn, syn–anti, and anti–anti compounds, respectively. The calculated coupling constants range from weakly ferromagnetic (+10.2 cm−1 ) for the syn-anti case to moderately antiferromagnetic (−61.3 cm−1 ) in the anti–anti case, in good agreement with the experimentally available data [257]. The known compounds with an anti–anti coordination mode have coupling constants of around −50 cm−1 , while that with a syn–anti coordination presents a weak ferromagnetic behavior (J = +14 cm−1 ) [257]. For the model with syn–syn coordination an intermediate behavior (J = −11.3 cm−1 ) was predicted. Changing the nature of R, the group bound to the carboxylato bridge, in compounds with four carboxylato bridges has a dramatic effect on the coupling constant. Calculated and experimental coupling constants for several compounds of this family are presented in Table 8 (experimental data are average values for compounds with slightly different geometries and/or terminal ligands) [51, 249, 250, 259, 260]. Exchange coupling between two paramagnetic centers that are not directly bound is frequently rationalized by adding the contributions of the different superexchange pathways mediated by the bridging ligands. Since dimers with four, three, two, and one carboxylato bridges have been structurally and magnetically characterized, this family of compounds offers an excellent opportunity to test the validity of such an approach. To this end, exchange coupling constants have been calculated for model compounds in which a varying number of carboxylato bridges have been Table 8. Calculated coupling constants (cm−1 ) for [L2 Cu2 (µ-RCOO)4 ]. [222]

a

L

R = -SiH3

−H

−CH3

−CF3

−CCl3

H2 O NH3 Exptl

−806 −749 −1000a

−417 −393 −550b

−299 −284 −300c

−254 −241 −300d

−158 −142 −200e

Ref. [259]; b Ref. [260]; c Ref. [51]; d Ref. [249]; e Ref. [250].

258

7 Electronic Structure and Magnetic Behavior

Fig. 7. Exchange coupling constants calculated for [(H2 O)5−n b Cu(µ-RCOO)n b Cu +

(H2 O)5−n b ](4−n b ) as a function of the number of bridging carboxylato ligands, n b [222].

replaced by two water molecules each. Fig. 7 shows that, regardless of the nature of the carboxylato bridge, the coupling constant exhibits a linear dependence on the number of bridges. The additive nature of the contribution of each bridge to J is also valid for hypothetical compounds with mixed bridges according to calculations. A compound with two formiato bridges and two trichloroacetato bridges, for example, should present a coupling constant of −301 cm−1 , which is in excellent agreement with the calculated values of −300 and −293 cm−1 for two model compounds in which the two formiato bridges are coordinated in a cis or trans disposition, respectively. Related questions, like the influence of the terminal ligands L on the coupling constant or the establishment of some magneto-structural correlations for these systems have also been tackled using the same computational methodology [222].

7.5.1.4

Azido-bridged Compounds

The azide ion is a versatile ligand that can bind to transition-metal atoms with different coordination modes, thus allowing for the assembly of dinuclear complexes with a wide range of magnetic behavior. When the N− 3 group acts as a bridging ligand with an end-on coordination mode (15), the resulting dinuclear complexes usually show ferromagnetic behavior. In contrast, when coordinated in an end-to-end fashion (16), antiferromagnetic coupling results. Although both situations have been computationally analyzed using the broken-symmetry approach, we will focus our discussion only on the ferromagnetically coupled dimers with azido bridges coordinated in an end-on fashion [218]. Details on magneto-structural correlations for compounds with N− 3 ligands with end-to-end coordination can be found in Ref. [219]. A large number of Cu(II) dinuclear complexes of such family have been reported, including end-on double-bridged compounds, all with ferromagnetic coupling. This behavior was formerly attributed by Kahn et al. to a spin polarization mechanism [261], a suggestion that has been ruled out more recently by the same authors in view of polarized neutron diffraction experiments [262] that indicate that the spin density at the bridging nitrogen atom has the same sign as that at the metal atoms. Calculation

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

259

Fig. 8. Magnetic coupling constants for a [Cu2 (µ-N3 )2 (C2 N2 H4 )2 ]2+ model complex as a function of the bridging angle θ with τ = 0◦ (empty circles) and 15◦ (empty squares). The black circles correspond to known experimental values [218].

of the coupling constant for the model compound [Cu2 (µ-N3 )2 (C2 N2 H4 )2 ]2+ with different Cu–N–Cu angles (Fig. 8) shows that the ferromagnetic behavior is only associated to the geometrical constraints that affect this angle in the experimentally known compounds. Antiferromagnetically coupled compounds can be expected in principle if one is able to force the Cu–N–Cu angle to adopt values larger than 104◦ , a prediction that had also been advanced by Thompson et al. based on the extrapolation of the experimental magneto-structural correlation [263]. The calculated spin densities at various atoms in [Cu2 (µ-N3 )2 (4-t Bu-py)4 ](ClO4 )2 (Table 9) are in good qualitative agreement with the experimental polarized neutron diffraction data [262]. It is interesting to note that the spin densities at the bridging (N1) and terminal (Nt ) nitrogen atoms have the same sign as in the copper atoms, indicating that the unpaired electron delocalization toward the donor atoms predominates over the spin polarization mechanism for this compound. There are some examples in the literature of end-on azido-bridged complexes with transition metals other than copper, the most common ones being those of Ni(II). From calculations [218] on a model compound (Fig. 9), J is seen to vary with the M–N–M angle in a different way than for the copper(II) model compound discussed above. The interaction is predicted to be ferromagnetic for all the range of angles

260

7 Electronic Structure and Magnetic Behavior

Table 9. Atomic spin densities obtained from experimental polarized neutron diffraction data [262] and calculated for [Cu2 (µ-N3 )2 (4-t Bupy)4 ](ClO4 )2 with the B3LYP functional using a Mulliken population analysis [218]. Atom

Experimental

Calculated

Cu N1 N2 N3 Nt

+0.78 +0.07 −0.02 +0.06 +0.07, +0.05

+0.60 +0.14 −0.04 +0.12 +0.09, +0.09

Fig. 9. Exchange coupling constant for the [Ni2 (µ-N3 )2 (NH3 )8 ]2+ model 15 as a function of the bridging angle θ (empty circles). The black circles correspond to known experimental values [218].

explored, with J increasing with the M–N–M angle, yielding a maximum at 104◦ . In this case, the strongest ferromagnetic coupling coincides with the most stable geometry and with the structures of the experimentally known compounds. For all of them a quintet ground state has been deduced from magnetic measurements, in good agreement with the calculations. A few compounds with transition metals other than copper or nickel have been also magnetically characterized. These compounds, with Co(II), Fe(III) or Mn(II) ions, show all ferromagnetic coupling. Calculations indicate that J increases with the M–N–M angle for a Mn(II) model, in a way analogous to that found for the nickel compound, although the maximum is now predicted at a larger M–N–M angle. Fig. 10 shows the calculated coupling constants as a function of the M–N–M angle for the three metals. A similar parabolic behavior is found in all cases, the main difference being the position of the maximum, which is found at 84◦ for Cu(II), 104◦ for Ni(II), and 114◦ for Mn(II). The range of low energy angles (outlined parts of the curves) is different for each metal, thus explaining why J decreases with the M–N–M angle for Cu(II), increases for Mn(II), and is practically constant for Ni(II).

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

261

Fig. 10. Total exchange constants (n 2 J , where n is the number of unpaired electrons per metal atom) for complexes of Cu(II), Ni(II) and Mn(II) with two end-on azido bridges as a function of the bridging angle θ. The outlined parts of the curves correspond to those geometries within 3 kcal mol−1 of the calculated minima [218].

7.5.2

Heterodinuclear Compounds

Among the dinuclear transition-metal complexes relevant to the field of molecular magnetism, those containing two different metal ions are especially interesting since it has been found that it is much easier to obtain ferromagnetic coupling in heterodimetallic compounds than in their homonuclear analogs [32]. Stabilization of high-spin states is much easier to achieve with two different paramagnetic centers because the unpaired electrons are more easily arranged in metal-centered orthogonal orbitals of such complexes. Although much progress has been made in recent years towards the accurate calculation of exchange coupling constants in homodinuclear transition-metal complexes, the field of heterodinuclear complexes remains relatively unexplored from the theoretical point of view. Only a few theoretical papers have been devoted to the quantitative evaluation of coupling constants for this type of complexes [264]. The influence of the exchange coupling on the electronic configuration has been investigated [223] for the series of bis(oximato)-bridged Cu(II)–M compounds (17) with M = Cu(II), Ni(II), Mn(II), Mn(III) or Cr(III). In all model compounds, ammonia molecules were used as terminal ligands. Due to the varying electronic configuration of the M atom, different coupling situations are expected in this family. Calculated exchange coupling constants for these model compounds (Table 10) are

262

7 Electronic Structure and Magnetic Behavior

Table 10. Exchange coupling constants (cm−1 ) for the series of model bis(oximato)-bridged Cu(II)–M compounds. Experimental values are provided for comparison [265]. M Cu(II) Ni(II) Mn(II) Mn(III) Cr(III)

Jcalc −648 −201 −67 127 43

Jexp −596 −198 −83 109 37

in excellent agreement with the experimental ones reported by Birkelbach et al. [265], showing that the broken-symmetry method combined with the B3LYP functional is also able to reproduce accurately the coupling constant for heterodinuclear compounds. As found for other compounds, the molecular structure is in this case important for the determination of the exchange coupling constant. The N–O distance at the bridge is one of the important geometrical parameters for oximato-bridged dimers in this respect. Fig. 11 shows the variation of J with such distance for the Cu(II)–Cu(II) compound. The antiferromagnetic coupling is significantly weakened when the N–O bonds are stretched, a finding that can be easily rationalized with the help of the HTH model. The in- and out-of-phase combinations of the metal x2 − y2 orbitals interact with a combination of N–O non-bonding and π ∗ orbitals of the ligand (18). When the bond is elongated, the π ∗ orbital is significantly stabilized, resulting in a poorer energy match with the metal x2 − y2 orbitals that effectively reduces the splitting of the SOMOs and, according to the HTH model, the antiferromagnetic contribution to J .

Fig. 11. Exchange coupling constants calculated for the model bis(oximato)-bridged Cu(II)–Cu(II) compound as a function of the N–O distance on the bridging ligand [223].

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

263

The influence of the terminal ligands, L, on the coupling constant or the analysis of some magneto-structural correlations for these systems have also been studied using the same computational methodology [223].

7.5.3

Polynuclear Compounds

Polynuclear transition metal compounds have attracted much attention during the last years due to their magnetic properties [266–268]. These complexes are particularly interesting because their intermediate size between the simplest binuclear complexes and bulk materials may result in completely new magnetic properties, being good candidates to behave as nanometer-sized magnetic particles. The presence of a larger number of metal atoms compared to binuclear complexes results in an increase of the complexity, since the polynuclear complexes have usually more than one J value, besides the intrinsic increase in computational resources required. Following the same strategy as for the binuclear complexes, we have chosen simple polynuclear systems in order to check the accuracy of the B3LYP method for such systems. One of the simplest and commonest cases of polynuclear complexes is that of the compounds containing a Cu4 O4 cubane core [269]. We have proposed to use the Cu · · · Cu distances as a classification criterion for the different structures adopted by these compounds [270]. We propose to divide this family of compounds in three classes: The first one contains complexes with two short and four long Cu · · · Cu distances, and we call them 2 + 4 (19). The second class presents four short and two long Cu · · · Cu distances, and will be labeled from here on as 4 + 2 (those with S4 symmetry, 20). Finally, for compounds in the third class all six Cu · · · Cu distances are similar, and they will be termed 6 + 0.

264

7 Electronic Structure and Magnetic Behavior

Two systems were selected for such a study: a complex with a 2 + 4 type structure and hydroxo bridging ligands [271] (21) and an alkoxo-bridged complex [272] that belongs to the 4 + 2 category (22).

The first molecule can be described as two binuclear complexes linked by weak interactions. In this case, due to the presence of two different exchange pathways, we have employed the following Hamiltonian to estimate the energy of the low-lying spin states, where the numbering of the paramagnetic centers and the definition of J and J are given in 23:

Hˆ = −J ( Sˆ1 Sˆ2 + Sˆ3 Sˆ4 ) − J ( Sˆ1 + Sˆ2 )( Sˆ3 + Sˆ4 )

(35)

In this model Cu1 and Cu2 are in the same binuclear unit and Cu3 and Cu4 in the other one. Using this expression, we can obtain the equations to calculate the two exchange coupling constants from the energies corresponding to three spin distributions (for more details of the procedure to calculate the J values, see Ref. [270]). The results indicate that the high spin state (S = 2) is the ground state and the two calculated coupling constants, corresponding to the intra-dimer (J ) and inter-dimer coupling (J ), are +68.0 and +0.6 cm−1 , respectively, to be compared with experimental values of +15.1 and +0.2 cm−1 . It is worth noting that this complex shows a relatively small experimental J value compared with other hydroxo-bridged binuclear complexes with similar Cu–O–Cu angles, [271] even if it shows a small roof-shape

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

265

distortion of the Cu2 O2 framework that was found to enhance the ferromagnetic behavior [217]. The second complex containing a Cu4 O4 core has alkoxo-bridging ligands and adopts a 4 + 2 structure with approximate S4 symmetry (22). This structure can be modeled by a square of copper atoms (24) with two exchange coupling constants:.

Hˆ = −J ( Sˆ1 + Sˆ2 ) · ( Sˆ3 + Sˆ4 ) − J ( Sˆ1 Sˆ2 + Sˆ3 Sˆ4 )

(36)

The calculated J and J values are +44.1 and +6.2 cm−1 , respectively. Comparison with the experimental values [272], +44.9 and −16.3 cm−1 , indicates a discrepancy in the sign of J . The analysis of the superexchange pathway for the J constant shows the presence of two long Cu · · · O distances of 2.5 Å at each side of the bridge. Equivalent dinuclear complexes with long bridging Cu · · · O distances present rather small exchange coupling constants, of ca. ±1 cm−1 [273, 274]. An alternative procedure to calculate the coupling constants for this complex is to transform the Cu4 O4 core in to a simple Cu(II) dinuclear complex by replacing two Cu(II) centers by diamagnetic Zn(II) ones. The new hypothetical systems generated in this way are equivalent to Cu(II) dinuclear complexes and, by using Eq. 30 we can obtain estimates for the two coupling constants depending on the sites occupied by the Cu(II) atoms (25). This approach gives values of J and J of +49.4 and +2.9 cm−1 , respectively, very close to those obtained from the calculation for the original Cu4 O4 core.

Let us point out that even if the two studied complexes with the Cu4 O4 core show ferromagnetic coupling between first neighbors, the origin of the ferromagnetism is different in each case. In the hydroxo-bridged complex, the nature of the hydroxo bridge combined with a Cu–O–Cu bond angle of 96.5◦ determines the ferromagnetism. For the alkoxo-bridged Cu4 O4 complex, on the other hand, the distortion

266

7 Electronic Structure and Magnetic Behavior

imposing long bridging Cu-O distances reduces the overlap between the dx 2 −y 2 orbitals of the Cu(II) atoms resulting in a weakening of the antiferromagnetic contribution compared to the undistorted alkoxo-bridged Cu(II) binuclear complexes. We have used a similar theoretical approach to study the magnetic behavior of hexanuclear Cu(II) and Ni(II) polysiloxanolato complexes [275, 276]. Our results confirm the crucial role of a guest chloride anion within the Ni(II) complex in the magnetic behavior resulting in a S = 0 ground state in contrast with the high spin state (S = 3) found for the Cu(II) complex [277]. For these two complexes, the agreement between the calculated and experimental exchange constants is really impressive. In summary, although little studies have been carried out so far for polynuclear complexes, the results found show promise that the DFT methods employing the B3LYP functional might be an excellent tool for the theoretical exploration of the magnetic behavior of complexes with high nuclearity.

7.5.4

Solid-state Compounds: The Case of Cu2 (OH)3 NO3

Given the wealth of theoretical results on hydroxo-bridged dinuclear Cu(II) molecular compounds presented in Section 5.1.1, it is worthwhile to focus on the behavior of solid copper hydroxonitrate Cu2 (OH)3 (NO3 ), which can be considered the solidstate counterpart of such dinuclear systems. The question arises on the applicability to periodic systems of the trends established for molecular ones. In certain situations, this approximation appears well founded and allows a considerable simplification of the theoretical treatment, combined with a reduction of the computational effort. For instance, in the case of CuGeO3 , the use of a binuclear molecular model was legitimated by invoking the covalent nature of the bonds and the local character of the exchange coupling [278] and the calculated coupling constants were found to be close to those obtained from experimental measurements. However, we shall see how the existence of several exchange pathways, the role played by interlayer interactions and the occurrence of frustration effects prevents magnetism in solid Cu2 (OH)3 NO3 from being investigated on the basis of isolated molecular entities. Copper hydroxonitrate provides an example of a layered solid featuring a planar array of transition-metal ions. In this system, the interlayer distance between stacked brucite-type Cu2 (OH)4 sheets is modulated via the replacement of one fourth of the OH− ions with NO− 3 molecular units. As revealed by an X-ray structural determination [279], the copper atoms in Cu2 (OH)3 NO3 are octahedrally coordinated to the neighboring OH− and NO− 3 groups in two non-equivalent sites. Cu1 atoms have four OH− and two oxygen atoms of the NO− 3 groups as nearest neighbors, while Cu2 atoms are coordinated by four OH− , an additional OH− at a longer distance, and one oxygen atom belonging to one NO− 3 group (see Fig. 12). Concerning its magnetic behavior, the dependence of magnetic susceptibility with the temperature has been measured until 350 K. Upon cooling, the susceptibility slightly increases, reaching a maximum at T = 12 K. An antiferromagnetic character within each plane is suggested by the concomitant decrease of the χ T product. The first attempt to gain insight into the exchange coupling in this system relied on extended Huckel ¨ (EH) calculations [280]. The effect of spin coupling between

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

267

Fig. 12. Top: a view of the unit cell of Cu2 (OH)3 NO3 as obtained according to the X-ray determination of Ref. [279]. The labeling is taken also from the same reference. Bottom: projection in the ab plane of Cu2 (OH)3 NO3 (including Cu and O atoms only). Cu and O atoms are labeled so as to indicate their different coordination [244].

268

7 Electronic Structure and Magnetic Behavior

the various pairs of Cu(II) ions (Cu2–Cu2, Cu1–Cu2 and Cu1–Cu1) on the magnetic behavior of Cu2 (OH)3 NO3 was analyzed by calculating spin multiplet energies and magnetic susceptibilities for a cluster of Cu(II) ions with S = 1/2 spin. This model consists of parallel Cu1 and Cu2 chains, taken to reproduce the crystal geometry along the b direction and corresponding to four exchange coupling constants. The temperature dependence of the susceptibility, investigated as a function of the sign and magnitude of the four exchange parameters, was interpreted as a clear evidence for antiferromagnetic interactions within one of the chains, coexisting with a much weaker antiferromagnetic coupling in the other chain. According to the EH calculations, it is the Cu2–Cu2 coupling which plays the dominant role, while Cu1–Cu1 and Cu1–Cu2 interactions are much weaker. This result was presented as a convincing argument in favor of a one-dimensional model, claimed to be more consistent with the experiments than a two-dimensional model. The most intriguing aspect of these results is the disregard of ferromagnetic coupling between Cu1–Cu1 and Cu2–Cu2 pairs. Interestingly, a recent NMR study devoted to this same compound [281], combined with a spin Heisenberg model, has been consistently interpreted in terms of both ferromagnetic and antiferromagnetic interactions among the Cu1 and Cu2 ions respectively. Furthermore, it should be recalled that ferromagnetic Cu–Cu site interactions play a major role in other Cu(II)-based layered compounds [282]. Two-dimensional models based on a planar triangular arrangement have also been proposed, providing a good description of the susceptibility in the paramagnetic regime. However, this amounts to neglecting the coupling between the layers, while ambiguities remain on the sign of the exchange interactions. Thus, it appears that neither the two-dimensional approach, nor the one-dimensional one, can satisfactorily describe the magnetism in Cu2 (OH)3 NO3 . Moreover, a priori assumptions on the model dimensionality appear unsuitable since they rule out variations in the spin distribution among different layers. To go beyond phenomenological models, calculations were performed on Cu2 (OH)3 NO3 within the Kohn-Sham density functional framework. A generalized gradient approximation due to Becke [89] and Perdew [283] for the exchange and correlation in the S = 0 spin polarized case (i. e. the equivalent of the brokensymmetry approach referred in the molecular compounds sections of this paper) was used. Only the valence electrons (including the 3d electrons of each Cu atom) were treated explicitly, while norm-conserving pseudopotentials were used to account for the core-valence interaction [284]. The system examined consisted of 96 atoms in a periodically repeated crystal made of four monoclinic unit cells (two in the a and c directions and one in the b direction) at the experimental lattice parameters. By taking as initial configuration the experimental geometry, [279] the electronic structure is relaxed to its ground state by minimization of the total energy with respect to the coefficients of the plane-wave expansion. Optimization of the structure is then carried out by means of the first-principle molecular dynamics scheme [243, 285]. The agreement found between theory and experiment was found quite satisfactory, as illustrated in Table 11 for the Cu-O distances and the Cu2–O–Cu2 angles, demonstrating the reliability of our structural model.

7.5 Exchange Coupling in Polynuclear Transition-metal Complexes

269

Table 11. Experimental (left column) and calculated (right column) Cu2 interatomic distances (given in Å) and Cu2–O–Cu2 angles in Cu2 (OH)3 NO3 . The unit cell is made of 24 atoms having positions defined with respect to the primitive coordinates x, y, z and −x, y + 1/2, −z [279]. Oh1, Oh21, Oh22 and O1 define the O atoms with respect to their coordination to Cu1 or Cu2 atoms.

Cu2–Oh1 Cu2–Oh21 Cu2–Oh22 Cu2–O1 Cu2–Oh21–Cu2 Cu2–Oh22–Cu2

Experimental

Calculated

2.30 (1) 1.99 (1) 2.00 (1) 2.36 (1) 99.8 (1) 99.7 (1)

2.31 1.99 2.00 2.39 99.5 99.2

Figure 13 shows the projections of the spin density distribution on the two (a, b) planes of our simulation cell. Our calculated distribution of spin densities yields an antiferromagnetic character on each (a, b) plane, but it differs from one (a, b) plane to the other. Focusing on the spin densities on the Cu2 sites, which were conjectured to be mostly antiferromagnetic on the basis of isolated chain models, one notes that both parallel and antiparallel alignments occur along the b direction. This feature is confirmed by the patterns observed for the four possible superexchange interactions among Cu centers, which are Cu2–Cu2 via two OH− groups (labeled “a” in Fig. 13), − Cu2–Cu1 via two OH− groups (labeled “b”), Cu1–Cu2 via a NO− 3 group and a OH − − group (labeled “c”) and Cu1–Cu1 via a NO3 group and a OH group (labeled “d”). All of these couplings exhibit either parallel or antiparallel spin alignments, as it can be deduced by moving along and across the b direction on the different rows of Cu sites forming the triangular lattice on the (a, b) planes. This suggests that the conjecture of a correlation existing between the nature of the bridging groups and the sign of the exchange interaction is likely not to hold. A delicate point concerns the role played by frustration effects in determining the topology of the spin densities. On the one hand, they may stabilize parallel or antiparallel alignments between neighboring spin moments, without systematic agreement with the actual signs of the competing interactions. On the other hand, the calculations were performed within the spin collinear framework and do not allow ascertaining the extent of non-collinearity in this system. The different topology of spin densities between the Cu planes provides compelling evidence on the fact that models for magnetism relying on only one single, isolated layer are insufficient for this compound. Moreover, the existence of a nonunique distribution of spin densities along the c direction has consequences on the dimensionality of magnetism in Cu2 (OH)3 NO3 . Indeed, it implies that the distribution of local magnetic moments on the (a, b) layers has a periodicity along the c direction which differs from that of the lattice. A precise assessment of this latter issue calls for simulations on even larger systems and on the possibility of assigning and controlling the distribution of spin densities on the different atoms.

270

7 Electronic Structure and Magnetic Behavior

Fig. 13. Spin density distribution on the two different (a, b) planes composing our model for Cu2 (OH)3 NO3 . Positive spin density is dark grey and negative spin density is in light grey. O atoms are black, Cu atoms are dark grey, N atoms are light grey and H atoms are white. Note that for sake of clarity the system of 96 atoms has been duplicated along the b direction. On the upper panel, the white lines and the black labels define the four different exchange pathways along which magnetic interactions occur in this compound. These are Cu2–Cu2 via two OH− groups (labeled by “a”), Cu1–Cu2 via two OH− groups (labeled by “b”), Cu1–Cu2 via a NO− 3 group and a OH− group (labeled by “c”) and Cu1–Cu1 via − a NO− 3 group and a OH group (labeled by “d”) [244].

Acknowledgments We are indebted to Dr. Coen de Graaf for performing the MOLCAS and DDCI calculations of the copper acetate. Our work was supported by DGES through project number PB98-1166-C02-01 Additional support came from CIRIT through grant 1997SGR-072. The computing resources were generously made available in the Center de Computacio´ de Catalunya (CESCA) with a grant provided by Fundacio´ Catalana per a la Recerca (FCR) and the Universitat def Barcelona. For the calculations on solid state compounds, we are grateful to the IDRIS computer center of CNRS (France). The collaboration between the Barcelona and Strasbourg groups has taken advantage of a PICASSO bilateral project between France and Spain. We also want to acknowledge in general to all the colleagues whose names appear in the references.

References

271

References [1] J. H. Van Vleck, The Theory of Electric and Magnetic Susceptibilities, Oxford University Press, Oxford, 1932 [2] B. C. Guha, Proc. Roy. Soc. (London) Ser. A 1951, 206, 353. [3] B. Bleaney, K. D. Bowers, Proc. Roy. Soc. (London) Ser. A 1952, 214, 451. [4] O. Kahn, Molecular Magnetism, VCH Publishers, New York, 1993. [5] R. Boca, Theoretical Foundations of Molecular Magnetism, Elsevier, Lausanne, 1999. [6] T. A. Albright, J. K. Burdett, M.-H. Whangbo, Orbital Interactions in Chemistry, J. Wiley, New York, 1985. [7] F. Jensen, Introduction to Computational Chemistry, J. Wiley & Sons, Chichester, 1999. [8] R. G. Parr, W. Yang, Density-Functional Theory of Atoms and Molecules, Oxford University Press, New York, 1989. [9] W. Koch, M. C. Holthausen, A Chemist’s Guide to Density Functional Theory, WileyVCH, Weinheim, 2000. [10] P. W. Anderson, Phys. Rev. 1959, 115, 2. [11] J. B. Goodenough, Phys. Rev. 1955, 100, 564. [12] J. B. Goodenough, Magnetism and Chemical Bond, Interscience, New York, 1963. [13] J. Kanamori, J. Phys. Chem. Solids 1959, 10, 87. [14] A. Szabo, N. S. Ostlund, Modern Quantum Chemistry, McGraw Hill, New York, 1989. [15] J. C. Slater, The Self-Consistent Field for Molecules and Solids, Vol. 4, McGraw-Hill, New York, 1974. [16] P. J. Hay, J. C. Thibeault, R. Hoffmann, J. Am. Chem. Soc. 1975, 97, 4884. [17] R. Hoffmann, W. N. Liscomb, J. Chem. Phys. 1962, 36, 2179. [18] R. Hoffmann, W. N. Lipscomb, J. Chem. Phys. 1962, 36, 3469. [19] R. Hoffmann, W. N. Lipscomb, J. Chem. Phys. 1962, 37, 2872. [20] R. Hoffmann, J. Chem. Phys. 1963, 39, 1397. [21] R. Broer, W. J. A. Maaskant, Chem. Phys. 1986, 102, 103. [22] W. A. Hatfield, Inorg. Chem. 1983, 22, 833. [23] R. Vicente, J. Ribas, S. Alvarez, A. Segui, X. Solans, M. Verdaguer, Inorg. Chem. 1987, 26, 4004. [24] A. Caneschi, D. Gatteschi, A. Grand, T. Laugier, L. Pardi, P. Rey, Inorg. Chem. 1988, 27, 1031. [25] S. Alvarez, M. Julve, M. Verdaguer, Inorg. Chem. 1990, 29, 4500. [26] F. Lloret, Y. Journaux, M. Julve, Inorg. Chem. 1990, 29, 3967. [27] A. Escuer, R. Vicente, J. Ribas, M. Salah el Fallah, X. Solans, Inorg. Chem. 1993, 32, 1033. [28] J. Ribas, M. Monfort, R. Costa, X. Solans, Inorg. Chem. 1993, 32, 695. [29] U. Koelle, H. Lueken, K. Handrick, H. Schilder, J. K. Burdett, S. Balleza, Inorg. Chem. 1995, 34, 6273. [30] P. Roman, ´ C. Guzman-Miralles, ´ A. Luque, J. I. Beitia, J. Cano, F. Lloret, M. Julve, S. Alvarez, Inorg. Chem. 1996, 35, 3741. [31] J. R. Hart, A. K. Rappe, S. M. Gorun, T. H. Upton, J. Phys. Chem. 1992, 96, 6255. [32] O. Kahn, Adv. Inorg. Chem. 1995, 43, 179. [33] O. Kahn, M. F. Charlot, Nouv. J. Chim. 1980, 4, 567. [34] J. J. Girerd, Y. Journaux, O. Kahn, Chem. Phys. Lett. 1981, 82, 534. [35] C. Kollmar, O. Kahn, Acc. Chem. Res. 1993, 26, 259. [36] M. F. Charlot, Y. Journaux, O. Kahn, A. Bencini, D. Gatteschi, Inorg. Chem. 1986, 25, 1060.

272

7 Electronic Structure and Magnetic Behaviour

[37] Y. Journaux, O. Kahn, H. Coudanne, Angew. Chem. Int. Ed. Eng. 1982, 21, 624. [38] Y. Journaux, O. Kahn, J. Zarembowitch, J. Galy, J. Jaud, J. Am. Chem. Soc. 1983, 105, 7585. [39] O. Kahn, J. Galy, Y. Journaux, J. Jaud, I. Morgenstern-Badarau, J. Am. Chem. Soc. 1982, 104, 2165. [40] A. Meyer, A. Gleizes, J. J. Girerd, M. Verdaguer, O. Kahn, Inorg. Chem. 1982, 21, 1729. [41] O. Kahn, H. Stumpf, Y. Pei, J. Sletten, Mol. Crys. Liq. Crys. 1993, 232, 591. [42] Y. Pei, Y. Journaux, O. Kahn, Inorg. Chem. 1989, 28, 100. [43] T. Mallah, S. Thiebaut, M. Verdaguer, P. Veillet, Science 1993, 262, 1554. [44] M. Verdaguer, A. Bleuzen, V. Marvaud, J. Vaissermann, M. Seuleiman, C. Desplanches, A. Scuiller, C. Train, R. Garde, G. Gelly, C. Lomenech, I. Rosenman, P. Veillet, C. Cartier, F. Villain, Coord. Chem. Rev. 1999, 192, 1023. [45] S. Ferlay, T. Mallah, R. Ouahes, P. Veillet, M. Verdaguer, Nature 1995, 378, 701. [46] H. Weihe, H. U. Gudel, J. Am. Chem. Soc. 1997, 119, 6539. [47] H. Weihe, H. U. Gudel, J. Am. Chem. Soc. 1998, 120, 2870. [48] I. von Seggern, F. Tuczek, W. Bensch, Inorg. Chem. 1995, 34, 5530. [49] H. Weihe, H. U. Gudel, Inorg. Chem. 1997, 36, 3632. [50] P. de Loth, P. Cassoux, J. P. Daudey, J. P. Malrieu, J. Am. Chem. Soc. 1981, 103, 7124. [51] H. U. Gudel, ¨ A. Stebler, A. Furrer, ¨ Inorg. Chem. 1979, 18, 21. [52] H. G. Gudel, ¨ in Magneto-Structural Correlations in Exchange Coupled Systems (Eds.: R. D. Willet, D. Gatteschi, O. Kahn), Reidel, Dordrecht, 1985, p. 329. [53] J. P. Daudey, P. de Loth, J. P. Malrieu, in Magneto-Structural Correlations in Exchanged Coupled Systems (Eds.: R. D. Willett, D. Gatteschi, O. Kahn), Reidel, 1985. [54] P. de Loth, J. P. Daudey, H. Astheimer, L. Waltz, W. Haase, J. Chem. Phys. 1985, 82, 5048. [55] M. F. Charlot, M. Verdaguer, Y. Journaux, P. de Loth, J. P. Daudey, Inorg. Chem. 1984, 23, 3802. [56] P. de Loth, P. Karafiloglou, J. P. Daudey, O. Kahn, J. Am. Chem. Soc. 1988, 110, 5676. [57] F. Nepveu, W. Haase, H. Astheimer, J. Chem. Soc. Faraday Trans. 1986, 2, 551. [58] H. Astheimer, W. Haase, J. Chem. Phys. 1986, 85, 1427. [59] S. Gehring, W. Haase, Mol. Cryst. Liq. Cryst. 1989, 176, 513. [60] C. Erasmus, W. Haase, Spectrochimica Acta 1994, 50A, 2189. [61] C. Erasmus, W. Haase, Inorg. Chim. Acta 1994, 221, 141. [62] J. Miralles, R. Caballol, J. P. Malrieu, Chem. Phys. 1991, 153, 25. [63] J. Miralles, J. P. Daudey, R. Caballol, Chem. Phys. Lett. 1992, 198, 555. [64] J. Miralles, O. Castell, R. Caballol, J. P. Malrieu, Chem. Phys. 1993, 172, 33. [65] O. Castell, J. Miralles, R. Caballol, Chem. Phys. 1994, 179, 377. [66] O. Castell, R. Caballol, V. M. Garcia, K. Handrick, Inorg. Chem. 1996, 35, 1609. [67] O. Castell, R. Caballol, Inorg. Chem. 1999, 38, 668. [68] I. de P. R. Moreira, F. Illas, C. J. Calzado, J. F. Sanz, J. P. Malrieu, N. Ben Amor, D. Maynau, Phys. Rev. B 1999, 59, R6593. [69] K. Fink, R. Fink, V. Staemmler, Inorg. Chem. 1994, 33, 6219. [70] C. Wang, K. Fink, V. Staemmler, Chem. Phys. 1995, 201, 87. [71] C. Kolczewski, K. Fink, V. Staemmler, Int. J. Quan. Chem. 2000, 76, 137. [72] C. Wang, K. Fink, V. Staemmler, Chem. Phys. 1995, 192, 25. [73] K. Fink, C. Wang, V. Staemmler, Int. J. Quan. Chem. 1997, 65, 633. [74] A. Ceulemans, G. A. Heylen, L. F. Chibotaru, T. L. Maes, K. Pierloot, C. Ribbing, L. G. Vanquickenborne, Inorg. Chim. Acta 1996, 251, 15. [75] A. Ceulemans, L. F. Chibotaru, G. A. Heylen, K. Pierloot, Mol. Phys. 1999, 97, 1197. [76] R. Fink, V. Staemmler, Theor. Chim. Acta 1993, 87, 129. [77] K. Fink, C. Wang, V. Staemmler, Inorg. Chem. 1999, 38, 3847.

References

273

[78] A. Ceulemans, L. F. Chibotaru, G. A. Heylen, K. Pierloot, L. G. Vanquickenborne, Chem. Rev. 2000, 100, 787. [79] DDCI code: O. Castell, R. Caballol, J. P. Malrieu, J. P. Daudey, Universitat Rovira i Virgili (Spain), Universite´ Louis Pasteur (France), 1998. [80] MOLCAS 4: O. K. Andersson, M. R. A. Blomberg, M. P. Fulscher, ¨ G. Karlstrom, ¨ R. Lindh, P.-A. Malmqvist, P. Neogrady, ´ J. Olsen, B. O. Roos, A. J. Sadlej, M. Schutz, ¨ L. Seijo, L. Serrano-Andres, ´ P. E. M. Siegbahn, P.-O. Widmark, University of Lund, Sweden, 1997. [81] Gaussian98: M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen, M. W. Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle, J. A. Pople, Gaussian, Inc, Pittsburgh, PA, 1998. [82] L. Noodleman, J, Chem. Phys. 1981, 74, 5737. [83] L. Noodleman, D. Post, E. J. Baerends, Chem. Phys. 1982, 64, 159. [84] L. Noodleman, E. R. Davidson, Chem. Phys. 1986, 109, 131. [85] L. Noodleman, D. A. Case, Adv. Inorg. Chem. 1992, 38, 423. [86] L. Noodleman, C. Y. Peng, D. A. Case, J. M. Mouesca, Coord. Chem. Rev. 1995, 144, 199. [87] L. Noodleman, J. Li, X. G. Zhao, W. H. Richardson, in Density Functional Methods: Applications in Chemistry and Materials Science (Ed.: M. Springborg), J. Wiley, New York, 1997. [88] S. H. Vosko, L. Wilk, M. Nusair, Can. J. Phys. 1980, 58, 1200. [89] A. D. Becke, Phys. Rev. A 1988, 38, 3098. [90] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785. [91] A. D. Becke, J. Chem. Phys. 1993, 98, 5648. [92] E. Ruiz, J. Cano, S. Alvarez, P. Alemany, J. Comp. Chem. 1999, 20, 1391. [93] R. Caballol, O. Castell, F. Illas, I. d. P. R. Moreira, J. P. Malrieu, J. Phys. Chem. A 1997, 101, 7860. [94] A. Bencini, F. Totti, C. A. Daul, K. Doclo, P. Fantucci, V. Barone, Inorg. Chem. 1997, 36, 5022. [95] J. H. Rodriguez, D. E. Wheeler, J. K. McCusker, J. Am. Chem. Soc. 1998, 120, 12051. [96] T. Soda, Y. Kitagawa, T. Onishi, Y. Takano, Y. Shigeta, H. Nagao, Y. Yoshioka, K. Yamaguchi, Chem. Phys. Lett. 2000, 319, 223. [97] F. Illas, I. d. P. R. Moreira, C. de Graaf, V. Barone, Theor. Chem. Acc 2000, 104, 265. [98] J. M. Wittbrodt, H. B. Schlegel, J. Chem. Phys. 1996, 105, 6574. [99] J. P. Perdew, A. Savin, K. Burke, Phys. Rev. A 1995, 51, 4531. [100] J. P. Perdew, M. Erzzerhof, K. Burke, A. Savin, Int. J. Quantum Chem. 1997, 61, 197. [101] J. Grafenstein, ¨ A. M. Hjerpe, E. Kraka, D. Cremer, J. Phys. Chem. A 2000, 104, 1748. [102] H. Fukutome, Int. J. Quantum Chem. 1981, 20, 955. [103] J. R. Hart, A. K. Rappe, S. M. Gorun, T. H. Upton, Inorg. Chem. 1992, 31, 5254. [104] J. R. Hart, A. K. Rappe, S. M. Gorun, T. H. Upton, J. Phys. Chem. 1992, 96, 6264. [105] A. A. Ovchinnikov, C. F. Bender, J. K. Labanowski, in Recent Developments and Applications of Modern Density Functional Theory, Elsevier Science, Amsterdam, 1998, p. 862.

274 [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137] [138] [139] [140] [141] [142]

7 Electronic Structure and Magnetic Behaviour A. Goursot, J. P. Malrieu, D. R. Salahub, Theor. Chim. Acta 1995, 91, 225. C. Daul, Int. J. Quan. Chem. 1994, 52, 867. M. Filatov, S. Shaik, J. Chem. Phys. 1999, 110, 116. M. Filatov, S. Shaik, Chem. Phys. Lett. 1999, 304, 429. M. Filatov, S. Shaik, J. Phys. Chem. A 2000, 104, 6628. M. G. Cory, M. Zerner, J. Phys. Chem. A 1999, 103, 7287. K. Yamaguchi, T. Tsunekawa, Y. Toyoda, T. Fueno, Chem. Phys. Lett. 1988, 143, 371. A. P. Ginsberg, J. Am. Chem. Soc. 1980, 102, 111. J. Li, M. R. Nelson, C. Y. Peng, D. Bashford, L. Noodleman, J. Phys. Chem. A 1998, 102, 6311. L. Noodleman, D. A. Case, J. M. Mouesca, B. Lamotte, J. Bio. Inorg. Chem. 1996, 1, 177. J. M. Mouesca, L. Noodleman, D. A. Case, Int. J. Quan. Chem. 1995, 95. J. M. Mouesca, L. Noodleman, D. A. Case, B. Lamotte, Inorg. Chem. 1995, 34, 4347. J. M. Mouesca, J. L. Chen, L. Noodleman, D. Bashford, D. A. Case, J. Am. Chem. Soc. 1994, 116, 11898. J. M. Mouesca, L. Noodleman, D. A. Case, Inorg. Chem. 1994, 33, 4819. L. Noodleman, Inorg. Chem. 1991, 30, 246. J. Jordanov, E. K. H. Roth, P. H. Fries, L. Noodleman, Inorg. Chem. 1990, 29, 4288. T. E. Westre, P. Kennepohl, J. G. DeWitt, B. Hedman, K. O. Hodgson, E. I. Solomon, , J. Am. Chem. Soc. 1997, 119, 6297. L. Noodleman, D. A. Case, A. Aizman, J. Am. Chem. Soc. 1988, 110, 1001. L. Noodleman, Inorg. Chem. 1988, 27, 3677. L. Noodleman, J. G. Norman, J. H. Osborne, A. Aizman, D. A. Case, J. Am. Chem. Soc. 1985, 107, 3418. J. M. Mouesca, B. Lamotte, Coord. Chem. Rev. 1998, 180, 1573. L. Noodleman, E. J. Baerends, J. Am. Chem. Soc. 1984, 106, 2316. C. L. Fisher, J. L. Chen, J. Li, D. Bashford, L. Noodleman, J. Phys. Chem. 1996, 100, 13498. J. Li, C. L. Fisher, J. L. Chen, D. Bashford, L. Noodleman, Inorg. Chem. 1996, 35, 4694. J. Li, C. L. Fisher, R. Konecny, D. Bashford, L. Noodleman, Inorg. Chem. 1999, 38, 929. E. A. Schmitt, L. Noodleman, E. J. Baerends, D. N. Hendrickson, J. Am. Chem. Soc. 1992, 114, 6109. X. G. Zhao, W. H. Richardson, J. L. Chen, J. Li, L. Noodleman, H. L. Tsai, D. N. Hendrickson, Inorg. Chem. 1997, 36, 1198. H. X. Wang, C. Bryant, D. W. Randall, L. B. LaCroix, E. I. Solomon, M. LeGros, S. P. Cramer, J. Phys. Chem. B 1998, 102, 8347. D. R. Gamelin, D. W. Randall, M. T. Hay, R. P. Houser, T. C. Mulder, G. W. Canters, S. deVries, W. B. Tolman, Y. Lu, E. I. Solomon, J. Am. Chem. Soc. 1998, 120, 5246. E. I. Solomon, L. B. LaCroix, D. W. Randall, Pure App. Chem. 1998, 70, 799. E. I. Solomon, M. D. Lowery, J. A. Guckert, L. B. LaCroix, Adv. in Chem. 1997, 253, 317. L. B. LaCroix, S. E. Shadle, Y. N. Wang, B. A. Averill, B. Hedman, K. O. Hodgson, E. I. Solomon, J. Am. Chem. Soc. 1996, 118, 7755. J. A. Guckert, M. D. Lowery, E. I. Solomon, J. Am. Chem. Soc. 1995, 117, 2817. S. E. Shadle, J. E. Pennerhahn, H. J. Schugar, B. Hedman, K. O. Hodgson, E. I. Solomon, J. Am. Chem. Soc. 1993, 115, 767. M. D. Lowery, E. I. Solomon, Inorg. Chim. Acta 1992, 200, 233. M. J. Baldwin, P. K. Ross, J. E. Pate, Z. Tyeklar, K. D. Karlin, E. I. Solomon, J. Am. Chem. Soc. 1991, 113, 8671. P. K. Ross, E. I. Solomon, J. Am. Chem. Soc. 1991, 113, 3246.

References

275

[143] M. J. Baldwin, D. E. Root, J. E. Pate, K. Fujisawa, N. Kitajima, E. I. Solomon, J. Am. Chem. Soc. 1992, 114, 10421. [144] F. Tuczek, E. I. Solomon, Inorg. Chem. 1993, 32, 2850. [145] D. E. Root, M. MahroofTahir, K. D. Karlin, E. I. Solomon, Inorg. Chem. 1998, 37, 4838. [146] D. E. Root, M. J. Henson, T. Machonkin, P. Mukherjee, T. D. P. Stack, E. I. Solomon, J. Am. Chem. Soc. 1998, 120, 4982. [147] M. J. Henson, P. Mukherjee, D. E. Root, T. D. P. Stack, E. I. Solomon, J. Am. Chem. Soc. 1999, 121, 10332. [148] T. C. Brunold, E. I. Solomon, J. Am. Chem. Soc. 1999, 121, 8277. [149] T. C. Brunold, E. I. Solomon, J. Am. Chem. Soc. 1999, 121, 8288. [150] T. C. Brunold, N. Tamura, N. Kitajima, Y. Morooka, E. I. Solomon, J. Am. Chem. Soc. 1998, 120, 5674. [151] T. E. Westre, P. Kennepohl, J. G. DeWitt, B. Hedman, K. O. Hodgson, E. I. Solomon, J. Am. Chem. Soc. 1997, 119, 6297. [152] D. R. Gamelin, E. L. Bominaar, M. L. Kirk, K. Wieghardt, E. I. Solomon, J. Am. Chem. Soc. 1996, 118, 8085. [153] C. A. Brown, G. J. Remar, R. L. Musselman, E. I. Solomon, Inorg. Chem. 1995, 34, 688. [154] T. C. Brunold, D. R. Gamelin, T. L. Stemmler, S. K. Mandal, W. H. Armstrong, J. E. PennerHahn, E. I. Solomon, J. Am. Chem. Soc. 1998, 120, 8724. [155] D. R. Gamelin, M. L. Kirk, T. L. Stemmler, S. Pal, W. H. Armstrong, J. E. Pennerhahn, E. I. Solomon, J. Am. Chem. Soc. 1994, 116, 2392. [156] K. Yamaguchi, S. Yamanaka, M. Nishino, Y. Takano, Y. Kitagawa, H. Nagao, Y. Yoshioka, Theor. Chem. Acc. 1999, 102, 328. [157] M. Nishino, Y. Yoshioka, K. Yamaguchi, K. Mashima, K. Tani, A. Nakamura, Bull. Chem. Soc. Jpn. 1998, 71, 99. [158] M. Nishino, S. Yamanaka, Y. Yoshioka, K. Yamaguchi, J. Phys. Chem. A 1997, 101, 705. [159] M. Nishino, M. Tanaka, S. Takeda, K. Mashima, W. Mori, K. Tani, A. Nakamura, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1996, 286, 193. [160] M. Fujiwara, S. Takamizawa, W. Mori, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1996, 278, A1. [161] M. Fujiwara, T. Matsushita, K. Yamaguchi, T. Fueno, Synth. Met. 1991, 43, 3267. [162] K. Yamaguchi, T. Fueno, M. Ozaki, N. Ueyama, A. Nakamura, Chem. Phys. Lett. 1990, 168, 56. [163] K. Yamaguchi, T. Fueno, N. Ueyama, A. Nakamura, M. Ozaki, Chem. Phys. Lett. 1989, 164, 210. [164] Y. Yoshioka, S. Kubo, S. Kiribayashi, Y. Takano, K. Yamaguchi, Bull. Chem. Soc. Jpn. 1998, 71, 573. [165] M. Nishino, Y. Yoshioka, K. Yamaguchi, Chem. Phys. Lett. 1998, 297, 51. [166] M. Nishino, S. Kubo, Y. Yoshioka, A. Nakamura, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1997, 305, 109. [167] M. Nishino, S. Takeda, W. Mori, A. Nakamura, K. Yamaguchi, Synth. Met. 1997, 85, 1763. [168] M. Nishino, M. Tanaka, S. Takeda, K. Mashima, W. Mori, K. Tani, A. Nakamura, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1997, 306, 463. [169] M. Nishino, M. Tanaka, S. Takeda, K. Mashima, W. Mori, K. Tani, A. Nakamura, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1996, 286, 201. [170] M. Fujiwara, M. Nishino, S. Takamizawa, W. Mori, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1996, 285, 507. [171] T. Kawakami, S. Yamanaka, Y. Takano, Y. Yoshioka, K. Yamaguchi, J. Mol. Struct. (Theochem) 1998, 451, 89.

276

7 Electronic Structure and Magnetic Behaviour

[172] M. Kamachi, A. Kajiwara, W. Mori, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1995, 273, 117. [173] T. Kawakami, H. Nagao, W. Mori, K. Yamaguchi, Synth. Met. 1997, 85, 1753. [174] T. Kawakami, H. Nagao, K. Ueda, W. Mori, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1996, 285, 499. [175] H. Nagao, H. Kawabe, T. Kawakami, M. Okumura, W. Mori, K. Nishikawa, K. Yamaguchi, Mol. Cryst. Liq. Crys. 1996, 285, 493. [176] H. Nagao, M. Nishino, Y. Shigeta, T. Soda, Y. Kitagawa, T. Onishi, Y. Yoshioka, K. Yamaguchi, Coord. Chem. Rev. 2000, 198, 265. [177] H. Nagao, S. Yamanaka, M. Nishino, Y. Yoshioka, K. Yamaguchi, Chem. Phys. Lett. 1999, 302, 418. [178] Y. Shigeta, T. Kawakami, H. Nagao, K. Yamaguchi, Chem. Phys. Lett. 1999, 315, 441. [179] A. Bencini, S. Midollini, Coord. Chem. Rev. 1992, 120, 87. [180] L. Banci, A. Bencini, D. Gatteschi, J. Am. Chem. Soc. 1983, 105, 761. [181] A. Bencini, D. Gatteschi, J. Am. Chem. Soc. 1983, 105, 5535. [182] A. Bencini, D. Gatteschi, J. Am. Chem. Soc. 1986, 108, 5763. [183] A. Bencini, J. Chim. Phys. 1989, 86, 763. [184] A. Bencini, C. Benelli, D. Gatteschi, C. Zanchini, A. C. Fabretti, G. C. Franchini, Inorg. Chim. Acta 1984, 86, 169. [185] C. Albonico, A. Bencini, Inorg. Chem. 1988, 27, 1934. [186] C. Adamo, V. Barone, A. Bencini, F. Totti, I. Ciofini, Inorg. Chem. 1999, 38, 1996. [187] A. Bencini, C. A. Ghilardi, A. Orlandini, S. Midollini, C. Zanchini, J. Am. Chem. Soc. 1992, 114, 9898. [188] A. Bencini, M. G. Uytterhoeven, C. Zanchini, Int. J. Quan. Chem. 1994, 52, 903. [189] A. Bencini, C. A. Ghilardi, S. Midollini, A. Orlandini, U. Russo, M. G. Uytterhoeven, C. Zanchini, J. Chem. Soc. Dalton Trans. 1995, 963. [190] F. Cecconi, C. A. Ghilardi, S. Midollini, A. Orlandini, P. Zanello, A. Cinquantini, A. Bencini, M. G. Uytterhoeven, G. Giorgi, J. Chem. Soc. Dalton Trans. 1995, 3881. [191] A. Bencini, D. Gatteschi, F. Totti, D. N. Sanz, J. A. McCleverty, M. D. Ward, J. Phys. Chem. A 1998, 102, 10545. [192] V. Barone, A. Bencini, I. Ciofini, C. A. Daul, F. Totti, J. Am. Chem. Soc. 1998, 120, 8357. [193] A. Bencini, D. Gatteschi, M. Mattesini, F. Tottia, I. Ciofini, Mol. Cryst. Liq. Crys. 1999, 334, 1377. [194] A. Bencini, I. Ciofini, C. A. Daul, A. Ferretti, J. Am. Chem. Soc. 1999, 121, 11418. [195] Z. Chen, J. Bian, L. Zhang, S. Li, J. Chem. Phys. 1999, 111, 10926. [196] A. Caneschi, F. Fabrizi de Biani, L. Kloo, P. Zanello, Int. J. Quantum Chem. 1999, 72, 61. [197] P. Belanzoni, N. Re, M. Rosi, A. Sgamellotti, E. J. Baerends, C. Floriani, Inorg. Chem. 1996, 35, 7776. [198] R. Stranger, T. Lovell, J. E. McGrady, Inorg. Chem. 1999, 38, 5510. [199] J. E. McGrady, R. Stranger, T. Lovell, Inorg. Chem. 1998, 37, 3802. [200] R. Stranger, J. E. McGrady, T. Lovell, Inorg. Chem. 1998, 37, 6795. [201] J. E. McGrady, R. Stranger, T. Lovell, J. Phys. Chem. A 1997, 101, 6265. [202] J. E. McGrady, T. Lovell, R. Stranger, Inorg. Chem. 1997, 36, 3242. [203] T. Lovell, J. E. McGrady, R. Stranger, S. A. Macgregor, Inorg. Chem. 1996, 35, 3079. [204] R. Stranger, S. A. Macgregor, T. Lovell, J. E. McGrady, G. A. Heath, J. Chem. Soc. Dalton Trans. 1996, 4485. [205] G. A. Heath, J. E. McGrady, R. G. Raptis, A. C. Willis, Inorg. Chem. 1996, 35, 6838. [206] G. A. Medley, R. Stranger, Inorg. Chem. 1994, 33, 3976. [207] R. Stranger, P. W. Smith, I. E. Grey, Inorg. Chem. 1989, 28, 1271.

References [208] [209] [210] [211] [212] [213] [214] [215] [216] [217] [218] [219] [220] [221] [222] [223] [224] [225] [226]

[227] [228] [229] [230] [231] [232] [233] [234] [235] [236] [237] [238] [239] [240] [241] [242] [243] [244]

277

C. D. Delfs, R. Stranger, Inorg. Chem. 2000, 39, 491. J. E. McGrady, R. Stranger, Inorg. Chem. 1999, 38, 550. J. E. McGrady, R. Stranger, J. Am. Chem. Soc. 1997, 119, 8512. W. Plass, Angew. Chem. Int. Ed. Eng. 1996, 35, 627. C. Blanchet-Boiteux, J. M. Mouesca, J. Phys. Chem. A 2000, 104, 2091. C. Blanchet-Boiteux, J. M. Mouesca, J. Am. Chem. Soc. 2000, 122, 861. C. Blanchet-Boiteux, J. M. Mouesca, Theor. Chem. Acc. 2000, 104, 257. R. Boca, H.-F. Klein, A. Schmidt, M. Valko, W. Linert, Chem Phys. Lett. 2000, 323, 243. E. Ruiz, P. Alemany, S. Alvarez, J. Cano, J. Am. Chem. Soc. 1997, 119, 1297. E. Ruiz, P. Alemany, S. Alvarez, J. Cano, Inorg. Chem. 1997, 36, 3683. E. Ruiz, J. Cano, S. Alvarez, P. Alemany, J. Am. Chem. Soc. 1998, 120, 11122. F. Fabrizi de Biani, E. Ruiz, J. Cano, J. J. Novoa, S. Alvarez, Inorg. Chem. 2000, 39, 3221. J. Cano, P. Alemany, S. Alvarez, M. Verdaguer, E. Ruiz, Chem. Eur. J. 1998, 4, 476. J. Cano, E. Ruiz, P. Alemany, F. Lloret, S. Alvarez, J. Chem. Soc. Dalton Trans. 1999, 1669. A. Rodr´ıguez-Fortea, P. Alemany, S. Alvarez, E. Ruiz, Chem. Eur. J. 2001, 7, 627. J. Cano, A. Rodr´ıguez-Fortea, P. Alemany, S. Alvarez, E. Ruiz, Chem. Eur. J. 2000, 6, 327. I. Castro, M. L. Calatayud, J. Sletten, F. Lloret, J. Cano, M. Julve, G. Seitz, K. Mann, Inorg. Chem. 1999, 38, 4680. L. Gutierrez, G. Alzuet, J. A. Real, J. Cano, J. Borras, A. Castineiras, ˜ Inorg. Chem. 2000, 39, 3608. CRYSTAL98: V. R. Saunders, R. Dovesi, C. Roetti, M. Causa, ` N. M. Harrison, R. Orlando, C. M. Zicovich-Wilson, Universita´ di Torino, CCLRC Daresbury Laboratory, 1999. W. C. Mackrodt, N. M. Harrison, V. R. Saunders, N. L. Allan, M. D. Towler, E. Apra, ` R. Dovesi, Philos. Mag. A 1993, 68, 653. W. C. Mackrodt, E. Williamson, J. Phys.: Condens. Matter 1997, 9, 6591. M. D. Towler, N. L. Allan, N. M. Harrison, V. R. Saunders, W. C. Mackrodt, E. Apra, ` Phys. Rev. B 1997, 50, 5041. M. Catti, G. Valerio, R. Dovesi, Phys. Rev. B 1996, 51, 7441. M. Catti, G. Sandrone, G. Valerio, R. Dovesi, J. Phys. Chem. Solids 1996, 57, 1735. M. Catti, G. Valerio, R. Dovesi, Phys. Rev. B 1997, 54, 16122. M. Catti, G. Sandrone, Faraday Discuss. 1997, 106, 189. J. M. Ricart, R. Dovesi, C. Roetti, V. R. Saunders, Phys. Rev. B 1995, 52, 2381. R. Dovesi, J. M. Ricart, V. R. Saunders, R. Orlando, J. Phys.: Condens. Matter 1995, 7, 7997. M. D. Towler, R. Dovesi, V. R. Saunders, Phys. Rev. B 1995, 52, 10150. R. Dovesi, F. Freyria Fava, C. Roetti, V. R. Saunders, Faraday Discuss. 1997, 106, 189. G. Valerio, M. Catti, R. Dovesi, R. Orlando, Phys. Rev. B 1995, 52, 2422. F. Freyria Fava, I. Baraille, A. Lichanot, C. Larrieu, R. Dovesi, J. Phys.: Condens. Matter 1997, 9, 10715. F. Freyria Fava, P. D’Arco, R. Orlando, R. Dovesi, J. Phys.: Condens. Matter 1997, 9, 489. A. Chartier, P. D’Arco, R. Dovesi, V. R. Saunders, Phys. Rev. B 1999, 60, 14042. V. Laget, C. Hornick, P. Rabu, M. Drillon, P. Turek, R. Ziessel, Adv. Mater. 1998, 10, 1024. R. Car, M. Parrinello, Phys. Rev. Lett. 1985, 55, 2472. C. Massobrio, P. Rabu, M. Drillon, C. Rovira, J. Phys. Chem. B 1999, 103, 9387.

278

7 Electronic Structure and Magnetic Behaviour

[245] V. H. Crawford, H. W. Richardson, J. R. Wasson, D. J. Hodgson, W. E. Hatfield, Inorg. Chem. 1976, 15, 2107. [246] E. Ruiz, S. Alvarez, P. Alemany, Chem. Commun. 1998, 2767. [247] M. Julve, M. Verdaguer, A. Gleizes, M. Philoche-Levisalles, O. Kahn, Inorg. Chem. 1984, 23, 3808. [248] M. Verdaguer, O. Kahn, M. Julve, A. Gleizes, Nouv. J. Chim. 1985, 9, 325. [249] M. Meln´ık, Coord. Chem. Rev. 1981, 36, 1. [250] M. Kato, Y. Muto, Coord. Chem. Rev. 1988, 92, 45. [251] E. Colacio, J.-P. Costes, R. Kivekas, ¨ J.-P. Laurent, J. Ruiz, Inorg. Chem. 1990, 29, 4240. [252] E. Colacio, J.-M. Dom´ınguez-Vera, J.-P. Costes, R. Kivekas, J.-P. Laurent, J. Ruiz, M. Sundberg, Inorg. Chem. 1992, 31, 774. [253] E. Colacio, J.-M. Dom´ınguez-Vera, R. Kivekas, ¨ J. Ruiz, Inorg. Chim. Acta 1994, 218, 109. [254] P. K. Coughlin, S. J. Lippard, J. Am. Chem. Soc. 1984, 106, 2328. [255] M.-C. Lim, W. Chen, H. M. Ali, Trans. Met. Chem. 1994, 19, 409. [256] S. P. Perlepes, E. Libby, W. E. Streib, K. Folting, G. Christou, Polyhedron 1992, 11, 923. [257] M. Inoue, M. Kubo, Inorg. Chem. 1970, 9, 2311. [258] E. Colacio, J. M. Dom´ınguez-Vera, M. Ghazi, R. Kivekas, ¨ M. Klinga, J. M. Moreno, Eur. J. Inorg. Chem. 1999, 441. [259] O. M. Steward, R. C. McAfee, S.-C. Chang, S. R. Piskor, W. J. Schreiber, C. F. Jury, C. E. Taylor, J. F. Pletcher, C.-S. Chen, Inorg. Chem. 1986, 25, 771. [260] R. J. Doedens, Prog. Inorg. Chem. 1976, 21, 209. [261] M.-F. Charlot, O. Kahn, M. Chaillet, C. Larrieu, J. Am. Chem. Soc. 1986, 108, 2574. [262] M. A. Aebersold, B. Gillon, O. Plantevin, L. Pardi, O. Kahn, P. Bergerat, I. von Seggern, ¨ T. F., L. Ohrstr om, ¨ A. Grand, E. Lelievre-Berna, ` J. Am. Chem. Soc. 1998, 120, 5238. [263] S. S. Tandon, L. K. Thompson, M. E. Manuel, J. N. Bridson, Inorg. Chem. 1994, 33, 5555. [264] F. Yan, Z. Chen, J. Phys. Chem. A 2000, 104, 6295. [265] F. Birkelbach, M. Winter, U. Florke, ¨ H. J. Haupt, C. Butzlaff, M. Lengen, E. Bill, A. X. Trautwein, K. Wieghardt, P. Chaudhuri, Inorg. Chem. 1994, 33, 3990. [266] D. Gatteschi, K. O., J. S. Miller, F. Palacio, Magnetic Molecular Materials, Kluwer Academic, Dordrecht, 1991. [267] D. Gatteschi, Current Opinion in Solid State & Materials Science 1996, 1, 192. [268] A. Muller, F. Peters, M. T. Pope, D. Gatteschi, Chem. Rev. 1998, 98, 239. [269] W. Haase, J. Mol. Cat. 1984, 23, 331. [270] E. Ruiz, A. Rodr´ıguez-Fortea, P. Alemany, S. Alvarez, Polyhedron 2001, in press. [271] J. Sletten, A. Sorensen, M. Julve, Y. Journaux, Inorg. Chem. 1990, 29, 5054. [272] X. S. Tan, Y. Fujii, R. Nukada, M. Mikuriya, Y. Nakano, J. Chem. Soc., Dalton Trans. 1999, 2415. [273] H. L. Shyu, H. H. Wei, G. H. Lee, L. Wang, Inorg. Chem. 1996, 35, 5396. [274] I. Unamuno, J. M. Gutierrez-Zorrilla, A. Luque, P. Roman, L. Lezama, R. Calvo, T. Rojo, Inorg. Chem. 1998, 37, 6452. [275] E. Rentschler, D. Gatteschi, A. Cornia, A. C. Fabretti, A.-L. Barra, O. I. Shchegolikhina, A. A. Zhdanov, Inorg. Chem. 1996, 35, 4427. [276] A. Cornia, A. C. Fabretti, D. Gatteschi, G. Palyi, ´ E. Rentschler, O. I. Shchegolikhina, A. A. Zhdanov, Inorg. Chem. 1995, 34, 5383. [277] E. Ruiz, J. Cano, A. Caneschi, D. Gatteschi, S. Alvarez, 2001, to be published. [278] E. Ruiz, J. Cano, S. Alvarez, P. Alemany, M. Verdaguer, Phys. Rev. B 2000, 61, 54. [279] H. Effenberger, Z. Kristallog. 1983, 165, 127. [280] G. G. Linder, M. Atanasov, J. Pebler, J. Solid. State Chem. 1995, 116, 1.

References

279

[281] S. Takeda, G. Maruta, K. Terasawa, N. Fukuda, K. Yamaguchi, Mol. Cryst. Liq. Cryst. 1999, 335, 11. [282] P. Rabu, S. Rouba, V. Laget, C. Hornick, M. Drillon, Chem. Commun. 1996, 1107. [283] J. P. Perdew, Phys. Rev. B 1986, 33, 8822. [284] N. Trouiller, J. L. Martins, Phys. Rev. B 1991, 43, 1993. [285] CPMD code: J. Hutter, O. Ballone, M. Bernasconi, P. Focher, E. Fois, S. Goedecker, D. Marx, M. Parrinello, M. Tuckerman, MPI fur ¨ Festkoerperforschung and IBM Research, 1990-98.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

8

Valence Tautomerism in Dioxolene Complexes of Cobalt David A. Shultz

8.1

8.1.1

Introduction – Bistability, Hysteresis, and Electronically Labile Materials Bistability and Hysteresis

There are several classes of molecules and materials whose electronic structures are a dramatic function of temperature, pressure, or other external stimuli. Such materials have been termed “electronically labile” [1] and are important in developing our understanding of electron transfer, conductivity, magnetism, light absorption, and related phenomena. Bistable molecules are those that can exist in two chemically distinct forms. Electronically labile molecules and materials are inherently bistable and therefore can serve as the basis for molecular electronic devices [2–10]. However, for such compounds to exhibit non-volatile memory, they must exhibit hysteresis, as illustrated in Fig. 1. Consider a bistable molecular material whose molecules can exist in two forms, A and B. Form A is stable at low temperatures and form B is stable at high temperatures. The temperature at which the mole fractions of A and B are equal (x(A) = x(B)) is called the critical temperature, T1/2 . If the interactions between molecules (λ) are very weak, then the transition from A to B is said to be non-cooperative, and the increase in x(B) as a function of increasing temperature is both single-valued and smooth as shown above. Thus, the non-cooperative transition can be viewed as a random distribution of B appearing as the temperature is increased. If however there is a moderate intermolecular interaction, λ, and if the interaction between molecules of A is different than the interaction between molecules of B, then the transition from A to B can be abrupt, and hysteresis can be observed [11]. Hysteresis is characterized by two T1/2 values: one observed upon cooling (T1/2 ↓), and one observed upon warming (T1/2 ↑). Between T1/2 ↓ and T1/2 ↑, A and B coexist. The coexistence of two chemically distinct species, and a mechanism for their interconversion is a recipe for molecular-level switching and information storage.

8.1.2

Electronically Labile Materials

Examples of electronically labile molecules are those that exhibit mixed-valence (MV) [12–17], valence tautomerism (VT) [18], or spin crossover (SC) [11, 19–27].

282

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 1. Bistability and hysteresis.

Mixed valence in inorganic complexes has been studied for over thirty years and can be described as a ligand-mediated interaction between two metal ions with different oxidation states, as shown in Fig. 2. Mixed-valent organic systems have also been described [28–41]. The study of MV complexes is interesting because of the opportunity to study electron transfer in well-characterized, single molecules. Depending on the degree of interaction between the metal centers, as mediated by bridging ligand(s), three classes of MV are possible [17]: Class I complexes are those in which no interaction

8.1 Introduction – Bistability, Hysteresis, and Electronically Labile Materials

283

Fig. 2. Mixed valence.

exists between metal ions; Class II complexes are those characterized by a weak interaction (β12 ); Class III complexes are delocalized. Class II is unique in that a particular valence state can be trapped if the thermal energy is less than the barrier to electron transfer, E TH . In addition, Class II complexes exhibit an electronic transition, E IT , called the intervalence transition [42]. Spin crossover, Fig. 3, refers to the change in spin multiplicity (high-spin = hs; low-spin = ls) of a transition metal ion (e. g., CrII , MnIII , FeIII , FeII , CoIII , or CoII ) as a function of temperature, pressure, light, or composition. Techniques used to study SC include Mossbauer ¨ spectroscopy, vibrational spectroscopy, electronic absorption spectroscopy, and magnetic susceptibility. SC compounds have a variety of interesting properties including photorefractivity and the propensity for bistability. Hauser has shown that an FeII SC complex when kept below 50 K can be trapped in its hs form by irradiating the ls form. The effect is called light induced excited state spin trapping (LIESST) [27, 43]. As indicated in Fig. 3, when two spin-states (e. g., 1 A and 5 T) have similar enthalpies but very different entropies, G ◦ can change sign within a readily attainable temperature range. Thus, depending on the temperature, either the hs form or ls form will predominate. We will call such equilibria “entropy driven.” In SC, the entropy driven equilibrium favors the hs-FeII form at high temperatures, and the ls-FeII form at low temperatures. Again, organic species with similar properties have been described [44–46]. The entropic driving force is primarily vibrational: occupied σ antibonding orbitals in the hs-FeII form result in longer Fe–L bond lengths and thus

Fig. 3. Spin crossover.

284

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 4. Valence tautomerism.

a high density of vibrational states. Indeed, large differences in Fe–L stretching frequencies have been measured for FeII complexes in the hs and ls forms [47], and these frequency changes account for the large S ◦ reported from variable-temperature heat capacity measurements [48]. Valence tautomerism is presented graphically in Fig. 4. A combined intramolecular metal-ligand electron transfer and spin crossover characterizes VT. To date, VT involves quinone-type ligands in the dianion form (catecholate, Cat) and radical anion form (semiquinone, SQ). The quinone ligand orbitals generally lie close in energy to the metal orbitals and can donate electrons to or accept electrons from the metal. As in SC, VT complexes undergo changes in d-orbital occupation as a function of temperature. The ls-CoIII form has the metal electronic configuration (π )6 (σ ∗ )0 , and the hs-CoII form has the metal electronic configuration (π )5 (σ ∗ )2 . Occupation of the σ ∗ orbitals occurs only in the hs-CoII form and results in longer Co–ligand bond lengths. This transition to longer bond lengths is accompanied by a large entropy increase due to a higher density of vibrational levels in the hs-CoII form. Therefore VT equilibria, like SC equilibria, are entropy driven.

8.2 8.2.1

Valence Tautomerism in Dioxolene Complexes of Cobalt Valence Tautomerism – A General Chemical Description

Valence tautomerism (VT) has been reported for complexes containing a variety of metal ions (Mn [49–54], Rh and Ir [55, 56], and Co) ligated to dioxolene ligands [18]; however, the focus of this chapter is centered on cobalt-containing materials. VT can be thought of as a special case of SC – one that involves electroactive ligands. Redox-active ligands open the possibility for metal-ligand electron transfer to accompany SC. A generic VT equilibrium is shown in Fig. 5 (N–N is a chelating diamine ligand): a ls-CoIII (SQ)(Cat) complex at low temperature is transformed into a hs-CoII (SQ)2 complex at high temperatures. The spin crossover is defined by the low-spin (d6 CoIII ) to high-spin (d7 CoII ) multiplicity change at the metal ion, and

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

285

Fig. 5. Generic VT equilibrium.

Fig. 6. Semiquinone and catecholate derivatives most commonly used in VT complexes.

the electron transfer involves either the reduction of CoIII by a Cat or the reduction of SQ to Cat by CoII . Most of the VT complexes reported to date are composed of two dioxolene ligands, a cobalt ion, and a diamine ligand. The most common dioxolene ligands are 3,5-dit-butylsemiquinone (3,5-DBSQ), 3,6-di-t-butylsemiquinone (3,6-DBSQ), and their corresponding catecholate forms (3,5-DBCat and 3,6-DBCat), Fig. 6. DBQ denotes di-t-butylquinone in either the SQ or Cat form. Semiquinones and catecholates are one- and two-electron reduction products of orthoquinones, and the coordination chemistry of these ligands has been reviewed [57–59]. As be discussed below, the important features of metal complexes of these dioxolene ligands include rather dramatic differences in bond lengths between metalSQ and metal-Cat forms. The magnetic properties of the two tautomeric forms are quite different, as shown in Fig. 7. Typical values of χ T , the paramagnetic susceptibility-temperature product, for the hs-CoII form range from ca. 2.4 to 3.8 emu K mol−1 (this corresponds to a spinonly effective magnetic moment between 4.4 and 5.5). The χ T value for the ls-CoIII tautomer is ca. 0.4 emu K mol−1 (this corresponds to a spin-only effective magnetic moment of ca. 1.8). Note that in the ls form the DBQ ligands are mixed-valent (one Cat and one SQ). In addition to magnetic differences, the optical properties of the VT tautomers differ substantially as shown in Fig. 8. The ls-CoIII tautomer is characterized by a band near 16 666 cm−1 (600 nm) and a ligand-based intervalence transition near 4000 cm−1 (2500 nm) [61]. The hs-CoII tautomer, on the other hand, shows a MLCT band near 13 000 cm−1 (770 nm) and no transition near 4000 cm−1 . There has been disagreement on the origin of the NIR band, but both computations [61] and analogy with ls-CoIII -monocatecholate complexes [62] point to a ligand-based intervalence transition rather than a LMCT band. The conversion of tautomeric forms can be followed quite easily using variabletemperature electronic absorption spectroscopy. The transformation is marked by several isosbestic points, consistent with two species only, i. e. no intermediates are involved.

286

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 7. Magnetic moment for Co(3,5-DBQ)2 (bpy) as a function of temperature. Reprinted with permission from Ref. [60]. Copyright 1980 American Chemical Society.

Fig. 8. Electronic absorption spectra for Co(3,5-DBSQ)2 (phen).toluene as a function of temperature in a polystyrene matrix. Reprinted with permission from Ref. [61]. Copyright 1997 American Chemical Society.

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

8.2.2

287

Valence Tautomerism – A Simplified MO Description

Figure 9 shows the important frontier metal and dioxolene orbitals for the M(DBQ)2 (N–N) unit. For simplicity, the N–N orbitals have been ignored. The dioxolene π ∗ orbital is singly occupied for SQ and doubly occupied for Cat. A complete MO description of VT complexes has been presented by Hendrickson and Noodleman [61]. The metal t2g d-orbitals (dxy , dxz , dyz ) are of the correct symmetry to mix with the dioxolene π ∗ -MOs in a π-fashion, while the remaining eg metal d-orbitals mix with dioxolene oxygen lone pair orbitals in a σ -fashion. Because the dioxolene π ∗ MOs are slightly higher in energy than the t2g orbitals, the latter are transformed into metal-ligand π-bonding MOs. Conversely, the dioxolene oxygen lone pair orbitals are low-lying (oxygen is electronegative) and mix with the metal eg orbitals to produce σ ∗ MOs. Slightly higher in energy than these metal-based MOs are dioxolene-based MOs of metal-π ∗ -antibonding character. Since the ligand-metal orbital mixing is moderate, the resulting MOs retain metalbased and ligand-based character, while still allowing measurable interaction between metal and ligand. This fortuitous separation of metal and ligand orbital energies accounts for the possibility of simultaneous electron transfer and spin-crossover. If metal–ligand mixing were extensive, electron transfer would be impossible because the metal-based and ligand-based orbitals would loose their individuality – the electrons would be delocalized. If metal–ligand mixing were negligible, spin crossover would be impossible because the metal-centered orbitals would be neither bonding nor antibonding with respect to dioxolene orbitals. The net result, illustrated in Fig. 10, is a set of metal-based and ligand-based orbitals whose close energy spacings allow electron transfer, and the differential bonding character of the metal-based orbitals permits SC. Typical bond lengths are given in Table 1 for VT complexes with N–N ligands shown in Fig. 11. As can be, ls-CoIII –O and ls-CoIII –N bond lengths are 0.1–0.15 Å

Fig. 9. Dioxolene and metal frontier orbitals for M(DBQ)2 (N–N).

288

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 10. Relative frontier orbital energies.

Fig. 11. N–N ligands for complexes in Table 1 [60, 63, 64]. Table 1. Bond lengths for VT complexes. N–N ligands shown in Fig. 11. Dioxolene ligand 3,5-DBSQ 3,5-DBCat 3,6-DBSQ 3,6-DBCat 3,6-DBSQ 3,6-DBSQ 3,6-DBSQ 3,6-DBSQ

Metal ion configuration

Ancillary ligand

Co–O (Å)

Co–N (Å)

Ref.

ls-CoIII

bpy

1.851–1.906

1.940–1.957

60

ls-CoIII

tmeda

1.852–1.899

2.021–2.031

63

hs-CoII

dafl

2.005–2.050

2.278–2.518

63

hs-CoII

NO2 phen

2.066

2.165

63, 77

shorter than the corresponding bond lengths for hs-CoII . This is because the ls-CoIII form has the metal electronic configuration (πyz )2 (πxz )2 (πxy )2 , and the hs-CoII form has the metal electronic configuration (πyz )2 (πxz )2 (πxy )1 (σx∗2 −y2 )1 (σz∗2 )1 , see Fig. 9. Occupation of the (σx∗2 −y2 and σz∗2 orbitals occurs only in the hs-CoII form and results in longer Co–N/Co–O bond lengths. As presented in the next section, the differences in metal–ligand bonding accounts for the large values of S ◦ .

8.2.3

VT Thermodynamics

Now that we have introduced a simple MO picture and pointed out salient bonding interactions within each MO, the stage is set for understanding the thermodynamics of the VT equilibrium and correlating thermodynamic parameters to features of our MO description.

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

289

Typically, H ◦ must be on the order of kT for both tautomers to be thermally accessible, and the equilibrium between ls-CoIII (SQ)(Cat)N–N and hs-CoII (SQ)2 is characterized by a positive H ◦ because the more Lewis acidic ls-CoIII has stronger metal-ligand bonds. Typical ls-CoIII –O (O = dioxolene oxygen) bond lengths are ∼1.85 Å, while those for hs-CoII –O are ∼2.05 Å. This is corroborated by the MO picture: metal electrons completely fill metal–ligand bonding orbitals in the ls-CoIII tautomer, while σ ∗ orbitals are occupied in the hs-CoII tautomer. The VT equilibrium is also characterized by a large, positive S ◦ . The origins of the positive S ◦ are primarily electronic and vibrational, as solvent reorganizational entropy is predicted to be very small [65]. Thus, S ◦ = Sel + Svib Before calculating the electronic term, a few comments on magnetic coupling of SQ and metal ion spins are in order. Exchange coupling between semiquinones and transition metal ions has been thoroughly studied [57, 58, 66–75]. Whether antiferromagnetic (low spin) or ferromagnetic (high spin) coupling is observed depends on metal orbital occupation and symmetry [11]. Basically, for octahedral metal complexes, a semiquinone spin is ferromagnetically coupled to unpaired electrons in the eg orbitals and antiferromagnetically coupled to unpaired electrons in the t2g orbitals. Recall that the SQ SOMO and eg orbitals are orthogonal, while the SQ SOMO and the t2g orbitals are of the same symmetry. The latter situation simultaneously gives rise to π-bonding and antiferromagnetic coupling since a bond is the quintessential example of antiferromagnetically-coupled electrons. The origin of ferromagnetic coupling between a SQ spin and unpaired metal electrons in the eg set can be traced back to Hund’s rule: electron-electron repulsion is reduced for spin-aligned electrons in orthogonal yet overlapping orbitals. For hs-CoII –SQ coupling, Hendrickson and Noodleman have calculated a rather substantial antiferromagnetic exchange parameter (J = −1700 cal mol−1 ), indicating that the SQ-π antiferromagnetic contribution is substantially greater than SQ-σ ∗ ferromagnetic contribution. This calculated J is much larger than that measured for the tetramer [Co4 (3,5-DBSQ)8 ], for which J = −86 cal mol−1 [72]. Increased entropy for the hs-CoII tautomer is expected due to (1) hs-metal ion configuration, and (2) spin–spin coupling between CoII and ligand spins. High-spin CoII has three unpaired electrons, and the two SQ ligands have one each. Interaction amongst the spins gives rise to 4 × 2 × 2 = 16 states: a sextet (S = 5/2), two quartets (2 × S = 3/2), and a doublet (S = 1/2). The doublet is the lowest-energy state due to antiferromagnetic coupling as described in the previous paragraph, and the electronic degeneracy (ghs ) for the CoII tautomer is 16. The CoIII tautomer on the other hand is a ground-state doublet with mixed-valent dioxolene ligands (SQ-Cat ↔ Cat-SQ). The electronic degeneracy (gls ) for this tautomer is therefore 2 × 2 = 4, and the electronic contribution to the entropy change can be calculated as, Sel = R ln(ghs /gls ) = R ln(16/4) ≈ 3 cal K−1 mol−1 As we will show below, estimated values for S ◦ are typically greater than 25 cal K−1 mol−1 , thus the electronic term makes only a small contribution to S ◦ .

290

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 12. Cartoons of ls-CoIII and hs-CoII potential wells.

As in SC, the antibonding character of the σx∗2 −y2 and σz∗2 orbitals is critically important for VT, as they are the origin of the large positive S ◦ . Consider the energy diagrams shown in Fig. 12. The CoII tautomer, with its occupied antibonding orbitals has a much greater density of vibrational levels, owing to the weaker metal– ligand bonds. The vibrational contribution to the entropy change may be written as: (Svib = R ln (1 − exp(−νls /RT )/(1 − exp(−νhs /RT ) where [1 − exp(−νls /RT )] represents the product of the vibrational partition functions for ls-CoIII . Considering only the metal ion and the immediate coordination sphere, there are 3(7) − 6 = 15 normal modes to consider. For Svib = 22 cal K−1 mol−1 , the ratio of the products of the partition functions must be ca. 60 000! This corresponds to a striking decrease (ca. twofold) in vibrational frequencies in the hs-CoII tautomer. As noted in the Introduction, large decreases in metal– ligand stretching frequencies have been measured in hs-FeII SC complexes relative to ls forms. In VT complexes, additional low-frequency ligand modes might also contribute to Svib [1].

8.2.4

Experimental Determination of Thermodynamic Parameters

Quite often, magnetic or optical data can be transformed to illustrate the fraction of CoII as a function of temperature. Consider the generic VT equilibrium, ls-CoIII (SQ)(Cat)(N–N)

K VT

hs-CoII (SQ)2 (N–N)

If x is the fraction of CoII complexes, (1 − x) is the fraction of CoIII complexes, (χ T )exp is the measured paramagnetic susceptibility–temperature product, (χ T )Co(II) is the paramagnetic susceptibility–temperature product for hs-CoII , and (χ T )Co(III) is the paramagnetic susceptibility–temperature product for ls-CoIII , then:

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

291

(χ T )exp = x(χ T )Co(II) + (1 − x)(χ T )Co(III) = x(χ T )Co(II) + (χ T )Co(III) − x(χ T )Co(III) = x[(χ T )Co(II) − (χ T )Co(III) ] + (χ T )Co(III) so: x = [(χ T )exp − (χ T )Co(III) ]/[(χ T )Co(II) − (χ T )Co(III) ] Therefore, if the (χ T )exp is measured and if (χ T )Co(II) and (χ T )Co(III) are known (or estimated), a plot of x = f (T ) can be generated. Furthermore, since K VT = [hs-CoII ]/[ls-CoIII ] = exp(−G ◦ /RT ), then: K VT = x/(1 − x) = exp(−H ◦ /RT + S ◦ /R) and x = exp(−H ◦ /RT + S ◦ /R) − x[(−H ◦ /RT + S ◦ /R)] so: x[1 + exp(−H ◦ /RT + S ◦ /R)] = exp(−H ◦ /RT + S ◦ /R) and x = exp(−H ◦ /RT + S ◦ /R)/[1 + exp(−H ◦ /RT + S ◦ /R)] or x = 1/[exp(H ◦ /RT − S ◦ /R) + 1] Shown in Fig. 13 are plots of x = f (T ) for H ◦ = 5 kcal mol−1 and S ◦ = 15, 20, 25, and 30 e. u..

Fig. 13. Dependence of mole fraction of hs-CoII tautomer on S ◦ .

292

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 14. N–N ligands for VT complexes in Table 2 [1, 76–80].

The critical temperature, T1/2 , is defined as the temperature where [CoII ] = [CoIII ], or x = 0.5. It is obvious that the equilibrium is entropy-driven. Experimentally, x = f (T ) plots can be generated from variable-temperature magnetic data, and the data fit to extract the thermodynamic parameters. Table 2 shows thermodynamic parameters for several CoIII /CoII VT complexes with different N–N ligands shown in Fig. 14.

8.2.5

Dependence of K VT Equilibrium on Ancillary Ligands

In 1980, Pierpont reported the structure and properties of Co(3,5-DBSQ)(3,5DBCat)(2,2 -bipyridine) · 0.5 toluene [60]. Since that time, a considerable amount of work has been reported on the mechanism and structure-property relationships for VT complexes. The majority of reports concern synthesis and characterization of CoII /CoIII VT compounds that differ in the chemical structure of the ancillary ligands. The thermodynamic parameters for KVT are intimately linked with the structure of the ancillary ligand.

8.2.5.1

Redox Potential

Hendrickson and coworkers found that the critical temperature of a series of complexes depends on the electronic structure of ancillary, aromatic N–N ligands [78, 81]. As estimated from variable-temperature electronic absorption spectra, the T1/2 values decrease in the order ∼348 K (dpbpy, Entry 7, Table 2), ∼298 K (dmbpy, Entry 6, Table 2), ∼273 K (bpy, Entry 5, Table 2), ∼230 K (phen, Entry 4, Table 2), ∼190 K (bpym, Entry 12, Table 2), and <190 K (bpyz, Entry 13, Table 2) [81]. As noted by these authors, this is the same order as decreasing reduction potential. These data are shown graphically in Fig. 15. The reason that the critical temperatures depend on reduction potential lies in the fact that more positive reduction potentials signify low-lying N–N π ∗ LUMOs.

293

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

Table 2. Critical temperatures and thermodynamic parameters for selected VT complexes. Entry

N–N

Dioxolenesa

T1/2 (K)

H ◦ (kcal mol−1 )

S ◦ (cal K−1 mol−1 )

Ref.

1

tmmda

3,6-DBQ

–

–

76, 82

2

tmeda

3,6-DBQ

–

–

3

tmpda

3,6-DBQ

phen

3,5-DBQ

5

bpy

3,5-DBQ

– 3.39 6.40 – 8.47

– 19.1 28.2 – 31.8

63, 82 76 76, 82

4

6 7 8

dmbpy dpbpy py2 O

3,5-DBQ 3,5-DBQ 3,6-DBQ

9.17 5.10 –

32.0 14.5 –

1 77 1, 79 63 1 1 80

9

py2 S

3,6-DBQ

–

–

80

10

py2 Se

3,6-DBQ

py2 Te

3,6-DBQ

12 13

bpym bpyz

3,5-DBQ 3,5-DBQ

– 12.4f – 8.37f – –

– 41.6f – 39.0f – –

80

11

280 >400b 310 >400b ca. 110 178b 226.6 240b,e 277.0 327b 286.6 350.0 270f 110 (T1/2 ↓)b,c 330 (T1/2 ↑)b,d 255f 370b 225f 290b <200f 210b 190 <190

80 1 1

a

DBQ denotes the catecholate/semiquinone forms of di-t-butylorthobenzoquinone. T1/2 determined on a solid sample. c T 1/2 determined while cooling the sample. d T 1/2 determined while heating the sample. e Toluene solvate. f C.G. Pierpont, personal communication. b

The N–N π ∗ LUMO is of the correct symmetry to interact with π -type (t2g ) metal orbitals. Since the empty N–N π ∗ orbital interacts only weakly with the occupied metal orbital, the overall binding energy is lowered while maintaining a small ligandfield-splitting necessary to sustain a hs-CoII ion. Thus, π -back-bonding stabilizes the electron rich hs-CoII tautomer and lowers T1/2 . This idea has been further developed using spin-polarized density functional calculations by Hendrickson and Noodleman [61].

8.2.5.2

Torsional/Vibrational Rigidity

Pierpont and coworkers reported a fascinating series of complexes that undergo VT in both the solid state and in solution (Entries 8–11, Table 2) [80]. As can be seen in

294

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 15. Dependence of T1/2 on N–N reduction potential.

Fig. 16. Magnetic moments for VT complexes having dipyridyl ether ancillary ligands. Reprinted with permission from Ref. [80]. Copyright 1997 American Chemical Society.

Fig. 16, the series of py2 S-, py2 Se-, and py2 Te-containing complexes, T1/2 decreases in steps of 80 K in the solid state and in steps of 25 K in the solution. These results are consistent with the redox potential dependence presented in Section 2.5.1: if chalcogen π-donating ability decreases from O to Te, then the π -

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

295

acceptor ability should increase due to the well-known affect of electron donors on π -MO energies (donors increase EHOMO and ELUMO , acceptors decrease EHOMO and ELUMO ). Thus as a result of π ∗ interaction, the hs-CoII tautomer should be stabilized most in the py2 Te complex and least in the py2 O complex and T1/2 should be lowest for the latter complex. The py2 O-containing complex does not fit the observed trend in T1/2 , however, if py2 O is a weaker π -acceptor (increasing T1/2 ), then the pyridyl nitrogens may also be weaker σ -donors due to the electronegativity of oxygen. The latter effect would decrease T1/2 . Thus, for the py2 O-containing complex, the opposing σ - and π-affects might cancel. Because of the lower electronegativities for the other chalcogens, this σ -effect is inoperative hence the observed trend in T1/2 . In addition, differences in carbon-chalcogen bond lengths will cause a change in bite angle which could also affect H ◦ in this series. In addition to the trend in T1/2 values, Pierpont reports huge hysteresis, shown in Fig. 17 for the py2 O-containing compound. In fact, they report two different crystal structures, shown in Figs. 18 and 19, having CoIII (3,6-DBCat)(3,6-DBSQ)(py2 O) and CoII (3,6-DBSQ)2 (py2 O) charge distributions! Interestingly, the smaller ls-CoIII tautomer is chelated by a folded py2 O ligand, while the larger hs-CoII tautomer is chelated by a planar py2 O ligand. In the solid state, intermolecular interactions between all folded py2 O ligands are different from interactions between all planar py2 O ligands (Fig. 20) and a large hysteresis results (hysteresis = 220 K, see Table 2). Prior to this report only one VT complex exhibited a small hysteresis of ca. 5 K [78]. Thus, Pierpont’s group has found an excellent bistable VT molecule.

Fig. 17. Hysteresis loop for Co(3,6-DBQ)2 (py2 O). Reprinted with permission from Ref. [80]. Copyright 1997 American Chemical Society.

296

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 18. Crystal structure for Co(3,6-DBSQ)2 (py2 O). Reprinted with permission from Ref. [80]. Copyright 1997 American Chemical Society.

Fig. 19. Crystal structure for Co(3,6-DBSQ)(3,6-DBCat)(py2 O). Reprinted with permission from Ref. [80]. Copyright 1997 American Chemical Society.

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

297

Fig. 20. Py2 O stacking in solid Co(3,6-DBSQ)2 (py2 O). Reprinted with permission from Ref. [80]. Copyright 1997 American Chemical Society.

8.2.5.3

Chain Length

Pierpont reported the preparation and characterization of VT complexes with α,ωdiaminoalkane ancillary ligands (Entries 1–3, Table 2) [63, 76, 82]. As seen in Table 2, and Fig. 21, the solution and solid-state T1/2 values are dramatically higher for the shorter chelating ligands, tmmda and tmeda. However for tmpda that forms a sixmembered ring chelate, the T1/2 is 178 K in the solid state. Pierpont ascribes the lower T1/2 value for Co(3,6-DBQ)2 (tmpda) to increased entropy changes in the case of the tmpda ligand. He notes that there is considerable disorder in the ancillary ligand in the crystal structure, that shows hs-CoII (3,6DBSQ)2 (tmpda) charge distributions at both 298 and ca. 150 K. No structural determination was made at temperatures low enough to observe the CoIII tautomer. Perhaps differences in metal-ligand bonding in this series affect changes in T1/2 . The origin of differences in bonding could lie in the different chelate ring sizes.

8.2.6

Pressure-induced VT

Hendrickson and Verdaguer studied the effect of temperature and pressure on KV T for solid samples of solvated and non-solvated hs-CoII (3,5-DBSQ)2 (phen) [83]. The solvated complex has T1/2 = 240 K, while the non-solvated form does not convert to the ls-CoIII tautomer even at 2 K. However, both solids could be converted reversibly to the smaller ls-CoIII tautomer with application of pressure. The solvated sample

298

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 21. Magnetic moments as a function of temperature for VT complexes with saturated diamine ligands. Figure taken from Ref. [76].

Fig. 22. Effect of temperature and pressure on thermodynamic parameters for VT complexes. Reprinted with permission from Ref. [83]. Copyright 1996 American Chemical Society.

exhibited a P1/2 of 0.37 GPa at 298 K, while the non-solvated sample exhibited a much higher P1/2 of 1.1 GPa also at 298 K, consistent with the variable-temperature results. The effect of both temperature and pressure on the thermodynamic parameters is shown in Fig. 22.

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

8.2.7

299

Light-induced VT and Rates of VT

In 1995 and 1996, Hendrickson reported studies of the photophysics of several VT complexes [1, 84]. A toluene solution of the ls-CoIII tautomer of Co(3,5-DBSQ)(3,5DBCat)(dpbpy) was exposed to laser excitation at 532 nm (90 ps pulse width), resulting in the formation of the LMCT excited state. A transient bleach was noted at 600 nm (λmax of the ls-CoIII tautomer) concomitant with a transient absorption at 720 nm (λmax of the hs-CoII tautomer). Satisfactory fits to the transient bleach and absorption were achieved using the same time constant, τobs = 1.2 ns (kobs = 8.3 × 108 s−1 = 1/τobs ), thus linking the decay of the ls-CoIII tautomer to the formation of the hs-CoII tautomer. No rise time for the CoII tautomer was observed, thus it is formed within the lifetime of the pulse and kisc > 1010 s−1 . Furthermore, because: K VT = kfvt /kbvt

and

kobs = kfvt + kbvt

then: kobs = K VT kbvt + kbvt and kobs = kbvt [K VT + 1] so: 1/kbvt = (1/kobs )[K VT + 1] and τbvt = (obs [K VT + 1] For (Co(3,5-DBSQ)(3,5-DBCat)(dpbpy) at 298 K, K VT = 0.24, so τbvt = 1.2 ns × (1.24), = 1.49 ns (kbvt = 6.7×108 s−1 ). For this complex, and for (Co(3,5-DBSQ)(3,5DBCat)(dmbpy) the following Jablonski diagram in Fig. 23 was constructed. Similar experiments for the two other complexes studied were not possible since they are in their CoII forms at 298 K.

Fig. 23. Jablonski Diagram for VT complexes.

300

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Table 3. Arrhenius parameters for decay of hs-CoII tautomers. DBQ denotes the catecholate/semiquinone forms of di-t-butylorthobenzoquinone. N–N

Dioxolene

E a (cm−1 )

A

phen bpy dmbpy dpbpy

3,5-DBQ 3,5-DBQ 3,5-DBQ 3,5-DBQ

858 582 1088 856

1.01 × 109 5.58 × 108 1.26 × 1010 4.32 × 1010

Hendrickson and Adams then completed variable-temperature (150–300 K) nanosecond transient absorption measurements for the four complexes shown in Table 3. Using an established model for radiationless decay, they extracted Arrhenius parameters for kbvt , also compiled in Table 3. At lower temperatures, the relaxation rate was found to be temperature-independent, consistent with quantum mechanical tunneling, and accounted for by their mechanism. Though there are no obvious correlations between Arrhenius parameters and ancillary ligands, there are important conclusions from this work. First, the mechanistic details of VT are very similar to SC, but the limiting factor in the rate of relaxation of hs-CoII to ls-CoIII is orthogonality of the SQ π ∗ MO and the metal σ ∗ (eg ) orbitals – the MOs involved in electron transfer from metal to ligand. In contrast, relaxation of hs-FeII to ls-FeII is limited by spin forbiddeness. Second, isosbestic points observed in both steady-state and time-resolved absorption experiments confirm the involvement of only two species in the tautomeric equilibrium, i. e., no ls-CoII forms are involved. The light-induced conversion of the ls-CoIII (S = 1/2) form to the hsCoII form is spin-allowed since hs-CoII -SQ exchange coupling produces the S = 1/2 ground state of the hs-CoII form. The implications of this study are quite significant if VT is to be used as a mechanism to create bistable compounds. In complexes such as those studied by Hendrickson and Adams, VT can only be controlled if fundamental changes are made in the chemical composition of the complexes. That is, bistability can only be achieved if the spin-allowed feature of the relaxation is removed. Note that it would be difficult to control the rate of relaxation by addressing issues of electron transfer between metal and ligand since the orbitals involved are already orthogonal. Thus, slowing kbvt and creating bistability can only be addressed by having different spin multiplicities in the ls-CoIII and hs-CoII forms. An example of this is described in the next section.

8.2.8

VT Complexes of Other Quinone Ligands and Redox Chemistry of VT Complexes

In 1998, Dei and coworkers reported VT in the complex shown in Fig. 24, with H ◦ = 10.0 kcal mol−1 and S ◦ = 33.5 cal K−1 mol−1 in toluene solution [85]. The complex features two different oxidation states of an iminoquinone ligand first prepared by Balch [86]. As shown below, the Cat-N-SQ form of the ligand is a radical dianion, while the Cat-N-BQ form is a diamagnetic anion.

8.2 Valence Tautomerism in Dioxolene Complexes of Cobalt

301

Fig. 24. Iminoquinone complexes.

This finding is important for three reasons. First, the complex represents a new structural type for cobalt VT compounds, and suggests that the Co(dioxolene)2 (N–N) formulas are not the only complexes to exhibit VT. Second, the magnetic properties of each tautomer are simple compared to previously described VT complexes, and the ground-state multiplicities of each tautomer are different. The magnetic simplicity lies in the fact that the hs-CoII tautomer is bound to two diamagnetic (S = 0) Cat-N-BQ ligands. Thus, the only electronic state to consider in the hs-CoII tautomer is that of the S = 3/2 metal ion, while the ls-CoIII tautomer has a single, ligand-centered unpaired electron. The fact that the CoIII (Cat-N-BQ)(Cat-N-SQ) and CoII (Cat-N-BQ)2 ground states are of different multiplicities means that the rate of relaxation (kbvt ) should decrease relative to the compounds discussed in the previous section. Thus, iminoquinone ligands might be used to control VT and create bistable materials. Third, the well-known redox activity of iminoquinone ligands suggests the possibility of modulating VT parameters through ligand oxidation or reduction. Redox modulation of VT has been hinted at in[Cp2 Co][CoIII (3,5-DBCat)2 (bpy)], shown in Fig. 25, prepared by cobaltocene reduction of CoIII (3,5-DBCat) (3,5-DBSQ)(bpy) [87]. As can be seen, redox chemistry and thermal activation yields four compounds with different optical and magnetic properties. There are two interesting, important findings in this report. First, from the reported variable-temperature electronic absorption spectra, T1/2 for [Cp2 Co][CoIII (3,5-DBCat)2 (bpy)] in solution appears to be about 30 K lower than for CoIII (3,5-DBCat)(3,5-DBSQ)(bpy), indicating enhanced stability of the reduced hs-CoII form. Second, ferromagnetic exchange between metal and ligand is reported contrary to some other hs-CoII -SQ couplings. Indeed, the magnetic moment of a solid sample of [Cp2 Co][CoII (3,5-DBSQ)(3,5-DBCat)(bpy)]

302

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

Fig. 25. New VT complexes prepared by reduction of existing VT complexes.

varies from 4.24 µB at 320 K to 4.04 µB at 70 K but then increases to 5.14 µB at 20 K. Not only is this data consistent with ferromagnetic exchange, but in the solid state the ls-CoIII form is not observed. Also contrary to CoIII (3,5-DBCat)(3,5-DBSQ)(bpy), the hs-CoII form of the reduced complex should exhibit the intervalence transition for the mixed-valent dioxolene ligands; however, spectra were not reported past 820 nm. Redox modulation of VT is a promising area of research, and a full study of this fascinating series is anxiously awaited.

8.2.9

Polymeric VT Materials

Pierpont has reported several examples of materials that form an extended structure due to either coordination polymerization, or metal-metal interactions induced by irradiation [51, 55, 88, 89], including the cobalt-containing coordination polymers having the structures shown in Fig. 26 [88, 89]. The n-hexane-solvated pyrazine (pyz) polymer exists as the hs-CoII tautomer at 350 K, and in the ls-CoIII form at 298 K. The material crystallizes as thin plates, with

Fig. 26. Polymeric VT complexes.

8.3 Future Directions

303

the polymer strands perpendicular to the surface of the crystal. Irradiation causes bending of the crystals due to elongation of Co–N bonds along the polymer main chain, similar to the “photomechanical effect” observed for crystals of Rh(CO)2 (3,6DBSQ) [55]. Pierpont reasons that increasing Co–N by 0.2 Å should cause a 0.06 mm elongation per millimeter of polymer length. However, due to the crystal morphological differences in Rh(CO)2 (3,6-DBSQ) and Co(3,6-DBSQ)(3,6-DBCat)(pyz), the bending effect is smaller in the latter material. This coordination polymer also degrades due to solvent loss, but the finding is still significant due to the illustration of rational design of a polymeric material that exhibits VT.

8.3

Future Directions

VT is a fascinating area of research. Future developments will include the use of new redox-active molecules, like the bis(semiquinone) ligands shown in Fig. 27 [90–96]. Various features of these ligands could allow the VT equilibrium to be tuned, and the dinuclear ligands will facilitate the formation of coordination polymers. New ligands can also be formed by ligand-centered reduction or oxidation of existing complexes, as demonstrated by Hendrickson and coworkers [87]. A second area for further study involves generating and controlling hysteresis. Earlier work demonstrated the need for solvate molecules that “soften” the lattice and facilitate intermolecular interactions [77]. Design principles that focus on solvate structure-property relationships might prove tedious, and there are other mechanisms to generate and/or enhance hysteresis. Perhaps the most obvious mechanism for generating hysteretic VT complexes is that demonstrated by Pierpont and coworkers: incorporating an ancillary ligand conformational change into the VT equilibrium [80]. This is an area that might prove particularly fruitful.

Fig. 27. New SQ-type ligands.

304

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

References [1] Adams, D.M.; Hendrickson, D.N. J. Am. Chem. Soc. 1996, 118, 11515. [2] Molecular Electronics and Molecular Electronic Devices; Sienick, K., Ed.; CRC Press: Boca Raton, FL, 1994. [3] Aviram, A. Int. J. Quant. Chem. 1992, 42, 1615. [4] Kahn, O.; Krobert, J.; Jay, C. Adv. Mater. 1992, 4, 718. [5] Molecular Electronics: Materials and Methods; Lazarev, P.I., Ed.; Kluwer Academic Publishers: Drodrecht, The Netherlands, 1991. [6] Kahn, O. Conjugated Polymeric Materials: Oppurtunities in Electronics, Optoelectronics, and Molecular Electronics; Kluwer Academic Publishers: Netherlands, 1990. [7] Hush, N.S.; Wong, A.T.; Bacskay, G.B.; Reimers, J.R. J. Am. Chem. Soc. 1990, 112, 4192. [8] Joachim, C.; Launay, J.P. J. Mol. Electron. 1990, 6, 37. [9] Kahn, O.; Launay, J.P. Chemtronics 1988, 3, 140. [10] Launay, J.P. Molecular Electronic Devices II; Marcel Dekker: New York, 1987. [11] Kahn, O. Molecular Magnetism; VCH: New York, 1993. [12] Astruc, D. Electron Transfer and Radical Processes in Transition-Metal Chemistry; WileyVCH: New York, 1995. [13] Hendrickson, D.N. In NATO ASI Series; K. Prassides, Ed.; Kluwer Publishing Co.: Dordrecht, The Netherlands, 1991; pp 67. [14] Richardson, D.E.; Taube, H. Coord. Chem. Rev. 1984, 60, 107. [15] Creutz, C. In Prog. Inorg. Chem.; S. J. Lippard, Ed.; Wiley: New York, 1983; Vol. 30; pp 1. [16] Day, P. Int. Rev. Phys. Chem. 1981, 1, 149. [17] Robin, M.B.; Day, P. Adv. Inorg. Radiochem. 1967, 10, 247. [18] Gutlich, ¨ P.; Dei, A. Angew. Chem. Int. Ed. Engl. 1997, 37, 2734. [19] Zarembowitch, J. New J. Chem. 1992, 16, 255. [20] Toftlund, H. Coord. Chem. Rev. 1989, 94, 67. [21] Maeda, Y.; Takashima, Y. Comments Inorg. Chem. 1988, 7, 41. [22] Bacci, M. Coord. Chem. Rev. 1988, 86, 245. [23] Beattie, J.K. Adv. Inorg. Chem. 1988, 32, 1. [24] Konig, E. Prog. Inorg. Chem. 1987, 35, 527. [25] Rao, C.N.R. Int. Rev. Phys. Chem. 1985, 4, 19. [26] Gutlich, ¨ P. Struct. Bonding (Berlin) 1981, 44, 83. [27] Gutlich, ¨ P.; Hauser, A.; Spiering, H. Angew. Chem. Int. Ed. Engl. 1994, 33, 2024. [28] Utamapanya, S.; Rajca, A. J. Am. Chem. Soc. 1991, 113, 9242. [29] Le Mest, Y.; L’Her, M.; Hendricks, N.H.; Kim, K.; Collman, J.P. Inorg. Chem. 1992, 31, 835. [30] Bonvoisin, J.; Launay, J.P.; Rovira, C.; Veciana, J. Angew. Chem. Int. Ed. 1994, 33, 2106. [31] Bonvoisin, J.; Launay, J.P.; Vanderauweraer, M.; DeSchryver, F.C. J. Phys. Chem. 1994, 98, 5052. [32] Domingo, V.M.; Castaner, J. J. Chem. Soc., Chem. Commun. 1995, 895. [33] Rajca, S.; Rajca, A. J. Am. Chem. Soc. 1995, 117, 9172. [34] Sedo, ´ J.; Ruiz, D.; VidalGancedo, J.; Rovira, C.J.; Bonvoisin, J.; Launay, J.P.; Veciana, J. Adv. Mat. 1996, 8, 748. [35] Bonvoisin, J.; Launay, J.P.; Verbouwe, W.; Vanderauweraer, M.; DeSchryver, F.C. J. Phys. Chem. 1996, 100, 17079. [36] Sedo, ´ J.; Ruiz, D.; VidalGancedo, J.; Rovira, C.J.; Bonvoisin, J.; Launay, J.P.; Veciana, J. Syn. Met. 1997, 85, 1651.

References

305

[37] Domingo, V.M.; Castaner, ˜ J.; Riera, J.; Brillas, E.; Molins, E.; Martinez, B.; Knight, B. Chem. Mater. 1997, 9, 1620. [38] Sedo, ´ J.; Ruiz, D.; VidalGancedo, J.; Rovira, C.J.; Bonvoisin, J.; Launay, J.P.; Veciana, J. Mol. Crys. Liq. Crys. 1997, A 306, 125. [39] Karafiloglou, P.; Launay, J.P. J. Phys. Chem. A 1998, 102, 8004. [40] Coronado, E.; Galan-Mascaros, J.R.; Gimenez-Saiz, C.; Gomez-Garcia, C.J.; Triki, S. J. Am. Chem. Soc. 1998, 120, 4671. [41] Shultz, D.A.; Lee, H.; Kumar, R.K.; Gwaltney, K.P. J. Org. Chem. 1999, 64, 9124. [42] Hush, N.S. Prog. Inorg. Chem. 1967, 8, 391. [43] Gutlich, ¨ P.; Hauser, A. Coord. Chem. Rev. 1991, 97, 1. [44] Shultz, D.A.; Boal, A.K.; Farmer, G.T. J. Am. Chem. Soc. 1997, 119, 3846. [45] Berson, J.A. Acc. Chem. Res. 1997, 30, 238. [46] Bush, L.; Maksimovic, L.; Feng, X.; Lu, H.; Berson, J. J. Am. Chem. Soc. 1997, 119, 1416. [47] Takemoto, J.H.; Hutchinson, B. Inorg. Chem. 1973, 12, 705. [48] Sorai, M.; Seki, S. J. Phys. Chem. Solids 1974, 35, 555-570. [49] Caneschi, A.; Dei, A. Angew. Chem. Int. Ed. 1998, 37, 3005. [50] Attia, A.S.; Pierpont, C.G. Inorg. Chem. 1998, 37, 3051. [51] Attia, A.S.; Pierpont, C.G. Inorg. Chem. 1997, 36, 6184. [52] Attia, A.S.; Pierpont, C.G. Inorg Chem. 1995, 34, 1172. [53] Lynch, M.W.; Hendrichson, D.N.; Fitzgerald, B.J.; Pierpont, C.G. J. Am. Chem. Soc. 1984, 106, 2041. [54] Lynch, M.W.; Hendrickson, D.N.; Fitzgerald, B.J.; Pierpont, C.G. J. Am. Chem. Soc. 1981, 103, 3961. [55] Lange, C.W.; Folde ¨ aki, ` M.; Nevodchikov, V.I.; V.K., C.; Abakumov, G.A.; Pierpont, C.G. J. Am. Chem. Soc. 1992, 114, 4220. [56] Abakumov, G.A.; Razuvaev, G.A.; Nevodchikov, V.I.; Cherkasov, V.K. J. Organomet. Chem. 1988, 341, 485. [57] Pierpont, C.G.; Lange, C.W. Prog. Inorg. Chem. 1994, 41, 331. [58] Pierpont, C.G.; Buchanan, R.M. Coord. Chem. Rev. 1981, 38, 45. [59] Kaim, W. Coord. Chem. Rev. 1987, 76, 187. [60] Buchanan, R.M.; Pierpont, C.G. J. Am. Chem. Soc. 1980, 102, 4951. [61] Adams, D.M.; Noodleman, L.; Hendrickson, D.N. Inorg. Chem. 1997, 36, 3966. [62] Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chim. Acta 1989, 163, 99. [63] Jung, O.-S.; Pierpont, C.G. Inorg. Chem. 1994, 33, 2227. [64] Jung, O.-S.; Pierpont, C.G. J. Am. Chem. Soc. 1994, 116, 1127. [65] Dei, A. Inorg. Chem. 1993, 32, 5730. [66] Dei, A.; Gatteschi, D. Inorg. Chim. Acta 1992, 189, 125. [67] Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1990, 29, 3409. [68] Benelli, C.; Dei, A.; Gatteschi, D.; Gudel, ¨ H.; Pardi, L. Inorg. Chem. 1989, 28, 3089. [69] Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1988, 27, 2831. [70] Kahn, O.; Reedijk, J.; Prins, R.; Thompson, J.S. Inorg. Chem. 1987, 26, 3557. [71] Lynch, M.W.; Valentine, M.; Hendrickson, D.N. J. Am. Chem. Soc. 1982, 104, 6982. [72] Lynch, M.W.; Buchanan, R.M.; Pierpont, C.G.; Hendrickson, D.N. Inorg. Chem. 1981, 20, 1038. [73] Kessel, S.L.; Emberson, R.M.; Debrunner, P.G.; Hendrickson, D.N. Inorg. Chem. 1980, 19, 1170. [74] Buchanan, R.M.; Kessel, S.L.; Downs, H.H.; Pierpont, C.G.; Hendrickson, D.N. J. Am. Chem. Soc. 1978, 100, 7894. [75] Buchanan, R.M.; Downs, H.H.; Shorthill, W.B.; Pierpont, C.G.; Kessel, S.L.; Hendrickson, D.N. J. Am. Chem. Soc. 1978, 100, 4318.

306

8 Valence Tautomerism in Dioxolene Complexes of Cobalt

[76] Jung, O.; Jo, D.; Lee, Y.; Sohn, Y.; Pierpont, C.G. Angew. Chem. Int. Ed. Engl. 1996, 35, 1694. [77] Adams, D.M.; Dei, A.; Rheingold, A.R.; Hendrickson, D.N. J. Am. Chem. Soc. 1993, 115, 8221. [78] Adams, D.M.; Dei, A.; Rheingold, A.L.; Hendrickson, D.N. Angew. Chem. Int. Ed. Engl. 1993, 32, 880. [79] Pierpont, C.G.; Jung, O.-S. Inorg. Chem. 1995, 34, 4281. [80] Jung, O.-S.; Jo, D.H.; Lee, Y.-L.; Conklin, B.J.; Pierpont, C.G. Inorg. Chem. 1997, 37, 19. [81] Hendrickson, D.N.; Adams, D.M.; Dei, A.; Rheingold, A.L. J. Am. Chem. Soc. 1993, 115, 8221. [82] Jung, O.-S.; Jo, D.W.; Lee, Y.-A.; Sohn, Y.S.; Pierpont, C.G. Inorg. Chem. 1998, 37, 5875. [83] Roux, C.; Adams, D.M.; Itie, ´ J.P.; Polian, A.; Hendrickson, D.N.; Verdaguer, M. Inorg. Chem. 1996, 35, 2846. [84] Adams, D.M.; Li, B.; Simon, J.D.; Hendrickson, D.N. Angew. Chem. Int. Ed. Engl. 1995, 34, 1481. [85] Caneschi, A.; Cornia, A.; Dei, A. Inorg. Chem. 1998, 37, 3419. [86] Balch, A.L.; Girgis, A.Y. Inorg. Chem 1975, 14, 2724. [87] Ruiz, D.; Yoo, J.; Guzei, I.A.; Rheingold, A.L.; Hendrickson, D.N. Chem. Comm. 1998, 2089. [88] Jung, O.-S.; Lee, Y.-A.; Pierpont, C.G. Syn. Met. 1995, 71, 2019. [89] Jung, O.-S.; Pierpont, C.G. J. Am. Chem. Soc. 1994, 116, 2229. [90] Shultz, D.A.; Bodnar, S.H. Inorg. Chem. 1999, 38, 591. [91] Shultz, D.A.; Lee, H.; Fico, R.M., Jr. Tetrahedron 1999, 55, 12079. [92] Shultz, D.A.; Bodnar, S.H.; Kumar, R.K.; Kampf, J.W. J. Am. Chem. Soc. 1999, 121, 10664. [93] Shultz, D.A.; Boal, A.K.; Lee, H.; Farmer, G.T. J. Org. Chem. 1998, 63, 9462. [94] Shultz, D.A.; Farmer, G.T. J. Org. Chem. 1998, 63, 6254. [95] Shultz, D.A.; Boal, A.K.; Driscoll, D.J.; Farmer, G.T.; Hollomon, M.G.; Kitchin, J.R.; Miller, D.B.; Tew, G.N. Mol. Cryst. Liq. Cryst. 1997, 305, 303. [96] Shultz, D.A.; Boal, A.K.; Driscoll, D.J.; Kitchin, J.R.; Tew, G.N. J. Org. Chem. 1995, 60, 3578.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

9

Molecule-based Magnets Derived from NiII and MnII , Azido Bridging Ligand and Related Compounds Joan Ribas, Albert Escuer, Montserrat Monfort, Ramon Vicente, Roberto Cort´es, ´ Luis Lezama, Teofilo Rojo, and Mohamed A. S. Goher

9.1

Introduction

Designing new high-dimensional magnetic molecular systems built from coordination compounds has recently been a point of attention for inorganic chemists. The variety of coordination chemistry provides the synthesizers with a useful tool to build magnetic molecular architectures interesting for their properties, which arise from the interaction among their subunits. Strategies based on the reaction of appropriate terminal and bridging ligands with paramagnetic metal ions allow the preparation of oligomeric species whose nuclearity and magnetic properties may, in any sense, be tailored. Research on molecular magnetism has focused on: – the synthesis and study of discrete polynuclear molecules, in an attempt to understand the magnetic coupling [1, 2] and the production of new paramagnetic clusters with high spin and strong anisotropy (superparamagnetic molecules) [3, 4]; and – the synthesis and characterization of new one-, two-, or three-dimensional materials which showing ferromagnetic long-range cooperative phenomena (molecular magnets) [5–9]. By using good superexchange bridging groups such as oxalate or cyanide, extended lattices of antiferromagnetic or ferrimagnetic systems which show magnetic order at low temperature have been achieved. In this way, the pseudohalide ligands represent a good choice for the design of new magnetic materials. Among these, the azido ligand is the most versatile in linking divalent metal ions. If NiII and MnII ions are considered, a large number of polynuclear MII complexes with azido bridging ligands have been reported in the literature. These complexes can be classified according to their dimensionality: discrete molecules, one-dimensional, two-dimensional, and three-dimensional systems. When the N− 3 anion acts as bridging ligand, there are two typical coordination modes: end-to-end (E E or 1,3) and end-on (EO or 1,1) (Fig. 1) The end-to-end (EE) coordination mode usually gives antiferromagnetic coupling and the end-on (EO) gives ferromagnetic coupling whereas, for very large M–(µ1,3 N3 )–M bond angles, the magnetic coupling may be reversed. When the azido bridges are exclusively end-to-end or end-on, not always all the bridges are structurally and crystallographically identical: within the same system, two or more different struc-

308

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 1. Common coordination modes of the azido bridging ligand.

tural bridges may be present, owing to the different distances and/or angles found in the structure. These systems, which may be antiferro- or ferromagnetic, belong to the “alternating magnetic systems” category. Thus, the use of two or more exchange coupling parameters (J) is required in order to fit the experimental results, whereas in the “uniform magnetic systems” all the azido bridges have the same coordination mode and the same structural parameters. Finally, for low-dimensional complexes, highly unusual systems simultaneously containing both coordination modes (EE and EO) have been reported. Their activity as alternating ferro–antiferromagnetic (F–AF) coupled systems should also be highlighted. To our knowledge, which coordination mode should be adopted, remains unknown. To decide which is the most suitable method for analyzing the superexchange interaction through an azido bridge has raised controversy recently. The first approach was carried out by means of MO extended-Huckel ¨ calculations [10] and, in 1986, Kahn and coworkers [11a] noted that the spin polarization model would describe the superexchange for the azido ligand more accurately, particularly for the EO coordination. More recently, spin density maps in the triplet ground state of [Cu2 (t-Bupy)4 (µ-N3 )2 ](ClO4 )2 (t-Buty = p-tert-butylpyridine) were published [11b] where it was stated that “the spin distribution in the triplet ground state in [Cu2 (tBupy)4 (µ-N3 )2 ](ClO4 )2 is dominated by a spin delocalization mechanism to which is superimposed a spin polarization effect within the π orbitals of the azido group” and “. . . can the 1,1-azido group be considered as an almost universal ferromagnetic coupler, or is the stabilization of the parallel spin state only achieved in a limited range of bridging angle values?”. However, studies on coupled dimers [12, 13] for EE azido bridges indicate that, at least for this coordination mode, the MO extended-Huckel ¨ calculations are appropriate. Recently, some ab initio calculations on these systems have been reported [11b, 14]. Magneto-structural correlations have been developed by the authors [15–23] and elsewhere [24–26].

9.2

Synthetic Procedures

All the polynuclear molecules have been synthesized from aqueous, methanol or dimethylformamide solution, because NiII and MnII are completely stable in these solvents. Some peculiarities should be highlighted:

9.3 Exchange-coupling Parameter

309

– The Ni/amine/N3 ratio determines the species obtained. – Increasing the coordination vacancies around the metal ions leads to an increase of the dimensionality of the resulting systems. For example, two bidentate amine leave two free positions; a tridentate amine, three free positions; a tetradentate macrocyclic amine two trans free positions, etc. In this sense, discrete mono-, bi-, or three-dimensional structures can be envisaged and obtained. – It is impossible, so far, to predict the coordination mode of the azido ligand (1,3 or 1,1 or mixing). – The counter-ion also has an effect. Different NiII or MnII salts with the same ligands and the same stoichiometry, result in the formation of different compounds. All the syntheses are completely reproducible.

9.3

Exchange-coupling Parameter

To compare all the results on magnetism reported, one Hamiltonian should be established to calculate the exchange coupling parameter (J ). In the literature, the two Hamiltonians most frequently used are H1 = − Ji,j Si Sj and H2 = −2Ji,j Si Sj . The resulting J value through H1 is twice that through H2 . Moreover, in some papers J is in K, although it is usually found in cm−1 . We thus used the Hamiltonian H = − Ji,j Si Sj (J in cm−1 ) and changed the values reported with the −2J Hamiltonian. The risk of adopting this assumption lies in those cases in which the authors do not clearly indicate which Hamiltonian has been used.

9.4

Molecular-based Magnetic Materials

Two methods can be followed for synthesizing molecular-based magnetic materials: (i) methods for synthesis of molecules with high spin and strong anisotropy; and (ii) methods for synthesis of small molecules or low-dimensional systems that can be linked in a 3D order ferro- or ferrimagnetically. With azido-bridging ligands the first strategy is so far unoperative. The polynuclear complexes with the highest spin are three NiII tetranuclear complexes with EOazido bridging ligands. The first has the formula [Ni4 (2-oxo-1,3-diaminopropane)2 (2hydroxy-1,3-diaminopropane)2 (µ1,1 -N3 )4 ](ClO4 )2 [27] (Fig. 2). The four NiII atoms are in a distorted octahedral environment and are related by an S4 symmetry axis forming a quasi-perfect square. The oxygen atom of the amine ligand (OHpn and Opn) also acts as a bridging ligand between two NiII ions. Magnetic measurements indicate ferromagnetic coupling with J = 21.3 cm−1 . Magnetization measurements show an S = 4 spin ground state. The second complex reported has a cubane-like structure (Fig. 3).

310

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 2. ORTEP plot showing the structure of the tetranuclear compound [Ni4 (2-oxo-1,3-diaminopropane)2 (2-hydroxo-2,3-diaminopropane)2 (µ1,1 -N3 )4 ](ClO4 )2 .

Fig. 3. Schematic drawing of [Ni4 (dbm)4 (EtOH)4 (η1 ,µ3 -N3 )4 ].

Its formulation is [Ni4 (dbm)4 (EtOH)4 (η1 ,µ3 -N3 )4 ] (dbm = dibenzoylmethane) [28, 29]. Each azido ligand is symmetrically bound to three NiII ions in an end-on arrangement. Magnetic behavior was well reproduced by the one-J model, giving J = 11.9 cm−1 . The formula of the third compound, which consists of tetranuclear units where the NiII ions are bridges by both µ-oxo and µ-end-on azido ligands, resulting in a face-share dicubane-like coordination core with two missing vertices (Fig. 4), is [Ni4 (dpkO)4 (µ-N3 )4 ] (dpk = di-2-pyridylketone) [30]. Magnetic measurements for this complex indicate a predominance of ferromagnetic interactions (Fig. 4). For the manganese ion, only one tetranuclear compound, with the formula [Mn2 (HL)(N3 )2 ]2 .3MeCN (H2 L = 1,7,14,20-tetramethyl-2,6,15,19-tetraza[7,7](2,6) pyridinaphe-4,7-diol) has been reported [31]. It comprises two dinuclear entities with

9.4 Molecular-based Magnetic Materials

311

Fig. 4. ORTEP plot (left) showing the structure of [Ni4 (dpkO)4 (µ-N3 )4 ] and χM T (•) and χM (◦) for this compound as a function of the temperature (right).

Fig. 5. Molecular structure of [Mn2 (HL)(N3 )2 ]2 (H2 L = 1,7,14,20-tetramethyl-2,6,15,19tetraaza[7,7](2,6)pyridiniphane-4,7-diol).

two oxo and one EO azido bridges, linked by a single EO azido bridge (Fig. 5). The global magnetic behavior measured between 300 and 4 K shows antiferromagnetic coupling. The second approach, consists in three parts: 1D, 2D, and 3D systems.

312

9.5

9 Molecule-based Magnets Derived from NiII and MnII

One-dimensional Systems

In uniform 1D systems, all the ions have the same magnetic pathway, i. e. only one type of azido bridging ligand and only one set of structural parameters. In alternating 1D systems, several types of azido bridging ligand and several sets of structural parameters are used.

9.5.1

With 1,3-Azido Bridging Ligands (AF, Uniform)

According to the position of the two adjacent azido bridging ligands, we can distinguish two 1D uniform systems: trans or cis, although these are very similar. The main structural data and the magnetic behavior (exchange coupled parameter, J ) of the trans complexes (Table 1) and of the cis complexes (Table 2) have been reported. A representative structural example of trans (A) and cis (B) chains are shown in Fig. 6. None of these systems shows long-range order. An extensive review of these systems and their correlations has been recently published [32]. On the other hand, these uniform one-dimensional systems obtained from NiII (S = 1) have been extensively studied by physicists as they provide useful examples for investigating the Haldane’s gap (a quantic gap between the ground state (S = 0) and the first excited state, when the S value is integer [33]). Unlike in the nickel ion, only one homogeneous chain with a double EE azide bridge, whose formula is [Mn(pyOH)2 (N3 )2 ]n (pyOH = 2-hydroxypyridine), has

Fig. 6. Example of a trans uniform chain, A, and a cis uniform chain, B.

9.5 One-dimensional Systems

313

Table 1. Structural and magnetic parameters for 1D trans-Ni-(µ-N3 )-Ni systems: Ni–N– N and dihedral angles (◦ ); J in cm−1 . (2,2 -mettn = 2,2 -dimethyl-1,3-propanediamine; 232-tet = N,N -bis(2-aminoethyl)-1,3-propanediamine; 3,2,3-tet = N,N -bis(aminopropyl)-1,3ethanediamine; macro = 2,3-dimethyl-1,4,8,11-tetraazacyclotetradeca-1,3-diene; cht = meso5,7,7,12,14,14-hexamethyl-1,4,8,11-tetraazacyclotetradecane); me-py = 5-methylpyrazole. Compound

Ni–N–N

τ

[Ni(tmd)2 (µ-N3 )]n (ClO4 )n [Ni(tmd)2 (µ-N3 )]n (PF6 )n

120.9 126.1

180 180

[Ni(2,2 -mettn)2 (µ-N3 )]n (ClO4 )n [Ni(2,2 -mettn)2 (µ-N3 )]n (PF6 )n

137.2 136.5

180 180

[Ni(cyclam)2 (µ-N3 )]n (ClO4 )n [Ni(232-tet)2 (µ-N3 )]n (ClO4 )n [Ni(323-tet)2 (µ-N3 )]n (ClO4 )n [Ni(macro)2 (µ-N3 )]n (ClO4 )n [Ni(cth)2 (µ-N3 )]n (ClO4 )n

∗∗

[Ni(me-py)4 (µ-N3 )]n (ClO4 )n

140.7 128.2 134.6 124.1 135.8 119.8 115.6 116.8 A) 128.5 130.7 B) 131.4 128.4 146.1

J

166.9

−100 −55.8∗ (−70.6) −49.4 −19.4∗ (−41.1) −39.2

142.4

−26.9

169.3

−62.7

175.7

−97.8

150.5

−41.1

146.2 75.7

−36.4 +6.9

∗ In these two complexes there is a phase transition when the temperature decreases, which changes the structural parameters and the magnetic results. The results at high temperature, (before the transition) are shown in parentheses while the results at low temperature (after the transition) are shown without parentheses. ∗∗ This complex shows two structurally distinct 1D uniform chains in the crystal net in a 2:1 ratio. These chains are indicated as A and B.

been reported for the manganese ion [34]. In this case, the pyOH ligand coordinates the manganese atom in a monodentate mode through the oxygen atom, resulting in a trans system. The coupling was moderately antiferromagnetic, with J = −7.0 cm−1 .

9.5.2

With 1,3-Azido Bridging Ligands (AF, Alternating)

Two types have been reported so far [32].

9.5.2.1

With Two Different Geometries in Adjacent 1,3-N3 Azido Bridges

This is the case of [Ni(333-tet)(µ-N3 )]ClO4 (333-tet = N ,N -bis(3-aminopropyl)1,3-propanediamine). This one-dimensional system can be schematized as –Ni–Na – Nb –Na –Ni–Nc –Nd –Nc –Ni–Na –Nb –Na –Ni–. Thus, the most interesting feature of this

314

9 Molecule-based Magnets Derived from NiII and MnII

Table 2. Structural and magnetic parameters for 1D cis-Ni–(µ-N3 )–Ni systems: Ni–N–Ni and dihedral angles (◦ ); J in cm−1 . (333-tet = N,N -bis(3-aminopropyl)-1,3-propanediamine; 2-methyl = 1,2-diamino-2-methylpropane; aep = 2-(aminoethylpyridine)). Compound

Ni–N–N

τ

[Ni(333-tet)2 (µ-N3 )]n (PF6 )n

151.8 151.3 131.8 125.9 135.0 122.6 123.7 120.1 122.6 122.6 127.8 126.2

142.8

−18.5

125.9

−16.8

146.8

−3.2

133.9 138.9 134.9 138.0 122.4

−33.0

[Ni(2-methyl)2 (µ-N3 )]n (ClO4 )n [Ni(2-methyl)2 (µ-N3 )]n (PF6 )n [Ni(bipy)2 (µ-N3 )]n (ClO4 )n [Ni(bipy)2 (µ-N3 )]n (PF6 )n [Ni(aep)2 (µ-N3 )]n (ClO4 )n

∗

∗

126.6 121.3 120.6 127.4

J

−22.4 <−1

∗

These two chains actually belong to the alternating AF chains because the asymmetric unit is formed by two Ni–N3 –Ni entities that are attached consecutively in a chain. The fit of the magnetic data was made assuming only one value of J , as a uniform chain, because the variation of Ni–N–N and dihedral angles is small.

Fig. 7. Schematic representation of the trans uniform chain [Ni(333-tet)(µN3 )]ClO4 (333-tet = N ,N -bis(3-aminopropyl)-1,3-propanediamine) A) showing the different set of structural parameters B) showing the torsion angle between the neighboring azido groups.

compound is the inversion centers at the central nitrogen of two consecutive (yet different) azido groups (Fig. 7). An equation for the analysis of data on the magnetic susceptibility of alternating chains with local S = 1, suggested by Borras et al. [35], provides calculations on a cyclic system of four spin pairs with an acceptable approximation up to kT /|J1 | = 0.4.

9.5 One-dimensional Systems

315

Fig. 8. ORTEP plot (top) showing the structure of the compound [Mn(3-Etpy)(N3 )2 ]n (3-Etpy = 3-ethylpyridine) and χM T (◦) and χM (◦) of this compound as a function of the temperature (bottom).

The coupling parameters have been thus optimized up to the values J = −80.7 cm−1 , J = −37.4 cm−1 and g = 2.4. The alternating parameter J /J = α is 0.46 and the theoretical ratio between the gaps calculated with the extended-Huckel ¨ M.O. model is 0.52, in agreement with the experimental value. This compound has been extensively studied as a model of the gapless point of the Affleck–Haldane conjecture [36]. As for manganese, one singular compound with an alternating double EE azido bridge has also been reported: [Mn(3-Etpy)2 (N3 )2 ]n (3-Etpy = 3-ethylpyridine) [34], which shows trans coordination and two different sets of bond parameters in each alternating Mn–(N3 )2 –Mn unit with a different degree of chair distortion (Fig. 8). The coupling was found to be antiferromagnetic.

316

9 Molecule-based Magnets Derived from NiII and MnII

9.5.2.2

With Consecutive Single and Double 1,3-Azido Bridging Ligands

This is the case of [(Ni2 (dpt)2 (µ-N3 )(µ-N3 )2 )]n (ClO4 )n (dpt = bis(3-aminopropyl)amine) [23] (Fig. 9). The structure consists of a 1D –Ni–(N3 )2 –Ni–N3 –Ni– system. The structure of the Ni–(N3 )2 –Ni fragment derives from the symmetric rotation of the coordination polyhedra of the nickel atoms around the x axis. As a result of this movement, the nickel atoms and the central nitrogen atoms of the azido groups remain in the x y plane, whereas the N(azido) atoms linked to the nickel atoms leave the original plane. The J values found following the equation suggested by Borras ´ [35] are: J = −84, 6 cm−1 and J = −41.4 cm−1 (α = 0.49). According to our previous discussion [32], these two J values can be unequivocally assigned to the single and double azido bridges respectively. In conclusion, all these antiferromagnetically coupled systems are not appropriate models for the study of molecular-based magnetic materials, unlike 1,1-azido bridging ligands which create ferromagnetic coupling.

Fig. 9. Structure of the 1D system with consecutive single and double 1,3-azido bridging ligands [Ni2 (dpt)2 (µ-N3 )(µ-N3 )2 ]n (ClO4 )n .

9.5.3

With 1,1-Azido Bridging Ligands (Ferromagnetic)

For the NiII ion, four complexes of this type have been reported but only three of them have been crystallographically studied [37]. The general formula is [Ni(L)(µ-N3 )2 ]n (L = ethylenediamine, 1,3-diaminopropane, 2,2 -dimethyl-1,3-diaminopropane and N -ethylethylenediamine; Fig. 10). The main structural and magnetic parameters were examined (Table 3) [32]. As expected, the four complexes are ferromagnetic (Fig. 11). For the first three chains the magnetic parameters were obtained assuming D = 0 or including the D parameter in the de Neef formula [38], which allows this kind of uniform chains to be fitted for S = 1. For this reason there are two J values in Table 3. The 1D complex with N -ethylethylenediamine is alternating and the J values have been obtained by a new model developed by M.S. El Fallah [37b]. This

9.5 One-dimensional Systems

317

Fig. 10. ORTEP plot showing the structure of [Ni(N -methylethylenediamine)(µ-N3 )2 ]n .

Table 3. Structural and magnetic parameters for 1D Ni–(µ1,1 -N3 )–Ni systems. Ni–N–Ni angles (◦ ); J in cm−1 . (en = ethylenediamine; tn = 1,3-diaminopropane; 2,2 -mettn = 2,2 -dimethyl1,3-diaminopropane; N -eten = N -ethylethylenediamine). Compound

Unit A

Unit B

[Ni(2,2 -mettn)(N3 )2 ]n

103.1 95.2 108.1 105.0 –

100.9 100.9 100.1 101.4 –

[Ni(N-eten)(N3 )2 ]n

103.3

100.6

[Ni(en)(N3 )2 ]n [Ni(tn)(N3 )2 ]n

∗

Ni–N–Ni

J∗

|D|

35.6 23.8 35.2 39.8 29.4 33.8 57.3∗∗ 28.5

6.3 0 4.4 0 6.7 0 –

The first J value is found by assuming |D| = 0; the second one by assuming |D| = 0 There are two J values because it is an alternating ferromagnetic chain

∗∗

complex shows a peculiar behavior as can be seen in Fig. 12, where the dependence of the magnetization against T is shown. At very low temperature, (2, 5 and 10 K) the curve is sigmoidal, which is indicative of the metamagnetic character below 10 K, approximately. The hysteresis loop (Fig. 13) at 2 K corroborates this behavior: for Hc = 10 KG, there is a change of the sign in δ 2 M/δ H 2 , which is typical of a metamagnetic system. With the same metal azide skeleton, the manganese derivative [Mn(2bzpy)(N3 )2 ]n chain has been characterized [39] (2-bzpy is 2-benzoylpyridine). The bridges are not equivalent but show little deviation from the mean Mn–N–Mn value of 100◦ . Owing to the large volume of the 2-bzpy ligand, the chains are well isolated with an interchain Mn · · · Mn distance of 9.93 Å, which minimizes the interchain coupling. As expected, the intrachain coupling was ferromagnetic and showed a χ T value of 32.8 emu K mol−1 at 2 K. No 3D ordering was found.

318

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 11. Plots of χM and χM T against T for [Ni(N -methylethylenediamine)(µ-N3 )2 ]n .

Fig. 12. Dependence of magnetization on T for [Ni(N -methylethylenediamine)(µ-N3 )2 ]n .

9.5 One-dimensional Systems

319

Fig. 13. Hysteresis loop for [Ni(N -methylethylenediamine)(µ-N3 )2 ]n .

9.5.4

With 1,3-N3 and 1,1-N3 bridges

In the synthetic procedure, when the Ni/bidentate diamine/azide ratio is 1/1/2, not only were new ferromagnetic chains obtained but, depending on the diamine used in the synthesis, new ferro-antiferromagnetic species were also synthesized. The alternation can be very simple, e. g. F–AF (Fig. 14A), or much more complicated e. g. F–F–F–AF (Fig. 14B). The magnetic data have been interpreted and fitted by numerical extrapolation of S = 1 rings of finite length [40, 41]. An isotropic Heisenberg system with quantic spin S = 1 has been assumed. The main structural and magnetic parameters for this kind of one-dimensional alternating chain are shown in Table 4, the AF coupling parameter J agrees with the results described elsewhere [32].

Fig. 14. Structure of 1D systems with consecutive F and AF azido bridging ligands (A) and with the F–F–F–AF azido bridging set (B).

320

9 Molecule-based Magnets Derived from NiII and MnII

Table 4. Structural and magnetic parameters for F–AF 1D systems. Ni–N–Ni, Ni–N3 –Ni δ and τ angles (◦ ), J in cm−1 . (N N -dmen = N ,N -dimethylethylenediamine; aep = 2-aminoethylpyridine, N N -dmen = N ,N -dimethylethylenediamine; Medien = Methyldiethylenetriamine). Compound

Type

[Ni(bipy)(µ-N3 )2 ]n

F–AF

[Ni(NN-dmen)(µ-N3 )2 ]n [Ni(aep)(µ-N3 )2 ]n

[Ni(Medien)(µ-N3 )]n (ClO4 )n [Ni(NN -dmen)(µ-N3 )2 ]n

Ni–N–N Ni–N–Ni δ

118.2 129.9 F–AF 121.1 139.4 F–AF 125.6 124.4 121.6 117.9 F–AF 138.1 125.7 F–F–F–AF 130.9 132.9

101.5

τ

35.2 180

98.2

≈0

180

98.7

*

*

100.8 98.9 100.5 101.2 102.9

–

J , J , J 26 −2.6 17 −156 5 −28

162.8 −34.7 38.2 8.4 180 −120 20 37

∗ This Ni–(N ) –Ni is fully asymmetric. Four nitrogen (azido) atoms and one Ni form a plane 3 2 while the other two nitrogen (azido) atoms and the other Ni are clearly separated from this plane. The distances to the mean plane are 1.32 Å (Ni) and 1.62 Å (N-azido).

The same kind of system has also been obtained for the MnII ion in the complexes [Mn(bpy)(µ1,1 -N3 )(µ1,3 -N3 )]n [42a] (Fig. 15a) and [Mn(3-Et,4-Mepy)(µ1,1 N3 )(µ1,3 -N3 )]n . [42b] (Fig. 15c). The azido ligands are arranged cis or trans respectively. Because of these structural features, alternating ferromagnetic (through EO bridges) and antiferromagnetic interactions (through EE bridges) should be expected. The thermal variation of both χM and χM T for the cis complex (Fig. 15b) are very similar to the trans complex. This behavior can only be explained assuming a positive J1 for EO and a negative J2 for the EE pathways. No formula was available in the literature for an S = 5/2 ferro-antiferromagnetic alternating chain and so a new model was developed [42b]. The best agreement obtained by least squares refinement is given by the following set of parameters: JF = 9.58 cm−1 ; JAF = −11.80 cm−1 , g = 2.01 for the cis complex and JF = 2.4 cm−1 ; JAF = −13.7 cm−1 , g = 2.036 for the trans complex. A new topology – double EO and single EE – has recently been reported for [Ni2 (Medien)(µ1,1 -N3 )2 (µ1,3 -N3 )]n (ClO4 )n [43] (Fig. 16). The structural parameters are unchanged for the two coordination modes and, as can be expected, the magnetic behavior corresponds to an F–AF alternating system. The data have been fitted with an expression built on the basis of the increasing ring systems calculations. Finally, a unique case with one end-to-end and three end-on bridges in alternation has been reported: the complex [Ni2 (tmeda)2 (µ1,1 -N3 )3 (µ1,3 -N3 )]n (tmeda = N , N , N , N -tetramethyl-ethylenediamine) [44] (Fig. 17). The main features occur in the Ni–N–Ni angles. The average value in the EO bridge is 84.2◦ , one of the lowest values for metal ions with an azido bridging ligand in EO mode. In the EE bridge the Ni–N–N angle

9.5 One-dimensional Systems

321

Fig. 15. Structure of [Mn(bpy)(µ1,1 -N3 )(µ1,3 -N3 )2 ]n (top); χM T ( ) and χM (◦) for this compound as a function of the temperature (middle) and structure of [Mn(3-Et,4-Mepy)2 (µ-N3 )2 ] (bottom).

9 Molecule-based Magnets Derived from NiII and MnII

322

Fig. 16. Structure of [Ni2 (Medien)(µ1,1 -N3 )2 (µ1,3 -N3 )]n (ClO4 )n .

Fig. 17. ORTEP plot showing the structure of the triply EO and single EE azido bridging 1D compound [Ni(tmeda)(µ-N3 )2 ]n .

is 138◦ with a Ni–N3 –Ni dihedral angle of 180◦ . The data have been fitted with the Borras ´ formula for an antiferromagnetic alternating S = 1 chain. This suggests that AF coupling takes place in both parts of the chain (J1 and J2 are negative).

9.6

Two-dimensional Systems

The magnetic properties of these compounds are especially interesting at low temperatures. To find ordered systems with the properties of weak molecular magnets is not unusual.

9.6.1 9.6.1.1

With Only Azido as Bridging Ligand With Single EE Bridges

No complexes of this kind have been reported for the NiII ion. For the MnII ion, two compounds with the general formula [Mn(R-py)2 (N3 )2 ]n where R-py is a pyridine derivative have been characterized. For 4-acpy (4-acetylpyridine) [45] or Minc (methylisonicotinate) [46], the compounds consist of manganese atoms with the two pyridinic ligands coordinated in trans and four end-to-end azido bridges, which link

9.6 Two-dimensional Systems

323

Fig. 18. Structure of the two-dimensional compound [Mn(4-acpy)2 (N3 )2 ]n (4-acpy = 4acetylpyridine) and χM T of this compound as a function of the temperature and field.

to the four neighboring manganese atoms, resulting in a quadratic layer (Fig. 18). As expected from the calculations, the coupling was antiferromagnetic for the two compounds, J = −3.83 cm−1 (4-acpy) and −2.24 cm−1 Minc). Owing to canting phenomena, the [Mn(4-acpy)2 (N3 )2 ]n compound shows long range order below the TC = 28 K and spontaneous magnetization and hysteresis loop at 2 K [47]. 9.6.1.2

EE and EO Bridges

With and the 2,2 -dimethylpropanediamine ligand (2,2 -mettn), a new twodimensional system with ferro-antiferromagnetic alternation has been obtained: [Ni(2,2-mettn)(µ-N3 )]n [48]. In this complex, each bridging azido ligand coordinates two NiII ions in an EO mode but, at the same time, this same azido ligand coordinates the neighboring NiII ion in an EE coordination mode (Fig. 19). NiII

324

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 19. Structure of the two-dimensional compound [Ni(2,2-mettn)(µ-N3 )]n (2,2-mettn = 2,2-dimethylpropane-1,3-diamine).

Fig. 20. Structure of the two-dimensional compounds [Ni(amine)(N3 )2 ]n (amine = N -isopropylethylenediamine, N,N-diethylethylenediamine and N ,N -diethyl-N -methylethylenediamine).

This linkage is extended through a layer and the layers are linked by hydrogen and Van der Waals forces. The magnetic properties of this system should be highlighted: a canting phenomenon, which takes place at low temperature, makes the complex a molecular magnet. For N -isopropylethylenediamine, N ,N -diethylethylenediamine and N,N-diethylN -methylethylenediamine, three new two-dimensional complexes of general formula [Ni(amine)(µ-N3 )2 ]n have been recently reported [49]. The structural arrangement, the same in each case, consists of dinuclear nickel units bridged by double EO azido bridges, linked to the four equivalent units by four single EE azido bridges (Fig. 20). In spite of these two kinds of azido bridging ligand, the global behavior is ferromagnetic (Fig. 21).

325

9.6 Two-dimensional Systems

Fig. 21. Plots of χM and χM T against T for [Ni(N ,N -diethyl-N -methylethylenediamine)(N3 )2 ]n . Table 5. Structural and magnetic parameters for 2D NiII ferromagnetic compounds. Ni–N–Ni angles (◦ ); Ni–N–N angles (◦ ); Ni–N3 –Ni τ angles (◦ ); J in cm−1 (N -ipen = N -isopropylethylenediamine; N ,N -deen = N ,N -diethylethylenediamine; N ,N -de-N -men = N ,N -diethyl-N methylethylenediamine). Compound

Ni–N–N

Ni–N–Ni

τ

J

[Ni(N -ipen)(µ-N3 )2 ]n

125.9 132.6 132.5 134.8 137.5 145.0

99.3

116.8

100.9

112.8

100.5

110.4

27.3 1.3 30.2 0.8 26.1 9.4

[Ni(N ,N -deen)(µ-N3 )2 ]n [Ni(N,N-de-N -men)(µ-N3 )2 ]n

To describe the magnetic data, a simplified pattern based on a sphere form of eight local S = 1 centers was designed following the full-diagonalization matrix method [50]. The two J values are ferromagnetic for the three complexes. Main structural and magnetic parameters are gathered in Table 5. Figure 22 shows the dependence of the magnetization on T for the N ,N -Et2 N -Meen derivative. For this derivative and for the N ,N -Et2 -en derivative, at very

326

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 22. Dependence of magnetization on T for [Ni(N ,N -diethyl-N -methyl ethylenediamine)(N3 )2 ]n .

Fig. 23. Hysteresis loop for [Ni(N ,N -diethyl-N -methylethylenediamine)(N3 )2 ]n .

low temperatures (2, 5 and 10 K) the curve is sigmoidal, which is indicative of the metamagnetic character below 10 K, approximately. The hysteresis loop (Fig. 23) at 2 K verifies this behavior: for Hc = 10 KG and 12 KG respectively, there is a change of the sign of δ 2 M/δ H 2 , which is typical of a metamagnetic system.

9.6 Two-dimensional Systems

327

Fig. 24. Typical properties derived from the canting phenomenon in [Mn(R-py)2 (N3 )2 ]n .

The same structure was also found in four MnII compounds with the general formula [Mn(R-py)2 (N3 )2 ]n , in which R-py is Etnic (ethylisonicotinate), 4-CNpy (4-cyanopyridine), 3-acpy (3-acetylpyridine) and Enic (ethylnicotinate) [47, 51, 52]. Bridging bond parameters are similar, with the EO Mn–N–Mn bond angle between 100.5◦ and 104.1◦ and Mn–N3 –Mn torsion angles (τ ) between 124.7◦ and 114.7◦ . J parameters were not calculated since the analytical expressions for alternating 2D systems were not available. However, as expected, the magnetic behavior is that of a ferro–antiferromagnetic system. The magnetic data for the Etnic compound were not reported, but for the other three compounds long range order was found at 16 K (Etnic and 3-acpy) and 28 K (4-CNpy). The typical properties derived from the canting phenomenon such as hysteresis loop and broadening of the EPR signal when T is close to TC , were characterized (Fig. 24). Another example of a 2D complex with both azido and organic-ligand bridges is the [Ni(4,4 bipy)(N3 )2 ]n system [53]. This compound consists of chains of NiII ions bridged by one double EO and two double EE azido bridges (this topology has not been reported so far), which are joined by 4,4 -bipyridine ligands leading to the two-dimensional arrangement (Fig. 25). The Ni–EO–Ni angle in the mixed

328

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 25. Structure and magnetism of [Ni(4,4 -bipy)(N3 )2 ].

bridge is almost 135◦ , larger than usual and is expected to give antiferromagnetic interactions. Preliminary magnetic measurements for this compound showed global ferromagnetic interactions (Fig. 25).

9.6.1.3

With only EO Bridges

The 2D complex [Mn2 (µ1,1 -N3 )4 (µ-bipym)]n [54] shows chains of MnII ions connected by double 1,1-N3 azido bridges, which are alternatively connected by bisbidentate bipyrimidine ligands (Fig. 26). The whole structure can also be described as a honeycomb sheet, consisting of an infinite hexagonal array of MnII ions bridged by bis-bidentate bipyrimidine ligands and double end-on azido bridges. The repetition of circular rings extends this 2D polymer across the xy plane (Fig. 26). The magnetic measurements for this 2D polymer indicate the presence of ferromagnetic interactions in the high temperature range, followed by antiferromagnetic interactions at low temperatures. So far, the lack of a theoretical model for this magnetic plane has hindered a deeper study of its magnetic behavior.

9.7 Three-dimensional Systems

329

Fig. 26. Structure of the “honeycomb” 2D complex [Mn(µ1,1 -N3 )(µ-bipym)]n .

A similar pattern has been reported for [Mn(pyrazine)(µ-N3 )2 ] [55]. The onset of spontaneous magnetization near 2 K suggests the presence of a weak ferromagnetic ground state.

9.7

Three-dimensional Systems

3D compounds are a challenge in the field of magnetic molecular systems, because of the need both to understand their magneto-structural correlations and to obtain 3D antiferro- or ferromagnets with possible technical applications. 3D systems with azido bridging groups could be obtained by either using metal and monodentate ligands in a 1:1 ratio, connecting 2D compounds through trans bidentate ligands or by preparation without the use of ancillary blocking ligands. [Mn(py)2 (µ1,3 -N3 )2 ]n , a 3D compound synthesized according to the second approach [56], shows a trans arrangement of the pyridine ligands and four EE bridges to the neighboring manganese atoms in a similar way as the 2D compounds with R = 4-acpy or Minc. In these 2D compounds, the pyridine-like ligands are placed above and below the two-dimensional sheet, whereas in the 3D arrangement, the py ligands are placed along channels. Bond angles lie in the normal range, Mn–N–N = 134.1◦ and 145.4◦ , and show the expected antiferromagnetic coupling, with J = −1.35 cm−1 (Fig. 27). The high TC of 40 K is noticeable for the long range order with the presence of a canting phenomenon. A second 3D compound with only azide bridges is [N(CH3 )4 ]n [Mn(µ1,3 -N3 )3 ]n [57]. The proportion of three azido ligands per divalent ion was achieved by using an appropriate cation. The crystal structure of the compound (Fig. 28) may be described as a distorted perovskite, where the MnII is shifted from the origin of the unit cell by approximately 0.25 Å along the b axis. The azido groups act as end-to-end bridging ligands between metals to form the 3D network. Each manganese ion shows a slightly distorted octahedral coordination sphere. The [N(CH3 )4 ]+ cations are located within the holes formed by the MnII -azido sublattice. The magnetic results for this complex clearly indicate the occurrence of AF interactions (Fig. 28), which can be explained assuming a regular 3D network as indicated by the crystal structure. The best least-square fit of the experimental data

330

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 27. Plot of χM T (•) and χM () of [Mn(py)2 (µ1,3 -N3 )2 ]n as a function of temperature showing the high TC of 40 K for the long range order.

Fig. 28. Structure of the three-dimensional compound [N(CH3 )4 ]n [Mn(µ1,3 -N3 )3 ]n (top) and χM T (♦) of this compound as a function of the temperature.

9.7 Three-dimensional Systems

331

Fig. 29. Structure of the three-dimensional compound Csn [Mn(µ1,3 -N3 )3 ]n (top) and magnetic data (experimental and calculated) of this compound as a function of the temperature.

to the expression of the magnetic susceptibility for a simple cubic Heisenberg antiferromagnetic system [58] is obtained with the parameters J = −1.21 cm−1 and g = 2.01. The similar Cs[Mn(µ-N3 )3 ] compound consists of alternating end-to-end/end-on layers similar to those above described, linked axially by end-to-end azido bridges [59] (Fig. 29). Three different magnetic interactions are present in the compound: one ferromagnetic and one antiferromagnetic in the layers and a third antiferromagnetic interaction between them. Monte Carlo calculations on a set of 10 × 10 × 10 classical spins 5/2 give the best agreement parameters J1 = 0.8, J2 = −4.3 and J3 = −3.3 cm−1 (Fig. 29). A new example of 3D MnII azido compound is a polymorph of the 2D [Mn2 (µ1,1 N3 )4 (µ-bipym)]n system, with the formula [Mn2 N3 (µ1,1 -N3 )(µ1,3 -N3 )(µ-bipym)]n ,

332

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 30. Structure of the three-dimensional compound [Mn2 N3 (µ1,1 -N3 )(µ1,3 -N3 )(µ-bipym)]n .

[54b]. The crystal structure consists of chains of manganese atoms alternately bridged by bipym and two end-on azido groups (Fig. 30). The chains are connected by end-toend azido groups in vicinal positions, which ensures linkages in two different directions. Magnetic properties are in accordance with predominant antiferromagnetic interactions, which are propagated by the bipym and the end-to-end azido bridges and which are larger than the ferromagnetic interactions corresponding to the end-on azido bridges. Another example of 3D MnII -azido systems is the [Mn(µ-4,4 -bpy)(µ1,3 -N3 )2 ] complex [60]. The determination of the crystal structure of [Mn(µ-4,4 -bpy)(µ1,3 N3 )2 ] revealed a complicated 3D network of manganese(II) ions bridged by both azido and 4,4 -bipy ligands (Fig. 31). In order to elucidate the structure, we have analyzed the subnets. In the 3D [Mn(N3 )2 ] subnet, it seems that the –Mn–N3 –Mn–N3 – connections show a helical propagation (Fig. 31b). However, a closer examination indicates that every MnII ion has four coordinated azido ligands with a distorted tetrahedral topology (Fig. 31c) which extends in space. Moreover, through the exploration of the global Mn–Mn a diamondoid conformation of this subnet can clearly be observed (Fig. 32a). The 4,4 -bipy ligands lie inside the adamantane cages which join opposite vertexes (Fig. 32b). The Mn–N distances range from 2.185(6) (azido) to 2.301(4) Å (bipyridine). The EE azido bridges form two asymmetric angles with the metal, with values of 153.0(5)◦ and 121.9(4)◦ . The pyridine rings of the 4.4 -bipy ligands form an angle of 30.1(3)◦ . The Mn · · · Mn distances through the azido and the 4,4 -bipy bridges are 5.945(4) and 11.637(2) Å, respectively. The 2–300 K temperature dependence of the magnetic susceptibility of [Mn(µ4,4 -bpy)(µ1,3 -N3 )2 ] – for a magnetic field of 50 G – was fit by the Curie–Weiss expression, with a Lande g value of 2.00 and a θ value of −79.8 K for T > 100 K, which is indicative of strong antiferromagnetic coupling between metal sites (Fig. 33). The room temperature χM T (assuming χM as M/H , which is the case at high temperatures) is 3.55 cm3 mol−1 K, reduced from 4.4 cm3 mol−1 K for uncoupled S = 5/2 ions because of antiferromagnetic coupling. Below 300 K, the χM T product decreases

9.7 Three-dimensional Systems

333

Fig. 31. Structure of the three-dimensional compound [Mn(µ-4,4 -bpy)(µ1,3 N3 )2 ]n complex (see text for explanation).

Fig. 32. Structure of the three-dimensional compound [Mn(µ-4,4 -bpy)(µ1,3 N3 )2 ]n complex (see text for explanation).

gradually, reaching a minimum below 45 K (Fig. 33), which is attributable to longrange magnetic ordering. These results are also confirmed by field-cooled magnetization. Below the minimum, a rapid increase in the χM T occurs because of an increase in the ferromagnetic component of weak ferromagnetic state. A maximum at 17 K is followed by swiftly decreasing moments due to saturation effects and/or increasing antiferromagnetic correlations until 2 K. In the presence of small applied dc magnetic fields, a spontaneous magnetization occurs in the compound. A low saturation magnetization value consistent with weak ferromagnetic behavior is observed for this compound. The 5 K field-dependence of the magnetization, M(H ), shows an initial sharp rise to approximately 750 emu Oe mol−1 at 2000 Oe (below the 27925 emu Oe mol−1 expected for an S = 5/2 system but larger than the

334

9 Molecule-based Magnets Derived from NiII and MnII

Fig. 33. Magnetic properties for [Mn(µ-4,4 -bpy)(µ1,3 N3 )2 ]n .

zero value expected for an antiferromagnetic ground state) and then becomes linear to 7 T, reaching 2340 emu Oe mol−1 . A transition to a weak ferromagnetic state could account for the observed magnetic behavior.

9.8

Conclusions

The azido ligand has been shown to be a good bridging ligand to build molecularbased magnets, derived from NiII and MnII complexes with different dimensionality. The versatility of this pseudohalide ligand is greater than that previously reported for other ligands, like oxalato and cyanide due to its two coordination modes (EE, EO and mixed). Concerning the magnetic behavior observed in these species, it can

References

335

be deduced, as expected, that antiferromagnetic interactions are associated with the EE bridging mode while ferromagnetic interactions are found in EO bridging mode. The dimensionality can be controlled, at least in part, by the reaction stoichiometry and the number of vacancies around the metal ions. The MnII has a greater tendency to give 3-dimensional structures than NiII complexes, whose greater tendency is to give one or bi-dimensional systems. In all cases, the canting phenomenon (noncompensated antiferromagnetism) is widely found in NiII and MnII molecular-based magnets.

Acknowledgments The authors thank the Spanish Government (Grant PB96/0163) and the Basque Government (Grant PI96/39) for the financial support.

References [1] R. D. Willet, D. Gattteschi, O. Kahn, (Eds.) Magneto-Structural Correlations in Exchange Coupled Systems. NATO ASI Series. Reidel, Dordrecth, 1985. [2] O. Kahn, Molecular Magnetism, VCH Publishers, Weinheim, 1993. [3] M. M. Turnbull, T. Sugimoto, L. K. Thompson, (Eds.) Molecular-Based Magnetic Materials: Theory, Techniques and Applications. ACS Symposium Series, n. 644, ACS, Washington, 1996. [4] A. Caneschi, D. Gatteschi, L. Pardi, R. Sessoli, Clusters, Chains and Layered Molecules: the Chemist’s Way to Magnetic Materials, in A. F. Williams, (Ed.) Perspectives in Coordination Chemistry, VCH, Weinheim, 1992. [5] E. Coronado, P. Delhaes, D. Gatteschi, J. S. Miller, (Eds.) Molecular Magnetism:from Molecular Assemblies to Devices, NATO ASI Series, Vol. 321, Kluwer, Dordrecht, 1995. [6] D. W. Bruce, D. O’Hare, (Eds.) Inorganic Materials, John Wiley, New York, 1992. [7] P. Delhaes, M. Drillon, (Eds.) Organic and Inorganic Low-Dimensional Materials, NATO ASI Series, Vol. 168, Plenum Press, 1987. [8] D. Gatteschi, O. Kahn, J. S. Miller, F. Palacio, (Eds.) Magnetic Molecular Materials. NATO ASI Series, Vol. 198. Kluwer, Dordrecht, 1993. [9] C. J. O’Connor, (Ed.) Research Frontiers in Magnetochemistry. World Scientific, Singapore, 1993. [10] J. Commarmond, P. Plumere, ´ J. M. Lehn, Y. Agnus, R. Louis, R. Weiss, O. Kahn, I. Morgensten-Badarau, J. Am. Chem. Soc., 1982 104, 6330. [11] (a) M. F. Charlot, O. Kahn, M. Chaillet, C. Larrieu, J. Am. Chem. Soc. 1986, 108, 2574; (b) M. A. Aebersold, B. Gillon, O. Platevin, L. Pardi, O. Kahn, P. Bergerat, I. von Seggern, F. Tuczek, L. Ohrstrom, A. Grand, E. Levievre-Berna. ` J. Am. Chem. Soc. 1998, 120, 5238. [12] A. Bencini, C. A. Ghilardo, S. Midollini, A. Orlandini, Inorg. Chem. 1989, 28, 1958. [13] O. Kahn, T. Mallah, J. Gouteron, S. Jeannin, Y. Jeannin, J. Chem. Soc. Dalton Trans. 1989, 1117.

336

9 Molecule-based Magnets Derived from NiII and MnII

[14] (a) O. Castell, R. Caballol, V. M. Garc´ıa, K. Handrick, Inorg. Chem, 1996, 35, 1609; (b) E. Ruiz, J. Cano, S. Alvarez, P. Alemany, J. Am. Chem. Soc. 1998, 120, 11122. [15] (a) M. I. Arriortua, R. Cortes, ´ L. Lezama, T. Rojo, X. Solans, M. Font-Bard´ıa, Inorg. Chim. Acta 1990, 174, 263; (b) T. Rojo, L. Lezama, R. Cortes, ´ J. L. Mesa, M. I. Arriortua, J. of Magn. Magn. Mat. 1990, 83, 519; (c) R. Cortes, ´ L. Lezama, T. Rojo, M. Karmele Urtiaga, M. Isabel Arriortua, IEEE Transactions on Magnetics, 1994, 30, 4728. [16] R. Cortes, ´ K. Urtiaga, L. Lezama, J. L. Pizarro, A. Goni, ˜ M. I. Arriortua, T. Rojo, Inorg. Chem. 1994, 33, 4009. [17] A. Escuer, R. Vicente, J. Ribas, M. S. El Fallah, X. Solans, M. Font-Bard´ıa, Inorg. Chem. 1993, 32, 3727. [18] A. Escuer, R. Vicente, J. Ribas, M. S. El Fallah, X. Solans, M. Font-Bard´ıa, Adv. Mat. Res. 1994, 1-2, 581. [19] A. Escuer, R. Vicente, M. S. El Fallah, X. Solans, M. Font-Bard´ıa, J. Chem. Soc. Dalton Trans., 1996, 1013. [20] A. Escuer, R. Vicente, X. Solans, M. Font-Bard´ıa, Inorg. Chem. 1994,33, 6007. [21] J. Ribas, M. Monfort, C. Diaz, C. Bastos, X. Solans, Inorg. Chem. 1993, 32, 3557. [22] J. Ribas, M. Monfort, C. Diaz, A. Escuer, R. Vicente, M. S. El Fallah, X. Solans, M. Font-Bard´ıa, Ad. Mat. Res. 1994, 1-2, 573. [23] (a) R. Vicente, A. Escuer, J. Ribas, X. Solans, Inorg. Chem. 1992, 3, 1726; (b) R. Vicente, A. Escuer, Polyhedron 1995, 14, 2133. [24] P. Chaudhuri, T. Weyhermuller, ¨ E. Bill, K. Wieghardt, Inorg. Chim. Acta, 1996, 252, 195. [25] D. M. Duggan, D. N. Hendrickson, Inorg. Chem. 1974, 13, 2929. [26] C. G. Pierpont, D. N. Hendrickson, D. M. Duggan, F. Wagner, E. K. Barefield, Inorg. Chem. 1975, 14, 604. [27] J. Ribas, M. Monfort, R. Costa, X. Solans, Inorg. Chem. 1993, 32, 695. [28] M. A. Halcrow, J. C. Huffman, G. Christou, Angew. Chem. Int. Ed. Engl., 1995, 34, 889. [29] M. A. Halcrow, J.-S. Sun, J. C. Huffman, G. Christou, Inorg. Chem., 1995, 34, 4167. [30] K. Urtiaga, Z. E. Serna, G. Barandika, L. Lezama, M. I. Arriortua, R. Cortes, ´ Proceedings of the XXXIII ICCC, 1998, p. 421. [31] S. Brooker, V. McKee, J. Chem. Soc. Chem. Commun. 1989, 619. [32] J. Ribas, A. Escuer, M. Monfort, R. Vicente, R. Cortes, ´ L. Lezama, T. Rojo, Coord. Chem. Rev. 1999, 193-195, 1027. [33] T. Takeuchi, T,Yosida, K. Inoue, M. Yamashita, T. Kumada, K. Kindo, S. Merah, M. Verdaguer, J. P. Renard, J. Magn. Magn. Mat. 1995, 140-144, 1633 and references therein. [34] A. Escuer, R. Vicente, M. A. S. Goher, F. A. Mautner, Inorg. Chem. 1998, 37, 782. [35] J. J. Borras-Almenar, ´ E. Coronado, J. Curely, ´ R. Georges, Inorg. Chem. 1995, 34, 2699. [36] M. Hagiwara, Y. Narumi, K. Kindo, M. Kohno, H. Nakano, R. Sato, M. Takahashi, Phys. Rev. Lett. 1998, 80, 1312 and references therein. [37] (a) J. Ribas, M. Monfort, C. Diaz, C. Bastos, X. Solans, Inorg. Chem. 1994, 33, 484; (b) I. Resino, PhD Thesis, University of Barcelona, 1999. [38] T. De Neef, Ph. D. Thesis, Eindhoven, 1975. [39] A. Escuer, R. Vicente, F. A. Mautner, M. A. S. Goher, Proceedings of the XXXIII ICCC, 1998, p. 356. [40] J. Ribas, M. Monfort, I. Resino, X. Solans, P. Rabu, F. Maingot, M. Drillon, Angew. Chem. Int. Ed. Engl. 1996, 35, 2520. [41] (a) G. Viau, M. G. Lombardi, G. De Munno, M. Julve, F. Lloret, J. Faus, A. Canneschi, J. M. Clement-Juan, J. Chem. Soc. Chem. Commun. 1997, 1195; (b) J. J. Borras-Almenar, ´ J. M. Clemente-Juan, E. Coronado, F. Lloret, Chem. Phys. Lett. 1997, 274, 79.

References

337

[42] (a) R. Cortes, ´ L. Lezama, J. L. Pizarro, M. I. Arriortua, X. Solans, T. Rojo, Angew. Chem. Int. Ed. Engl. 1994, 33, 2488; (b) R. Cortes, ´ M. Drillon, X. Solans, L. Lezama, T. Rojo, Inorg. Chem. 1997, 36, 677; (c) M. A. M. Abu-Yussef, A. Escuer, M. A. S. Goher, F. A. Mautner, R. Vicente, Eur. J. Inorg. Chem. 1999, 687. [43] A. Escuer, R. Vicente, M. S. El Fallah, S. B. Kumar, F. A. Mautner, D. Gatteschi, J. Chem. Soc. Dalton Trans. 1998, 3905. [44] J. Ribas, M. Monfort, B. Kumar-Ghosh, X. Solans, Angew. Chem. Int. Ed. Eng. 1994, 33, 2087. [45] A. Escuer, R. Vicente, M. A. S. Goher, F. A. Mautner, Inorg. Chem. 1995, 34, 5707. [46] A. Escuer, R. Vicente, M. A. S. Goher, F. A. Mautner, J. Chem. Soc, Dalton Trans. 1997, 4431. [47] A. Escuer, R. Vicente, M. A. S. Goher, F. A. Mautner, Inorg. Chem. 1997, 36, 3440. [48] (a) M. Monfort, J. Ribas, X. Solans, J. Chem. Soc. Chem. Commun. 1993, 350; (b) J. Ribas, M. Monfort, X. Solans, M. Drillon, Inorg. Chem. 1994, 33, 742. [49] M. Monfort, J. Ribas, I. Resino, M. S. El Fallah, H. Stoeckly-Evans, Proceedings of the XXXIII ICCC, 1998, p. 386. [50] CLUMAG Program, D. Gatteschi, L. Pardi, Gazz. Chim. Ital. 1993, 123, 231. [51] A. Escuer, R. Vicente, M. A. S. Goher, F. A. Mautner, Inorg. Chem. 1996, 35. 6386. [52] M. A. S. Goher, N. A. Al-Salen, F. A. Mautner, J. Coord. Chem. 1998, 44, 119. [53] R. Cortes, ´ L. Lezama, T. Rojo, Inorg. Chem, in press. [54] (a) G. De Munno, M. Julve, G. Viau, F. Lloret, J. Faus, D. Viterbo, Angew. Chem. Int. Ed. Engl. 1996, 35, 1807. (b) R. Cortes, ´ L. Lezama, J. L. Pizarro, M. I. Arriortua, T. Rojo, Angew. Chem. Int. Ed. Engl. 1996, 35, 1810. (c) R. Cortes, ´ M. K. Urtiaga, L. Lezama, J. L. Pizarro, M. I. Arriortua, T. Rojo, Inorg. Chem. 1997, 36, 5016. [55] J. L. Manson, A. M. Arif, J. S. Miller, Chem. Commun. 1999, 1479. [56] M. A. S. Goher, F. A. Mautner, Croat, Chim. Acta 1990, 63, 559. [57] F. A. Mautner, R. Cortes, ´ L. Lezama, T. Rojo, Angew. Chem. Int. Ed. Engl. 1996, 35, 78. [58] G. S. Rushbrook, P. J. Wood, Mol. Phys. 1958 1, 257. [59] M. A. S. Goher, J. Cano, Y. Journaux, M. A. M. Abu-Yussef, F. A. Mautner, A. Escuer, R. Vicente, Chem. Eur. J. 2000, 6, 778. [60] (a) R. Cortes, ´ S. Mart´ın, L. Lezama, J. L. Pizarro, M. I. Arriortua, T. Rojo, Inorg. Chem. In press; (b) H-Y. Sen, D-Z. Liao, Z-H. Jiang, S-P. Yan, B-W. Sun, G-L. Wang, X-K. Yao, H-G. Wang, Chem. Lett., 1998, 469.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

10

Oxalate-based 2D and 3D Magnets Melanie Pilkington and Silvio Decurtins

10.1

Introduction

There are a wealth of structures and properties to be generated by combining supramolecular chemistry with materials science [1]. The field of molecule-based magnetism [2], by essence of its supramolecular nature has capitalized on the specific properties of structurally well-defined and versatile supramolecular materials. Two classes of functional supramolecular materials are the two- and three-dimensional network architectures, self-assembled from polyatomic ligands and metallic centers that carry a magnetic moment. bis-Bidentate chelating ligands, such as bipyrimidine and oxalate, among others, have proven to be very useful organic linkers between spin-carrying metal ions [2]. It is well known that the oxalate bridging ligand (ox = C2 O2− 4 ), in particular, is a good mediator of both antiferromagnetic and ferromagnetic interactions between identical and different metal ions. As a consequence, oxalates have been widely used to construct two- and three-dimensional polynuclear compounds in the search for new molecule-based magnets. The aim of this chapter is to give the reader an overview of the factors that control the way in which oxalate complexes self-assemble, highlighting the main structural characteristics of the extended molecular networks, discussing the magnetic properties which result as a consequence of the specific arrangement of the paramagnetic metal ions in space.

10.2

Basic Principles of Specific 2D and 3D Network Configurations

One of the modern challenges in the field of supramolecular materials is the controlled, spontaneous generation of well-defined architectures from both organic and inorganic building blocks. A high structural organization can be ensured through the multiple one- to three-binding of transition-metal ions giving rise to a variety of extended networks in one to three dimensions. The emphasis here lies in the design of the ligands together with the choice of the metal ions, since the overall topology of the network is strongly influenced by the coordination algorithm of the linking metal ions, as well as on the choice of the bridging ligands. The bis-chelating coordinating ability of the oxalate ion has made it a useful and versatile choice as a bridging ligand for the design of two- and three-dimensional networks. Indeed, the self-assembly of inorganic architectures incorporating oxalate ions as the bridging species have yielded various types of inorganic frameworks, which include one-dimensional chains [3, 4], two-dimensional layers [5, 6], and three-dimensional networks [7].

340

10 Oxalate-based 2D and 3D Magnets

Fig. 1. Schematic representation of the chiral (the isomer is shown) preorganized anionic coordination entity.

It is the concept of connectivity which becomes important to address, since it is this which defines the way in which a set of points are connected to construct a lattice which can be infinite in one- to three-dimensions. During the following discussion, we will focus on the fundamental ideas that are relevant to the formation of the two- and three-dimensional framework topologies [8]. Both classes of structure are formally composed of [Mz+ (ox)3 ](6−z)− building blocks, whereby each of these units represents a three-connected point as shown in Fig. 1. These tris-oxalato complexes can in principle polymerize in two ways. The first alternative leads to a 2D honeycomb layered structure, whereas the second possible arrangement results in the formation of an infinite 3D framework. In the former case, building blocks of opposite chirality are alternately linked, which confines the bridged metal ions to lie within a plane, as shown in Fig. 2a, and results in the formation of a two-dimensional layered motif. In contrast, an assembly of building blocks of the same chiral configuration leads to the 3D framework, as shown in Fig. 2b. After having considered the way in which the chirality of the sub-units can influence the lattice dimensionality, it is then possible to apply a series of topological rules to define the number of building blocks that are needed to build closed circuits and hence, extended framework motifs. Generally, the formation of two-dimensionally linked assemblies affords subunits possessing either four-connected points or, as is

a)

b) [Mz+ (ox)3 ](6−z)−

Fig. 2. Chiral building blocks assembled with: (a) alternating chiral configuration, and (b) equal chiral configuration.

10.2 Basic Principles of Specific 2D and 3D Network Configurations

341

Fig. 3. (a) Two dimeric units of the alternating chirality type are necessary to form a closed hexagon ring; (b) the resulting planar network motif.

Fig. 4. (a) Two tetrameric units of the same chirality type are necessary to form a closed decagon ring; (b) a fragment of the 3D, chiral framework.

actually the case, a combination of two three-connected points. Fig. 3, illustrates the way in which subsequently two dimeric subunits may be combined to form a planar honeycomb network with the general stoichiometry [MII MIII (ox)3 ]n− n . In an analogous manner, it can easily be seen that two tetrameric subunits are needed to build closed circuits composed of ten metal centers, which in sum define a three-dimensional decagon framework, see Fig. 4. Such tetrameric subunits are necessary, because only four three-connected points (Z = 4) combined together have the necessary number of six free links to build the 3D net. Identically oriented links repeat at intervals of (Z + 1) points, so that circuits of 2(Z + 1) points are formed. Thus, the final structure represents a uniform net in the sense that the shortest path, starting from any point along any link and returning to that point along any other link, is a circuit of ten points. In addition, as explained

342

10 Oxalate-based 2D and 3D Magnets

Fig. 5. Stereoview of the 3-connected 10-gon (10,3) network topology.

above, the topological principle implies that for the 3D case, only subunits of the same chirality are assembled. Consequently, the uniform anionic 3D network-type 2n− or [MI MIII (ox) ]2n− (MII/III = transition-metal with a stoichiometry of [MII 3 n 2 (ox)3 ]n I ions and M = Li, Na) is chiral, since it is composed of 2n centers exhibiting the same kind of chirality. Naturally, this chiral topology is in line with the symmetry elements which are present in the crystalline state of these 3D frameworks, which in sum constitute either one of the enantiomorphic cubic space groups, P43 32 or P41 32 for the former, and the cubic space group P21 3 for the latter stoichiometry. Thereby, the 2n metal ion centers occupy special sites with a threefold symmetry axis. Fig. 5 shows a stereoview of this network topology. Extended helical geometries are also encountered through the three-dimensional repetitive assembling of subunits with helical chirality. Thus, this framework structure may alternatively be seen as composed of either right-handed (-chirality) or left-handed (-chirality) helices with 41 , 43 or 21 screw axes, running in three perpendicular directions, while simultaneously being covalently linked to each other. Fig. 6 exhibits such a helical strand extending parallel to a screw axis. Finally, the discrimination between the formation and crystallization of either 2D or 3D anionic frameworks with analogous network stoichiometries depends on the choice of the templating counter-ions. In particular, [XR4 ]+ , (X = N, P; R = phenyl, nalkyl) cations have been found to initiate the growth of 2D sheet structures containing [MII MIII (ox)3 ]n− n , network stoichiometries [6]. Fig. 7 shows two projections of the structure of a compound that crystallizes in 2D honeycomb layers. This thinking applies in particular, when planning the design of a chiral threedimensional supramolecular host-guest system, since the mutual interaction of the two distinct complementary molecular units or coordination entities is necessary. Examples of this methodology include the above described anionic, tris-chelated transition-metal oxalato complexes [Mz+ (ox)3 ](6−z)− which assemble to the host

10.2 Basic Principles of Specific 2D and 3D Network Configurations

343

Fig. 6. View of a helical strand along a 21 axis from a 3D type compound [9]. [MI MIII (ox)3 ]2n− n

Fig. 7. Sector from the [N(Bu)4 ][MnII FeIII (ox)3 ] layer compound. (a) [001] projection; (b) [010] projection [10].

344

10 Oxalate-based 2D and 3D Magnets

system together with the cationic, tris-chelated transition-metal diimino complexes, e. g. [M(bpy)3 ]2+/3+ , bpy = 2,2 -bipyridine, which play the role of the guest compounds.

10.3

Structural Studies on 2D Oxalato Bridged Compounds

Several X-ray crystal structures of two-dimensional oxalato bridged mixed-metal networks have been reported since 1993. The first structural information on 2D oxalates was obtained by Atovmyan et al. [5], who succeeded in growing single crystals of the compound [NBu4 ][MnII CrIII (ox)3 ] by slow diffusion of aqueous solutions of a mixture of NBu4 Br with K3 [Cr(ox)3 ].3H2 O and MnCl2 in an H-shaped tube. This structure revealed an anionic two-dimensional network consisting of µ-oxalato bridged MnII and CrIII ions, with NBu+ 4 cations located between the layers. One of the butyl groups of the cations penetrates a void in the neighboring anionic layer and the separation between two adjacent anionic layers is 8.95 Å. Shortly thereafter, Decurtins and et al. [6] obtained single crystals of the compound [PPh4 ][MnII CrIII (ox)3 ], the structure of which is comparable to that of Atovmyan described above (space group R3c). It comprises of an alternating assembly of the two types of chiral [M(ox)3 ] building blocks with and configurations, adopting a two-dimensional honeycombed layered structure with a network stoichiometry of [MnII CrIII (ox)3 ]n− n . There are six pairs of anionic and cationic layers per unit cell and the distance between two adjacent layers is reported to be 9.5 Å. Those phenyl groups of the cations that point vertically in the direction of the anionic layers just fit into the slightly ellipsoid-shaped vacancies. In a subsequent communication [10], the single-crystal X-ray crystal structures of two additional compounds with stoichiometries [NPr4 ][MnII CrIII (ox)3 ] and [NBu4 ][MnII FeIII (ox)3 ] were also reported. Fig. 8, shows the [110] projection of the former compound that crystallizes in the space group R3c. The compound with the latter stoichiometry crystallizes in the hexagonal space group P63 . Day and co-workers [11] have carried out single-crystal X-ray crystallographic studies on the compound [N(n-C5 H11 )4 ][MnII FeIII (ox)3 ] which exhibits a twodimensional network connectivity pattern in accordance with the above findings. In this case, the special disposition of the organic cations interleaving the honeycombed anionic layers (whereby the cavities are occupied by two methyl groups approaching from opposite sides) leads to the compound crystallizing in the orthorhombic space group C2221 . An alternative approach has been devised by Coronado et al. [12, 13] which involves the combination of two magnetically active sub-lattices, namely, the bimetallic oxalato-bridged honeycombed network [MII MIII (ox)3 ] (MII = Cr, Mn, Fe, Co, Ni, Cu; MIII = Cr, Fe) and the organometallic cation decamethylferrocenium. The crystal structure was solved for the MnII FeIII derivative in the monoclinic space group C2/m and in this case, the decamethylferrocenium cations intercalate the oxalato-bridged layers without penetrating into the vacancies.

10.3 Structural Studies on 2D Oxalato Bridged Compounds

345

Fig. 8. [110] projection of [N(n-C3 H7 )4 ][MnII CrIII (ox)3 ]. Black atoms indicate MnII positions [10].

Day et al. [14-16] have also reported detailed investigations of the synthesis and structural characterization of a series of mixed-valence compounds with stoichiometries of the type [A][CrII CrIII (ox)3 ] and [A][FeII FeIII (ox)3 ], together with mixedmetal compounds of stoichiometry [A][MnII FeIII (ox)3 ]. In these studies, all of the phases prepared with a wide variety of organic template cations [A] were structurally characterized by X-ray powder diffraction, and the reflections indexed and unit cell constants refined on a hexagonal cell; all of the these structures were proven to

346

10 Oxalate-based 2D and 3D Magnets

be consistent with a 2D honeycomb network. The systematic variation of [A] was aimed at modulating the separation between the layers and hence varying the physical properties.

10.4

Magnetic Studies on 2D Oxalato Bridged Compounds

These 2D polymeric materials all exhibit a wide range of magnetic behavior, and ferro-, ferri- or antiferromagnetic long-range ordering processes, as well as shortrange correlations and spin glass-like phenomena have all been observed. Generally, the magnetic interactions and magnetic ordering in the [A][MII MIII (ox)3 ] materials is strongly dependent on the character of the cations [A] as well as the combination of MII MIII metal ions. In the early 90’s Okawa et al. [17, 18]. reported a synthetic and magnetic study on ferri- and ferromagnetic mixed metal assemblies with the stoichiometries [NBu4 ] [MII MIII (ox)3 ], (MII = Mn, Fe, Co, Ni, Cu, Zn; MIII = Cr, Fe). During this time, these compounds were falsely assumed to have a 3D network structure, but in subsequent reports by Atovmyan [5] and Decurtins [6], the verification of the two-dimensional class of structure has since been proven. The series of CrIII derivatives show ferromagnetic phase transitions with TC < 14 K and magnetic hysteresis loops with coercive fields HC < 320 G. The FeIII compounds with MII = Fe, Ni have been shown to behave as ferrimagnets with magnetic phase transition temperatures of 43 K and 28 K, respectively [17]. In 1994, Day et al. [15] reported the results from a detailed investigation of the mixed valence ferrimagnets [A][FeII FeIII (ox)3 ] (A = quaternary ammonium or phosphonium). For this series of compounds, large negative values of the Weiss constant θ indicate that the near-neighbor FeII –FeIII magnetic exchange interaction is antiferromagnetic. Surprisingly, for the A = NBu4 sample, a magnetization study revealed a strongly negative magnetization value below 30 K (see Fig. 9). This effect originates from a temperature dependence of the magnetization that varies differently for the two magnetic sub-lattices. This is the first reported example of this phenomenon in the field of molecule-based magnets. In a later communication [19], these authors present a more comprehensive study on the temperature dependent magnetization in these [A][FeII FeIII (ox)3 ] layer compounds, and discuss the manifestation of Neel ´ N and Q type ferrimagnetism. The major aim of these studies being to map and identify the reason which accounts for the presence of negative magnetization at low temperature for the NBu4 salt. According to Neel, ´ the ground state of a ferrimagnet is determined by the saturation magnetization of each magnetic sub-lattice and their relative ordering rates with respect to temperature. For the honeycomb lattice with alternating FeII and FeIII ions, this situation corresponds to an initially steeper ordering on the FeII sub-lattice, as shown schematically in Fig. 10. From a related study [20] namely, neutron spin polarization analysis of the magnetic ordering of [P(C6 D5 )4 ][FeII FeIII (ox)3 ], it was concluded that for this

10.4 Magnetic Studies on 2D Oxalato Bridged Compounds

347

Fig. 9. Plot of magnetization against temperature for AFeII FeIII (ox)3 [A = NBun4 (); NPrn4 ()] cooled in a field of 100 G [15].

Fig. 10. Magnetic sublattice ordering in a ferrimagnet containing FeII and FeIII with Neel ´ type order [19].

compound, no long-range magnetic order exists as a consequence of the random anisotropy effects introduced by the different electronic configurations of the FeII and FeIII ions. The mixed valence compounds with stoichiometry [A][CrII CrIII (ox)3 ], (A = NBu4 , P(Ph)4 ) have also been synthesized and fully characterized structurally and magnetically [14]. Again, although short-range antiferromagnetic correlations are observed below 100 K, no transitions to a long-range ordered state were observed above 2 K. For the network of stoichiometry [MnII FeIII (ox)3 ], one encounters the rare situation of a bimetallic lattice in which both metal ions have the same electronic ground state namely, 3d5 , S = 5/2. In this case, the compound has been found to mimic antiferromagnetic behavior [11]. Thus, the temperature dependence of the magnetic

348

10 Oxalate-based 2D and 3D Magnets

Fig. 11. Proposed magnetic structure of [P(C6 D5 )4 ][MnII FeIII (ox)3 ] [21].

susceptibility is that of a classic 2D antiferromagnet, reaching a broad maximum at 55 K. Definitive evidence for this behavior comes from a neutron powder diffraction study [21] by Day et al., which revealed a simple collinear antiferromagnetic alignment of MnII and FeIII moments parallel to the c-axis, or in other words, perpendicular to the layers, see Fig. 11. The first published work from elastic neutron scattering studies on the magnetic structure of a layered bimetallic tris-oxalate salt comes from Decurtins et al. [10]. and refers to a MnII CrIII phase which orders ferromagnetically at Tc = 6 K. The best agreement between the observed and calculated magnetic neutron intensities of the diffraction pattern was achieved with a collinear ferromagnetic arrangement of both of the MnII and CrIII spins along the c-axis. This picture is consistent with the results of a single-crystal magnetization experiment that showed the c-axis to be the easy axis of magnetization. In fact, the same conclusions had already been drawn by Atovmyan et al. in two earlier publications [5, 22]. In addition, a thermal and magnetic study [23] by calorimetric and magnetic ac and dc techniques on the compounds [P(Ph)4 ][MII CrIII (ox)3 ] (MII = Mn, Fe) focuses on the elucidation of the actual magnetic dimensionality of these compounds. These results show that the MnII derivative displays low dimensional magnetic character and provide evidence for the onset of 3D magnetic ordering. In contrast, experimental results for the FeII derivative indicate a spinglass-like behavior. The difference in magnetic behavior observed for these two compounds has been assigned to the effect of random anisotropy which could arise as a consequence of structural disorder prevalent in this family of compounds.

10.5 Structural Studies on 3D Oxalato Bridged Compounds

349

In 1998, Kahn et al. [24]. reported on the magnetic properties of the 2D bimetallic compounds with stoichiometries [NBu4 ][MII RuII (ox)3 ], (MII = Mn, Fe, Cu). The MnII derivative shows ferromagnetic RuII -MnII interactions, but no long-range ordering was observed down to 2K. The FeII and CuII compounds exhibit a ferromagnetic behavior, whereas the former sample reveals a long-range magnetic ordering at Tc = 13 K. The susceptibility and magnetization data of the latter compound could not yet be interpreted satisfactorily.

10.5

Structural Studies on 3D Oxalato Bridged Compounds

The first structure of a 3D transition metal network incorporating the oxalate ion was that of [Ni(phen)3 ][KCoIII (ox)3 ].2H2 O, phen = 1,10-phenanthroline, reported by Snow and co-workers in 1971 [25]. The true dimensionality of this compound however went unrecognized during this period, and the potential of oxalate ions to form 3D networks was not fully realized until 1993, when Decurtins et al. published the crystal structure of the iron(II)oxalato complex with tris(2,2 -bipyridine)iron(II) cations [7]. II 2n− and This compound has an overall stoichiometry of [FeII (bipy)3 ]2+ n [Fe2 (ox)3 ]n forms a 3D anionic polymeric network that is best described with the 3-connected 10-gon network topology. A stereoview of this anionic network has already been given in Fig. 5. The possibilities for linking paramagnetic metal ions via oxalate bridges in threedimensional lattices and their possible applications in the field of molecule-based magnets have since provided the momentum for subsequent investigations into the self-assembly of 3D oxalato bridged metal complexes. In a subsequent communication, Decurtins et al. [9] reported the structural characterization of three additional 2n compounds, one homometallic with the same network stoichiometry ([MnII 2 (ox)3 ]2 ) II as the Fe compound above, and two bimetallic with network stoichiometries of and [LiCrIII (ox)3 ]2n− [NaFeIII (ox)3 ]2n− n n , respectively. Both types of anionic networks form an analogous 3D pattern, but their differences in stoichiometry result in them 2n− crystallizing in different space groups. The [MnII network is iso-structural 2 (ox)3 ]2 with the iron(II) compound and the structure of the two heterometallic compounds only differs in the respect that the overall space group symmetry is lowered to P21 3. According to the lower symmetry, the asymmetric unit contains a complete oxalate ligand, both metals of the network and the complete bipyridine ligand. The planar oxalate ligands repeatedly bridge the metal ions in all three dimensions producing chiral left- or right-handed helical strands. Fig. 12 shows the [100] projection of the net. The tris-chelating [MII (bipy)3 ]2+ cations occupy the vacancies [LiCrIII (ox)3 ]2n− n within the network. In 1996, a third related study came from Decurtins et al. [26], the aim of which was to investigate the factors which contribute to the structural and chemical flexibility of this type of three-dimensional network. Since it had previously been established that the tris-chelated cations play an important role in initiating the formation and crystallization of the 3D nets, the structural response towards varying the cations, i. e.

350

10 Oxalate-based 2D and 3D Magnets

Fig. 12. [100] projection of 3D network of [FeII (bipy)3 ][LiCrIII (ox)3 ] [9].

using a [MIII (bpy)3 ]3+ (MIII = Cr, Co) template or the bulkier [NiII (phen)3 ]2+ cations was investigated. In the former case, the system achieves this with an elaborate inclusion of an additional complex anion, while keeping the known three-dimensional topology (described above), crystal symmetry and lattice parameters fixed. It does so by encapsulating these anions in cubic-shaped empty spaces formed by six of the planar bipyridine ligands from three adjacent cations. Fig. 13 shows a view of the packing arrangement of three adjacent tris-chelated cations encapsulating a perchlorate anion. In the latter case, the framework demonstrates a marked flexibility in modifying both the shape and width of the cavities in order to accommodate the steric requirements of the bulky cations. The possibility of using 1,2-dithiooxalate as an isosteric replacement to the oxalate bridging ligand is also demonstrated, leading to the first reported example of a three-dimensional dithiooxalate-bridged compound. This compound with stoichiometry [NiII (phen)3 ][NaCoIII (dto)3 ] has a topology which is once again consistent with the chiral three-dimensional (10,3) network. In contrast to the previously reported structures, this compound crystallizes in the orthorhombic space group P21 21 21 and as a consequence, there is no longer a threefold crystallographic axis imposed on the tris-chelated metal ions and the cations within the network. The 1,2-dithiooxalate ligands (C2 O2 S2− 2 ) are approximately planar but. One important structural difference worth comment is the discrimination in the co-ordination behavior of the bridging 1,2-dithiooxolate ligand, since bonding to cobalt(III) occurs through the sulfur atoms, whereas the sodium atoms bind to the oxygen ends of the bridging ligand. The interatomic Na-O distances with a mean value of 2.376(11) Å can be compared with the mean value of 2.319(3) Å in the oxalate bridged network [NaFeIII (ox)3 ]2n− n

10.5 Structural Studies on 3D Oxalato Bridged Compounds

351

Fig. 13. View of three adjacent [Cr(bipy)3 ]3+ complexes exhibiting the cavity formed by three pairs of parallel oriented bipyridine ligands. The center of the cage is occupied by a [ClO4 ]− anion [26].

[9]. The cation packing arrangement is similar to that in the cubic oxalate-bridged compounds, namely, three adjacent cations with their V-shaped pockets form a sufficiently large cavity to accommodate additional molecules. These series of compounds clearly illustrate that the ten-gon framework is indeed flexible and can readily distort to account for changes in the nature of the ligand or the cations. In recent years Decurtins et al. [27, 28] have reported details regarding the crystal structures of several additional anionic 3D polymeric networks. The crysIII 2n− have a regular tetrahedral tals of the compound [FeII (bipy)3 ]2+ n [AgCr (ox)3 ]n morphology that is consistent with the cubic space group P21 3 and indeed, this compound is isostructural with the previously described heterometallic [MI MIII (ox)3 ]2n− n chiral three-dimensional polymeric networks. Furthermore, the above authors have repeatedly demonstrated that it is possible to apply a straightforward series of topological rules to realize the same 3D structural type for many different transition metal ions, the most recent being the compounds of stoichiometries III 2n− with MIII = RuIII , RhIII . The study of these com[RuII (bipy)3 ]2+ n [NaM (ox)3 ]n pounds was of particular interest for their photophysical properties [29]. Applying this methodology, Julve and co-workers have recently synthesized and reported the crystal structure of the chiral three-dimensional oxalatobridged compound [CoIII (bpy)3 ][CoII 2 (ox)3 ]ClO4 [30]. As expected this compound is isostructural with the compounds of formula [CrIII (bpy)3 ][MnII 2 (ox)3 ]ClO4 and [CrIII (bipy)3 ][MnII 2 (ox)3 ]BF4 reported in 1996 by Decurtins et al. [26]. All of the above described chiral three-dimensional structures were obtained from racemic starting materials. Very recently however, Verdaguer et al. [31] have applied this methodology to exploit the feasibility of the self-assembly of chiral two- and three-dimensional oxalato-bridged polymeric networks starting from homochiral subunits namely resolved [CrIII (ox)3 ]3− and [MII (bpy)3 ]2+ (MII = Ni, Ru) building blocks. Naturally, the topology of the three-dimensional systems is consistent with the chiral three-dimensional three-connected ten-gon (10,3) network

352

10 Oxalate-based 2D and 3D Magnets

configuration achieved from the enantiomeric mixture of the precursors. Solid-state circular dichroism (CD) measurements demonstrate the enantiomeric character of the resulting polymers.

10.6

Magnetic Studies on 3D Oxalato Bridged Compounds

Three-dimensional homo- and heterometallic oxalato bridged frameworks are examples of supramolecular host/guest compounds which show long-range magnetic ordering, as well as various kinds of photophysical properties [29, 32]. The existence of a magnetically ordered phase was first deduced from magnetic susceptibility measurements on the three-dimensional compound [FeII (bpy)3 ][MnII 2 (ox)3 ] [9]. The experimental data revealed a rounded maximum at about 20 K in the plot of χM against T (thus TN < 20 K) and a Weiss constant θ of −33 K in the plot of 1/χM against T . This long-range magnetic ordering originates from the exchange interactions between neighboring MnII ions, mediated by the bridging oxalate ligands. This compound was the first three-dimensional oxalato network for which the magnetic structure could be solved [33]. Neutron diffraction experiments were performed on a polycrystalline sample in the temperature range of 30 K to 1.8 K. As anticipated, an increase in the magnetic neutron intensities due to the long-range antiferromagnetic ordering of the spins of the MnII ions was detected. Fig. 14 illustrates the observed [difference I (1.8 K) − I (30 K)], calculated and difference magnetic neutron diffraction patterns. The increase of the neutron intensities corresponds to a propagation vector K = 0, thus the magnetic unit cell is equal to the chemical cell. The temperature dependence of the dominant magnetic intensity of the diffraction peak at (210) (2θ = 21.1◦ ), indicates an ordering temperature TN = 13(0.5) K, in good agreement with the magnetic susceptibility experiments. The best agreement between observed and calculated magnetic neutron intensities was achieved with a collinear, antiferromagnetic arrangement of the MnII moments according to the three-dimensional irreducible representation τ4 , which is derived from the enantiomorphic pair of the chiral, cubic crystallographic space groups P43 32/P41 32 [34]. The ordered magnetic moment at 1.8 K amounts to µMn = 4.6(1) µB , where µB is the electron Bohr magneton. The saturation magnetization, MS , is related to the equation MS = g × µB × N × S, where S is the spin quantum number, N the Avogadro number and g the electron g-factor. Thus, a g × S value corresponding to the number of unpaired electrons of 4.6 is obtained, which is compatible with the expected five unpaired electrons (g = 2) from the MnII ions. Figure 15 depicts the pattern of the magnetic structure with the 3D manganese(II) network. Despite the three-dimensional helical character of the framework structure incorporating the magnetic ions, a two-sublattice antiferromagnetic spin arrangement has been proven, hence no helimagnetic structure was apparent and this behavior is in accordance with the typical isotropic character of the Heisenberg MnII ion.

10.6 Magnetic Studies on 3D Oxalato Bridged Compounds

353

Fig. 14. Observed [difference I (1.8 K) − I (30 K)] calculated, and difference magnetic neutron diffraction patterns of a polycrystalline sample of [Fe(bpy)3 ][MnII 2 (ox)3 ] [33].

Fig. 15. Scheme of antiferromagnetic collinear configuration of the magnetic moments originating from the MnII ions, which constitute the chiral 3D network compound [33].

Magnetic studies for the isostructural three-dimensional networks of stoichiomeII II II tries [CoIII (bpy)3 ][CoII 2 (ox)3 ]ClO4 (1) and [M (bpy)3 ][Co2 (ox)3 ] [M = Fe (2) and Ni (3)] recently carried out by Julve et al. [30], show the occurrence of a weak ferromagnetism at low temperatures for compounds (1) (Tc = 9.0 K) and (2) (Tc = 6.5 K), whereas no magnetic ordering was observed for compound (3). This magnetic ordering at very low temperatures is attributed to the spin canting of the magnetic moments of the metal ions in the chiral three-dimensional oxalate-bridged cobalt(II) net. These results also suggest that the magnetic ordering of the three-dimensional 2n− anionic framework is strongly dependent on the size and diamag[CoII 2 (ox)3 ]n netic or paramagnetic character of the tris-chelated counter ions used to achieve the electroneutrality. In this report, the influence of the size of the [MII/III (bpy)3 ]2+/3+

354

10 Oxalate-based 2D and 3D Magnets

complex on the spin canted magnetic structure is also discussed, but further work is needed to investigate these ideas in greater detail.

10.7

Summary

In the middle of the last century, Watson and Crick solved the structure of DNA which provides us with one of the most important demonstrations of how nature has evolved an efficient self-assembly process for the synthesis of a complex functional supramolecular system from simple molecules. Following in the footsteps of nature, the scientific challenge, to find new methods for the controlled self-assembly of complex molecules from simple building blocks opens up exciting prospects for the future design of molecular materials with useful sets of properties such as magnetism, photophysics and electronics. Until fairly recently however, these aims have been frustrated by the absence of reliable and general structural paradigms needed for the systematic design of crystal lattices with predictable structure and desirable functions. The field of oxalate-based 2D and 3D molecule-based magnets highlights the progress chemists have made in a relatively new area of materials chemistry, namely molecule-based magnets to address the concept of controlling intermolecular interactions in order to design crystal lattices with predictable structures. It has clearly been demonstrated that it is certainly possible to develop a simple set of topological rules for the design of a supramolecular system that exhibits self-organization, and as a consequence this work offers exciting prospects for the future design of new supramolecular materials. These two classes of compounds have also provided scientists with new sources of materials for study, and there is indeed a wealth of experimental results reported in the chemical literature from magnetic susceptibility and magnetization experiments. Furthermore, the physical methods themselves for characterizing these new materials have also advanced, with more experience gained in the use of elastic neutron diffraction methods to elucidate magnetic structures, which moves forward the frontiers of materials research. This review has demonstrated the versatility of the oxalate ion which enables it to play a key role in the formation of a whole class of transition-metal based supramolecular host/guest systems. The future in this field lies in the exploitation of these host-guest solids, where each component is able to contribute its own set of physical characteristics. The search for synergistic properties within this class of multifunctional materials is currently an active and ongoing area of research and should provide many novel and interesting classes of compounds in the future.

References

355

References [1] J.-M. Lehn. Supramolecular Chemistry, Concepts and Perspectives, VCH, Weinheim, 1995. [2] O. Kahn. Molecular Magnetism, VCH, Weinheim, 1993. [3] J.J. Girerd, O. Kahn, M. Verdaguer, Inorg. Chem. 1980, 19, 274–276. [4] H. Oshio, U. Nagashima, Inorg. Chem. 1992, 31, 3295–3301. [5] L.O. Atovmyan, G.V. Shilov, R.N. Lyubovskaya, E.I. Zhilyaeva, N.S. Ovanesyan, S.I. Pirumova, I.G. Gusakovskaya, JETP Lett. 1993, 58, 766–769. [6] S. Decurtins, H.W. Schmalle, H.R. Oswald, A. Linden, J. Ensling, P. Gutlich, ¨ A. Hauser, Inorg. Chim. Acta 1994, 216, 65–73. [7] S. Decurtins, H.W. Schmalle, P. Schneuwly, H.R. Oswald, Inorg. Chem. 1993, 32, 1888– 1892. [8] A.F. Wells, Structural Inorganic Chemistry, Clarendon Press, Oxford, 1984. [9] S. Decurtins, H.W. Schmalle, P. Schneuwly, J. Ensling, P. Gutlich, ¨ J. Am. Chem. Soc. 1994, 116, 9521–9528. [10] R. Pellaux, H.W. Schmalle, R. Huber, P. Fischer, T. Hauss, B. Ouladdiaf, S. Decurtins, Inorg. Chem. 1997, 36,2301–2308. [11] S.G. Carling, C. Mathoniere, ` P. Day, K.M.A. Malik, S.J. Coles, M.B. Hursthouse, J. Chem. Soc., Dalton Trans. 1996, 1839–1843. [12] M.Clemente–Leon, ´ E. Coronado, J. Galan–Mascaros, C. Gomez–Garcia, J. Chem. Soc., Chem. Commun. 1997, 1727–1728. [13] E. Coronado, J. Galan–Mascaros, C. Gomez–Garcia, J. Ensling, P. Gutlich, ¨ Chem. Eur. J. 2000, 6, 552–563. [14] C.J. Nuttall, C. Bellitto, P. Day, J. Chem. Soc., Chem. Commun. 1995, 1513–1514. [15] C. Mathoniere, ` S.G. Carling, D. Yusheng, P. Day, J. Chem. Soc., Chem. Commun. 1994, 1551–1552. [16] C. Mathoniere, ` C.J. Nuttall, S.G. Carling, P. Day, Inorg. Chem. 1996, 35, 1201–1206. [17] H. Tamaki, M. Mitsumi, K. Nakamura, N. Matsumoto, S. Kida, H. Okawa, S. Iijima, Chem. Letts. 1992, 1975–1978. [18] H. Tamaki, Z.J. Zhong, N. Matsumoto, S. Kida, M. Koikawa, N. Achiwa, Y. Hashimoto, H. Okawa, J. Am. Chem. Soc. 1992, 114, 6974–6979. [19] C.J. Nuttall, P. Day, Chem. Mater. 1998, 10, 3050–3057. [20] D. Visser, S.G. Carling, I.D. Watts, P. Day, K.H. Andersen, Physica 1996, B267–268, 266–269. [21] C.J. Nuttall, P. Day, Inorg. Chem. 1998, 37, 3885–3888. [22] L.O. Atovmyan, G.V. Shilov, N.S. Ovanesyan, A.A. Pyalling, R.N. Lyubovskaya, E.I. Zhilyaeva, Y.G. Morozov, Synthetic Metals 1995, 71, 1809–1810. [23] G. Antorrena, F. Palacio, M. Castro, R. Pelaux, S. Decurtins, J. Magn. Magn. Mater. 1999, 196–197, 581–583. [24] J. Larionova, B. Mombelli, J. Sanchiz, O. Kahn, Inorg. Chem. 1998, 37, 679–684. [25] K.R. Butler, M.R. Snow, J. Chem. Soc. (A) 1971, 565–569. [26] S. Decurtins, H.W. Schmalle, R. Pellaux, P. Schneuwly, A. Hauser, Inorg. Chem. 1996, 35, 1451–1460. [27] H.W. Schmalle, R. Pellaux, S. Decurtins, Z. Kristallogr. 1996, 211, 533–538. [28] R. Pellaux, S. Decurtins, H.W. Schmalle, Acta Crystallogr. 1999, C55, 1057–1079. [29] M.E. von Arx, L. van Pieterson, E. Burattini, A. Hauser, R. Pellaux, S. Decurtins, J. Phys. Chem. A, 2000, 104/05, 883–893. [30] M. Hernandez–Molina, F. Lloret, C. Ruiz, M. Julve. Inorg. Chem. 1998, 37, 4031.

356

10 Oxalate-based 2D and 3D Magnets

[31] R. Andres, ´ M. Gruselle, B. Malezieux, ´ M. Verdaguer, J. Vaissermann, Inorg. Chem. 1999, 38, 4637–4646. [32] M.E. von Arx, A. Hauser, H. Riesen, R. Pellaux, S. Decurtins. Phys. Rev. B. 1996, 54, 1500–15807. [33] S. Decurtins, H.W. Schmalle, R. Pellaux, R. Huber, P. Fischer, B. Ouladdiaf, Adv. Mater. 1996, 8, 647–651. [34] W. Sikora, PC–program MODY in Collected abstracts of a “Workshop on Magnetic Structures and Phase Transitions”, Krakow, Poland, 1994.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

11

Hybrid Organic-Inorganic Multilayer Compounds: Towards Controllable and/or Switchable Magnets Pierre Rabu, Marc Drillon, Kunio Awaga, Wataru Fujita, and Taketoshi Sekine

11.1

Introduction

The past quarter of a century has witnessed substantial efforts in the development of intercalation chemistry, with the combination at the nanometer scale of molecular species and inorganic layered networks. Several layered host structures, e. g. clays, chalcogenides or layered double hydroxides (LDHs), with structural flexibility as regards the interlayer separation, have been investigated. The capacity of the inorganic network to participate in redox or acid-base reactions, and that of the intercalated organic molecule to undergo selective, controllable, and reversible reactions, have generated attractive chemical applications [1, 2]. Likewise, solid-state physicists have been interested by magnetic phase transitions and related phenomena, including the Kosterlitz-Thouless transition and the spin crossover [3]. There is usually no real chemical bond between the guest and host sub-networks in cation-exchangeable layered materials. The cohesion is basically due to electrostatic or van der Waals interactions, and, when allowed, weak hydrogen-bonding. Anion-exchangeable layered compounds appear in turn well-adapted to favor covalent links between organic and inorganic constituents, because the anions are usually bonded to the metal ions and form the framework of the crystal. This has been achieved in series of layered transition metal hydroxides in which large organic anions have been inserted [4]. Hybridization of the organic and inorganic components to obtain cooperative property materials, i. e. whose properties differ from those of the individual assemblies, is a major objective. Striking results have reported hybrid compounds combining two basic properties, e. g. conductivity and magnetism or optics and magnetism [5–8]. Thus, a multi-functional system has been achieved by intercalation of the hyperpolarizable dimethylamino-N-methyl stilbazolinium (DAMS) chromophore in between MnPS3 inorganic layers [9]. This compound associates both spontaneous magnetization below 40 K, and remarkable second-order non-linear optical properties (two orders of magnitude better than those of urea). In addition, it is possible to control and/or switch the physical properties of the inorganic layer by a structural modification of the organic layer. In this paper, we report first the ion-exchange reactions and correlations between structure and magnetic properties in the layered metal hydroxides M2 (OH)3 X, where M is a divalent metal and X an exchangeable anion. The structural and magnetic influence of large organic anions (i. e. aliphatic chain anions) coordinating the metal ion is discussed. Striking results are obtained when: (i) well separated inorganic

358

11 Hybrid Organic-Inorganic Multilayer Compounds

layers have a net moment, favoring a ferromagnetic-like 3D order; (ii) radical-based anions are linked to the hydroxide layers giving hybrid multilayer magnets; and (iii) there are connections between the magnetic layers which favor through-bond spin-polarization interactions. The structural transformations of the organic molecules in layered copper hydroxides, triggered by an external stimulus (chemical reagent, light illumination), and the drastic change of the magnetic properties are then discussed. Likewise, the influence of hydrostatic high pressure on the magnetic properties is reported for the layered compounds (RNH3 )2 CuCl4 , consisting of a two-dimensional infinite square lattice of copper(II) bridged by halide ions. It is shown that, because of the local anisotropy of the metal ions, pressure-induced ferromagnetism may be realized. The results demonstrate that the organic/inorganic layered compounds open interesting prospects in the field of controllable and/or switchable magnetic materials.

11.2

Hydroxide-based Layered Compounds

− Botallackite-type layered hydroxides, Cu2 (OH)3 X (X = NO− 3 , CH3 COO , etc.), are known as two-dimensional magnetic materials. Fig. 1 schematically depicts the 2D network of [Cu2 (OH)3 ]+ [10]. This is similar to the well-known structure of Cd(OH)2 , from which the structure of Cu2 (OH)3 X is obtained by replacing periodically every fourth OH− with X− . There are two crystallographically distinct copper atoms lying in 4 + 2 (OH− + X− ) and 4 + 1 + 1 (OH− + OH− + X− ) elongated octahedral environments with Jahn–Teller distortions. It is characteristic that the neighboring copper(II) ions are bridged by two oxygen atoms of OH− and/or X− and that the magnetic system in the layer can be regarded as a 2D triangular lattice

Fig. 1. 2D network of Cu2 (OH)3 X along the c axis.

11.2 Hydroxide-based Layered Compounds

359

Fig. 2. Magnetic behavior of Co2 (OH)3 (NO3 ) (squares) and of Co(OH)2 (triangles), illustrated as plots of χ T = f (T ) and M = f (H ) (inset).

of Heisenberg S = 1/2 spins. The anion X− coordinating via the dz2 orbital of the copper(II) ion makes a layer which is sandwiched by the copper hydroxy layers. From the magnetic study of Cu2 (OH)3 (NO3 ), there is clear evidence of antiferromagnetic (AF) in-plane interactions, and of long-range magnetic ordering below TC = 12 K [11]. Basically, such a compound can be regarded as a good prototype of a frustrated 2D system, which theoretically is still controversial as regards to the nature of the ground state. The magnetic behavior of the parent compound Co2 (OH)3 (NO3 ), isostructural with the above compound, is totally different [12]. On cooling, the product χ T (Fig. 2) increases significantly up to a sharp maximum (at ca. 10 K), characteristic of ferromagnetic in-plane interactions; it then drops to zero on further cooling. The field-dependent magnetization has a characteristic metamagnetic transition and a saturation at ca. 1.45 µB , in agreement with the expected value. Similar behavior is observed for Co(OH)2 [12, 13]. Clearly, this magnetic behavior is indicative of two regimes; above the critical temperature TC , ferromagnetic in-plane interactions dominate, while at lower temperatures the interlayer interactions are no longer negligible, and long-range antiferromagnetic order occurs. According to the variation of the basal spacing, from 4.65 Å (hydroxide) to 6.95 Å (hydroxide nitrate), the twodimensional character is expected to be more efficient for Co2 (OH)3 (NO3 ) than for Co(OH)2 . This agrees with the occurrence of a metamagnetic transition for both compounds, with critical fields at T = 4 K ranging from 0.17 T for the hydroxide nitrate to 1.5 T for the hydroxide. Finally, if interlayer interactions are needed for the occurrence of long range 3D ordering, the divergence of the in-plane correlation length ξ close to TC seems to be the driving force, as given by the relationship kTC ≈ ξ 2 j S 2 [14].

360

11.3

11 Hybrid Organic-Inorganic Multilayer Compounds

Anion-exchange Reactions

Exchange of interlamellar anions by organic species (dicarboxylic acid anions) was first reported by Miyata and Kumura [15]. The interlayer arrangement of anionic surfactants has been extensively studied in layered double hydroxides (LDHs) [MII 1−x x+ − MIII x (OH)2 ] Ax , consisting of positively charged hydroxide layers separated by A− anions counter-balancing the positive charge of the layers [4, 16]. Likewise, Yamanaka et al. have investigated exchange reactions in the layered copper(II) compound Cu2 (OH)3 (CH3 COO).H2 O (1), in the hope of isolating highly active oxidation catalysts [17]. The ion-exchange is achieved by dispersing 1, as a polycrystalline powder, in an aqueous solution containing an excess of the relevant sodium salt. The mixture is stirred at room temperature, then the material obtained is filtered, washed with water, and dried under vacuum. When the sodium salt is not easily dissolved in water, the ion-exchange can be carried out in methanol or in an acetate-water solution. It is likely that the reaction occurs more easily for small anions, as noted for 2-naphthoate and 2-anthracene carboxylate, which are easily intercalated, while the parent isomers, 1-naphthoate and 9-anthracene carboxylate, are not obtained [18]. The acetate ion is usually totally exchanged after a few hours, giving X-ray diffraction patterns with intense (ool) reflections, in agreement with layered structures. The other reflections (hkl with h and k = 0) are much weaker, and have the usual enlargement for disordered pillared compounds. During the exchange process the completeness of the reaction is checked by the abrupt change of the (ool) reflections, relative to those of the starting compound. We show here that the layer compounds M2 (OH)3 X.zH2 O, in which the exchange− able anion, X = NO− 3 or OAc , is coordinated to the divalent metal ion, are well adapted for substitution reactions. The use of n-alkyl sulfate or n-alkyl carboxylate anions affears to be convenient to tune the basal spacing over a wide range of values [11, 19, 20]. These compounds are similar to the weak ferromagnets CuCl4 (nCm H2m+1 NH3 )2 , whose correlations between the structure and magnetic properties have been investigated [21]. Further results are reported in this paper. In such hybrid systems, the basal spacing is shown to be related to the carbon chain length (m) through the relationship d(Å) = d0 + η(1.27m cos θ), where η = 1 or 2, depending on the chain packing, and θ is the angle of the chains with respect to the normal to the layers [4]. The distance d0 involves: (i) the size of the bridging group, (ii) the van der Waals distance between either facing methyl groups or methyl groups and hydroxide layers, and (iii) the thickness of the inorganic layer. For an homogeneous series, because the molecular area of the chains is constant whatever the m value, variation of d0 will point to a change of inorganic layer thickness. Likewise, a change of the tilt angle (θ ) will be usually related to a slight structural deformation of the coordination polyhedra, and accordingly of the structure of the layers. Basically, because of the nature of the chemical bond between the organic and inorganic sub-networks, the M2 (OH)3 X.zH2 O compounds (M = Co, Cu; X is an organic anion) are expected to differ from the LDHs. Indeed, for the former, the anion X, covalently linked to the metal ion M, might induce a significant variation of the exchange coupling, and as a result a change of the bulk magnetic properties.

11.4 Influence of Organic Spacers in Hydroxide-based Compounds

361

We discuss hereafter the properties of a series of copper(II) and cobalt(II) hybrid layered compounds in which mono- or difunctional organic species are inserted, with the aim of modulating the distance between magnetic layers, and accordingly the properties of the bulk material.

11.4 11.4.1

Influence of Organic Spacers in Hydroxide-based Compounds The Cu2 (OH)3 X Series

It is known that the interlayer X anion in Cu2 (OH)3 X is exchangeable and various intercalation compounds can be prepared. By ion-exchange reaction with nalkyl sulfate anions, two series of hybrid compounds, with the general formulation Cu2 (OH)3 (n-Cm H2m+1 SO4 ).zH2 O, with z = 0 or 1, have been isolated. The variation of the basal spacing, illustrated in Fig. 3 for the anhydrous compounds, increases linearly with the aliphatic chain length m, according to the relationship d(Å) = 12.01 + 1.27m. This implies unambiguously that the n-alkyl chains are organized in a zip-like fashion (monolayers) and most probably orientated normal to the inorganic layers (Fig. 4a). In this configuration, the terminal methyl groups of the alkyl chains interact hydrophobically with the hydrogen atoms of the hydroxide layers. The temperature-dependent susceptibility for copper(II) n-alkyl sulfates is plotted as χ T against T in Fig. 5. At high temperatures the value observed is that expected for two copper(II) per mole (∼0.8 emu K mol−1 ). At low temperatures the decrease of χ T indicates that AF interactions dominate. Owing to the large separation between magnetic layers, 19.1 Å (n = 6) to 26.7 Å (n = 12), it can be stated that 2D short-range correlations are responsible for the behavior observed. Slight structural

Fig. 3. Variation of basal spacing with m (number of carbon atoms) in the copper(II) n-alkyl carboxylate (full squares and circles) and n-alkyl sulfate (open squares) series.

362

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 4. Structural models for layered compounds involving organic chains separating the inorganic layers. Mono and bilayer packings of the chains are shown in (a) and (b), respectively, whereas (c) deals with bridging dianions.

Fig. 5. Temperature-dependence of χ T for the layered copper(II) n-alkyl sulfates with different m values.

modifications of the layers probably explain the difference between χ T = f (T ) variations. The carboxylate anions are shown to be well adapted for intercalation reactions into hydroxide layer compounds [4, 17, 19, 20, 22]. The substitution of nCm H2m+1 COO− with m ≥ 4 for CH3 COO− gives two series of hybrid compounds (noted α and β), depending on the synthetic conditions [19]. The α-type compounds are hydrated with the formula Cu2 (OH)3 (n-Cm H2m+1 COO).zH2 O (with z = 0.3– 0.5), whereas the β-type compounds are anhydrous. The varieties studied had the same Cu/OH/Cm H2m+1 COO ratios [19]. Note that the β compounds can have a OH/Cm H2m+1 COO ratio <3 (down to 2.3), indicating that after the substitution of CH3 COO− a part of the hydroxyl ions are exchanged. This is confirmed by pH measurement [23], showing a characteristic increase after ca. 24 h, which agrees with the increase of OH− ions in solution. Then the pH decreases after 40 h, according to a slight dehydration of the products and decomposition into CuO. For short n-alkylcarboxylate chains (m = 4, 5), only α-type compounds are isolated. Figure 3 shows the dependence of the basal spacing on the carbon chain length for both n-alkyl carboxylate series. The linear variations, d(Å) = d0 + 2.54m cos θ, agree with a bilayer packing of the alkyl chains and a tilt angle θ = 22◦ . The difference between d0 values (ca. 5.4 Å) suggests different structures for the inorganic layers. The d0 value obtained for the α series agrees well with the layer thickness in cop-

11.4 Influence of Organic Spacers in Hydroxide-based Compounds

363

per(II) hydroxide nitrate [10], so that Brucite-like inorganic layers may be assumed. In turn, the β derivatives have d0 value very similar to that of n-alkyl sulfate compounds, suggesting a large deviation from the Brucite-like structure. Together with the pH variation, these considerations imply topotactic exchange during the first step of the reaction (α-derivative) followed by a dissolution-crystallization process (β-derivative). Clearly, the lack of crystal structure for these series is a limitingfactor, but some insight is provided by IR spectroscopy experiments [23]. As for the hydroxide acetate [24], it seems unambiguously that COO groups coordinate the metal ion. Most probably, the two oxygen atoms of COO are involved, one with a copper(II) ion, the other in a strong hydrogen-bond with neighboring OH− ion. In fact, the character of the COO groups is shown to be different in the two series. Two carboxylates doublets are superimposed in the α derivatives, one with unidentate character, the other with bridging character. The former disappears completely in the β derivatives. For more insight on the local structures, Cu K-edge EXAFS spectra of Cu2 (OH)3 (NO3 ) and Cu2 (OH)3 (n-Cm H2m+1 COO) compounds were measured [25]. The comparison of the observed spectra indicates that the local structures of Cu2 (OH)3 (NO3 ) and the m = 1 short chain compound (α-type) are very similar, while significant differences are noticed with those of the β-type long chain materials. The most significant change in the latter is concluded to be the contraction of the axial Cu–O bond. The change seems to result from the so called “chemical pressure” produced by alkyl carboxylate chains, which also causes the cooperative structural modification in the [Cu2 (OH)3 ]+ layers. The temperature dependence of χ T for the Cu2 (OH)3 (n-Cm H2m+1 COO) series is illustrated in Fig. 6 for m = 1 (acetate) and m = 7, which are representative of short and long chain compounds [19]. Upon cooling, the hydroxide acetate shows a constant value of χ T , then a slight increase down to ca. 10 K, indicating the dominance of weak ferromagnetic (F) intralayer interactions. At lower temperature, the drop of χ T suggests an AF interlayer coupling. The field dependent magnetization,

0.4

100

0.3

m=7(β)

10

M / µBmol

χT / emu K mol

-1

-1

0.2 0.1 0.0

-0.1 -0.2 -0.3 -0.4 -15

1

-10

-5

0

H / kOe

5

10

15

m=1 m=7(α)

0.1

0

50

100

T/K

150

200

Fig. 6. Temperature-dependence of χ T for the copper(II) nalkyl carboxylates with m = 1 and m = 7. The isothermal magnetization for m = 7 (β phase) is given in the inset.

364

11 Hybrid Organic-Inorganic Multilayer Compounds

which has an S-shaped dependence on the field, highlights the influence of weak interlayer couplings, in agreement with the findings for metal hydroxides, M(OH)2 with M = Fe, Co, Ni [12, 26]. Different behaviors are observed for the long chain derivatives, α and β. Whatever the m value, the α compounds show a decrease of χ T down to 2 K, denoting the AF character of the exchange interactions. In turn, for the β compounds, a decrease of χ T is first observed upon cooling down to a minimum at ca. 60 K, then a very sharp increase at lower temperature, pointing to a ferrimagnetic or non-collinear spin-arrangement in copper(II) layers. The very large value of χ Tmax is the signature of a net magnetic moment in the ground-state [19, 20]. The occurrence of long-range ferromagnetic order is illustrated by a characteristic hysteretic effect in the M(H) curve at T = 4 K (inset of Fig. 6). Similar behavior is observed for the different β compounds, even for very large basal spacing (e. g. up to 40.7 Å for m = 12). Note that the magnetization in high field confirms the stabilization of a non-collinear or a ferrimagnetic order. Finally, the ordering temperatures, determined in an applied field H = 0.5 Oe, range from 21.2 K for m = 7 to 19 K for m = 12, without clear relationship with the chain length. The above structural considerations are relevant for explaining the discrepancy in the magnetic behavior of the α and β derivatives. Given the presence of side groups bound to oxygen ligands and their influence on the bond angles, the intralayer exchange interactions are expected to be antiferromagnetic [23, 27]. However, the organic moieties modulate the exchange process by inducing crucial changes in the site symmetry around the metal ion, and as a result in the exchange pathways [28]. Moreover, the coexistence of different coordination sites, as already mentioned for other copper(II) hydroxy-salts [29], may promote an uncompensated magnetic ground state, and induce a 2D ferrimagnetic-like behavior, as observed for the β derivatives. This is supported by the high temperature variations of the χ T product which show the typical features of low dimensional ferrimagnets [30].

11.4.2

The Co2 (OH)3 X Series

The parent cobalt(II) compounds exhibit very similar structure-property correlations. The hybrid compounds Co2 (OH)4−x (n-Cm H2m+1 COO)x .zH2 O (m = 1, 7, 9, 10, 12) and Co2 (OH)4−x (n-Cm H2m+1 SO4 )x .zH2 O (m = 6, 9, 12) are prepared by ion-exchange reaction. A suspension of Co2 (OH)3 (NO3 ) in aqueous solution is used with the appropriate sodium salt of the exchangeable anion [31]. Chemical analysis gives, for the n-alkyl carboxylate series, x = 0.8 ± 0.05, z = 1–1.5, and for the n-alkyl sulfate series, x = 0.55 ± 0.05, z = 1.5–2. The X-ray powder diffraction patterns of the substituted compounds show intense (ool) reflections, in agreement with a layered structure. As might be anticipated, the basal spacing is closely related to the n-alkyl chain length (Fig. 7), while the in-plane parameters remain nearly constant [31]. Note that all compounds exhibit hexagonal symmetry, with space group P3m, similarly to Co(OH)2 . The cobalt(II) hydroxide acetate (m = 1) exhibits two phases, a hydrated phase with a basal spacing of 12.7 Å, and an anhydrous one, obtained by further heating at

11.4 Influence of Organic Spacers in Hydroxide-based Compounds

365

Fig. 7. Variation of basal spacing with m for the cobalt(II) n-alkyl carboxylates (squares) and n-alkyl sulfates (circles).

50◦ C under vacuum, with layers 9.3 Å apart. A structural model may be proposed for the anhydrous variety from the X-ray diffraction results and van der Waals spheres of the constituent species. In such a model, the cobalt(II) ions in distorted octahedral surrounding form a Brucite-like array. The acetate anion, coordinated to cobalt(II) ion by an oxygen atom, occupies the interlayer spacing with a head to tail arrangement, according to the copper(II) hydroxide acetate [24]. Focusing on the carboxylate series, when the length of the coordinating anion (n-alkyl chain) becomes larger, a linear variation of the basal spacing, d, with m is observed (Fig. 7). A monolayer structure of the organic chains is deduced, corresponding to a tilt angle of the chains, θ = 52◦ , with respect to the c axis. Further, from the extrapolated value, d0 = 17.98 Å, it is deduced that the layer thickness is about 7.3 Å, if water molecules are located in between hydroxide layers and terminal methyl groups. Clearly, this model is ruled out for the short chain carboxylate, m = 1, whose basal spacing (12.7 Å) is much lower than expected. Owing to the chemical composition of the studied compounds, namely Co5 (OH)8 (n-Cm H2m+1 COO)2 .z(H2 O (z = 2.5 to 3.8), it can be stated that the structure derives from that predicted for the zinc analog Zn5 (OH)8 (NO3 )2 [32, 33], illustrated in Fig. 8. The presence of cobalt(II) in Td and Oh sites is further confirmed from UV-visible spectra by the presence of 4 A2 → 4 T1 (P) and 4 T1g (F) → 4 T (P) transitions [34]. In such a model, the aliphatic chains are connected through 1g the carboxylate moieties to the Co(Td) metal ions, and occupy the space in between the cobalt layers with a tilt angle θ . The n-alkyl sulfate series exhibits a quite similar variation of the basal spacing with m (Fig. 7). The linear dependence of d, namely d = 14.02 + 0.88m, agrees with a monolayer stacking of the chains and a tilt angle θ = 46◦ . The thickness of the inorganic layer is shown to be 8.6 Å, to be compared to ca. 4 Å for the Brucite-like layers [4], indicating that the structure likely involves, as above, cobalt(II) ions in Oh and Td surroundings. The cobalt(II) compounds with long chain anions, n-Cm H2m+1 SO− 4 or nCm H2m+1 COO− , have very similar magnetic behavior. Figure 9 shows the χ T = f (T ) plots for the n-alkyl carboxylate series. At high temperature, the χ T val-

366

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 8. Structural model for Zn5 (OH)8 (NO3 )2 showing the packing of NO3 groups in the interlayer spacing.

ues agree with the expected one for isolated cobalt(II) ion (∼3.0 emu K mol−1 ), while upon cooling, a sharp increase is observed, pointing to a ferro- or ferrimagnetic ground state. The maximum χ T value raises from ca. 55 emu K mol−1 to 220 emu K mol−1 as the basal spacing increases from 12.7 Å (n = 1) to 27.4 Å (n = 12). At very low temperature, the susceptibility depends closely on the applied field; the more the basal spacing increases, the more this effect is pronounced. Characteristic hysteretic effects are observed in the magnetization curves for the long chain compounds, with a large spontaneous magnetization and coercive field (inset in Fig. 9). The saturation value at 20 kOe (∼1 µB /Co mol) is lower than expected for cobalt(II) ion at low temperature. The presence of an out-of-phase signal in a. c. susceptibility measurements confirms the stabilization of a net moment at low temperature. Owing to the structural and magnetic findings, it can be thought that the ground state is ferrimagnetic-like, due to the non-compensation of spin sub-networks corresponding to cobalt(II) ions in Oh and Td symmetries. Finally, these results highlight the correlation between structures and magnetic properties in the hydroxide-based compounds. It is pointed out that for a ferro- or ferrimagnetic order within the layers, the situation depends to a large extent on the interlayer spacing. For small spacing (less than ca. 10 Å), the through-bond interlayer

11.4 Influence of Organic Spacers in Hydroxide-based Compounds

367

Fig. 9. Temperature-dependence of χ T for the layered cobalt(II) n-alkyl carboxylates with different m values.

interactions stabilize a 3D AF order. When the spacing is made larger (m = 6 to 12), the direct exchange between unpaired electrons in adjacent layers becomes negligible. Further, due to the nature of the chains, the spin polarization effect is also negligible. Nevertheless, the compounds exhibit a spontaneous magnetization and a characteristic hysteresis cycle. Clearly, the large TC values (16.5 to 22.7 K) and the weak dependence on the basal spacing can hardly be related to quantum interlayer interactions. So, the question which arises deals with the dimensionality of the magnetic network, namely, to what extent 2D long range order may explain the observed behavior. Basically, Ising-like anisotropy may promote 2D ordering, but owing to the large hysteretic effects, characteristic of bulk magnets, the influence of through-space dipolar interaction has to be considered to get more insight on the origin of the magnetic order.

11.4.3

Dipolar Interaction and Long-range Magnetic Order

Recently, Gˆırtu et al. [35] have reported a model of triangular quantum Heisenberg antiferromagnet to explain the behavior of the β-type copper(II) compounds Cu2 (OH)3 (n-Cm H2m+1 COO). From the frequency dependence of the a. c. magnetic susceptibility, and field-cooled/zero-field-cooled magnetization, slow relaxation and a spin-glass like behavior are suggested. Considering the interplay between the Heisenberg AF exchange, giving frustration, and a weak additional Dzyaloshinskii– Moriya interaction, leading to spin canting, the authors propose a new unusual state with both canted antiferromagnetic and glassy characteristics. Spin relaxation process has also been discussed for Fe monolayer stripes deposited on a vicinal W(110) substrate [36]. It is demonstrated that a ferromagnetic phase transition induced by dipolar coupling between adjacent stripes occurs before freezing temperature. Even if the 3D character of the long range order is well established in several compounds involving chains or layers [37, 38], the balance between the through-space dipolar interactions and quantum exchange couplings may appear quite controver-

368

11 Hybrid Organic-Inorganic Multilayer Compounds

sial. The electrostatic exchange interaction is known to vanish extremely rapidly with the distance between metal centers, crudely as J ≈ r −n with n = 10 to 12 [39], while dipole-dipole interaction varying as r −3 has long range effect [40]. Clearly, the influence of the latter is usually negligible, except for an assembly of parallel spins, since it is shown to vary as r −2 between chains and r −1 between finite layers, for the distances under consideration [41]. As a result, dipolar interactions cannot be ignored in low-dimensional ferromagnets, specifically in the vicinity of TC when the spins become correlated on a large distance [38]. Let us consider a two dimensional square lattice of spins S, coupled by a ferromagnetic exchange interaction. At absolute zero, the magnetic layer exhibits a ferromagnetic alignment of the spins, and the ground state has the higher spin multiplicity. Upon increasing temperature, the spins become only correlated on a finite distance ξ . For a 2D Heisenberg ferromagnet, this one is related to the exchange constant J and the spin value S by the relationship [42]: ξ 2 = (J S/kT ) exp(4π J S 2 /kT ) The exponential divergence of ξ is well illustrated in Fig. 10 by the temperature dependence of the effective moment ξ 2 gS, deduced from magnetization data for Cu2 (OH)3 (n-C7 H15 COO). It is well fitted in the paramagnetic region, just above TC , using J = +44 K. As far as the actual exchange constant between copper(II) ions is concerned, this value should be re-normalized to take into account the trigonal symmetry and the spin density of the real compound. The basic idea is that the dipole interaction between magnetic layers leads to 3D ordering as soon as the in-plane correlation length ξ reaches a threshold value related to the basal spacing c. Because of the exponential divergence of ξ(T ), the temperature for which the threshold is reached should depend only weakly on c, as experimentally observed. In order to minimize the dipole and anisotropy energies, the order between layers is expected to be ferromagnetic if the spins are normal to the layers. In the proposed model [38], it is assumed that (i)

Fig. 10. Variation of ξ 2 gS as temperature is reduced for the m = 7 compound, determined from magnetization data above TC .

11.4 Influence of Organic Spacers in Hydroxide-based Compounds

369

the in-plane correlation length is driven by the in-plane exchange interaction, (ii) any exchange interaction between layers is negligible, and only through-space dipole coupling is available between moments located in different layers, and (iii) a small local anisotropy favors the spin orientation normal to the layers. For T = 0, each layer may be viewed as a chess board with alternating spin-up and spin-down squares, each one containing ξ 2 spins. By use of the point-dipole approximation, the dipolar field acting on a moment µ = gSµB , as a result of all the other moments apart ri j is calculated by use of the expression [38]: Hdip = µi j [3 cos2 (θi j ) − 1]/ri3j where µ stands for the thermal average over µ, and the summation extends over all pairs of sites. Clearly, Hdip depends closely on the correlation length ξ(T ), which diverges rapidly when temperature is lowered, and may be rewritten Hdip = µλ(ξ ). A picture of the critical region, i. e. corresponding to the transition to long range order, is deduced from the molecular field approximation, in which the molecular field is the dipolar field calculated above. The temperature dependence of the spontaneous magnetization is then obtained by solving the self consistent set of equations [38]:

µ = µ[coth(x) − 1/x] where x = ξ 2 µHdip /kT and Hdip = λ(ξ ) µ With the knowledge of ξ(T ) above TC (illustrated above for m = 7) the expected variation of the spontaneous magnetization µ/µ = f (T ) can be computed (Fig. 11). After scaling the lattice parameter a to account for the triangular network of the copper(II) layers and the actual moment per site, the critical temperature at which spontaneous magnetization occurs is shown to agree very well with the experimental value, ca. 20 K. Large change of the c/a ratio does not modify dramatically the critical temperature, as observed experimentally. In turn, modifications of the in-plane exchange interaction J affect directly the thermal dependence of ξ , and

Fig. 11. Theoretical variation of the zero-field magnetization for different c/a ratios. The influence of the in-plane exchange interaction on the TC value is given in inset. J0 corresponds to the result of the fit for m = 7.

370

11 Hybrid Organic-Inorganic Multilayer Compounds

induce very significant variations of the long range ordering temperature (Fig. 11). This result is very similar to that found for layered magnets with a weak interlayer exchange coupling j, the ordering temperature of which obeys TC = 4π J/ ln(J/j) [43]. Finally, these calculations point out that dipolar interaction between 2D ferromagnets with axial easy axis promotes a 3D ferromagnetic order, even for spacing as large as 40 Å between the magnetic layers. This coupling dominates over the classical AF exchange mechanism, so long as the bridging ligands do not allow electronic transfers, i. e. for organic chains with saturated carbon atoms. The basic idea that dipolar interactions are always negligible is thus clearly irrelevant in layer-based compounds.

11.5

Difunctional Organic Anions Connecting Magnetic Layers

Anionic exchange in divalent metal hydroxide compounds appears promising for the design of new multiproperty systems. It is however worthwhile to be able to control the connection between magnetic layers. For this purpose, recent works have focused on the design of dicarboxylate compounds [44], exhibiting difunctional alkanedioate or alkenedioate bridging anions coupling inorganic layers (Fig. 4c). The comparison of hybrid systems with saturated and unsaturated aliphatic chains, is expected to provide insight on the influence of π electrons in the interlayer exchange mechanism. The copper(II) hydroxydicarboxylates have been prepared by anionic exchange reaction starting from Cu2 (OH)3 (OAc).H2 O (1), as explained above. The acetate ions are exchanged for dicarboxylate anions X = (CO2 (CH2 )m CO2 )2− , with m = 1 to 8. Similar exchange reactions have been achieved with unsaturated dicarboxylate anions by using the fumaric (X = C2 H2 C2 O4 ), trans-hexenedioic (X = C4 H6 C2 O4 ) and muconic (X = C4 H4 C2 O4 ) acid salts [23, 44]. Following the usual process, a polycrystalline powder of 1 is dispersed into an aqueous solution of the sodium salt XNa2 , at room temperature. After 1 day, the exchange is completed by repeating once the treatment of the reaction product with a fresh dicarboxylate solution, for 2 days. The precipitate is washed with alcohol, then dried at 40◦ C. The general formulation of the compounds, given in Table 1, is Cu2 (OH)4−η Xη/2 .zH2 O, with η ranging from 1 to 2.4 and z from 0 to 1.6. The cobalt(II) derivative with the trans-hexenedioate (X = C4 H6 C2 O4 ) anion is synthesized from Co2 (OH)3 (NO3 ). The exchange reaction is carried out as above, with the trans-hexenedioic acid sodium salt. The obtained bluish-green crystal powder is washed with alcohol and dried at 40◦ C. The chemical composition corresponds to η = 0.68 and z = 1.57. The layer structure of the compounds is evidenced from XRD patterns, showing intense (ool) reflections. During the exchange process, the completeness of the reaction is checked by the abrupt change in peak position, with respect to the starting materials. The other reflections (with h and k = 0) are much smaller, and exhibit usual enlargement for disordered pillared compounds [45]. The diffraction patterns are analyzed with hexagonal or pseudo-hexagonal unit cells, like for the mono-carboxylate

11.04 10.45 13.20 13.15 14.73 15.60

X = C3 H6 C2 O4 (4) X = C4 H8 C2 O4 (5)

X = C5 H10 C2 O4 X = C6 H12 C2 O4 X = C7 H14 C2 O4 X = C8 H16 C2 O4

8.80 10.22 10.38 13.40

X = C2 H2 C2 O4 (10) X = C4 H6 C2 O4 (11)

X = C4 H4 C2 O4 (12)

Co7 (OH)11.6 (C4 H6 C2 O4 )1.2 .5.5H2 O (13)

Unsaturated dicarboxylates

(6) (7) (8) (9)

9.20 8.80

Basal spacing (Å)

X = CH2 C2 O4 (2) X = C2 H4 C2 O4 (3)

Saturated dicarboxylates

Cu2 (OH)4−η Xη/2 .zH2 O

AF/– F/F TC = 13 K F/Meta TN = 16 K F/F TC = 56.5 K

AF/– F/Meta T N = 11.2 K AF/– F/Meta T N = 15.7 K AF/– AF/– AF/– Ferri/Ferri TC = 17 K

Magnetic behavior, 2D/3D

Table 1. Basal spacing and physical properties of layered hydroxy dicarboxylates.

1572 1592 1563 1592

−80 −18 −126 −69

+12

+40

1548

1559

1584 1580

1536 1571

−40 +35

−26 +46

1577 1536

1382

1407

1383 1413

1400

1407 1398 1408

1408 1429

1432 1421

884

873

895

864

864

Characteristic IR bands (cm−1 ) C=O C–O γ OH–O

−1.7 +1.7

θ Values (K)

11.5 Difunctional Organic Anions Connecting Magnetic Layers

371

372

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 12. Variation of the basal spacing with m for the copper(II) hydroxy dicarboxylates. Linear variations are observed for odd and even series.

derivatives [31]. The similarity of XRD patterns with those of Cu2 (OH)3 (OAc).H2 O suggests that the structure of the copper(II) inorganic layers is unchanged after exchange reaction. The experimental basal spacing (d) for copper(II) n-alkanedioates is given in Table 1. The variation of d with the chain length m is illustrated in Fig. 12 for saturated organic anions. The basal spacing shows a step-like variation with m, which results from the alternating orientation of the carboxylate groups in the bridging unit. For a given parity, a linear variation of d is observed, in agreement with the length of the organic anion. The tilt angles of the bridging units are shown to be 42.9◦ and 25.7◦ for odd and even m, respectively. The d0 value deduced from extrapolation to m = 0 differs by ca. 2 Å for the odd (8.3 Å) and even (6.3 Å) series. The mean value (7.3 Å) is close to the basal spacing reported for the parent compound Cu2 (OH)3 NO3 (d = 6.95 Å) exhibiting a Brucite-like structure [10]. This suggests that the inorganic layers in the exchanged compounds may be viewed as quasi-planar triangular arrays of copper(II) ions in octahedral surrounding. For unsaturated dicarboxylate anions, very close basal spacings are observed (0–0.2 Å step aside), compared to the corresponding saturated derivative. The cobalt(II) compound with X = C4 H6 C2 O4 shows a larger basal spacing than the copper(II) analog (∼3.2 Å larger) [44]. Such a difference, already emphasized for the mono-carboxylate series, denotes a larger thickness of the inorganic layer. Owing to the color change after anion-exchange, pointing to the presence of cobalt(II) in Td symmetry, and the chemical composition, it can be predicted that the local structure is similar to that of Zn5 (OH)8 (NO3 )2 [32] or Co7 (OH)12 (NO3 )2 .zH2 O [46]. In such structures, the organic anions are connected through carboxylate moieties to the Co(Td) metal ions, and occupy the space in between the layers with water molecules. The evidence of chemical bonds between the dicarboxylate anion and two metal ions is confirmed par IR spectroscopy. A carboxylate doublet is identified at 1530–1590 cm−1 and 1400–1430 cm−1 . The former corresponds to the C=O stretch. From the position of the band, it is deduced that the carboxyl groups are bonded to the metal ion. The doublet is unique, indicating that the two carboxylate functions are linked to the layers, according to the charge balance between the organic and

11.5 Difunctional Organic Anions Connecting Magnetic Layers

373

Fig. 13. Temperature-dependence of χ T for the copper(II) hydroxy dicarboxylates with saturated (3 (diamonds) and 5 (circles)), and unsaturated (11 (squares) and 12 (triangles)) bridging anions. The inset shows the metamagnetic transition of M(H) for 5.

inorganic sub-networks. Note that a strong band, attributed to the out-of-plane γ OHO vibration [46], is recorded around 880 cm−1 for five copper(II) compounds, namely 3, 5, 9, 11, 12, only (see Table 1). The magnetic properties of the copper(II) compounds are shown to be very sensitive to the spatial extension of the alkanedioate chains, as summarized in Table 1, but no clear correlation can be drawn at first sight. Most of them exhibit a 2D AF behavior, and there is no evidence of bulk ordering at low temperature. The compounds 3 (m = 2), 5 (m = 4), 9 (m = 8) show in turn a ferro or ferrimagnetic 2D behavior. The χ T = f (T ) variation for the succinate derivative (3) is very similar to that of the acetate derivative (1) (Fig. 13) [19]. Below 50 K, a slight increase is observed upon cooling down to 11.2 K, suggesting weak ferromagnetic in-plane interactions, then a drop to zero which agrees with interlayer AF interactions. In turn, the unsaturated derivative (X = C2 H2 C2 O4 ) shows clearly an AF behavior over the whole temperature range. This variation cannot be explained by a change of the basal spacing, but most likely by slight modification in the local structure of the inorganic layer, inducing a change of sign of the exchange coupling. Such a magnetic dependence is well documented for copper(II) systems which appear to be very sensitive to the metal–oxygen–metal bridging angle [28]. The most striking results are emphasized for m = 4 with saturated (X = C4 H8 C2 O4 ) and unsaturated (X = C4 H6 C2 O4 and X = C4 H4 C2 O4 ) aliphatic chains, noted 5, 11 and 12, respectively. These compounds, which only differ by the number of double bonds of the anionic spacer (0 to 2), are characterized by very close basal spacing (10.45 Å to 10.22 Å), and accordingly analogous structures. A strong and regular increase of χ T is observed upon cooling down (Fig. 13), characteristic of significant in-plane ferromagnetic interactions, with a maximum at 15.7 K for 5, 12.8 K for 11 and 16.0 K for 12. The main difference occurs at low temperature, since 5 and 12 show an AF long range order with a metamagnetic transition for Hc = 2250 Oe and 2040 Oe, respectively (inset of Fig. 13), while 11 exhibits a characteristic ferromagnetic transition. This one is evidenced from the strong divergence of χ T at low temperature (factor 10 with respect to 5), and the variation of the a.c. susceptibility

11 Hybrid Organic-Inorganic Multilayer Compounds 2

50

1 -1

60 M / µBmol

χ', χ'' x 20 / emu mol

-1

374

40

0

30

-1

20

-2

-15 -10 -5

0 5 H / kOe

10 15

10 0 0

20

40

T/K

60

Fig. 14. Temperature-dependence of the a.c. susceptibilities for 11. The inset shows the M(H) variation at 4 K.

is found to be 13.0 K. The (Fig. 14). The TC value recorded from the maximum of χa.c. field dependence of the magnetization (inset of Fig. 14) agrees with a ferromagnetic order between copper(II) layers. The hysteretic effect at 4 K is small (ca. 100 Oe) indicating a weak anisotropy of the magnetic centers. A fit of the magnetic susceptibility has been performed in the paramagnetic regime to determine the in-plane exchange interaction between copper(II) ions. Using the high temperature series expansions for the spin1/2 2D Heisenberg triangular lattice [47], a very good agreement between theory and experiment is obtained for J = +27 K (H = −J Si S j ). Note that above TC , the χ T variation is well-fitted from the sum of Arrhenius-like contributions, a exp(E 1 /kT ) + 0.86 exp(E 2 /kT ) with a = 5 × 10−6 , E 1 /k = 287 K, and E 2 /k = 52.6 K, showing clearly two regimes for the magnetic behavior (0.86 gives the high temperature Curie constant). Thus, above 25 K, χ T is well described by an exponential decay whose activation energy corresponds roughly to 2J . In the vicinity of TC , the variation of χ T is depicted by a rapidly diverging function, with a strong activation energy, indicating that interlayer interactions are no longer negligible. Magnetic experiments performed with the cobalt(II) trans-hexenedioate (X = C4 H6 C2 O4 ) parent compound (13) show a slight decay of χ T from room temperature down to 100 K, because of the spin-orbit coupling effect, then a sharp divergence below 65 K with a maximum of 290 emu K mol−1 , pointing to a long range ferromagnetic ordering. The a.c. susceptibility measurements confirm the occurrence of a magnetic ground state with an out-of-phase signal below TC = 56.5 K. Finally, the long chain compound m = 8 (9) exhibits a very different magnetic behavior (Fig. 15). The χ T product decreases on cooling from 0.8 emu K mol−1 at room temperature down to 0.44 emu K mol−1 at 40 K, then displays a rapid increase below 20 K. As already mentioned for copper(II) n-alkyl carboxylate derivatives [19], such a χ T variation is characteristic of a ferrimagnetic order within copper(II) layers. Owing to the in-plane exchange pathways, which involve oxygen atoms from both (OH)− and (C8 H16 C2 O4 )2− , it can be stated that the triangular network is likely strongly distorted and ordered magnetic sub-networks take place. At low temperature, the

11.5 Difunctional Organic Anions Connecting Magnetic Layers

375

Fig. 15. Temperature-dependence of χ T for the long chain copper(II) hydroxy dicarboxylate 9. The best fit from the sum of exponential laws (see text) is plotted in full line. The inset shows the hysteresis of the M(H ) curve at 2 K.

transition towards a magnetic ground state is evidenced from a.c. susceptibility experiment. An out-of-phase signal occurs below TC = 17 K, which is slightly higher than for 11. Some insight on the correlation with the local structure is given by IR spectroscopy. Thus, a band is observed in the range 864–895 cm−1 for copper(II) compounds exhibiting ferro- or ferrimagnetic layers. Further, all these compounds involve aliphatic chains with even m values, i. e. characterized by the same tilt angle. For compounds with odd m values, the local environment of Cu(II) ion is likely modified, due to the different tilt angle, and as a result the intralayer exchange coupling can change dramatically. Let us focus now on the low temperature behavior of the saturated dicarboxylate compounds exhibiting ferro- or ferrimagnetic in-plane correlations. Basically, the situation is very similar to that reported in chapter 11.4 for monocarboxylate compounds. For small basal spacing (m = 2, 4), a very weak interlayer interaction stabilizes a 3D AF order, and a metamagnetic transition occurs in low field. When the spacing becomes larger (m = 8), a spontaneous magnetization is observed, due to dipolar through-space interaction between magnetic layers. This description fails for the unsaturated compound 11, since bulk ferromagnetism is stabilized although the basal spacing is very close to that of the saturated derivative. Even if the turning value of d is expected to be 10–11 Å, the mechanism is likely different, and must involve the electronic transfers by π electrons along the bridging anion. For the conjugated derivative (12), an AF ground state takes place with a metamagnetic transition in low field, according to the spin polarization in the aliphatic chain. The π character of the bridging species is thus an essential feature to control the coupling between magnetic layers. In this respect, these systems differ markedly from the classical layered hybrid compounds, mostly because they provide a new class of multilayer magnets with possible electronic transfers between adjacent layers.

376

11.6

11 Hybrid Organic-Inorganic Multilayer Compounds

Metal-radical-based Layered Magnets

The design of organic-inorganic materials exhibiting a large coupling between the properties of the sub-networks is one of the challenges in chemistry of the hybrid materials. This is well illustrated in hydroxide-based compounds, in which the intercalation of organic radicals, such as 3-carboxy-2,2,5,5-tetramethyl-1-pyrrolidine1-oxyl (PROXYL) or imino-nitroxide benzoate (INB) anions induce new magnetic properties. The intercalation of PROXYL in between [Cu2 (OH)3 ]+ layers has been reported by Fujita [48]. Cu2 (OH)3.5 (PROXYL)0.5 .H2 O (14) is prepared by ion-exchange in copper hydroxide acetate with PROXYL sodium salt. The exchange is carried out by stirring a suspension of Cu2 (OH)3 (OAc).H2 O and PROXYL.Na in water for a day. The precipitate is filtered, washed with water and dried under vacuum at room temperature. Namely, half of the acetate is replaced by the radical anion, the remainder by hydroxyl groups. The presence of PROXYL radical within the compound was checked by IR spectroscopy. X-ray diffraction patterns point to the strong increase of the basal spacing, from 9.3 Å (hydroxide acetate) to 16.1 Å with radical intercalation. The temperature dependence of χ T for compound 14 is plotted in Fig. 16, and compared with that of the parent hydroxide acetate 1 discussed above. It is worth noticing that the magnetic behavior is drastically modified after intercalation of the radical anion. χ T decreases monotonically with decreasing temperature, indicating the dominance of in-plane AF interaction between neighboring spins, contrary to the finding for copper(II) hydroxide acetate. The Curie constant, C = 1.05 emu K mol−1 , agrees with the theoretical value. EPR measurements were performed at various temperatures to determine the origin of the AF interaction. At room temperature, the g-factor 2.007 is assigned to PROXYL radical, while copper(II) is EPR silent in the whole temperature range, due to the exchange broadening. As temperature is decreased, the g-factor gradually decreases and becomes less than 2 below 50 K, while

Fig. 16. Temperature-dependence of the paramagnetic susceptibilities for Cu2 (OH)3 (CH3 CO2 ).H2 O (square) and the intercalation compound, Cu2 (OH)3.5 (PROXYL)0.5 . H2 O (circle).

11.6 Metal-radical-based Layered Magnets

377

Fig. 17. Cu K-edge EXAFS oscillation function k 3 χ(k) (a) and Fourier transform (b) for Cu2 (OH)3.5 (PROXYL)0.5 .H2 O.

the peak-to-peak line width increases up to about 145 G at 4 K. The temperature dependence of the EPR susceptibility allows to deduce that the interaction between the PROXYL radical and copper(II) layer is small. The Cu K-edge EXAFS study shows different features for compound 14 and the hydroxide acetate (1), suggesting different environments for the copper atoms. The Fourier transform (Fig. 17) exhibits two main peaks below 4 Å, assigned to the Cu–O and Cu–Cu contributions, and a peak at ca. 6.0 Å, assignable to the next Cu– Cu shell and suggesting a linear alignment of the metal ions. The EXAFS analysis indicates that the local structure is similar to that of the short chain compound Cu2 (OH)3 (CH3 COO).H2 O. In particular, the Cu–OH–Cu angles are shown to be increased by 1–3◦ , compared with those of the parent compound. It results that the antiferromagnetic character of the in-plane coupling pointed out by magnetic susceptibility experiment can be related to changes of the dihedral and bridge angles in the Cu–OH–Cu pathways. Similarly, intercalation of para- and meta-imino nitroxide benzoate anions ( pINB and m-INB, respectively) within layered hydroxides has been carried out by ion-exchange [11, 49]. For p-INB radicals intercalated in between copper(II) layers, the observed AF behavior is very similar to that reported for compound 14 [48]. In turn, the cobalt(II) parent compounds appear to be more adapted to give multilayer materials with ferrimagnetic properties. Three compounds were obtained by ion-exchange reaction in Co2 (OH)3 NO3 with p-INB, m-INB and the diamagnetic precursor of the former, namely bis-hydroxyimidazolidine (Fig. 18) [49]. From chemical analysis, the composition Co2 (OH)3.5 (X)0.5 .2H2 O with X = p-INB (15), m-INB (16) and the diamagnetic precursor (17) is deduced, which agrees with the PROXYL derivative (14). The absence of residual nitrate groups was checked by IR spectroscopy. Further, the characteristic carboxylate vibration bands are consistent with a η-coordination of an oxygen atom to cobalt(II) while the second one is involved in a hydrogen bond, likely with a water molecule. On the other hand, the UV spectra point to the presence of both Oh and Td sites for cobalt(II) ions. From the X ray diffraction patterns, a layer structure is found with basal spacing ranging from 20.2 Å for 15 and 16 to 22.8 Å for 17, in agreement with the size of the intercalated organic molecules.

378

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 18. Synthesis route to obtain the hybrid radical-based materials from Co2 (OH)3 NO3 [49].

Fig. 19. Temperature-dependence of χ T for the radical-based compounds 15 (full circles), 16 (open circles) and the starting hydroxide nitrate (squares). The susceptibility of 16 plotted in the inset shows the variation of the in-phase (χ ) and out-of-phase (χ ) signals.

Fig. 20. Temperature-dependence of the remnant magnetization, normalized to its T = 2 K value, for 15 (full circles), 16 (open circles) and 17 (diamonds). The field dependent magnetization for 16 in given in the inset at T = 1.8 K).

11.6 Metal-radical-based Layered Magnets

379

The magnetic properties of the hybrid compounds are given in Figs. 19 and 20, and compared to those of the hydroxide nitrate. Upon cooling from room temperature, the observed χ T product shows for the radical-based compounds 15 and 16 a minimum around 60 K, and a strong increase up to a maximum at 10 K. Unlike the hydroxide nitrate [12], both compounds order ferromagnetically, as evidenced by the out-of-phase signal in a.c. external field (Fig. 19). From the maximum of χ , the ordering temperatures are found to be 6.0 K and 7.2 K for 15 and 16, respectively. Field-dependent magnetization measurements confirm these results. The M(H ) curve for compound 16 (Fig. 20) exhibits a hysteresis loop characteristic of a ferromagnetic-like 3D order. At T = 1.8 K, the coercive field is found to be 510 Oe (340 Oe for 15), while the high field magnetization (ca. 3.5 µB mol−1 at 8 T) is lower than expected for a complete ferromagnetic order. The competition between inplane exchange interactions and/or the spin-orbit coupling effect for cobalt(II) ions may induce such a behavior. The thermal variation of the net moment shows clearly the similarity between compounds 15 and 16 (Fig. 20). Compound 17, whose basal spacing is very close to 15, exhibits qualitatively the same behavior, but a much higher Curie temperature (TC = 15.3 K). Thus, the presence of organic radicals grafted to the metal-based layers has a striking effect on the bulk behavior. The fact that the ordering temperature is strongly affected indicates that a significant metal-radical exchange interaction, rather than a simple polarization of the radicals by metal layers, takes place. The significant lowering of TC (ca. 9 K less) demonstrates that the coordination of cobalt(II) ions by nitroxide radicals modifies considerably the interaction between inorganic layers. The magnetic transition is clearly displayed in the first derivative of the EPR absorption spectra given in Fig. 21 for compound 16. A broad feature occurs at

Fig. 21. EPR spectra of 16 showing the evolution of the line shape between 4 and 30 K. The temperature dependent shift of the free radical line is given in the inset.

380

11 Hybrid Organic-Inorganic Multilayer Compounds

low field (not shown on the graph), below TC for all three compounds, which is attributed to the metal layers coupling ferromagnetically to yield the 3D long range order. Since the cobalt(II) layers are EPR silent at higher temperature, this low field hump corresponds to a ferromagnetic resonance. As expected, the Lorentzianshaped signal of the radical is located close to the free electron g-value at room temperature, and remains nearly unchanged down to TC . Drastic changes occur close to TC , as sketched in the low temperature spectra. A strong broadening of the EPR line and a large shift of the resonance field (inset in Fig. 21) are simultaneously observed with the onset of bulk ferromagnetism. Therefore, the organic radical is actually probing the cooperative alignment of the neighboring cobalt(II) moments. The shift of the radical resonance means that the resonance field is no longer the applied Zeeman field. In the paramagnetic regime the spin Hamiltonian includes the Zeeman interaction and the dipolar interaction with the surrounding dipole moments. Added to these, the exchange interaction between (i) radical spins, Jπ−π , and (ii) radical and cobalt spins, Jπ−Co , must be considered. None of these is strong enough to give striking effects at room temperature, e. g. strong exchange narrowing or line broadening, because of rapid relaxation processes. However, as TC is smaller in the radical-based compounds, it appears that the sandwiched radical layer is actually counteracting the driving interaction for the bulk ferromagnetic ordering. The dipolar interaction Hdip is assumed as the driving interaction for the set up of bulk ferromagnetism within the three compounds, as shown for n-alkyl carboxylate parent compounds. The lowering of the critical temperature for compounds 15 and 16 compared to 17 must be attributed to the presence of radical, since the interlayer distances are very close. Clearly, the radical spins do not simply line up within the internal field created upon ordering the cobalt(II) layers, since the ordering temperature, TC , would be constant. The observed behavior is most likely due to the interlayer exchange interactions Jπ−Co and Jπ−π counteracting the ferromagnetic through-space dipolar coupling. In this respect, this class of hybrid magnets differs markedly from the classical intercalated layer compounds, mostly because both subnetworks (inorganic and organic) are in close interaction. These results are promising for the design of multifunctional materials with a synergy between the properties of the organic and inorganic networks.

11.7

11.7.1

Controllable Magnetic Properties of Layered Copper Hydroxides Solvent-mediated Magnetism

There is considerable interest in the packing and dynamics of long-chain polymethylene compounds intercalated in layered compounds, because of their relevance to the functionality of biological membranes, liquid crystals and so on [50, 51]. In addition, some of these compounds exhibit characteristic transformation, such

11.7 Controllable Magnetic Properties of Layered Copper Hydroxides

381

as trans–gauche transitions of the alkyl chains and monolayer–bilayer phase transitions, which are triggered by heat or soaking in solvents [52-54]. From the viewpoint of magnetism, it is interesting to produce organic/inorganic hybrids between such long-chain organic molecules with structural flexibility and layered copper hydroxides, because Cu2 (OH)3 X shows a magnetic variety which drastically depends on the molecular shape and alignment of X [19, 20]. If the magnetic properties of the inorganic layer are affected by a structural modification in the organic layer which is introduced by a stimulus to it, the hybrid can be regarded as a controllable and/or switchable magnetic material, in which the inorganic layer carries magnetic moments and the organic layer plays the role of the organizer of the magnetic properties of the former. In this section, we describe the magnetic properties of Cu2 (OH)3 (8-(( p(phenylazo)phenyl)oxy)octanoate) (18) [22]. A reversible phase transition which takes place in two organic solvents and results in a drastic change in the magnetic properties is reported. The intercalation compound 18 was obtained from 1 by the ion-exchange method. Figure 22(a) shows the diffraction pattern of the virgin sample of 18. The interlayer distance was estimated to be 20.7 Å, subtracting the calculated thickness of the inorganic host layer from the basal spacing of 25.5 Å, obtained from the periodicity in the pattern. It well corresponds to the calculated anion height of 21.7 Å. This means a monolayer structure of the organic layer in which the molecules are aligned nearly perpendicular to the metal-based sheets, as shown in Fig. 23(a). Although the compound is insoluble in acetonitrile, it was dispersed in it and the mixture was stirred for two days. Figure 22(b) and (c) are the X-ray diffraction patterns of the compound after 24 and 48 h, respectively. The measurements were carried out, after the samples were dried in vacuum. The peaks of the original phase remarkably decrease in intensity after 24 h (starred in Fig. 22(b)) and completely disappear after 48 h. Instead of them, new (001) lines appear and intensify gradually. The line interval indicates that

Fig. 22. X-ray diffraction patterns of compound 18 obtained by soaking in acetonitrile for 0 h (a), 24 h (b) and 48 h (c), and by soaking in hot methanol for 24 h (d).

382

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 23. Packing of the organic layer in compound 18, in methanol (a) and acetonitrile (b).

the interlayer distance of the transformed compound in acetonitrile is 38.7 Å, which is almost twice as long as the molecular height of the anion. After the soaking, the organic anion is considered to form a membrane-like bilayer, as shown in Fig. 23(b). It is, furthermore, found that the bilayer phase reverts into the monolayer structure in hot methanol. Fig. 22(d) shows the powder X-ray pattern of the compound, obtained from the bilayer phase after 24 h of soaking in methanol at 50◦ C. It agrees with that of the original monolayer phase, shown in Fig. 22(a). The mono–bilayer transformation reversibly takes place in the two solvents. The observed mono–bilayer transformation is similar to solvent-induced interdigitation in phospholipid assemblies. Phospholipid molecules, which are principal lipid components of biomembranes, assemble spontaneously in aqueous environments, usually forming a bilayer structure. It is known that a number of lipids form an interdigitated monolayer in the presence of alcohol [55, 56]. The induction of interdigitation is dependent upon alcohol concentration, lipid chain length and temperature. Since the van der Waals energy of interaction in the closely-packed interdigitated monolayer form is greater than that in the bilayer one, this commonly favors the formation of the former structure. This interaction is counterbalanced by the repulsive interaction of the terminal hydrophobic group of the phospholipid molecule with the aqueous phase as they are exposed to the polar environment at the interface. When alcohol is added into the system, the phase transformation to the interdigitated form occurs. This takes place because the alcohol molecules cover the hydrophobic terminal and reduce the repulsive interaction. The solvent-induced behaviors found in this work may also occur by the similar mechanism, although the results of the elemental analyses of the monolayer phase which agree with those of the bilayer phase, indicate no alcohol at least in the dried sample. The open circles in Fig. 24 show the temperature dependence of χ T for the original monolayer phase. The value of χ T gradually decreases with decreasing temperature

11.7 Controllable Magnetic Properties of Layered Copper Hydroxides

383

Fig. 24. Temperature-dependence of the paramagnetic susceptibilities for the original (open circles) and revived (crosses) monolayer phases and for the bilayer phase (full circles). The inset shows the temperature-dependence of the a.c. susceptibilities for the bilayer phase; real (full circles) and imaginary parts (open circles).

down to 3 K, meaning dominance of an AF interaction. Above 140 K, the behavior can be explained in terms of Curie–Weiss law with C = 0.860 emu K mol−1 and θ = −151 K. The monolayer phase is paramagnetic in the whole temperature range. The closed circles in Fig. 24 show the temperature dependence of χ T for the bilayer phase obtained in acetonitrile. It shows a big anomaly below 100 K, compared with the monotonic variation of the original monolayer phase. The inset of Fig. 24 shows the temperature dependence of the a.c. susceptibilities of the bilayer phase, in which the circles and squares indicate the real (χ ) and imaginary (χ ) components, respectively. The plots of χ make an anomalous peak around 10 K, whose maximum value is 57 emu mol−1 . χ also shows an anomaly at 10.8 K, after temperature independent behavior above it. The divergence of the a.c. susceptibility indicates a ferromagnetic order at TC = 10.8 K. The magnetization curve at 4.5 K shows typical behavior of a weak ferromagnet; an abrupt increase at the lower fields, followed by a gradual increase without showing saturation at the higher fields (not shown). The residual magnetization is 1460 emu mol−1 which corresponds to a canting angle of 13.6◦ between two moments. We found the reversible mono–bilayer phase transition of the organic layer in the hybrid 18, which is activated by soaking in acetonitrile and hot methanol. Such a reversible mono-bilayer transition is quite unusual in organic/inorganic layered materials. In addition the interesting transformation of the organic layer results in the drastic change of the magnetic properties of the inorganic layer. While the monolayer phase is paramagnetic down to 3 K, the bilayer phase becomes ferromagnetic with TC = 10.8 K. The organic layer changes its structure, according to the environment in the organic solvents. This structural change affects the structure of the inorganic layer and modifies the magnetic properties. In addition, the story suggests the possibility of switchable and/or controllable physical properties, derived from cooperation between an inorganic host exhibiting physical properties, such as magnetism, conductivity and so on, and an organic guest playing a role of sensor.

384

11 Hybrid Organic-Inorganic Multilayer Compounds

11.7.2

Photoisomerism of Azobenzenes in Layered Copper Hydroxides

Recently, matrix effects for photoisomerization reactions have been extensively studied in media such as clay minerals [57, 58], polymer films [59], liquid crystals [60], micelles [61], membranes [62, 63], and LB multilayers [64]. It is interesting to elucidate such photochemical reactions in the layered copper hydroxides which show a magnetic variety governed by the exchanged organic molecules. In this section we describe photoisomerism of the azobenzene moiety in compound 18, that exhibits a phase transition between the interdigitated monolayer phase (18a) and the membrane-like bilayer phase (18b), together with the results on a reference sample, Cu2 (OH)3 (p-(phenylazo)benzoate) (19) [65]. The interlayer distance of 19 (14.6 Å) is almost equal to the anion height of 12.0 Å, so that the anion is thought to form an interdigitated monolayer structure. To examine the photo effects, the relative diffuse transmittance was measured on the KBr-diluted pellets of 18a, 18b and 19, before and after UV irradiation in the range 300–400 nm. The obtained transmittance was converted into the absorbance by the Kubelka-Munk equation. The absorption spectra for 1, 18a, 18b, and 19, are depicted in Figs. 25(a)–(d), respectively, where the solid curves show the spectra before irradiation and the broken curves do after it. As shown in Fig. 25(a), the parent compound 1 is transparent for λ > 300 nm. Before irradiation, the spectra for 18a, 18b, and 19 show a characteristic band at ca. 320 nm which is assigned to the π–π ∗ transition in the trans-azobenzene moiety. After irradiation, the spectra of 18a and 18b clearly suggest the trans-to-cis isomerization; the π –π ∗ transition decreases in intensity, although the isomerization is much more complete in 18b than in 18a. It is worth noting here that the cis-to-trans isomerization reaction in 18b does not occur by irradiation for λ > 400 nm. On the other hand, the reference sample 19 shows no photoisomerization. Presence of a spacer between the azobenzene moiety and the carboxylate seems essential for the photochemical reaction. We calculated the

Fig. 25. Absorption spectra for 1 (a), 18a (b), 18b (c) and 19 (d). Solid and broken curves show the spectra before and after UV irradiation, respectively.

11.8 Layered Perovskite Ferromagnets

385

packing coefficient of the intercalation compounds, using the results of the powder X-ray analyses and the structural parameters of Cu2 (OH)3 (NO3 ) [10]. The packing coefficients of 18a, 18b and 19 were shown to be 0.71, 0.39 and 0.62, respectively. The values for 18a and 19 are comparable to those of the molecular crystals. The trans-to-cis isomerization is considered to hardly take place in such dense packing, because it is associated with a large structural change [66, 67]. The packing in 18b seems low enough to allow the isomerization reaction. Magnetic measurements were carried out for the KBr-diluted pellets of 18b in both trans and cis forms. However there was no significant difference between the two varieties. Unfortunately, the photochemical reaction is not linked to the magnetic system. This could be related to the fact that the azobenzene part is dislodged from the inorganic layer with the alkyl chains. Finally, the study of Cu2 (OH)3 (8-(( p-(phenylazo)phenyl)oxy)octanoate) shows that a reversible mono–bilayer transition of the organic layer is observed in methanol and acetonitrile, which results in the drastic modification in magnetism. This is a good example of cooperative property between organic and inorganic parts. The trans-tocis isomerization reaction of the azobenzene moiety is observed by photoirradiation even in the copper hydroxide layer, although the reaction is irreversible and no magnetic change takes place.

11.8

Layered Perovskite Ferromagnets: High-pressure Effects and Spontaneous Magnetization

Compounds of the type (RNH3 )2 CuX4 , where R is an organic molecule and X is Cl or Br, have been known as 2D ferromagnets with TC of 10–15 K [14, 68, 69]. These compounds have also attracted recent interest as a related material to the high TC superconductors and as an organic/inorganic hybrid layered system. They crystallize in a layered perovskite structure, consisting of isolated layers of cornersharing CuX6 octahedra, sandwiched by the organic cations [69–71]. The structure of the CuX4 layer is schematically shown in Fig. 26. It is governed by the cooperative Jahn–Teller effect: each octahedra is prolonged along the Jahn–Teller z axis, that lies

Fig. 26. Schematic view of the structure of the [CuX4 ]2− layer. The arrows indicate the prolonged Jahn–Teller z axes.

386

11 Hybrid Organic-Inorganic Multilayer Compounds

in the CuX plane and has nearly orthogonal relations with the neighbors. The CuX plane includes longer and shorter Cu–Cl bond distances, which are indicated by DL and DS , respectively. The intralayer magnetic interaction is ferromagnetic, because of the orthogonal relation between the magnetic dx2 −y2 orbitals [72]. In this chapter, we describe new aspects of the layered perovskite ferromagnets: high pressure effects and spontaneous magnetization caused by magnetic anisotropy.

11.8.1

High-pressure Effects

Optical measurements on (C2 H5 NH3 )2 CuCl4 have been performed by Moritomo and Tokura under high pressure up to 10 GPa [73]. They found an unusual phase transition at 4 GPa, where rearrangement of the Jahn–Teller distortion takes place and the prolonged z axes become normal to the Cu–Cl plane after it. They speculated that the magnetic interaction between the neighboring Cu(II) ions turns over from ferro- to antiferromagnetic, because of overlaps between the magnetic dx2 −y2 orbitals in the high pressure phase. Therefore, these compounds are candidates for switchable or controllable magnet governed by pressure. To obtain this functionality, search for decreasing the transition pressure and basic study of the high pressure effects on the magnetic behavior are important. In this section we describe high pressure effects on a layered perovskite compound, ( p-cyanoanilinium)2 CuCl4 (20) [74], which shows one of the strongest ferromagnetic interactions in the (RNH3 )2 CuCl4 series. In addition, the lattice exhibits the smallest difference between DL and DS , which may result in a decrease in the critical pressure. For these reasons, we chose compound 20 as the first sample for the high pressure measurements. The closed circles in Fig. 27 show the temperature dependence of the magnetization (M) of 20 at ambient pressure, for an applied field H = 4280 Oe. The magnetization shows a quick increase below 30 K, which is accompanied with the ferromagnetic order. The transition temperature is found to be TC = 9.5 K by a.c. susceptibility mea-

Fig. 27. Temperature-dependence of the magnetization of 20 at different pressures.

11.8 Layered Perovskite Ferromagnets

387

Fig. 28. Pressure-dependence of the magnetic interaction J in 20. The J values are normalized by that at ambient pressure. The dashed curves show the pressuredependence of the square of the overlap integral between the dz2 and pz orbitals, assuming constant bond compressibilities of 2, 3, and 4%/GPa.

surements [74]. As the pressure increases, the quick increase of M because of the ferromagnetic transition, shifts to higher temperature, indicating an increase in TC . The in-plane exchange parameter, J , has been evaluated by fitting the magnetic data above 30 K to the theoretical expression for the two-dimensional ferromagnetic square lattice [75], χ T /C = 1 + 2(J/kT ) + 2(J/kT )2 + 3/4(J/kT )3 + 13/12(J/kT )4 + . . . where C is the Curie constant, and k the Boltzmann constant. Fig. 28 shows the pressure-dependence of the J value, normalized by the ambient pressure value, J = 24 K [74]. The exchange constant increases almost linearly with increasing pressure and becomes 1.4 times as large at 1 GPa. The applied pressure increases both TC and the in-plane coupling J . Figure 29 shows the J values of the various (RNH3 )2 CuCl4 derivatives plotted against DL , the longest Cu–Cl bond distance in the octahedra [71, 74, 76–78]. Such plots have been initiated by Willett et al. [77]. The open circle in this figure shows the position of compound 20, whose DL = 2.877 Å. It is well known that J becomes larger with decreasing DL . The shorter Cu–Cl bond length DS of 20 is 2.298 Å. It is also known that DS depends little on the materials [74, 76–78]. Therefore, the observed pressure-induced enhancement of J would be caused mainly by the decrease in DL , keeping the orthogonal relation between the neighboring magnetic orbitals. The pressure-dependence of the ferromagnetic exchange can be semiquantitatively interpreted in terms of shrinkage of the CuCl lattice. As shown in Fig. 30, the charge transfer from the dz2 orbital in the site 2 to the dx2 −y2 orbital in the site 1, through the Cl pz orbital, is theoretically predicted to be responsible for the ferromagnetic interaction [72]. In this case, J is written as J ≈ (t 2 /U )J H /U , where t is the transfer integral between the dx2 −y2 and dz2 orbitals, U is the on-site Coulomb repulsion of electrons on the Cu(II) ion, and JH is the intraatomic exchange. U and JH depend little on the pressure, because they are intraatomic parameters. Roughly speaking, t is proportional to the two overlap integrals, between the dx2 −y2 and pz orbitals, and between the pz and dz2 orbitals [77]. Because DL is expected to be much more affected by the applied pressure than DS , the observed change of J would originate in the change of the latter overlap integral. The dashed curves in Fig. 28 show pressure dependence of the square of the overlap integral between the pz and dz2

388

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 29. Dependence of the variation of the in-plane exchange interaction on the long Cu–Cl distance DL for various (RNH3 )2 CuCl4 derivatives.

Fig. 30. Superexchange interaction in the CuCl lattice.

orbitals, s 2 , linear decreases in DL with several bond compressibilities of 2, 3 and 4%/GPa being assumed. In this figure, the value of s 2 is normalized by that at the ambient pressure. The enhancement of J can be well understood by the decrease of 3% in DL at the pressure of 1 GPa, which is a reasonable value for the compressibility of molecular crystals. It is also possible to estimate the pressure effect from Fig. 29. If we assume a linear relationship between J and DL in the range DL < 3.2 Å, the increase in the J value at 1 GPa corresponds to a decrease of 6% in DL . The two different analyses give compressibilities of the same order of magnitude. Finally, the pressure-induced enhancement of the ferromagnetic interaction in [CuCl4 ]2− layers is analyzed as a decrease of the longest Cu–Cl bond distance in the Cu–Cl–Cu superexchange pathway. Note that the phase transition discovered by Moritomo and Tokura [73] has not been confirmed by magnetic measurement, because of the limited pressure used during the experiments. However, the concluded structural change, namely the decrease in the longest CuCl bond distance or relaxation of the in-plane Jahn–Teller distortion, is consistent with their optical results [73].

11.8 Layered Perovskite Ferromagnets

11.8.2

389

Spontaneous Magnetization in Layered Perovskite Ferromagnets

Regarding the layered perovskite ferromagnets, the interlayer magnetic coupling is much weaker than the intralayer coupling, and usually antiferromagnetic. Therefore, the magnetic moments cancel out each other, and there is no spontaneous magnetization in the ground state. A few derivatives have been reported to possess a ferromagnetic interlayer interaction [21, 79]. However, there has been no observation of spontaneous magnetization so far, presumably because the interlayer interaction is too weak to make the magnetic hysteresis loop observable. In this section we describe the existence of spontaneous magnetization in ( p-chloroanilinium)2 CuBr4 (21), comparing with the isostructural copper chloride derivatives, ( p-chloroanilinium)2 CuCl4 (22) and ( p-nitroanilinium)2 CuCl4 (23). The crystallographic parameters for the three compounds are given in Table 2. Compounds 21 and 23 crystallize in the orthorhombic Pbca space group, while 22 does in the monoclinic P21 /c space group. Although 22 belongs to a different crystal system, the three compounds are isostructural. Figure 31 shows a schematic comparison of the unit cells of the two crystal systems, where the anilinium ions are omitted for the sake of clarity. The unit cell of 21 or 23 includes two organic-inorganic-organic sandwich layers, while that of 22 includes one. The cell volume of 21 or 23 is twice as large as that of 22. The alignment of the anilinium cations in the organic bilayer seems to cause the difference in crystal system [74]. Figure 32 shows the temperature dependence in the range 5–25 K of χa.c. (real part) for the above compounds, measured in an a.c. field (125 Hz) of 1–5 Oe. The compounds 21–23 exhibit magnetic ordering with an abrupt increase in χa.c. at 15 K, 9.1 K and 7.0 K, respectively. Fig. 33(a) shows the field dependence of χa.c. , namely the differential susceptibility (dM/dH )T , for a polycrystalline sample of 21, at 7 K. Upon increasing the field from zero, the data exhibit a maximum at 30 Oe, at which the magnetization curve has the largest gradient. After passing the maximum, χa.c. goes down to zero, showing saturation of the magnetization. When the magnetic field is decreased, a maximum of χa.c. appears at −30 Oe. The behavior points to a hysteresis loop of magnetization caused by the spontaneous magnetic moment at 7 K. Table 2. Crystal parameters for layered perovskites 21–23.

Formula Crystal system Space group a/Å b/Å c/Å β/◦ V /Å3 Z

21

22

23

(ClC6 H4 NH3 )2 CuBr4 Orthorhombic Pbca 7.551(2) 32.082(10) 7.879(2)

(ClC6 H4 NH3 )2 CuCl4 Monoclinic P21 /c 16.436(4) 7.397(1) 7.265(1) 101.51(2) 865.5(3) 2

(O2 NC6 H4 NH3 )2 CuCl4 Orthorhombic Pbca 7.063(1) 32.460(6) 7.925(2)

1908.8(9) 4

1816.9(6) 4

390

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 31. Schematic comparison of the unit cells of 21, 22, and 23.

Fig. 32. Temperature-dependence of χa.c. for 21 (circles), 22 (triangles), and 23 (diamonds).

The presence of the hysteresis loop is confirmed by dc magnetization measurements (not shown), in which the magnitude of the remnant magnetization is estimated to be 33% of the saturation magnetization. The field-dependent susceptibilities χa.c. for 22 and 23 are plotted in Figs. 33(b) and (c), respectively. The data also show two maxima at ca. −15 and 15 Oe, but no hysteresis is found. The absence of hysteresis loop agrees with a model of bulk antiferromagnet, with ferromagnetic layers coupled through a weak AF interaction. Then, the maximum in χa.c. is considered to correspond to a spin-flop transition. Basically, ferromagnetic systems may also exhibit a negligible hysteretic effect, but this needs fully isotropic spin moments. As a result, compounds 22 and 23 are very similar to previously studied Cu(II) layered perovskites. It is concluded that the substitution of chloride with bromide for X in ( p-chloroanilinium)2 CuX4 leads to the occurrence of spontaneous magnetization. To our knowledge, this is the first example of spontaneous magnetization in the Cu(II) layered perovskite materials. The field dependence of χa.c. for a single crystal of 21 was carefully examined, to enable understanding of the spontaneous magnetization (Fig. 34). When the field

11.8 Layered Perovskite Ferromagnets

391

Fig. 33. Field-dependence of χa.c. for polycrystalline samples of 21 at 7 K (a), 22 at 6 K (b) and 23 at 5 K (c).

is applied parallel to the b axis, χa.c. shows a hysteretic variation similar to that of the polycrystalline sample. The compound behaves along the b direction as a ferromagnet. For a field parallel to the a axis, the χa.c. values exhibit sharp peaks at 130 and −130 Oe, but no hysteresis is detected. Such peaks have been observed for (C2 H4 NH3 )2 CuCl4, and assigned to spin-flop transitions [79]. The spin flop transition occurs for a field parallel to the AF easy axis. It is known that the magnetic susceptibility parallel to the easy axis is non-zero and constant in the spin flop phase, but begins to decrease suddenly at a critical field, at which the transition from spin flop phase to paramagnetic phase takes place. The plots of χa.c. for 21 make no plateau after the sharp peaks, suggesting the absence of spin flop phase. The sharp peaks of χa.c. are probably due to metamagnetic transitions from an AF state to a paramagnetic one. This transition is usually realized under the condition that the magnetic anisotropy is larger than the antiferromagnetic coupling energy. It is concluded that

392

11 Hybrid Organic-Inorganic Multilayer Compounds

Fig. 34. Field dependence of χa.c. for the single crystal of 21, in the field parallel to the b axis (a), the a axis (b) and the c axis (c).

compound 21 behaves as an antiferromagnet in a field parallel to the a axis. Along the c axis, χa.c. is non-zero and depend little on the field. This clearly indicates that the c axis coincides with the magnetic hard axis. These results indicate that the easy axis is parallel to the a axis, and the direction of the spontaneous magnetization is parallel to b. Owing to the orthogonality between both directions, canted ferromagnetism is reasonably concluded for the mechanism of the spontaneous magnetization. The insets in Figs. 34(a), (b) and (c) schematically give the alignment of the “magnetic moments” on two neighboring layers in 21, with increases in field parallel to the b, a and c axes, respectively. Here, “the moment” means the effective intralayer magnetic moment resulting from in-plane ferromagnetic interactions. In the insets of Figs. 34(b) and (c), the spin canting toward the b axis is ignored. The expected shape of the magnetization curves is also drawn there. The spin canting is not allowed in 22 because the adjacent layers are related by a translational operation [80], while the space group (Pbca) of 21 and 23 allows it. Although there is no symmetry difference between 21 and 23, the canted ferromagnetism is not observed in 23. Further, (C2 H5 NH3 )2 CuCl4 also belongs to the same crystal system, but it has been reported not to show spontaneous magnetization [81]. The crystal structure of 21 has no extraordinary feature compared with those of other layered perovskites. Since there is no single-ion anisotropy on the copper(II) ion (SCu = 1/2), the so-called antisymmetric spin coupling [82] should cause the spin canting. It is concluded that an enhancement of the spin–orbit interaction caused by bromide in 21 is likely responsible for the spin canting [81].

References

393

Acknowledgments It is a pleasure to acknowledge the collaboration of Raymond Ziessel, Philippe Turek, and Pierre Panissod on hydroxide-based compounds and the technical assistance of Alain Derory and Richard Poinsot for magnetic measurements.

References [1] D. O’Hare, Inorganic Materials (Eds. D.W. Bruce, D. O’Hare) John Wiley & Sons, Chichester, 1993, chapter 4. [2] R. Brec, D. Schleich, G. Ouvrard, A. Louisy and J. Rouxel, J. Inorg. Chem. 1979, 18, 1814. [3] J.M. Kosterlitz and D. Thouless, J. Phys., 1973, C6, 1186. [4] M. Meyn, K. Beneke and G. Lagaly, Inorg. Chem. 1990, 29, 5201; Inorg. Chem. 1993, 32, 1209. [5] J.F. Nicoud, Science 1994, 263, 636. [6] P. Day, Mol. Cryst. Liq. Cryst. 1997, 305, 533. [7] P. Gutlich, ¨ Mol. Cryst. Liq. Cryst. 1997, 305, 17. [8] L. Ouahab, Chem. Mater. 1997, 9, 1909. [9] R. Clement, ´ P.G. Lacroix and D. O’Hare, J. Evans, Adv. Mater. 1995, 6, 794; S. Benard, ´ P. Yu, T. Coradin, E. Riviere, ` K. Nakatani and R. Clement, ´ Adv. Mater. 1997, 9, 981. [10] H. Effenberger, Z. Krist., 1983, 165, 127. [11] M. Drillon, C. Hornick, V. Laget, P. Rabu, F.M. Romero, S. Rouba, G. Ulrich and R. Ziessel, Mol. Cryst. Liq. Cryst. 1995, 273, 125. [12] P. Rabu, S. Angelov, P. Legoll, M. Belaiche, and M. Drillon, Inorg. Chem. 1993, 32, 2463. [13] T. Takada, Y. Bando, M. Kiyama, H. Miyamoto and T. Sato, J. Phys. Soc. Jpn. 1966, 21, 2745. [14] L.J. de Jongh, Magnetic Properties of Layered Transition Metal Compounds, Kluwer Academic Publishers, 1990. [15] S. Miyata and T. Kumura, Chem Lett. 1973, 843. [16] M. Bujoli-Doeuff, L. Force, V. Gadet, M. Verdaguer, K. el Malki, A. de Roy, J.P. Besse and J.P. Renard, Mat. Res. Bull. 1991, 26, 577. [17] S. Yamanaka, T. Sako and M. Hattori, Chem. Lett. 1989, 1869; Solid State Ionics, 1992, 53, 527. [18] W. Fujita and K. Awaga, unpublished results. [19] P. Rabu, S. Rouba, V. Laget, C. Hornick and M. Drillon, J. Chem. Soc., Chem. Comm. 1996, 1107. [20] W. Fujita and K. Awaga, Inorg. Chem. 1996, 35, 1915. [21] L.J. de Jongh, W.D. van Amstel and A.R. Miedema, Physica, 1972, 58, 277. [22] W. Fujita and K. Awaga, J. Am. Chem. Soc. 1997, 119, 4563. [23] V. Laget, C. Hornick, P. Rabu and M. Drillon, J. Mater. Chem., 1999, 9, 169. [24] N. Masciocchi, E. Corradi, A. Sironi, G. Moretti, G. Minelli, and P. Porta, J. Solid State Chem. 1997, 131, 252. [25] W. Fujita, K. Awaga and T. Yokoyama, Inorg. Chem. 1997, 36, 196. [26] T. Takada, Y. Bando, M. Kiyama and H. Miyamoto, J. Phys. Soc. Jpn. 1966, 21, 2726.

394

11 Hybrid Organic-Inorganic Multilayer Compounds

[27] V. Laget, M. Drillon, C. Hornick, P. Rabu, F. Romero, P. Turek and R. Ziessel, J. Alloys Compounds 1997, 262, 423. [28] O. Kahn, Molecular Magnetism, VCH Publishers, Weinheim, 1993. [29] M. Atanasov, N. Zotov, C. Friebel, K. Petrov and D. Reinen, J. Solid State Chem. 1994, 108, 37. [30] E. Coronado, M. Drillon and R. Georges, Research Frontiers in Magnetochemistry (Ed. C.J. O’Connor), World Scientific, 1993, 27. [31] V. Laget, S. Rouba, P. Rabu, C. Hornick and M. Drillon, J. Magn. Magn. Mater., 1996, 154, L7; V. Laget, C. Hornick, P. Rabu, M. Drillon and R. Ziessel, Coord. Chem. Rev. 1998, 178, 1533. [32] W. Stahlin ¨ and H.R. Oswald, J. Solid State Chem., 1971, 2, 252. [33] M. Louer, ¨ D. Louer, ¨ D. Grandjean, Acta Cryst. 1973, B29,1696. [34] V. Laget, These ` Universite´ Louis Pasteur, 1988, Strasbourg, France. [35] M. Gˆırtu, C.M. Wynn, W. Fujita, K. Awaga, A.J. Epstein, J. Appl. Phys. 1998, 83, 7378. [36] J. Hauschild, H.J. Elmers and U. Gradmann, Phys. Rev. B 1998, 57, 677. [37] C.M. Wynn, M.A. Gˆırtu, W.B. Brinckerhoff, K.I. Sugiura, J.S. Miller and A.J. Epstein, Chem. Mat. 1997, 9, 2156. [38] M. Drillon and P. Panissod, J. Magn. Magn. Mat. 1998, 188, 93. [39] D. Bloch, J. Phys. Chem. Solids 1963, 27, 881; L.J. de Jongh and R. Block, Physica 1975, 79B, 568. [40] N.W. Aschcroft and N.D. Mermin, Solid State Physics (Ed. Hold-Saunders Int.) 1981. [41] P. Panissod and M. Drillon, Magnetism Molecules to Materials, same series. [42] M. Takahashi, Phys. Rev. Letters 1987, 58,168. [43] V.L. Pokrovsky, G.V. Uimin, Magnetic Properties of Layered Transition Metal Compounds, (Ed. L.J. de Jongh) Kluwer Academic Publishers, Dordrecht,1990. [44] C. Hornick, P. Rabu and M. Drillon, Polyedron, 2000, 19, 259. [45] D.L. Bish, J.E. Post, Modern Powder Diffraction, Reviews in Mineralogy vol. 20 (Ed. P.H. Ribbe) BookCrafters, Inc., Michigan, 1989. [46] N. Zotov, K. Petrov and M. Dimitrova-Pankova, J. Phys. Chem. Solids 1990, 51, 1199. [47] G.S. Rushbrooke, G.A. Baker and P.J. Wood, Phase Transition and Critical Phenomena, (Ed. C. Domb and M.S. Green) Academic Press, 1974. [48] W. Fujita, PhD thesis, The University of Tokyo (Japan), 1996. [49] V. Laget, C. Hornick, P. Rabu, M. Drillon, P. Turek, and R. Ziessel, Adv. Mater. 1998, 10, 1024. [50] G. Weissman, R. Claiborne, Cell Membranes, HP Publishing, New York, 1975. [51] T. Kajiyama, J. Macromol. Sci. Chem. 1988, A25, 583. [52] G. Lagaly, Angew. Chem. Int. Ed. Engl. 1976, 15, 575. [53] Y.Mathey, R. Setton, C. Mazieres, Can. J. Chem. 1977, 55, 17. [54] H. Kopka, K. Beneke, G. Lagaly, J. Colloid Interface Sci. 1988, 123, 427. [55] S. A. Simon, T. J. McIntosh, Biochim. Biophys. Acta 1984, 773, 169. [56] J. L. Slater, C.-H. Huang, Prog. Lipid Res. 1988, 27, 325. [57] M. Ogawa, K. Kuroda, Chem. Rev. 1995, 95, 399. [58] M. Ogawa, K. Kuroda, Bull. Chem. Soc. Jpn. 1998, 70, 2593. [59] G. Smets, Adv. Polymer Sci. 1983, 50, 17. [60] V. Ramesh, M.M. Labes, J. Am. Chem. Soc. 1987, 109, 3228. [61] J. Sunamoto, K. Iwamoto, M. Akutagawa, M. Nagase, H. Kondo, J. Am. Chem. Soc. 1982, 104, 4904. [62] D.A. Holder, H. Ringsdorf, V. Deblauwe, G. Smets, J. Phys. Chem. 1984, 88, 716. [63] T. Seki, H. Ichimura, J. Chem. Soc. Chem. Commun. 1987, 1187.

References

395

[64] E. Ando, J. Miyazaki, K. Morimoto, H. Nakahara, K. Fukuda,Thin Solid Films 1985, 133, 21. [65] W. Fujita, K. Awaga, Mol. Cryst. Liq. Cryst. 1998, 315, 29. [66] K.Takagi, H. Usami, H Fukaya, Y. Sawaki, J. Chem. Soc. Chem. Commun. 1988, 1174. [67] K. Takagi, T. Shichi, H. Usami, Y. Sawaki, J. Am. Chem. Soc. 1993, 115, 4339. [68] R. D.Willett, in Magneto-Structual Correlations in Exchange Coupled Systems; R.D. Willett, D. Gatteschi and O. Kahn (Eds.; D. Reidel) Publishing Company: 1985; 269. [69] R.Willett, H. Place, M. Middleton, J. Am. Chem. Soc. 1988, 110, 8639. [70] J. P.Steadman, R. D. Willett, Inorg. Chem. Acta 1970, 4, 367. [71] K. P. Larsen, Acta Chem. Scand. 1974, A28, 194. [72] K. I. Kugel, D. I. Khomskii, Sov. Phys. Usp. 1982, 25, 231. [73] Y. Moritomo and Y. Tokura, J. Chem. Phys. 1994, 101, 1763. [74] T. Sekine, T. Okuno and K. Awaga, Mol. Cryst. Liq. Cryst. 1996, 279, 65. [75] G. A. Baker, H. E. Gilbert, J. Eve and G. S. Rushbrooke, Phys. Lett. 1967, 25A, 207. [76] C. P. Landee, K. E. Halvorson and R. D. Willett, J. Appl. Phys. 1987, 61, 3295. [77] R. D. Willett, F. H. Jardine, I. Rouse, R. J. Wong, C. P. Landee and M. Numata, Phys. Rev. B 1981, 24, 5372. [78] R. Wong and R. D. Willett, J. Magn. Reson. 1981, 42, 446. [79] J. J. M.Steijger, E. Frikkee, L. J. de Jongh, W. D. Huiskamp, Physica 1984, B123, 271. [80] P. Zhou, J. E. Drumheller, B. Patyal, R. D. Willett, Phys. Rev. B 1992, 45, 12365. [81] P. Bloembergen, P. J. Berkhout, J. J. M. Franse, Int. J. Magnetism 1973, 4, 219. [82] T. Moriya, in Magnetism; Vol. 1 (Eds Rado and Suhl) 1963, 85.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

12

Intercalation-induced Magnetization in MPS3 Layered Compounds Ren´e Cl´ement and Anne L´eaustic

12.1

Introduction and Scope

Magnetic interactions are observed for many low-dimensional compounds [1–6] and these acquire spontaneous magnetization at low temperature either because of ferroand ferrimagnetism or because of canted antiferromagnetism. Many consist of inorganic sheets separated by organic layers and are often referred to as hybrid organicinorganic compounds. Examples are the alkylammonium tetrachlorocuprates or chromates and the layered heterometallic oxalates. In other cases, layers are separated by water molecules, or grafted with organic chains [7–10]. In all these cases, the compounds usually crystallize out of solution and their magnetic properties are determined once they are formed. The scope of this chapter is limited to the case where the process of inserting foreign species into a host structure causes such profound modifications of the host as to deeply affect its magnetic properties and induce previously non-existent magnetization. To the best of our knowledge, the only such case reported so far concerns the MPS3 layered compounds (M = Mn, Fe, Ni), hence the scope of this chapter will be limited to these compounds. The MPS3 compounds form a vast family which has received much attention over the past two decades for their unusual reactivity (redox and cation-exchange intercalation reactions) [11–13] and for the numerous properties of both the pure and intercalated phases. Application of the MPS3 compounds as cathodic materials [11] catalysts [14], ferroelectric [15], non-linear optical [16] and magnetic materials have been reported. Magnets with critical temperatures as high as 90 K [17] as well as multi-property materials [18] have been obtained in this family of compounds.

12.2

Structural and Electronic Aspects

The MPS3 (or M2 P2 S6 ) compounds, where M is a transition metal in the +II oxidation state (M = Mn, Fe, Co, Ni, Zn, Cd, etc.), form a vast class of lamellar semiconductors which has been rediscovered in the early seventies [19]. These compounds are typically synthesized by heating a stoichiometric mixture of metal, sulfur and red phosphorus around 700◦ C in an evacuated sealed quartz ampoule. This procedure affords polycrystalline powders as well as monocrystalline thin platelets (typically

398

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 1. Side view of the MPS3 layered structure.

10 mm2 ) which grow by vapor phase transport. In addition to the homometallic compounds, a large variety of heterometallic compounds such as Fe1−x Nix PS3 can be formed [11], which have the same structure and can be viewed as solid solutions. More complex compounds have also been described, in which two MII cations are replaced by a (MI + MIII ) combination, for example the ferroelectric InIII CuI P2 S6 [15]. All the layered MPS3 phases have a common monoclinic structure based on the CdCl2 type (space group C2/m ) shown in Fig. 1 [19–21]. This consists of a cubic close-packed sulfur array with alternate layers of cation sites vacant. The atomic arrangement within a slab can be described in two ways. The representation in Fig. 2a emphasizes the juxtaposition of sulfur octahedra sharing edges, with metal ions and P–P pairs (nearly perpendicular to the layer plane) located inside the sulfur octahedra. Alternatively, a slab can be described as an extended polynuclear coordination complex formed by MII ions held together by (P2 S6 )4− bridging ligands (Fig. 2b). This “coordination view” of the MPS3 compounds is actually relevant to many features of their reactivity, and the compounds have been even sometimes considered 4− ionic salts, hence the “hexathiohypodiphosphates” name often enas M2+ 2 (P2 S6 ) countered in contrast with the “metal phosphorus trisulfides” terminology. When dealing with the magnetic properties of the MPS3 compounds, it is useful to realize that the MII ions are located at the corners of a honeycomb lattice and that the superexchange pathway between two next-neighbor ions is mediated by two bridging sulfur atoms belonging to two different P2 S6 groups. There is a trigonal distortion of the sulfur S6 octahedra surrounding each MII ion. In the case of FeII , the combination of spin-orbit coupling and trigonal distortion results in strong anisotropy of the magnetic properties. Various models of the electronic structure of the MPS3 compounds have been suggested [22, 23]. These materials have very low conductivity (usually ca 10−8 −1 cm−1 except FePS3 whose conductivity reaches 10−5 −1 cm−1 ) and they have been described as broad-band semiconductors [24]. Their color depends on the nature of

12.3 Magnetic Properties of the Pristine MPS3 Phases (M = Mn, Fe, Ni)

399

Fig. 2. (a) Schematic top view of a single MPS3 layer, which consists of close-packed P2 S4− 6 ions. The MnII ions are located in the octahedral sites of sulfur atoms (represented by circles) generated by every three adjacent P2 S4− 6 ions. The P–P bonds are perpendicular to the plane of the figure. The a and b axes of the monoclinic unit cell are shown (a = 6.077 ˚A, b = 10.524 ˚A). (b) Schematic top view of a single MPS3 layer regarded as a polynuclear complex made up of MII cations coordinated to P2 S4− 6 (ethane-like) bridging ligands.

the metal (pale green for Mn, colorless for Zn and Cd, black for Fe and Ni). Tight binding band structure calculations [25] have shown that there is an important gap, about 10 eV, between the upper sulfur based bonding orbitals of the valence band and the antibonding conduction bands. The metal d orbitals remain quite localized and form narrow, partially filled eg and t2g bands that lie slightly above the top of the valence band. The optical absorption which determines the color is because of transitions between the top of the sulfur-rich valence band to the eg metal states. Because the d crystal orbitals are narrow, electrons do not pair up, hence the MPS3 compounds can be considered as Mott insulators.

12.3

Magnetic Properties of the Pristine MPS3 Phases (M = Mn, Fe, Ni)

The magnetic properties of the pristine (unintercalated) MPS3 compounds have been the subject of numerous investigations [24, 26–30]. The susceptibility in the paramagnetic region shows that the divalent metallic ions are in their high spin configuration where applicable (Fig. 3a, b). Because of the Van der Waals gap separating adjacent slabs, strong magnetic interactions occur only between in-plane ions. The shortest interlayer M–M distance is of the order of the third-neighbor M–M distance within the same layer and therefore the magnetic lattice is nearly bidimensional. The

400

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 3. (a) Temperature-dependence of the reciprocal magnetic susceptibility χ −1 for pristine MPS3 phases (M = Mn, Fe) (b) Temperature dependence of χ −1 for pristine NiPS3 .

three pristine MPS3 (M = Mn, Fe, Ni) compounds are antiferromagnets with Neel ´ temperatures of 78, 123 and 155 K, respectively [27]. Magnetic structures have been determined by Neutron diffraction [11, 31–33]. Although the compounds are isostructural and antiferromagnets, the type of magnetic ordering is quite different (Fig. 4). In MnPS3 , the nearest-neighbor interactions between the MnII ions are all antiferromagnetic, so that each spin “up” (S = 5/2) is surrounded by three spins “down” and vice versa. In FePS3 however, each FeII ion is coupled ferromagnetically to two of the nearest neighbors and antiferromagnetically to the third one. Thus, the FeII spins form ferromagnetic chains (running along the a axis) coupled antiferromagnetically to each other along the b direction (Fig. 4). In both compounds all magnetic moments are perpendicular to the layer planes. The magnetic unit cell of FePS3 is a superstructure of the crystallographic one, the c parameter being doubled. This arises because two ferromagnetic chains of adjacent layers are not in registry along the c stacking direction. There is some controversy as to the magnetic structure of NiPS3 : Brec and Ouvrard have suggested a structure

Fig. 4. Schematic diagram of the magnetic structures of pristine MnPS3 and FePS3 . Only the metal atoms are shown. The white and black circles represent up and down magnetic moments.

12.4 Ion-exchange Intercalation into the MPS3 Compounds

401

similar to that of FePS3 where the magnetic moments are perpendicular to the layer planes [11]. More recently, Joy et al. have studied the anisotropy of the susceptibility and suggested that the magnetic moments have directions parallel to the plane of the layers [29]. Although the three mentioned compounds are isostructural, the spin dimensionalities are very different. In the paramagnetic range, the susceptibility of MnPS3 is isotropic [29, 34]. It has been fitted down to 80 K by a Rushbrook and Wood expansion for a honeycomb lattice of spins 5/2 assuming a two-dimensional Heisenberg Hamiltonian [34, 35]. In contrast, the susceptibility of FePS3 exhibits considerable anisotropy and the magnetic properties are most effectively treated by an Ising model. The magnetic ordering in FePS3 is accompanied by a discontinuous change of the lattice parameters and it appears as a first-order phase transition in a 2D Ising antiferromagnet [32]. The strong anisotropy arises from a combination of spin–orbit coupling and trigonal distortion (elongation) of the FeS6 octahedra. Magnetic exchange constants and zero-field splitting energies of the ground state of the FeII ions have been evaluated from the anisotropic paramagnetic susceptibilities [29]. An unusual feature is the large difference of the Weiss constants deduced from the measurements of the susceptibilities χ|| ( = 53 K) and χ⊥ ( = −54 K). This dependence on the direction provides an explanation for the discrepancies in the values of found in several articles on powdered FePS3 samples, which is attributed to various preferential orientations of the microcrystals [24, 26, 28, 31]. In the case of NiPS3 , Joy et al have claimed that the magnetic properties are best described by an anisotropic X Y Heisenberg Hamiltonian [29].

12.4

Ion-exchange Intercalation into the MPS3 Compounds

The MPS3 compounds possess an unusual intercalation chemistry that involves either redox or ion-exchange reactions. The studies dealing with the redox route are dominated by the chemical and electrochemical intercalation of lithium into NiPS3 because of the application of this compound as a battery material [11, 36]. As no spectacular phenomena have been reported as far as spontaneous magnetization is concerned, we will not comment any further this class of intercalates. The basic observation of ion-exchange intercalation is the spontaneous reaction, at room temperature, of MnPS3 with solutions of a number of ionic compounds such as KCl, tetramethylammonium or cobalticinium chloride. These reactions lead to compounds Mn1−x PS3 (G)2x (H2 O) y , where the positive charge of the guest cations G+ is counterbalanced by the loss of an equivalent amount of intralayer MnII cations [37, 38]: MnPS3 + 2xKCl (excess in water) → Mn1−x PS3 K2x (H2 O)≈1 + xMnCl2 (x ≈ 0.2) This process appears rather unusual, as it implies that manganese cations are able to leave their intralamellar sites under very mild conditions. Intralayer vacancies

402

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

are created, which play an important role, see below. The guest species are solvated by solvent molecules. The intercalates are well crystallized, as their X-ray powder diffraction patterns exhibit sharp hkl reflections. Their composition cannot be modulated, i. e. the reaction always leads to a fully loaded intercalate where the guest species are close-packed. In case of an insufficient amount of the guest species, a diphasic mixture of MnPS3 and of the fully loaded intercalate is obtained. Insertion fails when the size of the guest species are too big. However, bulky species can be inserted in two steps: insertion of hydrated alkali metal ions followed by exchange of the alkali ions with the bulky species. For example, potassium ions have been exchanged with Ru(2,2 bipy)2+ 3 . Several other MPS3 phases (M = Mn, Cd, Zn, Fe) give rise to a similar insertion process but the reaction often requires assistance, which can be provided by a complexing agent such as EDTA that coordinates the leaving MII cations [39]. Each host material in the series has its own particularities, and some guest species obviously have more affinity towards the host lattice than others: for example, FePS3 inserts pyridinium or methylviologen cations under mild conditions (6 h in methanol at 60◦ C), but the presence of EDTA is required to insert tetraethylammonium species. The ion-exchange intercalation process afforded by the MPS3 compounds is unique in the field of inorganic layered materials. The high mobility of the intralamellar MII cations is very difficult to account for if insertion were to take place by diffusion in the solid state. We have suggested that the MII cation transfer and intercalation process occur via local microdissolution and subsequent reconstruction of the host lattice, thanks to an heterogeneous equilibrium between solid MPS3 and 4− the constituting solvated species M2+ aq and P2 S6aq [40]. Mn2 P2 S6 (s)

+4x G+

4− Mn2+ aq + P2 S6

Mn2−2x P2 S6 G4x (H2 O)y (s)

This process is favored by the small charge +2 of the metallic cations and the ionicity of the metal-sulfur bonding. Ion exchange has not been observed with NiPS3 in which the M–S bonding is more covalent.

12.5 12.5.1

The Magnetic Properties of the MnPS3 Intercalates General

The temperature-dependence of the magnetic susceptibility (measured with a Faraday magnetometer) of pure MnPS3 and of several ion-exchanged intercalates is shown in Fig. 5 as 1/χ = f (T ) [41]. Above 100 K, all the intercalates exhibit excellent agreement to a Curie–Weiss law with strongly negative Weiss constants, which reveal that the interactions are still antiferromagnetic (AF). However, the susceptibility of all intercalates is larger than that of MnPS3 and the maximum of χ that

12.5 The Magnetic Properties of the MnPS3 Intercalates

403

Fig. 5. Temperature-dependence of the reciprocal magnetic susceptibility χ −1 for different MnPS3 intercalates.

Fig. 6. Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization (RE), and zero-fieldcooled magnetization (ZFCM) of the Mn0.83 PS3 (CoCp2 )0.34 intercalate in a field of 30 Oe.

occurs around 100 K in MnPS3 is no longer seen. Therefore the AF interactions are weaker in the intercalates than in pristine MnPS3 . More striking, non-linearity in the intercalate plots develops as the temperature is lowered, until around a critical temperature TC a very large increase of the susceptibility is observed, indicating the onset of a spontaneous bulk magnetization. The dependence of the magnetization of the powdered intercalates on temperature has been measured using a SQUID and a low field. Representative results are shown in Figs. 6 and 7 in the case of the tetramethylammonium and the cobalticenium intercalates, respectively. The field-cooled magnetization (FCM) observed on cooling under 10 Oe shows a steep increase when T falls below 35 K. Then the field was switched off at 10 K and the sample allowed to warm up. The remnant magnetization is quite strong, particularly for the cobalticinium intercalate and it steadily decreases until it vanishes at TC . The zero-field-cooled magnetization (ZFCM) was obtained by cooling from 100 K to 10 K in zero field, then warming up under 10 Oe. In both cases, the ZFCM slightly increases on warming and drops rapidly at TC . The dependence of magnetization on applied magnetic field has been measured at 10 K for several intercalates. Data are shown in Fig. 8 for the tetramethylammonium intercalates. M rises steeply as the field is increased and reaches a saturation regime. However Msat (ca 3500 cm3 mol−1 Oe per mole of Mn) is only a fraction of the the-

404

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 7. Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization (RE), and zero-fieldcooled magnetization (ZFCM) of the Mn0.8 PS3 (Me4 N)0.4 intercalate within a field of 30 Oe.

Fig. 8. Dependence of the magnetization of the Mn0.83 PS3 (CoCp2 )0.34 intercalate (powder) on applied magnetic field at 10 K.

oretical value that would be attained if all the spins were aligned (27 900 cm3 mol−1 (per mole of Mn)). Upon cycling the applied magnetic field, the intercalates show a narrow hysteresis loop (width ∼200 Oe) characteristic of soft magnets. To appreciate the anisotropy of the magnetization, a monocrystalline platelet of the tetramethylammonium intercalate was synthesized. The dependence of the magnetization at 10 K on applied magnetic field is shown in Fig. 9 for two orientations (parallel and perpendicular) of the field with respect to the plane of the platelet (which is also the plane of the slabs). Examination of Fig. 9 suggests that the axis of easy magnetization is essentially perpendicular to the plane of the platelets, and that the spins in the magnetically ordered state are essentially perpendicular to the plane of the slabs, as in pure MnPS3 . The whole body of data is therefore characteristic of the occurrence of ferrimagnetism or weak ferromagnetism in both intercalates. Similar results have been obtained with other guest species and Table 1 reports the composition of the various compounds investigated as well as the values of interlayer spacings, saturated magnetizations Msat and critical temperatures TC . The main question that arises from the above experimental results is the mechanism by which intercalation turns the antiferromagnetic properties of MnPS3 into ferrimagnetic ones. A definite answer would necessitate the knowledge of refined

405

12.5 The Magnetic Properties of the MnPS3 Intercalates Table 1. Saturated magnetization at 10 K of selected MnPS3 intercalates. Intercalate

Spacing (Å)

Msat (cm3 mol−1 Oe)

TC (K)

Mn0.84 PS3 (Me4 N)0.32 (H2 O)∼1 Mn0.83 PS3 (CoCp2 )0.34 (H2 O)∼0.3 Mn0.81 PS3 (K)0.38 (H2 O)∼1 Mn0 .89 PS3 (octyl NH3 )0.22 (H2 O)∼0.5 Mn0 .86 PS3 (pyH)0.28 (H2 O)∼0.7 Mn0 .80 PS3 (NH4 )0.4 (H2 O)∼1

11.45 11.82 9.37 10.38 9.65 9.38

4000 4000 1000 1150 80 3100

35 35 20 45 35 15

Fig. 9. Dependence of the magnetization at 10 K of a single platelet of the Mn0.8 PS3 (Me4 N)0.4 intercalate on magnetic field applied perpendicular or parallel to the plane of the platelet.

crystallographic and magnetic structures of the intercalates. This is however out of reach, because intercalated crystals always have too many defects. Nevertheless, X-ray and neutron diffraction experiments have been carried out on the cobalticenium and tetramethylammonium intercalates, the latter being the best crystallized compound encountered in our study.

12.5.2

X-ray and Neutron-diffraction Study of Selected Intercalates

The X-ray powder diffraction spectra of Mn0.84 PS3 (Me4 N)0.32 (H2 0) y and of Mn0.83 PS3 [Co(C5 H5 )2 ]0.34 (H2 O) y at room temperature display reflections almost as sharp as those of MnPS3 . Interesting enough, these intercalates exhibit several reflections in the low angle region that cannot be indexed using the monoclinic unit cell of MnPS3 , even if the increase of interlayer distance is taken in account. We have shown previously that a superstructure involving the tripling of the a axis of pristine MnPS3 allowed complete indexation of the powder spectra [42]. A subsequent study involving X-ray powder and single crystal experiments has confirmed this early suggestion, but it has also shown that the overall structure of these intercalates was more complicated [43, 44]. For example, in the case of the cobalticenium intercalate, disorder in the layer stacking mode has been evidenced. The tetramethylammonium intercalate appears much more ordered, with the existence of a guest superlattice in

406

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 10. Neutron diffraction pattern of the deuterated Mn0.83 PS3 (CoCp2 )0.34 intercalate (reproduced from Ref. 45).

Fig. 11. Neutron diffraction pattern of the deuterated Mn0.84 PS3(Me4 N)0.32 intercalate (reproduced from Ref. 45).

addition to the host superlattice. A shift of the host lattice layers in the a direction by a/3 has also been evidenced in the tetramethylammonium intercalate, which results in a change of the monoclinic angle, β, from 107.35 to 90◦ . The combination of both the host and the guest superlattices results in a global hexagonal unit cell of dimensions a = b = 36.6 Å [44]. However, as far as the interpretation of a spontaneous magnetization in the intercalates is concerned, it will be sufficient to retain from the above discussion the appearance upon intercalation of a host superlattice with a triple a axis (as represented in Fig. 12). Powder neutron diffraction patterns for Mn0.83 PS3 [Co(C5 D5 )2 ]0.34 (D2 O) y and Mn0.84 PS3 [N(CD3 )4 ]0.32 (D2 O) y recorded above and below the magnetic ordering temperatures are shown in Figs. 10 and 11 [45]. In both cases, the low temperature patterns exhibit a number of additional reflections which can be attributed to the magnetic lattice and confirm that these materials are magnetically ordered. The magnetic reflections can all be indexed as hk0 reflections of the unit cell having a tripled a axis [45]. This appearance of only hk0 reflections suggests that the magnetic order is largely confined within a single layer of these materials. Examination of the peak profiles of the magnetic scattering (difference curve) of the cobalticenium intercalate shows the existence of broad peaks falling off at high angles. Such a peak shape is characteristic of turbostratically disordered layered materials [46]. The neutron and X-ray diffraction results suggest that the ferrimagnetic behavior of the intercalates is strongly connected to the superstructure evidenced, and also that the superstructure is related to the ordering of the metal vacancies. Significant

12.5 The Magnetic Properties of the MnPS3 Intercalates

407

support for this model can be inferred from consideration of relative intensities of the peaks observed. For example in the case of Mn0.83 PS3 [Co(C5 D5 )2 ]0.34 (H2 O) y the superstructure peak at d = 9.1 Å is observed to be strong in the room temperature Xray pattern, very weak in the high temperature neutron pattern, yet has a significant magnetic contribution below TC . Consideration of relative scattering powders reveals that a metal vacancy peak would appear strongly in X-ray diffraction (where heavy metal atoms dominate the overall scatter), only weakly in the high temperature neutron pattern (where scattering is dominated by the deuterated guest molecules), yet have a high magnetic contribution below TC since magnetic scatter is due to the electrons of the metal atoms. Further support that an ordered array of metal vacancies can exist and give rise to observable effects in the diffraction pattern of these materials comes from the related compound InIII 2/3 PS3 , which adopts a structure very similar to that of MnPS3 , but in which one third of the possible metal sites are vacant. In this compound, evidence has been brought that the unit cell was tripled in the a direction [47, 48].

12.5.3

A Ferrimagnetic Model of the MnPS3 Intercalates: Imbalancing of Spins

To summarize, there is a strong support for a link between metal vacancy sites and magnetic ordering in the ion-exchanged intercalates of MnPS3 . The observed similarities between the magnetic scatter in the low temperature neutron experiments and the peaks due to the metal vacancy lattice in the high temperature X-ray patterns have shown that the formation of a metal vacancy superstructure and transition to the magnetically ordered state are intimately linked. This provides evidence that the observation of a spontaneous magnetization results from the imbalancing of spins on the two sublattices of the AF ordered host lattice, leading to a net magnetization. It is possible to propose a tentative spin arrangement to explain the observation of ferrimagnetism, which has a reasonable quantitative agreement with the values of Msat found for the ion-exchanged intercalates. The process of intercalation gives rise to the loss of approximately one metal ion in six, leading to the observed superstructure because of the metallic vacancy ordering. One can therefore imagine a situation as depicted in Fig. 12 whereby the ordered removal of ions on only one spin sublattice gives rise to an excess of spin on the second sublattice, resulting in an overall net moment. The arrangement of Fig. 12 would result in a saturated magnetization of 1/6(N gβ S) which for an isotropic MnII system would be 4000 cm3 mol−1 Oe. The experimental values of Msat for the cobalticinium and tetramethylammonium intercalates (Table 1) are very close to these values. From the observation that the 001 reflections have no magnetic contributions to their intensity and also from the anisotropy of magnetization, it is tempting to conclude that the easy axis of magnetization in these intercalates remains the same as in the pristine host, and therefore that the magnetic moments should be perpendicular to the layer planes. We note however that the magnetic structure factors for compounds of symmetry C2/m (that of the pristine MnPS3 ) has the form

408

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 12. Imbalanced spin arrangement in a ferrimagnetic host layer of the cobalticenium and tetramethylammonium MnPS3 intercalates. Only the honeycomb lattice of spins is represented. The shaded squares represent the ordered manganese vacancies. The axes 3a and b of the superlattice cell are shown.

F = µ⊥ 4i sin(2π k/3) [where k is a Miller index and µ⊥ is the component of the moment perpendicular to the scattering vector] such that no magnetic contribution to 001 reflections is expected.

12.5.3.1

Consideration of other MnPS3 Intercalates

Examination of Table 1 shows that some intercalates, for example the octylammonium intercalate, do not have a value of Msat as high as 4000 cm3 mol−1 Oe. This feature can be qualitatively understood. In the frame of the above suggested model, the saturated magnetization of metal deficient MPS3 intercalates is expected to depend crucially on the extent of metal loss and above all on the perfectness of ordering of the created vacancies. Obviously there would be no net magnetization if the vacancies were located randomly. Because of a much smaller metal loss, the octylammonium intercalate Mn0.89 PS3 (n-octyl NH3 )0.22 (H2 O) y cannot give rise to the same superstructure as the cobalticenium or tetramethylammonium intercalates. Furthermore, the pyridinium intercalate shows a very small value of Msat , which suggests that the vacancies are almost totally disordered. This is in agreement with the X-ray diffraction pattern of the pyridinium intercalate which does not show any superlattice peaks. As an example of an even more subtle magnetic behavior, the temperature dependence of the magnetization of the methylviologen intercalate Mn0.82 PS3 (MV)0.18 (H2 O) y is shown in Fig. 13. This dependence is markedly different from the cases examined so far. The FCM curve shows a sharp peak at 32 K, the remnant magnetization is nearly zero and the ZFCM curve is identical to the

12.5 The Magnetic Properties of the MnPS3 Intercalates

409

Fig. 13. Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization (RE), and zero-fieldcooled magnetization (ZFCM) of the Mn0.8 PS3 (MV)0.2 intercalate in a field of 30 Oe.

FCM one. It is clear that in this case intercalation does not induce any spontaneous magnetization. Comparison with the other intercalates suggests that the increase of M on cooling below 40 K may be due to the onset of spontaneous magnetization within each slab, and that the sudden drop between 32 and 34 K is because of the antiferromagnetic coupling of the magnetic moments of adjacent layers. More structural data would be necessary to understand the reasons for this particular behavior.

12.5.3.2

The Dependence of T C on the Nature of the Intercalates

It has been shown that a perfectly planar, purely 2D Heisenberg ferromagnet cannot order at a temperature higher than 0 K [5]. However, if some anisotropy exists, ordering can occur below a temperature which must be higher than the so called Stanley-Kaplan temperature TSK = 1/5(Z − 1)(2S(S + 1) − 1) J/K, where Z is the number of nearest neighbors [49]. Magnetic ordering in pure MnPS3 occurs at 78 K. Because the creation of vacancies reduces the average value of Z and J , the ordering temperature of any intercalate derived from MnPS3 is expected to be significantly lower than 78 K. The critical temperatures TC of the various intercalates investigated actually decrease when the amount of metallic vacancies increases (Table 1), an effect evidently due to dilution of the interacting spins. To confirm this fact, a sodium intercalate Mn0.5 PS3 (Na)1.0 (H2 O)∼4 has been synthesized by treating a potassium intercalate with a concentrated solution of NaCl in the presence of EDTA. This intercalate no longer exhibits any magnetic transition (it remains paramagnetic down to 10 K at least). Therefore magnetic ordering is subject to a “percolation threshold”, probably around a metal content close to Mn0.7 PS3 . The above considerations also show that it cannot be reasonably expected to obtain a magnet having TC above liquid nitrogen temperature among the MnPS3 composites.

410

12.6 12.6.1

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

The Magnetic Properties of the FePS3 Intercalates General

When compared with MnPS3 from a magnetic point of view, FePS3 appears interesting for two reasons: (i) the magnetic interactions between FeII ions are stronger than between MnII ions, as illustrated by the much higher Neel ´ temperature of FePS3 , and therefore this host lattice should permit to obtain magnetic materials with higher ordering temperatures; and (ii) the electronic structure of the FeII is much more anisotropic and causes the magnetic behavior of FePS3 to be of the Ising type [28, 29]. Ion-exchange intercalation into FePS3 is less easy than into MnPS3 because of a weaker ability of the lattice to release FeII ions, possibly due to a larger crystal field stabilization energy. Nevertheless, several iron deficient Fe1−x PS3 G2x (solv) y intercalates have been synthesized and studied in our group (G = cobalticenium, tetraethylammonium, 4-picolinium, 3,5-lutidinium, N -methylpyridinium . . . ), but it turned out that most of them do not acquire any magnetization at low temperature, that is they order in an overall antiferromagnetic way. We will focus in this paragraph on two particular FePS3 intercalates, those with pyridinium and methylviologen guest cations. Both of them have been synthesized by treating FePS3 with a concentrated solution of the appropriate chloride in ethanol at 60◦ C overnight.

12.6.1.1

The Pyridinium Intercalate

The temperature dependence of the magnetization of the pyridinium intercalate Fe0.88 PS3 (pyH)0.24 (solv) y (powder) has been measured in low and zero field [50]. Results are shown in Fig. 14. The field-cooled magnetization (FCM) curve obtained by cooling under 10 Oe shows a rapid increase of M when T decreases below 100 K. Then the field was switched off at 40 K and the remnant magnetization measured as the sample was warmed up to 120 K. The remnant magnetization is very strong, virtually indistinguishable from the FCM curve, and it vanishes at TC (90 K). Finally we measured the zero-field-cooled magnetization (ZFCM) by cooling down from 120 K to 40 K in zero field, then warming up under 10 Oe. The ZFCM remains practically equal to zero at low temperature, then rises and exhibits a maximum just below TC . These data demonstrate the occurrence of spontaneous magnetization in this intercalate beyond liquid nitrogen temperature. A striking demonstrative experiment is the following: when the bottom of a glass tube containing the powdered intercalate is immersed into a liquid nitrogen containing beaker sitting on top of a stirring plate, the grains of the intercalate rotate along with the stirring rod; the phenomenon stops once the liquid Nitrogen has evaporated. The dependence of the magnetization of Fe0.88 PS3 (pyH)0.24 (solv) y on applied magnetic field has been studied at various temperatures between 40 K and TC (40, 60, 70, 80 and 85 K) [50]. Results are presented in Fig. 15. When the temperature is close to TC , the magnetization increases rapidly for low values of the applied field, then reaches a saturation regime with a constant slope. However as the temperature

12.6 The Magnetic Properties of the FePS3 Intercalates

411

Fig. 14. Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization (RE), and zero-fieldcooled magnetization (ZFCM) of the Fe0.88 PS3 (PyH)0.24 intercalate within a field of 10 Oe (reproduced from Ref. 50).

Fig. 15. Dependence of the magnetization of Fe0.88 PS3(PyH)0.24 on applied magnetic field at different temperatures (reproduced from Ref. 50).

is lowered, a threshold applied field is necessary before the magnetization increases. Also, for a given magnetization to be reached, it is necessary to apply increasing values of the field as the temperature is lowered. At 10 K, virtually no magnetization was detected, even in an applied field as strong as 40 kOe. From the curves at 85, 80, 70 and 60 K, we note that the slopes in the high field region are all the same, whatever the temperature, and also that the magnitude of the magnetization decreases as the temperature approaches. To appreciate the anisotropy of the magnetization, we carried out measurements on single platelets oriented parallel or perpendicular to the external magnetic field. Measurements were carried out at 80 K to limit the importance of the “threshold” effect. The field dependence of the magnetization is shown in Fig. 16 for both orientations [50]. The results show considerable anisotropy. When the platelet is set up with its plane parallel to the applied magnetic field, only a very weak magnetization appears even for high values of the field. In contrast, a rapid increase of the magnetization occurs ca 800 Oe in the perpendicular orientation. These results strongly suggest that the axis of easy magnetization is nearly perpendicular to the slabs.

412

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 16. Dependence of the magnetization at 80 K of a single platelet of Fe0.88 PS3 (PyH)0.24 on a magnetic field applied perpendicular or parallel to the plane of the platelet (reproduced from Ref. 50).

12.6.1.2

The Methylviologen Intercalate

The temperature dependence of the magnetization of the powdered methylviologen intercalate Fe0.83 P0.99 S3 (MV)0.14 has been studied as above under 10 Oe [50]. Results are shown in Fig. 17. The FCM is very similar to that obtained in the case of the pyridinium intercalate. TC is somewhat lower (77 K) but the magnetization appears stronger. Switching off the field at 40 K and warming up again reveals a strong remnant magnetization that vanishes at TC . Correlatively, the ZFCM keeps a very small value. The field-dependence of the magnetization below TC was studied at 70 K and 40 K. Results are shown in Fig. 18. With respect to the pyridinium intercalate, the methylviologen one shows two interesting features: (i) the magnetization reaches more rapidly its saturation regime (which here is a true plateau) and its magnitude is stronger. A value of ca 2000 cm3 mol−1 Oe (per mole of Fe)) is reached, that is about 10% of the expected value if all the FeII spins were aligned. (ii) There is a weaker “threshold” effect. The M(H ) curve at 70 K rises very rapidly at low fields; the increase at 40 K is more sluggish than at 70 K, but nevertheless much more rapid than in the case of the pyH intercalate.

Fig. 17. Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization (RE), and zero-fieldcooled magnetization (ZFCM) of the Fe0.83 PS3 (MV)0.14 intercalate within a field of 10 Oe (reproduced from Ref. 50).

12.6 The Magnetic Properties of the FePS3 Intercalates

413

Fig. 18. Dependence of the magnetization of the Fe0.83 PS3 (MV)0.14 intercalate on applied magnetic field at different temperatures (reproduced from Ref. 50).

Fig. 19. Hysteresis loop of the Fe0.83 PS3 (MV)0.14 intercalate at 70 K (reproduced from Ref. 50).

Hysteresis of the methylviologen intercalate was sought at 70 K by cycling the applied field between +5000 Oe and −5000 Oe This material has an hysteresis loop with a width of ca 200 Oe, which is represented in Fig. 19.

12.6.2

Spectroscopic Characterization of the FePS3 Intercalates

The powder X-ray diffraction patterns of both the pyridinium and the methylviologen FePS3 intercalates show sharp reflections indicating a highly crystalline state. In both cases, all these reflections can be indexed in a monoclinic cell similar to that of pristine FePS3 , with an expansion in the direction by about 3.3 Å because of the size of the guest species which lie with their molecular planes parallel to the layers. In particular, it must be underscored that in contrast to the MnPS3 intercalates, no superstructure involving a triple a axis was found [50]. In the search for more structural information, we have investigated the Mossbauer ¨ spectra of these intercalates. The Mossbauer ¨ spectrum of FePS3 , shown in Fig. 20a, consists of a single doublet characterized by an isomer shift (IS) of 0.868 mm s−1 and a quadrupole splitting (QS) of 1.522 mm s−1 , in agreement with previously published data [51]. These values cor-

414

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 20. (a) 57 Fe Mossbauer ¨ spectrum ¨ spectrum of FePS3 . (b) 57 Fe Mossbauer of Fe0.83 PS3 (MV)0.14 intercalate (reproduced from Ref. 50).

respond to FeII ions occupying equivalent sites. The spectrum of the methylviologen intercalate at room temperature is shown in Fig. 20b [50]. It show two doublets, labeled A and B. The pyridinium intercalate shows a similar spectrum but the relative intensity of B to A is larger in the MV intercalate than in the pyH intercalate. The IS of both A and B are very close to the IS in pure FePS3 , but their QS are very different (1.65 mm s−1 for A and 2.34 mm s−1 for B). Therefore doublet A in these intercalates is almost identical to the doublet of pure FePS3 , whereas B indicates the occurrence of another kind of FeII ions in a more distorted site. A detailed study of the Mossbauer ¨ spectra of several other FePS3 intercalates loaded with various arylammonium species has shown that the relative intensity of peaks A and B was related to the amount of metallic vacancies in the host lattice [50]. More precisely, the relative intensity of B to A in all these intercalates could be accounted for by assuming that A arises from the FeII ions surrounded by three next-neighbor FeII ions whereas B arises from the fraction of FeII surrounded by only two next-neighbor ¨ spectroscopy brings indirect evidence FeII ions and one vacant site. Thus, Mossbauer (together with analytical data) for the presence of metallic vacant sites within the intercalated layers and for the occurrence of distorted metallic sites.

12.6 The Magnetic Properties of the FePS3 Intercalates

12.6.3

415

Discussion on the Role of Intercalation into FePS3

The observation of spontaneous magnetizations in iron-containing compounds is always subject to pitfalls because of the possibility of traces of ferromagnetic impurities such as metallic iron or oxides. Joy et al.[52] have recently inserted various neutral amines into FePS3 . Their intercalates remained antiferromagnets, but they nevertheless exhibited superparamagnetic properties due to the formation of nanoparticles of Fe2 O3 as an impurity. Our critical analysis of their experiment is that insertion of neutral amines into FePS3 is a complex process which also releases some FeII ions. Because of the basic pH and the presence of air, these ions are partially hydrolyzed and oxidized to hydroxides or oxides. We have carefully checked that the pyridinium and methylviologen intercalates described above are paramagnetic above their critical temperature and therefore we claim that these intercalates are genuine magnets below TC . As recalled in Section 3, the magnetic interaction in pure FePS3 is strongly anisotropic and of the Ising type. The strong anisotropy and remnant magnetization of the intercalates shows that these characteristics are obviously retained after intercalation. A striking property of the pyH intercalate (and to a lesser extent of the MV one) is the “threshold effect” observed in the M(H ) curves: the lower the temperature, the stronger the applied field has to be to reveal the magnetization. A possible hypothesis to explain this feature would be to assume the existence of a weak interlayer antiferromagnetic interaction that would have to be overcome by the applied field. However the FCM curve shows that the magnetization appears even in a very small external field, provided this field is present when the sample crosses the TC range. It is therefore likely that the “sluggish” increase of the M(H ) curves at temperatures significantly lower than TC is due to the strong anisotropy of the FeII ions which prevent the spins from aligning up easily. The critical temperature of the MV intercalate (77 K) is significantly lower than that of the pyH intercalate (90 K). As already pointed out in the case of the MnPS3 intercalates, this diminution arises from a “spin dilution” effect due to the fact that there is a greater amount of iron vacancies in Fe0.83 P0.99 S3 (MV)0.14 than in Fe0.88 P0.99 S3 (pyH)0.24 . Furthermore, the fact that the M(H ) curve rise more rapidly with smaller “threshold effect” for the MV intercalate (as compared to the pyH intercalate) might also be because of dilution of the FeII ions, which reduces anisotropic effects. The fundamental question which then arises is: why does intercalation turn the overall antiferromagnetic behavior of pure FePS3 into different behavior which gives rise to spontaneous magnetization? The parallelism between MnPS3 and FePS3 seems at first sight striking. We have shown above that the magnetization of the MnPS3 intercalates was due to the ordering of the MnII vacancies created by intercalation, which destroys the up/down balance that prevails in the pure MnPS3 . In the case of the FePS3 intercalates, intralayer metallic vacancies also form, but X-ray diffraction has not brought any evidence for the occurrence of a superlattice, hence the vacancies that appear upon intercalation do not order. It is therefore likely intercalation plays a different role in FePS3 than it does in MnPS3 . The pristine FePS3 possesses in-plane ferromagnetic interactions beside the antiferromagnetic ones. Therefore a possible model would be that intercalation of

416

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

electron accepting species such as pyridinium and methylviologen modifies the inplane interactions and favor the ferromagnetic interactions at the expense of the AF ones. To explore further this possibility, we have synthesized a series of cadmiumsubstituted Fe1−x Cdx PS3 layered compounds, in which the diamagnetic CdII ions are expected to dilute the FeII ions and prevent from magnetic ordering, an advantage which should permit studying the influence of intercalation on the intralayer magnetic coupling.

12.6.4

Magnetic Properties of Iron-diluted Fe1−x Cdx PS3 Compounds

FePS3 and CdPS3 are known to form solid solutions Fe1−x Cdx PS3 over the whole range between x = 0 and x = 1 [53, 54]. We have recently synthesized several Fex Cd1−x PS3 compounds (x = 0.1, 0.2, 0.35, 0.5, 0.8) as well as their methylviologen intercalates, and determined their magnetic properties [55]. X-ray powder diffraction confirmed that all compounds were monophasic. A monotonous decrease of the a and b parameters of the common monoclinic unit cell is observed on going from CdPS3 to FePS3 . Insertion of methylviologen resulted in all cases in an increase of the interlayer distance by about 3.3 Å, indicating that methylviologen species lie with the plane of the aromatic cycles parallel to the layers. The temperature dependence of the magnetic susceptibility of the various unintercalated Fex Cd1−x PS3 compounds is shown in Fig. 21 in the form of a plot of χM T against T , where χM is the susceptibility (per mole of Fe). Upon cooling, χM T for the iron-rich Fe0.8 Cd0.2 PS3 steadily decreases over the range 70–100 K, indicating predominant AF interactions. In contrast, χM T for the cadmium-rich compounds (x = 0.1, 0.2, 0.35) significantly increases upon cooling from 300 K and reach a maximum close to 6 cm3 mol−1 K around 50–70 K. Fe0.5 Cd0.5 PS3 has an intermediate behavior. All χM T products eventually drop towards much smaller values at very low temperature. The temperature dependence of χM T for the various Fex Cd1−x PS3 methylviologen intercalates is shown in Fig. 22. Remarkably, χM T for the intercalates with

Fig. 21. Temperature-dependence of χM T for unintercalated Fex Cd1−x PS3 compounds: ( ) x = 0.1, () x = 0.2, () x = 0.35, (•) x = 0.5, () x = 0.8 (reproduced from Ref. 55).

12.6 The Magnetic Properties of the FePS3 Intercalates

417

Fig. 22. Temperature-dependence of χM T for Fex Cd1−x PS3 methylviologen intercalates ( ) x = 0.1, () x = 0.2, () x = 0.35, (•) x = 0.5, () x = 0.8 (reproduced from Ref. 55).

x = 0.1, 0.2, 0.35 and 0.5 increases much more rapidly upon cooling than χM T for the corresponding pristine phases. The difference is particularly pronounced for the three last compositions, where (χM T )max reaches about 10 cm3 mol−1 K (x = 0.2 and 0.35) and even 15 cm3 mol−1 K (x = 0.5). It therefore appears that intercalating methylviologen cations into any of the Fex Cd1−x PS3 phases reinforces the ferromagnetic interactions between the FeII ions. The behavior of the iron-rich Fe0.8 Cd0.2 PS3 methylviologen intercalate is somewhat different: χM T increases only very slightly on cooling (whereas it decreases for the pristine compound), then very sharply increases below 60 K in a way that suggests the onset of spontaneous magnetization. This hypothesis is confirmed by the study of this compound in a weak magnetic field (30 Oe). The temperature dependence of the FCM, RE and ZFCM is represented in Fig. 23a. The dependence of the magnetization on an applied field (measured at 30 K) shows a steep rise at low fields followed by a saturation regime (Fig. 23b). Remarkably, the saturation value ∼2500 cm3 mol−1 Oe (per mole of Fe)) is somewhat stronger than for the FePS3 methylviologen intercalate (∼2000 cm3 mol−1 Oe (per mole of Fe)) [50]. All these data show that the Fe0.8 Cd0.2 PS3 methylviologen intercalate undergoes magnetic ordering below a critical temperature TC of about 60 K, slightly lower than the critical value (77 K) of the FePS3 MV intercalate.

12.6.5

Discussion

This study of the Fe1−x Cdx PS3 layer phases emphasizes the sensitivity of the magnetic interactions in this system towards substitution and intercalation: (i) The increase of χ T on cooling the pristine Fex Cd1−x PS3 compounds (x = 0.1, 0.2, 0.35 and 0.5) implies a predominance of in-plane ferromagnetic interactions in these systems. (ii) The more pronounced increase of χ T observed on cooling the various methylviologen Fex Cd1−x PS3 intercalates (x = 0.1, 0.2, 0.35 and 0.5) reveals even stronger ferromagnetic in-plane interactions between the FeII ions. This result is of importance, because it demonstrates that intercalation is capable of perturbing the magnetic coupling between the Fe ions sufficiently to reinforce the ferromagnetic

418

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

Fig. 23. (a) Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization (RE), and zero-field-cooled magnetization (ZFCM) of the Fe0.8 Cd0.2 PS3 MV intercalate within a field of 30 Oe. (b) Dependence of the magnetization of the Fe0.8 Cd0.2 PS3 MV intercalate on applied magnetic field at 30 K temperature.

components is at the expense of the AF ones. Therefore this result gives strength to the suggestion (Section 6.2) that spontaneous magnetization in the intercalates of FePS3 (and also of Fe0.8 Cd0.2 PS3 ) arises because intercalation affects the balance between F and AF interactions. It is worth emphasizing that such a mechanism is not to be considered in the case of the MnPS3 intercalates, because there are only AF interactions in pristine MnPS3 . The question which then arises is: why do the in-plane ferromagnetic interactions become stronger when a given Fex Cd1−x PS3 phase is intercalated? A Mossbauer ¨ study of the Fex Cd1−x PS3 phases and of their MV intercalates has shown that intercalation increases the distortion of the FeII coordination sphere in the intercalate [55]. The magnetic interactions between metallic ions are known to be very sensitive to small changes of the bridging geometry, and hence it seems reasonable to suggest that the reinforcement of the F vs AF magnetic in-plane interactions which accompany methylviologen intercalation is due to the distortions created by the insertion process. An unexpected and very interesting result of the study on the Fex Cd1−x PS3 systems is that strong in-plane magnetic interactions between the FeII ions are retained even at relatively high dilutions. The maxima of the χ T products of the MV intercalates reach about 10 cm3 mol−1 K (per mole of iron atom) for x = 0.2 and x = 0.35. To account for such values, it must be assumed that the iron ions are grouped in domains comprising at least four FeII ions, the interaction between the ions in a given domain being ferromagnetic. Such high-spin domains with a total spin S = 8 would give (χ T = 9 cm3 mol−1 K (per mole of iron) in the low temperature limit (if spinonly effects are considered and g = 2). As the metal atoms within a slab are arranged at the corners of a honeycomb lattice, such clusters might comprise one central FeII atom surrounded by three others ones. Actually χ T even reaches 15 cm3 mol−1 K in the Fe0.5 Cd0.5 PS3 intercalate, implying the occurrence of even larger high-spin FeII domains. Therefore the magnetic data of the iron-diluted Fex Cd1−x PS3 intercalates

12.7 The NiPS3 -Cobaltocene Intercalation Compound

419

imply a non-random population of the metallic sites on the nanometer scale: indeed, if the iron ions in Fe0.2 Cd0.8 PS3 were randomly dispersed, most FeII ions would have three CdII ions as nearest metallic neighbors. Another interesting point is that the χ T maxima for the intercalates with x = 0.2, 0.35 and 0.5 occur roughly at the same temperature despite very different iron dilutions. This feature indicates that intralayer AF interactions play a negligible role in the χ T decrease at very low temperature. Finally, it is worth nothing that the MV intercalate of the iron-rich compound Fe0.8 Cd0.2 PS3 behaves much in the same way as the MV intercalate of FePS3 itself. Interesting enough, the saturated magnetization of the Fe0.8 Cd0.2 PS3 methylviologen intercalate is significantly larger than the magnetization of the analog FePS3 methylviologen intercalate, which means that the contribution of the F interactions compared with the AF interactions in the former is larger than in the latter.

12.7

The NiPS3 -Cobaltocene Intercalation Compound

We give here only a very brief account on the effect of intercalation on the NiPS3 layer compound. In contrast to Mn- or FePS3 , NiPS3 has never been observed so far to give intercalates by the ion-exchange mechanism, a feature probably due to a more covalent metal-sulfur bonding. However NiPS3 inserts electron donating species thanks to a redox process. Thus, a cobaltocene intercalate NiPS3 (CoCp2 )0.37 has been prepared by reaction of NiPS3 with cobaltocene in toluene at 130◦ C over 2 days [56] and its magnetic properties have been studied [57]. The temperature dependence of the magnetization of NiPS3 (CoCp2 )0.37 in low field is shown in Fig. 24a. The field-cooled magnetization rapidly increases below

Fig. 24. (a) Temperature-dependence of the field-cooled magnetization (FCM), remnant magnetization, (RE), and zero-field-cooled magnetization (ZFCM) of the NiPS3 (CoCp2 )0.37 intercalate within a field of 30 Oe. (b) Dependence of the magnetization of the NiPS3 (CoCp2 )0.37 intercalate on applied magnetic field at 20 K.

420

12 Intercalation-induced Magnetization in MPS3 Layered Compounds

70 K. The field dependence of the magnetization at 20 K is shown in Fig. 24b. These data demonstrate the occurrence of spontaneous magnetization below an ordering temperature close to 60 K. This result is very interesting in light of the above interpretation of the occurrence of spontaneous magnetization in the FePS3 intercalates: despite the fact that no metallic vacancies are created in NiPS3 upon cobaltocene intercalation, NiPS3 (CoCp2 )0.37 does exhibit spontaneous magnetization. Therefore the “ordered vacancy” mechanism involved with MnPS3 definitely cannot apply and a different mechanism must exist. As in FePS3 , the NiII ions in NiPS3 undergo both F and AF in-plane interactions, and therefore the same mechanism as the one postulated above for FePS3 might also apply.

12.8

Multi-property Materials: Associating Magnetism and Non-linear Optics

Associating different physical properties in a single material and seeking synergy between these properties is certainly one of the important goals in the field of molecular and hybrid organic-inorganic materials. The intercalation chemistry of the optically transparent MnPS3 compound brings a convenient opportunity to associate magnetization and NLO properties: the host lattice can provide a spontaneous magnetization once intercalated, specific guest organic chromophores may generate second harmonic radiation if properly oriented within the interlayer galleries. Therefore it can be expected that inserting hyperpolarizable species into MnPS3 could lead to a compound that could act as a magnet able to generate second harmonic. We have synthesized a few years ago the Mn0.86 PS3 (DAMS)0.28 intercalate, where DAMS represents the dimethylaminostilbazolium cation shown in Scheme 1 [18].

Scheme 1

The DAMS cation is one of the most efficient NLO chromophore [58]. Although the pristine MnPS3 host lattice is centrosymmetric, this intercalate possesses a high efficiency at second harmonic generation, about 300 times larger than urea when irradiated by 1.34 µm laser radiation. The magnetic properties of Mn0.86 PS3 (DAMS)0.28 have been studied. This intercalate actually becomes a magnet below about 40 K [18] and it appears to be the first material to possess both strong spontaneous NLO properties and spontaneous magnetization up to a quite elevated temperature (TC ≈ 40 K). Other stilbazolium chromophores give similar effects. Extensive work has shown that the non-centrosymmetric arrangement of the chromophores in the galleries is caused by formation of very large J -type aggregates of these chromophores [59]. Much work remains to be done on these nanocomposites, in particular to study whether the magnetic and NLO phenomena can be coupled.

12.9 Conclusion and Perspectives

12.9

421

Conclusion and Perspectives

This chapter obviously leaves a number of questions unanswered. The difficulty of obtaining refined crystal structures of intercalation compounds constitutes a serious drawback that prevents full understanding of their magnetic properties. Nevertheless we have shown that the intercalation chemistry of the MPS3 compounds not only gives rise to interesting new magnetic materials, but also brings some perspectives in the field of molecular materials, in particular to associate magnetism to other properties in the same material. Intercalation therefore provides an alternative to other methods, such as sol–gel chemistry, for the synthesis of hybrid organic–inorganic nanocomposites, bringing the advantage of a better crystallinity, a factor which can be of crucial importance when cooperative interactions are required. The ideas developed in Section 8 have already lead to other NLO active magnets in the field of layered oxalates [60]. Intercalation into the MPS3 phases might also permit the synthesis of other types of multiproperty materials utilizing energy migration between the host lattice and the guest species [61], photoinduced charge separation between co-intercalated donor and acceptor molecules [62] or spin crossover of intercalated complexes [63]. Of particular interest are the “high spin” nanometer-sized domains formed in the iron diluted Fex Cd1−x PS3 layers, especially because of the large anisotropy of these systems. If the total spin of such clusters could be increased, they could meet the criteria searched in the emerging field of high spin clusters, which is currently studied as an approach to a new type of magnetic memory.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

O. Kahn, Molecular Magnetism, VCH, New York, 1993. P. Day, Acc. Chem. Res. 1979, 12, 236. J.S. Miller, A.J. Epstein, W.M. Reiff, Acc. Chem. Res. 1988, 21, 114 J.P. Renard in Organic and Inorganic Low-Dimensional Crystalline Materials, NATO ASI Series B, Vol 168 (Eds.: P. Delhaes, M. Drillon), Plenum Press, New York, 1987. L.J. De Jongh, A.R. Miedema, Adv. Phys. 1974, 23, 1-260. R.L. Carlin, A.J. Van Duyneveldt, Magnetic Properties of Transition Metal Compounds, Springer, New York, 1977. S.G. Carling, P. Day, D.Visser, J. Sol. St Chem. 1993, 106, 111-119. a) S.G.Carling, P. Day, D. Visser, Inorg. Chem. 1995, 34, 3917. b) P. Day, Phil. Trans. Roy. Soc. 1985, A314, 145. M. Drillon, P. Panissod, J. Magn. Magn. Mater. 1998, 188, 93. V. Laget, C. Hornick, P. Rabu, M. Drillon, P. Turek, R.. Ziessel, Adv.Mater. 1998, 10, 1024. R. Brec, Solid State Ionics 1986, 22, 3. D. O’Hare in Inorganic Materials; (Eds.: D.W. Bruce, D. O’Hare), John Wiley & Sons, Chichester, 1993; Chapter 4. R. Clement, ´ O. Garnier and J. Jegoudez, Inorg. Chem. 1986, 25, 1404. E. Manova, C. Severac, ´ A. Andreev, R. Clement, ´ J. Catalysis 1997, 169, 503.

422 [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42]

[43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

12 Intercalation-induced Magnetization in MPS3 Layered Compounds A. Simon, J. Ravez, V. Maisonneuve, C. Payen, V.B. Cajipe, Chem. Mater. 1994, 6, 1575. T. Coradin, R. Clement, ´ P. G. Lacroix, K. Nakatani, Chem. Mater. 1996, 8, 2153. R. Clement, ´ L. Lomas, J.P.Audiere, ` Chem. Mater. 1990,2, 641. P.G. Lacroix, R. Clement, ´ K. Nakatani, J. Zyss, I. Ledoux, Science 1994, 263, 658. a) W. Klingen, R. Ott, H. Hahn, Z. Anorg. Allg. Chem. 1973, 396, 271. b) W. Klingen, G. Eulenberger, H. Hahn, Z. Anorg. Allg. Chem. 1973, 401,97. R. Brec, G. Ouvrard, A. Louisy, J. Rouxel, Ann. Chem. Fr. 1980, 5, 499–512. G. Ouvrard, R. Brec, J. Rouxel, Mat. Res. Bull. 1985, 20, 1181. F.S. Khumalo, H.P. Hugues, Phys. Rev.B 1981, 23, 5375. M. Piacentini, F.S. Khumalo, C.G. Olson, J.W. Anderegg, D.W. Lynch, Chem. Physics 1982, 65, 289. R. Brec, D. Schleich, G. Ouvrard, A. Louisy, J. Rouxel, Inorg. Chem. 1979, 18, 1814. M.H. Whangbo, R. Brec, G. Ouvrard, J. Rouxel, Inorg. Chem.. 1985, 24, 2459. B.I. Taylor, J. Steger, A. Wold, J. Solid State Chem. 1973, 7, 461. J. Berthier, Y. Chabre, M. Minier, Solid State Commun. 1978, 28, 327. G. Le Flem, R. Brec, G. Ouvrard, A. Louisy, P. Segransan, J. Phys. Chem. Solids, 1982, 43, 455. P.A. Joy, S. Vasudevan, Phys. Rev. B, 1992, 46, 5425. P.A. Joy, S. Vasudevan, J. Chem. Phys. 1993, 99, 4411. K. Kurowasa, S. Saito Y. Yamaguchi, J. Phys. Soc. Japan 1983, 52, 3919. P. Jernberg, S. Bjarman, R. Wappling, J. Magn. & Magn. Mater. 1984, 46, 178. A. Wiedenmann, J. Rossat-Mignod, A. Louisy, R.Brec, J.Rouxel, J. Solid State Commun. 1981, 40, 1067. R. Clement, ´ J.J. Girerd, I. Morgenstern-Badarau, Inorg. Chem. 1980, 19, 2852. G.G. Rushbrooke, P.J. Wood, Mol. Phys. 1958,1, 257. G. Ouvrard, in Chemical Physics of Intercalation II, NATO ASI Series, (Eds P. Bernier et al.) Vol. 305, Plenum Press, New York, 1993, p. 315. R. Clement, ´ Chem. Commun. 1980, 647 R. Clement, ´ R. J. Amer. Chem. Soc. 1981,103, 6998. R. Clement, ´ O. Garnier, J. Jegoudez, Inorg. Chem. 1986, 25, 1404. R. Clement, ´ M. Doeuff, C. Gledel, J. Chim. Phys. 1988, 85, 1053. R. Clement, ´ J. P. Audiere, J. P. Renard, Rev. Chim. Min.,1982, 19, 560. R. Clement, ´ I. Lagadic, A. Leaustic, ´ J.P. Audiere, ` L. Lomas, in Chemical Physics of Intercalation II, NATO ASI, Vol.305, (Eds P. Bernier et al. ), Plenum Press, New York, 1993, p. 315. John Evans, PhD Thesis, University. of Oxford (UK), 1994. J.S.O. Evans, D. O’Hare, R. Clement, ´ J. Am. Chem. Soc. 1995, 117, 4595. J.S.O. Evans, D. O’Hare, R. Clement, ´ A. Leaustic, ´ P. Thuery, Adv. Mater. 1995, 7, 735. B.E. Waren, Phys. Rev. 1941, 59, 693–698. S. Soled, A. Wold, Mater. Res. Bull., 1976, 11, 657. R. Diehl, C.D. Carpentier, Acta Cryst. 1978, B34, 1097. H. Stanley, T.A. Kaplan, Phys. Rev. Lett. 1966, 17, 913. A. Leaustic, ´ J.P. Audiere, ` D. Cointereau, R. Clement, ´ L. Lomas, F. Varret, H. ConstantMachado, Chem. Mater., 1996, 8, 1954. G.A. Fatseas, M. Evain, G. Ouvrard, R. Brec, M.H. Whangbo, Phys. Rev. B, 1987, 35, 3082. P.A. Joy, S. Vasudevan, Chem. Mater. 1993, 5, 1182. S. Lee, P. Colombet, G. Ouvrard, R. Brec, Inorg. Chem. 1988, 27, 1291. A. Bhowmick, B. Bal, S. Ganguly, M. Bhattacharya, M.L. Kundu, J. Phys. Chem. Solids 1992, 53, 1279.

References

423

[55] A. Leaustic, ´ E. Riviere, ` R. Clement, ´ E. Manova, I. Mitov, J. Phys. Chem. B, 1999, 103, 4833. [56] R. Clement, ´ O. Garnier, Y. Mathey, Nouv. J. Chim. 1982, 6, 13. [57] E. Manova, A. Leaustic, ´ I. Mitov, D. Gonbeau, R. Clement, ´ Mol. Cryst. Liq. Cryst. 1998, 311, 155. [58] S.R. Marder, J.W. Perry, W.P. Schaefer, Science 1989, 245, 626. [59] T. Coradin, R. Clement, ´ P. G. Lacroix and K. Nakatani, Chem. Mater. 1996, 8, 2153. [60] S. Benard, ´ P. Yu, T. Coradin, E. Riviere, ` K. Nakatani, R. Clement ´ Adv. Mater. 1997, 9, 981 [61] E Lifshitz, R. Clement, L.C. Yu-Hallada, A H Francis, J. Phys. Chem. Solids 1991, 52, 1081–1086. [62] R. Jakubiak, A.H. Francis, J. Phys. Chem. 1996, 100, 362–367. [63] C. N. Field, M.-L. Boillot, R. Clement, ´ J. Mater. Chem., 1998, 8, 283. [64] D. Gatteschi, A. Caneschi, L. Pardi, R. Sessoli, Science 1994, 265, 1054.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

13

Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets Carlo Bellitto

13.1

Introduction

Hybrid organic-inorganic layered solids, consisting of alternating inorganic and organic layers are interesting materials for several reasons. They can, in fact, be used as ion-exchangers, catalysts, hosts in intercalation compounds, etc. [1–3], and they have also provided several examples of low-dimensional magnets [4]. Typical layered materials are represented by the “intercalation” [5] and the “molecular composite” [6] compounds. In contrast with intercalation compounds, which can exist both with and without organic molecules between layers of the lattice, in molecular composite solids, the organic groups are ionically or covalently bound to the inorganic layers. Examples of the latter are the layered perovskite halides of formula (Cn H2n+1 NH3 )2 [MX4 ] [1, 7], (M = divalent metal ion, X = halogen) and the metal(II) phosphonates [2, 3], i. e. M[Cn H2n+1 PO3 ].H2 O and M2 [O3 P(CH2 )n PO3 ].2H2 O, (M = Mg, Cd, Zn, Cu, Mn, n = 1, 2, . . . ). In the latter, the ligand is covalently bonded to the metal ion through the phosphonate group, and the resulting compounds are thermally stable in a wide range of temperatures [3]. In layered metal phosphonates, the metal ions are bridged by the oxygens of the phosphonate ligands, and they form sheets that are separated one from the other by the organic substituents of the ligands [8, 9]. These materials merit attention for separate points of view. They provide an opportunity for tailoring new and functional materials for basic science and technological applications. The inorganic part, in fact, typically characterized by covalent interactions, offers the possibility of combining physical properties, like ferromagnetism, luminescence, semiconducting behavior, with the properties typical of the organic groups, like mesomorphism, non-linear optics, polymerization, plastic mechanical properties, etc., Phosphonate ligands are also interesting, because they can be functionalized by active groups, such as: –NH2 , –CO2 H, –SH, –PO3 H2 . Since the compounds can be prepared with different 3d-block elements (V, Cr, Mn, Fe, Co, Ni, Cu), and the layers can be separated in a controllable way, by varying the length of the alkyl side chains of the phosphonic acid, they can provide interesting examples of low-dimensional magnetic materials. The two-dimensional nature of the lattice favors the next-neighbors magnetic exchange interactions in metal phosphonates containing paramagnetic ions, and, at low temperatures a long-range magnetic ordering has been observed [10]. Further, by choosing a suitable long-chain phosphonic acid, and using Langmuir-Blodgett technique, it is possible to obtain thin-film magnets [11]. Depending on the metal ion and on the nature of the R group of the

426

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

ligand, metal phosphonates could also crystallize in different structures from that described above, and they will presented and commented here. It is also important to point out that excellent reviews describing the state of art in the field have been recently reported [2, 3, 12], but no one focuses the attention on the magnetic properties of the metal phosphonates containing paramagnetic ions, and especially, on those which show cooperative magnetic behavior at low temperatures. This review will attempt to give an overview on the magnetic properties of the known paramagnetic transition metal-ion phosphonates and bis(phosphonates), and, where possible, the correlation between the observed magnetic phenomena and the crystal structures will be also shown.

13.2

Synthesis of the Ligands

Simple phosphonic acids are commercially available, but in many cases the ligand with a special required R group must be prepared in the laboratory, and, the most versatile method to prepare it, is the so-called Michaelis–Arbuzov reaction [13]. It involves the reaction of an ester of trivalent phosphorus with an aryl or alkyl halide, according to Eq. (1): (RO)3 P + R X → (RO)2 P(O)R + RX(1) (1) R = CH3 , C2 H5 ; and R = CH3 (CH2 )n –; C6 H5 –; (n = 0, 1, 2 . . .). The ester is then hydrolyzed in concentrated HCl. The phosphonic acid, obtained as a white powder, is soluble in water. Alkylene-bis(phosphonic) acids are synthesized using the same method and by starting with an alkylene dihalide.

13.3 13.3.1

Vanadium Phosphonates Preparation

Three recent reviews describing a large body of vanadium phosphonates have been recently reported [2, 3, 12]. In this section we will only briefly describe the synthesis of oxovanadium phosphonates, the only magnetically well-characterized derivatives of this metal ion. They were prepared by a direct reaction in solution of the reagents at room pressure and by hydrothermal method. Vanadyl phosphonates are prepared mainly by addition of V2 O5 to an hot alcoholic solution of the phosphonic acid. The alcohol acts as a solvent and as reducing agent of V5+ to V4+ ions. Recently, a detailed description of the preparation of vanadyl phosphonates were reported by Brody and Johnson [14]. The solvent is incorporated in the lattice, but it can be removed by

13.3 Vanadium Phosphonates

427

washing the compound with water. (VO)[C6 H5 PO3 ].2H2 O is a light blue–green polycrystalline solid. The hydrothermal synthesis provided single-crystals of the monohydrated (VO)[C6 H5 PO3 ].H2 O. Light green crystals were isolated by mixing V2 O3 , the phenyl phosphonic acid and water in a Teflon-lined autoclave and maintained at 200◦ C for a few days [15]. (VO)[Cn H2n+1 PO3 ].yH2 O (y = 1.5, 1 ≤ n ≤ 3; y = 1.0, 4 ≤ n ≤ 8) were also prepared by hydrothermal reaction of V2 O3 and the corresponding alkylphosphonic acid in water at 200◦ C [16]. This method has also been used by Nocera et al. [17] in preparing a series of layered vanadyl phosphonates, of formula (VO)[X-C6 H4 PO3 ].nH2 O, n = 1, 1.5, where the phenylphosphonate ligand was substituted in para and meta positions of the phenyl ring by various pendants, X, i. e. X = p-NO2 , m-F, p-Cl, p-F, H, CH3 .

13.3.2

Crystal Structures

Two compounds of formulas (VO)[C6 H5 PO3 ].nH2 O (n = 1, 2) have been isolated and they show different crystal structures [15]. The monohydrate, (VO)[C6 H5 PO3 ].H2 O, consists of layers of corner-sharing [VO6 ] octahedra and O3 P–C tetrahedra. Phenyl groups extend out of both side of the inorganic layers, resulting in a repeat distance of 14.14(2) Å, and they are anchored to the layer by a P–C covalent bond. Within the inorganic layer the [VO6 ] octahedra share an axial oxygen with the neighboring vanadium to form chains of · · · –V=O–V=O– · · · along the b-axis. These chains are not uniform, and they have alternating short 1.610(9) Å and long 2.14(1) Å V–O bonds. The ligand bridges across two vanadium atoms in the same chain and one single metal atom in the neighboring chain as shown in Fig. 1. The coordination around the vanadium is completed by the oxygen atom of the water. The phenyl groups form bilayers perpendicular to the (bc)-plane [15]. The bis-hydrate, (VO)[C6 H5 PO3 ].2H2 O, has also been isolated and its structure is closely related to that of the mineral newberyite Mg[HPO4 ].3H2 O [18] (See Fig. 2). In the structure V atom replaces Mg, a phenyl group replaces the OH group on P, and the stippled water molecule is replaced by the vanadyl oxygen atom. [VO6 ] octahedra are made by two water molecules, coordinated in cis and trans positions to the vanadyl oxygen atoms, and by one oxygen of the [PO3 ] groups in a such a way that one oxygen of the ligand connects one V atom; the phenyl groups from adjacent layers interpenetrate. The adjacent vanadium atoms do not show and short contact, but they are linked by O–P–O bridges.

13.3.3

Magnetic Properties

The magnetic properties of the layered vanadyl phosphonates (LVP, hereafter) have been investigated in relation to their structures. The compounds are all characterized by containing the paramagnetic d 1 vanadyl centers, but they crystallize in different types of chain-like structures [15, 16]. The magnetic susceptibility, χM , of (VO)[C6 H5 PO3 ].2H2 O was found to be linear over the whole measured temperature range. The fit of the magnetic susceptibility data at high temperatures confirmed

428

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 1. The crystal structure of VO[C6 H5 PO3 ].H2 O: view of the (bc)-plane illustrating the –V–O–Vchains of the corner-sharing [VO6 ] octahedra. (Reproduced from Ref. 15 by permission of the American Chemical Society.)

Fig. 2. The crystal structure of MgHPO4 .3H2 O (Reproduced from Ref. 18 by permission of Munksgaard Int. Pu. Copenhagen).

that the vanadium is present in the compound as a V4+ ion, i. e. S = 1/2, g = 2.0. The magnetic behavior of the mono-hydrate (VO)[C6 H5 PO3 ].H2 O is different at low temperatures. In fact, an increase of the susceptibility is observed in the plot of χM against T , and a maximum is reached at Tmax ≈ 15 K. This behavior can be explained by assuming antiferromagnetic coupling along the · · · –V=O–V– · · · chain. Magnetic studies were performed on the series (VO)[Cn H2n+1 PO3 ].yH2 O (y = 1.5, 1 ≤ n ≤ 3; y = 1.0, 4 ≤ n ≤ 8) [16]. The magnetic susceptibility data show distinct differences between C1 –C3 and C4 –C8 compounds. The maximum observed in the

13.3 Vanadium Phosphonates

429

Fig. 3. Plots of d. c. molar susceptibility against temperature between 5 and 20 K for VO[X-C6 H4 PO3 ].H2 O, [X = p-NO2 (square), m-F (triangle), p-F(diamond) and H (circle)] showing the broad maxima. (Reproduced from Ref. 17 by permission of the American Chemical Society.)

plots of χM against T for (VO)[C2 H5 PO3 ].yH2 O and (VO)[C3 H7 PO3 ].yH2 O is a typical signature of antiferromagnetically exchange-coupled dimers. In the case of n = 5, 7, the plot is typical of a simple paramagnet. A detailed magnetic study on the series VO[X-C6 H4 PO3 ].nH2 O, n = 1, 1.5 has been done by Nocera et al. [17]. The objective of that work was the study of the correlation of the magnetic properties with the Hammett σ parameter. The temperature dependent magnetic susceptibility plots of the compounds are reported in Fig. 3, and two different type of magnetic behavior are observed. When the substituent X is p-NO2 , the compound is a simple paramagnet, with a negligible Weiss constant, i. e. 0.1 K. Conversely, if X = H, m-F and p-F, the behavior is a typical of systems having intra-layer antiferromagnetic couplings between V4+ centers. In fact, a broad maximum in the plot of χ against T is observed between T = 4 K and T = 7 K. The values of the Curie constants, as obtained from the high-temperature data fit, are all close to 0.37 cm3 K mol−1 , thus confirming that the compounds all contain V4+ (S = 1/2) ions. An attempt to explain the lowtemperature magnetic behavior was done by applying the Bleaney–Bowers model [19] for dimers, containing two S = 1/2 cations, with isotropic g-tensor, according to Eq. (2): χ = (1 − f )[(N g 2 µ2B /kT ) × (1/(3 + exp(−2J/kT ))] + f × (C/T )

(2)

where f is the amount of paramagnetic impurities taken as isolated V4+ ions, g = 1.98 and J/k is the nearest-neighbor exchange interaction. The J/k values, obtained from the fit together with the structural parameters, are reported in Table 1.

430

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Table 1. Selected structural, vibrational, and magnetic data for VO[X-C6 H4 PO3 ].nH2 O, n = 1, 1.5 and Hammett values for C6 H4 -X.a

a (Å) b (Å) c (Å) β (◦ ) d-spacing (Å) C (emu K mol−1 )b θ (K) b Tχ max (K) µeff (B.M.) b J/k (K) c σd

C6 H5

C6 H4 - p-F

C6 H4 -m-F

C6 H4 - p-NO2

28.50 7.18 9.42 97.1 14.14 0.360 −2.3 7 1.7 −5.5 0

28.59 7.16 9.45 97.6 14.17 0.359 −1.1 5.5 1.69 −4.5 0.15

28.60 7.15 9.44 97.5 14.18 0.368 −3.5 4.0 1.72 −3.3 0.34

30.5 6.98 9.55 97.1 15.13 0.370 −0.1 Not found 1.73 0 0.81

a

From Ref. 17. Fit over the range 50 < T < 100 K. c From Bleaney-Bowers equation to data over the range 2 < T < 100 K. d Substituent constants from J. March, Advanced Organic Chemistry, J. Wiley and Sons, NY, 1992. b

The trend suggests that the coupling becomes weaker as electron-withdrawing ability rises, due to the cation Lewis acidity or aryl substituent σ value. LVPs with the substituent X = p-Cl, and p-CH3 are not included in Table 1, because they fall outside the above structural groups. Temperature dependence magnetic susceptibility plots of these compounds show antiferromagnetic downturns at temperatures higher than those observed in Fig. 3, and the values of T (χmax ) were found at T = 54 K (X = p-Cl) and at T = 58 K (X = p-CH3 ). The corresponding J/k coupling constants are one order of magnitude higher, thus indicating a stronger antiferromagnetic coupling. These compounds, appear to have the same framework of the layered (VO)[HOPO3 ].0.5H2 O [20], where the vanadyl centers interact directly through µ2 bridging oxygens of a V(µ2 -O)2 V dimers. It is believed that this can be ascribed to the steric bulk of the substituents: H, F and NO2 groups are small, while CH3 and Cl are both thicker than an aryl ring.

13.4 13.4.1

Divalent Metal Phosphonates Synthesis

Metal(II) phosphonates were prepared for the first time in 1979 by Cunningham et al. [21], and then synthesized and characterized independently by the groups of Mallouck [8a] and Clearfield [8b]. The compounds are obtained by mixing stoichiometric amounts of the divalent metal halide (or sulfate) and the corresponding phosphonic

13.4 Divalent Metal Phosphonates

431

Table 2. Coordination of the metal ion in M(II)[RPO3 ]H2 O. R group

Cr

Mn

Fe

Co

Ni

Cu

Zn

Mg

Cd

CH3 C2 H5 C3 H7 C4 H9 C6 H5 CH3 (CH2 )17

Oh – – – – –

Oh Oh Oh – Oh Oh

Oh Oh – – Oh –

Oh – – – – –

Oh – – – Oh –

C4V C4V – – – –

Oh – – – Oh –

Oh – – – Oh –

Oh – – – Oh –

acid in water. The solution with the adjusted pH, (which ranges between 3 to 7, depending on the metal ion) is allowed to react for several days under reflux, according to Scheme 1: MX2 + RPO3 H2 → M[O3 PR].H2 O + 2HX R = CH3 , C2 H5 , . . . , Cn H2n+1 , C6 H5 . . . ; M = Cr, Mn, Fe, Co, Ni, Cu, Zn, Cd, Mg Scheme 1

With Cr(II) [22] and Fe(II) [23, 24], the reaction is carried out under inert atmosphere in presence of urea, and by refluxing the solution over 80◦ C for several days. The compounds were obtained mainly by precipitation or, occasionally, by evaporation of a solution containing the acid, metal salt and urea at 60◦ C, as it was in the case of Cd derivatives [25]. Nickel(II) alkylphosphonates were prepared by a direct reaction between the Ni(OH)2 and the ligand: the stoichiometric mixture placed in an ampoule was heated at the melting point temperature of the ligand for three days [26]. The metal(II) phosphonates which have been isolated and characterized are listed in Table 2. Paramagnetic metal bis(phosphonates) were also prepared, (see below), but they are less investigated than the monoalkylphosphonates. Both families of compounds offer the possibility of lattice engineering, by varying both the metal ion and the length of the chain, by changing the number of carbon atoms in the organic R group. This gives the possibility of studying the influence of these factors on the physical properties and, in particular, on the magnetic ones. As indicated in Table 2, not all the compounds corresponding to the general formula M[RPO3 ].H2 O have the divalent metal ion, M, in a distorted octahedral coordination. There are examples, in fact, where the metal ion is five or even four coordinated, resulting in different crystal structures. The color of the compounds depends on the electronic configuration and on the type of coordination of the metal ion. For example, Zn(II), Cd(II) and Mn(II) phosphonates are white, Fe(II) light yellow, Cr(II) light blue, the Co(II) are blueviolet and Ni(II) light green. In a few cases single crystals were obtained, but, in general, they are isolated as polycrystalline powders. Many of the reported crystal structures have been therefore determined from X-ray powder diffraction studies, i. e. ab-initio and Rietveld method refinements (see below).

432

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

13.4.2

Crystal Structures

The crystal structures of M[RPO3 ].H2 O (M = Mn(II), Mg(II), Zn(II), Cd(II); R = Cn H2n+1 and C6 H5 ) were reported in 1988, by Mallouck et al. [8]. and the most studied were the Mn(II) derivatives. The series show layer structures, quite similar to that found in ternary metal phosphates of formula M I M I I PO4 H2 O, (M I = K+ , II NH+ 4 ; M = Mn, Fe, Co, Ni) [27]. They crystallize mainly in the orthorhombic space group Pmn21 (see Fig. 4) [8, 10, 25]. The b-axis of the unit cell results to be related to interlayer separation, and thus it increases with increasing number of the carbon atoms in the ligand (Table 3). In metal(II) phosphonates the metal ion is six-coordinated. Because the ligand to metal ion ratio is 1:1, to obtain coordination six, two of the phosphonate oxygens chelate the metal atom and, at the same time, bridge across adjacent atoms in the same row. The oxygens are then three-coordinate. The third phosphonate oxygen bridges to an adjacent row, thus creating the layer arrangement shown in Fig. 5. The sixth position is occupied by the oxygen of the water molecule. The symmetry of the six-coordination is highly distorted, and the resulting layer is kinked or crenellated. The organic layer is made by the organic R groups, which extend through the C-P bond perpendicular to the inorganic planes. In Mn[C6 H5 PO3 ].H2 O [8], it has been found that the phenyl groups are disordered between two orientations. Van der Waals contacts are present between these groups. It should also be added that axial photographs showed additional reflections indicating the doubling of the cell along the a- and c-axes, but the refinements were not satisfying. A similar layered structure has been observed in two structurally characterized Fe(II) phosphonates, i. e. Fe[C2 H5 PO3 ].H2 O [24a] and Fe[C6 H5 PO3 ].H2 O [28]. The crystal structure of the former has been determined on a single crystal, while that of the latter from X-ray diffraction pattern by a Rietveld refinement. Fe[C2 H5 PO3 ].H2 O crystallizes in the monoclinic space group P1n1, with the unit-cell parameters a = 4.856(8) Å, b = 10.33(1) Å, c = 5.774(3) Å, and β = 91.0(1)◦ . It has a lamellar structure, made of alternating organic and inorganic layers. The latter consist of the iron atoms coordinated octahedrally by five phosphonate oxygen atoms and

Table 3. Unit-cell dimensions of Mn[RPO3 ].H2 O salts indexed in the space group Pmn21 .

a b c

Compound

a (Å)

b (Å)

c (Å)

Mn[CH3 PO3 ].H2 Oa Mn[C2 H5 PO3 ].H2 O Mn[C3 H7 PO3 ].H2 O Mn[C4 H9 PO3 ].H2 O Mn[C6 H5 PO3 ].H2 Ob Mn[C10 H21 PO3 ].H2 O Mn[C17 H38 PO3 ].H2 Oc

5.819(2) 5.83(6) 5.84(4) 5.84(4) 5.734(2) 5.86 5.7

8.818(2) 10.24(6) 11.71(2) 14.72(2) 14.33(3) 30.6 48.5

4.897(2) 4.87(8) 4.91(2) 4.91(8) 4.945(2) 4.91 4.9

From Ref. 10. From Ref. 8a. From Ref. 11e.

13.4 Divalent Metal Phosphonates

433

Fig. 4. The crystal structure of M[C6 H5 PO3 ].H2 O, M = Mg, Mn, viewed down the a-axis, showing the arrangement of the organic group. (Reprinted from Ref. 8 by permission of the American Chemical Society.)

one from the water molecule. The site symmetry around the metal ion is distorted octahedral. The P-C bonds are nearly perpendicular to these planes and the ethyl groups make the van der Waals contacts between the inorganic layers. Fe[C6 H5 PO3 ].H2 O crystallizes in the orthorhombic space group Pmn21 , and it is isomorphous and isostructural to the Mn(II) and Cd(II) analogs [8, 25]. Few Co(II) phosphonates have been synthesized and structurally characterized. The methylphosphonate has been found to be isomorphous and isostructural with the Zn derivatives [29]. In the case of Co[(CH3 )3 CPO3 ].H2 O, however, the cobalt ions are present both in tetrahedral and octahedral coordination. Co[(CH3 )3 CPO3 ].H2 O crystallizes in the monoclinic space group P21 /c, with the unit-cell parameters a = 12.256(1) Å, b = 17.939(1) Å, c = 10.769(1) Å, and β = 93.57(1)◦ [30]. The unit cell contains three sites of Co(II) ions: two in tetrahedral and one in octahedral coordination. The structure is still layered and made of corrugated sheets in the (bc)-plane. Co(1) has the octahedral symmetry, made from three phosphonate oxygens and three water molecules. These [Co(1)O6 ] octahedra form pairs with a common edge, resulting in Co(1)–O(2)– Co(1)–O(2) parallelograms. Tetrahedral coordination is observed for the other two cobalt positions, Co(2) and Co(3) bonded only to the oxygens of the phosphonate ligand. The [Co(3)O4 ] tetrahedra are arranged in edge-sharing pairs. Each oxygen of

434

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 5. The crystal structure of M[C6 H5 PO3 ].H2 O, M = Mg, Mn, viewed along the b-axis, showing the arrangement of the inorganic layer. (Reproduced from Solid State Ionics 1988, 26, 63 by permission of Elsevier).

the three types of [PO3 ] groups is bonded to the metal atoms, ensuring the connection of the [Co(1)O6 ], [Co(2)O4 ] and [Co(3)O4 ] polyhedra within the layer, according to the sequence of [Co(2)O4 ] chains (parallel to c-axis) intercalated by chains based on alternating [Co(1)O6 ] pairs and [Co(3)O4 ] pairs (parallel to c-axis) see Fig. 6. The (CH3 )3 C– groups extend into lamellar space, roughly perpendicular to the corrugated layers. The arrangement is different from its methyl-phosphonate analog [29], and this is ascribed to the bulkiness of the of tert-butyl group. An interesting functionalized Co(II) phosphonate of formula Co3 [(O2 C(CH2 )2 PO3 ]2 has been recently synthesized and studied by Bujoli et al. [31]. The crystal structure is similar to that of Zn3 [(O2 CC2 H4 PO3 ]2 , which has been found to be layered and pillared [32]. In the latter, the Zn atoms were found to have two different coordinations: the first one is tetrahedral and the metal is coordinated by three oxygens from the phosphonate groups and one oxygen from the carboxyl group. These [ZnO4 ] tetrahedra are arranged in edge-sharing pairs and forming Zn–O–Zn–O parallelograms. The other type of Zn center is six-coordinated, by two oxygens from two carboxylic groups and four phosphonate oxygen atoms. Each oxygen of the [PO3 ] groups is bonded to a metal atom, ensuring the connection of the [ZnO4 ] and [ZnO6 ] polyhedra within the layers of (bc)-plane see Fig. 7. The organic groups extend perpendicular to the inorganic layers and linked together via the carboxylic moieties, thus leading to a 3D network. An interesting feature is the identification of chains parallel to c-axis, formed by pairs of [ZnO4 ] tetrahedra interconnected by [ZnO6 ] octahedra via corner sharing. As mentioned above, copper(II) phosphonates have a different coordination around the metal ion. In Cu[RPO3 ].H2 O the structure is still layered, but the copper

13.4 Divalent Metal Phosphonates

435

Fig. 6. The crystal structure of Co[(CH3 )3 CPO3 ].H2 O viewed down the a-axis, showing the arrangement of the inorganic groups (reproduced from Ref. 30 by permission of the Royal Chemical Society).

Fig. 7. The inorganic layer arrangement of M3 [O2 CC2 H4 PO3 ]2 , M = Zn, as viewed perpendicular to the a-axis. Carbon atoms not bound to phosphorus have been omitted for clarity. (Reproduced from Ref. 32 by permission of the American Chemical Society.)

ions are five coordinated and in distorted tetragonal pyramidal symmetry [33–35]. Cu[CH3 PO3 ].H2 O crystallizes in the monoclinic space group P21 /c, with the unitcell parameters a = 8.495(4) Å, b = 7.580(4) Å, c = 7.289(4) Å, and β = 90.08(4)◦ . One oxygen of each phosphonate bonds to two copper atoms forming chains, while the other two phosphonate oxygens bond to two copper atoms in an adjacent chain. The base of the pyramid is made by three oxygens of the ligands and one from the water molecule. The fifth oxygen atom, is nearly perpendicular to the plane and has

436

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

a bond longer the average of the four equatorial bond lengths. The three oxygens of the phosphonate group are all bonded to Cu atoms. One of them bridges two Cu atoms which are 3.13 Å apart, thus identifying copper dimers forming four-membered parallelogram-shaped rings Cu–O–Cu–O in the plane. Two next-nearest neighboring copper atoms are linked by O–P–O bridges. Cu[C6 H5 PO3 ].H2 O [33] crystallizes in the orthorhombic space group Pbca, with the unit-cell parameters a = 7.5547(4) Å, b = 7.4478(6) Å, and c = 27.982(1) Å. The coordination around the copper atoms and the layer structure is similar to that of the methyl-phosphonate. The difference arises from the orientation of adjacent phenyl groups, the main plane of which are approximately perpendicular to each other (see Fig. 8). The dehydration of the above mentioned materials gives Cu[RPO3 ] salts, which show different structures. For example α-Cu[C2 H5 PO3 ] could be prepared hydrothermally or, by heating at 180◦ C, Cu(C2 H5 PO3 ).H2 O [34]. It crystallizes in the monoclinic space group P21 /c with

Fig. 8. The layer arrangement in Cu[C6 H5 PO3 ].H2 O, as viewed down the a-axis. (Reproduced from Ref. 33 by permission of the American Chemical Society.)

13.4 Divalent Metal Phosphonates

437

the unit-cell parameters a = 10.7773(3) Å, b = 5.6960(2) Å, c = 7.6273(2) Å, and β = 94.47(1)◦ , and the copper ion is still pentacoordinated, but in a trigonal bipyramidal symmetry. The three oxygens of the phosphonate group are all bonded to the Cu atoms, one of them bridges two copper atoms. One of them, O(1) bridges two copper atoms, which are 3.115(3) Å apart, with a short and long bond. The same copper atoms are bridged to a second O(1) atom from another phosphonate ligand, thus forming a four membered parallelogram-shaped ring. On the other hand, the O(3) oxygens also bridge two Cu atoms (at the Cu–Cu distance of 3.203(3) Å, providing a zigzag Cu–O–Cu linking along the b-axis, with again a short and long bond to the adjacent copper atoms. In conclusion, the inorganic layer can be described as infinite rows consisting of four rectangular rings connected together by two Cu–O(2)–P– O(3)–Cu bridges, running parallel to c-axis. These chains are linked together in the b-axis direction by alternating CuO(3) and P–O(1) bonds. The anhydrous product obtained by thermal treatment of Cu[C2 H5 PO3 ].H2 O at 200◦ C in air for three hours can be also prepared hydrothermally. If α-Cu[CH3 PO3 ] is placed in a Teflon cell of an autoclave and sealed, placed in an oven at 200◦ C for 15 days, two different compounds of formulas β-Cu[CH3 PO3 ] and Cu3 O[CH3 PO3 ].2H2 O are obtained. β-Cu[CH3 PO3 ] is found to be rhombohedral (R-3, no. 148) and it shows an original channel-type arrangement. Each Cu is still fivecoordinated in a distorted tetragonal pyramidal coordination. All the three oxygens of the phosphonate groups are all bonded to Cu atoms. Two of them bridge copper atoms which are 3.0102(5) Å apart. The structure consists of infinite zigzag chains of copper running parallel to the c-axis (see Fig. 9). Each copper is connected to two adjacent copper atoms by forming Cu-O(2)-Cu-O(3) rings developing along the c-axis.

13.4.3

Magnetic Properties

On the basis of their magnetic properties, metal(II)phosphonates can be broadly classified as dimeric systems, one-dimensional magnetic systems, or two-dimensional magnetic systems. The first two classes are paramagnetic at high temperatures, and they do not show long-range magnetic ordering, at least at the lowest measured temperatures. The third class, on the other hand, contains interesting examples of magnetic materials. They are paramagnetic at room temperature, and, on cooling, show long-range magnetic ordering. Two classes of compounds have been identified, i. e. “weak-ferromagnetic” and “antiferromagnetic” metal(II) phosphonates. The different magnetic behavior reflects the crystal structures of the compounds.

13.4.3.1

Dimeric Systems

The magnetic properties of various copper(II) phosphonates have been reported by Bujoli et al. [35]. They observed for Cu[RPO3 ].H2 O (R = CH3 ,C2 H5 ,C6 H5 ) that plots of the product χM T against T are very similar, see Fig. 10. The room temperature magnetic moments range from 2 to 2.2 µB /Cu. Below 50 K the χM T decreases

438

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 9. Perspective representation of β-Cu[CH3 PO3 ], viewed down the c-axis, showing the channel type arrangement. The carbon atoms are depicted in black. Visualization of the copper chain along the c-axis on the bottom. (Reproduced from Ref. 35 by permission of the American Chemical Society.)

Fig. 10. Plots of χT against T for Cu[RPO3 ].H2 O, R = CH3 , C2 H5 , C6 H5 , in the temperature range 5–300 K. (Reproduced from Ref. 35 by permission of the American Chemical Society.)

13.4 Divalent Metal Phosphonates

439

drastically, revealing antiferromagnetic interactions between the near-neighbor copper ions. According to the crystal structure described above, two adjacent copper(II) ions are bridged by oxygen atoms so to form a four-membered parallelogram-shaped Cu–O–Cu–O rings in the layer. This suggests that the presence of magnetically coupled Cu–Cu dimers could be responsible of the magnetic behavior. The susceptibility data could be fitted, by using the Bleany-Bowers dinuclear model [19] for a Heisenberg Hamiltonian, H = −2J S1 S2 , and the magnetic susceptibility varies with the temperature according to Eq. (3): χ = (N g 2 µ2B /kT )[1/(3 + exp(−2J/kT )]

(3)

where |2J |/k represents the singlet–triplet energy gap and N , g, µB , k, and T have the usual meaning. The results of the fit are reported in the Table 4; the J/k are small in value and negative, and this explains why no maximum is detected in plots of χM against T . Table 4. Best-fit values of J/k and g for Cu[RPO3 ].H2 O derivatives.a

a

Compound

g

J/k

Cu[CH3 PO3 ].H2 O Cu[C2 H5 PO3 ].H2 O Cu[C6 H5 PO3 ].H2 O

2.19(1) 2.23(1) 2.15(1)

−3.45(5) −3.78(4) −4.69(5)

From Ref. 35.

13.4.3.2

One-dimensional Magnetic Systems

Copper(II) phosphonates exist also in an anhydrous state, and the crystal structure is profoundly affected when the water molecule is removed (see above). The plots of magnetic susceptibility against temperature are reported in Fig. 11, and what is immediately apparent is the presence of a broad maximum at Tmax ∼ 35 K for R = CH3 and for R = C2 H5 , and at Tmax ∼ 50 K for R = C6 H5 [35]. Because in these systems there are more than one possible superexchange pathway when compared with the hydrates, and, these paths form no simple 2D network, it was difficult for the authors to perform high-temperature series expansion. The broad maxima are the signature of the presence of low-dimensional antiferromagnetic interactions inside the inorganic layers. The magnetic behavior of β-Cu[CH3 PO3 ] [34] has also been studied and the temperature dependence of the susceptibility is reported in Fig. 12. This has interpreted on the basis of S = 1/2 antiferromagnetic chain model, by using the equation, proposed by Hatfield [36], for an alternating magnetic chain. The susceptibility is expressed as follows: χ = (N g 2 µ2B /kT )[(A + Bx + C x 2 )/(1 + Dx + E x 2 + F x 3 )]

(4)

where x = |J |/kT , J/k is the exchange constant, and A to F are functions of α, the alternation parameter [37]. The best fit, nevertheless not entirely satisfying, was

440

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 11. Plots of χ against T for α-Cu[RPO3 ], R = CH3 , C2 H5 , C6 H5 , in the temperature range 5–300 K. (Reproduced from Ref. 35 by permission of the American Chemical Society.)

Fig. 12. Plots of χ against T for β-Cu[RPO3 ] in the temperature range 5–300 K. The dashed line is a fit. (Reproduced from Ref. 35 by permission of the American Chemical Society.)

13.4 Divalent Metal Phosphonates

441

obtained for the parameters: J/k = −10(2) K, α = 0.20, and g = 2.4. This is probably because of the distorted chain structure of the compound.

13.4.3.3

Two-dimensional Magnetic Systems

The Mn[RPO3 ].H2 O series has been the object of extensive magnetic studies by the research groups of Day and of Talham [10, 11]. Mn(II) phosphonates were found to order antiferromagnetically at TN ≈ 15 K [10]. At temperatures above the critical temperature, TN , all the compounds show a negative deviation from the Curie law, thus indicating antiferromagnetic nearest-neighbor exchange interactions within the inorganic layers, see Fig. 13. The nearest-neighbor exchange constant, J/k, for a two-dimensional antiferromagnetic system, could be estimated by measuring the static magnetic susceptibility in the paramagnetic temperature range, and by applying a high-temperature series expansion appropriate to a quadratic layer Heisenberg magnet [38]. This method utilizes the relationship between the inverse of the magnetic susceptibility and the temperature for different spin quantum numbers, and provides an accurate method for determining the exchange parameter J/k. In Mn(II) phosphonates, however, the layer structure is crenellated, and therefore, the 2D quadratic Heisenberg model represents an approximation. The values of the nearest-neighbor exchange constants, J/k, for Mn(II) compounds were then determined from the value of the temperature of the maximum in the susceptibility, T (χmax ), according to the relation: kT (χmax )/J = 2.06S(S + 1) [4] and these values are reported in Table 5. The super-exchange pathway responsible for magnetic exchange is assumed the same for each member of the Mn series [11].

Fig. 13. Plots of χ against T for Mn[RPO3 ].H2 O, R = CH3 , C2 H5 , C4 H9 , in the temperature range 5–300 K. (Reproduced from Ref. 10 with permission from Academic Press, NY).

442

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Table 5. Magnetic parameters for Mn[RPO3 ].H2 O derivatives. Compound

C (cgs)

θ (K)

Tχ max (K)

|J |/k (K)

TN (K)

TN /θ

Mn[CH3 PO3 ].H2 O Mn[C2 H5 PO3 ].H2 O Mn[C3 H7 PO3 ].H2 O Mn[C4 H9 PO3 ].H2 O Mn[C6 H5 PO3 ].H2 O Mn[C5 H11 PO3 ].H2 O Mn[C17 H38 PO3 ].H2 O

3.56(1) 3.84(2) 3.68(1) 3.68(1) 4.56(1) 3.83(3) −

−48.9(7) −59.7(7) −51.7(7) −54.4(9) −46.0(1) −48.9(1) −34

24.3(2) 25.0(2) 22.4(4) 22.4(4) 19.(1) 25.5(3) 25

2.70(2) 2.78(2) 2.48(4) 2.48(4) 2.2(2) 2.85 2.80

14.9 15.1 14.9 15.0 12.1 14.6 13.5

0.305 0.253 0.288 0.276 − − 0.397

Scheme 2. The potential exchange pathways (1–3) in Mn[RPO3 ].H2 O (Reprinted from Ref. 11a, by permission of the American Chemical Society.)

Scheme 2 shows three possible pathways connecting each of the manganese ions to each of its four nearest neighbors. The first involves the interaction through the Mn–O–Mn bridge, and it is the most direct one and thought to be dominant, compared to the other two. Below the critical temperature, the magnetic behavior is more complicated. For example, on cooling Mn[C4 H9 PO3 ].H2 O in presence and absence of an applied magnetic field and, then taking the plots of d. c. M against T in both modes, (see Fig. 14) what it is observed is that the plots do not overlap. The irreversible magnetization, obtained as a difference of field-cooled (FC) and zero-field cooled (ZFC) plots, collapses to zero near the critical temperature, TN . This is an indication of a non-zero magnetization at zero field, below the critical temperature, TN , similarly to that observed in a ferromagnet. But, there is a difference between this behavior and that of the normal ferromagnet. The measured saturation moment is only a small percentage of that expected for that ion in a completely aligned ferromagnetic material [39]. The phenomenon is known as a “canted antiferromagnetism” or “weak ferromagnetism”, and it has been observed for the first time in α-Fe2 O3 [39]. Two mechanisms have been suggested to operate. The first requires the presence in the crystals of two equivalent sites for the magnetic ions, but the direction of their anisotropy axes should be different, and this is the case of NiF2 and related crystals [40]. The second mechanism is known as a antisymmetric exchange, and it is a combined effect of the spin–orbit coupling and of the isotropic superexchange interactions. When a magnetic ion is placed in a low-symmetry ligand field, spin-orbit coupling mixes a small fraction of excited states into the ground state, so that a zero-field splitting of the lowest electronic levels results. This mimics

13.4 Divalent Metal Phosphonates

443

Fig. 14. ZFC (♦) and FC (+) plots of χ against T in 100 G for the Mn[C4 H9 PO3 ].H2 O in the temperature range 5–25 K. (Reproduced from Ref. 10 by permission of the Academic Press, NY.)

the effect of a large static magnetic field along the principal axis of the local ligand field. The moments are therefore constrained to that direction, and this is the singleion anisotropy. In the presence of the near-neighbor antiferromagnetic isotropic exchange, the moments will not be aligned in an exactly antiparallel fashion, but they will make a small angle to each other! This implies that in the antiferromagnetically ordered state there is a small net moment! The symmetry requirements for a spin-canting phenomenon exclude the presence of an inversion center between the interacting magnetic ions. The space group in Mn[RPO3 ].H2 O is Pmn21 , (see above) and then it is sufficiently low that the antisymmetric exchange mechanism for weak ferromagnetism, proposed by Dzyaloshinsky–Moriya [41], can occur. The exchange, which operates in addition to the isotropic Heisenberg exchange, is described on a microscopic level as: Haniso = d · [s1 × s2 ]

(5)

where d is a constant vector that depends on the symmetry of the crystal. The antisymmetric exchange derives from the anisotropic spin interaction, which can be described as a sum of symmetric and antisymmetric components: V1,2 = s1 · KS · s2 + s1 · KA · s2

(6)

The two tensors KS and KA are approximated as: K s ∼ (λ/)2 J ∼ (g/g)2 J and K A ∼ (λ/)J ∼ (g/g)J

(7)

444

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

and λ is the spin–orbit coupling and is the ligand field splitting, g = g − 2. Now only antisymmetric component can cause spin canting. In Mn2+ ion the ground state is 6 S5/2 , and the ion should be isotropic. In a crystal field, however second order spin-orbit coupling mixes the 4 T1g excited state into 6 A1g ground state, resulting in anisotropy in the g-tensor. The above discussion suggests that the low-crystalline symmetry, present in these molecule-based substances, is responsible for the weak ferromagnetism [10, 27]. The Mn(II) phosphonates series has also been studied by Talham et al. [11a], by using the antiferromagnetic resonance technique. The latter study confirmed the observed canted antiferromagnetism [10] and they were able to determine the canting angle from the analysis of the frequency and the field dependence of the AFMR signals. They have also prepared and studied Langmuir–Blodgett films of Mn(II) octadecylphosphonate thin films, which undergo to a magnetic ordering transition at T = 13.5 K, resulting in a weak ferromagnet below that temperature [11b]. Quite recently, the same authors prepared and studied Langmuir-Blodgett films of Mn(II)4-(4 -tetradecyloxyphenyl)butylphosphonate. The compound has a similar structure to the Mn alkyl phosphonates, and thin films undergo to a long-range magnetic transition at TN = 14.8 K, resulting again in a “canted antiferromagnet” [11c]. The other transition metal ions, which provide examples of “weak ferromagnetic” layered compounds, are the Cr(II), Fe(II) and Co(II) ions. The most recent studied metal(II) phosphonate series is that of Fe(II) ion [22–24, 28]. Unfortunately, only a few crystal structures have been reported in literature up to now, and therefore a detailed correlation between the structure and the magnetic properties is not yet possible. Bujoli et al. reported on the crystal structure and studied the magnetic properties of Fe[C2 H5 PO3 ].H2 O [24]. This compound is a “canted antiferromagnet” below TN = 24.5 K, (see Fig. 15), and the compounds reported in Table 6. The “memory” effect observed in Fe[C6 H5 PO3 ].H2 O [28] below the critical temperature, TN , can be nicely seen in the hysteresis loop reported in Fig. 16, measured at T = 10 K. The value of the coercive field, Hc , is 6400 G, unusually large for this kind of molecule-based magnets, while the remnant magnetization, Mr , is 740 emu G mol−1 . The magnetic long-range ordering along the series appears at TN = 25.0 K, a temperature which is higher than the ones observed in the corresponding Mn(II) phosphonates. No estimation of the values of J/k has been made for the Fe(II) phosphonates, by using the high temperature series expansion to a quadratic layer Heisenberg magnet. In Fe[C2 H5 PO3 ].H2 O it has been given a value of J/k = −5 K, as estimated from Table 6. Structural and magnetic properties of iron(II) phosphonates. Compound

TN (K)

θ (K)

TN /θ

Space group

Interlayer distance (Å)

(NH4 )FePO4. H2 O Fe[CH3 PO3 ].H2 O Fe[C2 H5 PO3 ].H2 O Fe[C4 H9 PO3 ].H2 O Fe[C6 H5 PO3 ].H2 O Fe2 [O3 P(CH2 )2 PO3 ].2H2 O

26.0 25.0 24.5 25 21.5 25.0

−65 −59 −43 −45 −56 −52

0.40 0.42 0.57 0.56 0.38 0.48

Pmn21 Pna21 P1n1 − Pmn21 −

8.8213(2) 8.77(1) 10.33(1) − 14.453(2) 15.23

13.4 Divalent Metal Phosphonates

445

Fig. 15. Temperature-dependence of the ZFC () and FC (•) magnetization of Fe[C6 H5 PO3 ]H2 O below 25 K. The full line is the irreversible magnetization as obtained from the difference of FCM and ZFCM plots. (Reprinted from Ref. 28 by permission of the American Chemical Society.)

Fig. 16. Hysteresis loop for Fe[C6 H5 PO3 ].H2 O measured at T = 10 K. (Reproduced from Ref. 28 by permission of the American Chemical Society.)

the relation kT (χmax )/J = 2.06S(S + 1) [4], taking from the plot T (χmax ) = 35 K. But, this result may be altered by the fact that the system a 2D square Heisenberg model is an approximation and that the susceptibility increase down to the critical temperature, TN . There are now two observations to be made:

446

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 17. Temperature-dependence of the reciprocal of the magnetic susceptibility, 1/(, of polycrystalline Cr[CH3 PO3 ].H2 O in the range 5–300 K. (Reproduced from Ref. 22b with permission from the American Chemical Society.)

assuming the same crystal structure in both series, why is the critical temperature higher in the ferrous salts than in manganese ones? and the critical temperature in both the Mn(II) and Fe(II) series is independent on the interlayer distance. The answer to the first question relies on the fact that the critical temperature is proportional to the near-nearest neighbor exchange constant, J/k, according, as a first approximation, to the molecular field approximation [38, 42]. The J/k of Fe(II) ethylphosphonate is higher in absolute value, i. e. 1.8 times than the one observed in the Mn(II) ones, just as it is the ratio between the two critical temperatures. Concerning the second question, there are now several examples of layered ferro- and ferrimagnetic materials reported in literature, where the critical temperature is found to be independent from the interlayer spacing [7, 10, 43, 44]. The magnetic properties of several Cr(II) phosphonate have been recently reported [22], but the lack of the knowledge of the crystal structures has prevented a detailed magnetostructural study of the series. The first member of the series, i. e. Cr[CH3 PO3 ]H2 O has been isolated as a polycrystalline light blue powder. The microcrystalline sample was zero-field cooled to T = 5 K, and the magnetization measured on heating the sample to room temperature. The temperature dependence of the reciprocal of the molar magnetic susceptibility is linear above 100 K, (see Fig. 17) and it follows the Curie–Weiss law. The Curie constant, C, is 2.92(5) emu K mol−1 , as fitted to the high temperature susceptibility data, using the equation χ = N µ2eff /3kB (T − θ), and this corresponds to an effective magnetic moment of 4.87 µB , consistent with the presence of Cr(II) in a high-spin d 4 (S = 2) electronic configuration. The large negative value of Weiss

13.4 Divalent Metal Phosphonates

447

Fig. 18. Hysteresis loop for Cr[CH3 PO3 ].H2 O measured at T = 5 K. (Reproduced from Ref. 22b with permission from the American Chemical Society.)

constant θ = −230(6) K, indicates strong antiferromagnetic near-neighbor exchange between the adjacent chromium(II) ions. Deviation from Curie–Weiss occurs below 100 K, where the magnetic susceptibility increases until a peak at T ≈ 34 K is observed. At lower applied fields the sharpness of the peak increases as well as the susceptibility values (χmax = 2.96 emu mol−1 ). The value of TN was found to be 35.8(3) K. The magnetic hysteresis loop measured at T = 5 K is reported in Fig. 18. The value of the remnant magnetization, Mr , and coercive field, Hr , are 865 emu G mol−1 and 2380 G, respectively. The isothermal magnetization increases slowly up to a threshold field, HT , value of 50 mT, where it rises and it reaches a maximum value of 1400 emu G mol−1 at 4 Tesla. The values corresponds to the 6% of the saturation magnetization value, expected for a S = 2 system, as calculated from the relation: Ms = N gµB S

(8)

where N is Avogadro’s number. This corresponds then to the saturation of weak ferromagnetic moments. Hysteresis phenomena disappear at temperatures above TN . Unexpected, the Cr(II) amino(hydrogen)ethylphosphonate chloride is paramagnetic down to 5 K, but below this temperature it shows a canted antiferromagnetism [45]. An interesting functionalized cobalt(II)phosphonate of formula Co3 [O3 P (CH2 )2 CO2 ]2 , has been recently synthesized by Bujoli et al. [31]. and the magnetic properties studied. As reported above the compound is isostructural with the pillared layered metal phosphonates M3 [O3 P(CH2 )2 CO2 ]2 , M = Zn(II) and Mn(II)

448

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 19. ZFC and FC plots of χ against T for Co3 [O2 C(CH2 )2 PO3 ]2 in the temperature range 5–150 K. (Reproduced from Ref. 31 by permission of the Royal Society of Chemistry.)

Fig. 20. Hysteresis loop for Co3 [O2 C(CH2 )2 PO3 ]2 measured at T = 5 K. (Reproduced from Ref. 31 with permission from the Royal Society of Chemistry.)

[32]. The ac and dc magnetic susceptibilities and magnetization measurements as well the specific heat measurements show that Co3 [O3 P(CH2 )2 CO2 ]2 , is again a “canted antiferromagnet”, with the antiferromagnetic long-range order at TN = 15.5 K (see Fig. 19). In the high temperature range (T > 200 K) the Curie–Weiss law is obeyed and C and θ were found to be 7.65 emu K mol−1 and −13.5 K, respectively. The deduced C constant is in agreement with the presence of one octahedral and two tetrahedral Co(II) ions in the unit cell. The negative value of the Weiss constant, θ , is related to the antiferromagnetic interactions between the metal ions. The magnetization curve below TN , i. e. at T = 2 K, exhibits an hysteresis effect with a coercive field, Hc , and a small remnant magnetization, Mr , of 5000 Oe and 0.43 µB , respectively (see Fig. 20). In this compound the magnetic coupling is rather complex if compared with the previous reported examples, due to the presence of two types of Co(II) ions: i. e. tetrahedral and octahedral. In fact the two neighboring Co(1) tetrahedral sites within the pairs (see Fig. 7) are coupled through two oxygen bridges [O(1)] and are also coupled with the Co(2) octahedral sites via an additional bridge O(2). The canting effect may be caused, in this case, from the single-ion anisotropy [31]. The absence of neutron diffraction studies prevents a definitive conclusion. Only a few examples of metal(III) phosphonates have been synthesized. One example is represented by HFe[CH3 PO3 ]2 [46]. The compound contains Fe(III) ions,

13.5 Metal(II) Diphosphonates

449

probably in octahedral symmetry, but the crystal and molecular structure is unknown. Magnetic and Mossbauer ¨ studies show that the compound is an antiferromagnet with a critical temperature TN = 18 K.

13.5 13.5.1

Metal(II) Diphosphonates Synthesis

Layered metal bis(phosphonates) were prepared by mixing stoichiometric amounts of the divalent metal halide (or sulfate) and the corresponding bis-phosphonic acid in water. As in the case of the metal(II) phosphonates, the pH of the solution was adjusted and the solution heated under reflux for several days. 2MX2 + H2 O3 PRPO3 H2 → M2 [O3 PRPO3 ].2H2 O + 4HX R = CH2 , C2 H4 , . . . ; M = Cr, Mn, Fe, Co, Ni, Cu, Zn (9)

Scheme 3.

Several metal diphosphonates were also prepared the hydrothermal method. The metal (II) diphosphonates, which have been isolated and characterized are listed in Table 7. It is important to point out that often, depending on the reaction conditions, mainly at pH < 2, the protonated bis-phosphonic acid could coordinate the metal ions, thus giving compounds of formulas M(II)[HO3 PRPO3 H] or M3 (II)[(HO3 PRPO3 )2 ] [47]. The main difficulty with describing the nature of these solids is the lack of definitive crystal structure data, owing the very low solubility of the compounds, and the possibility of having a mixture of metal diphosphonates. Up to now, only in a few cases the crystal structure has been solved, and from X-ray powder diffraction, as it was in the case of Zn and Cu alkylene bis(phosphonates) [48]. In the case of Cr(II) [22] and Fe(II) bis(phosphonates) [23], the preparation of the compounds is carried out under inert atmosphere in presence the urea, and by refluxing the solution over 80◦ C for several days. Co2 [O3 PCH2 PO3 ].H2 O has been synthesized by hydrothermal method, by starting from the complex Co(NH3 )6 Cl3 , ethylenediamine, Table 7. Coordination of the metal ion in M(II)2 [O3 PRPO3 ].2H2 O.

a b

R group

Cr

Mn

Fe

Co

Nib

Cu

Zn

CH2 C2 H4 C3 H6 C6 H4

– Oh – –

– – – –

– Oh – –

Td /Oh – – –

– – – –

C4V C4V – C4v

Oh Oh Td a Td a

Without coordinated water molecules. See text.

450

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

methylenediphosphonic acid and boron trifluoride-ethylamine complex at 190◦ C, for a few days [49]. Quite recently, Ferey et al. reported on an interesting series of Ni4 [O3 PCH2 PO3 ]2 (H2 O)n , (n = 3, 2, 0) [50]. The Ni4 [O3 PCH2 PO3 ]2 (H2 O)3 has been prepared in a pure form by hydrothermal conditions. It dehydrates quasitopotactically, giving successively the bis-hydrate at 275◦ C and the anhydrous nickel methylene-bis(phosphonates) at 350◦ C.

13.5.2

Crystal Structures

The crystal structures of divalent metal diphosphonates, M2 [O3 PRPO3 ].2H2 O (M = Zn, Ni, Cu; R = CH2 , C2 H4 , . . . ), have been solved from X-ray powder data, using ab-initio method and Rietveld refinement. Zn(II) and Cu(II)alkylene bis(phosphonates), (R = C2 H4 ,C3 H6 ) show pillared structures. Zn2 [O3 PC2 H4 PO3 (H2 O)2 ] [48] is a layered compound, and the layer arrangement is similar to the Zn[C6 H5 PO3 ]H2 O (See Fig. 21). The zinc atom is octahedrally coordinated. Two of the oxygens chelate and bridge the metal ions and the third one binds to only one metal. The sixth position is occupied by the oxygen of the water molecule. The adjacent inorganic layers are bridged by the ethylene group of the bis(phosphonate) anion.

Fig. 21. The pillared structure of Zn2 [O3 P(CH2 )2 PO3 ].2H2 O. Note the chelation and bridging type of interactions of two of the phosphonate oxygens O1 and O3. (Reproduced from Ref. 48 by permission of the American Chemical Society.)

13.5 Metal(II) Diphosphonates

451

Fig. 22. Microporous structure of Co2 [O3 PCH2 PO3 ].H2 O, showing the channels along the c-axis (After Ref. 49 by permission of VCH, Weinheim).

In both Cu2 [O3 PC2 H4 PO3 (H2 O)2 ] and Cu2 [O3 PC3 H6 PO3 (H2 O)2 ].H2 O the copper (II) ion is five-coordinated with four of the binding sites are from the phosphonate oxygens and one from the water molecule in a square pyramidal geometry [48]. The structure of Co2 [O3 PCH2 PO3 ].H2 O was determined by single-crystal X-ray diffraction [49]. It crystallizes in the monoclinic space group C2/c with the unit-cell parameters a = 18.820(4) Å, b = 8.246(2) Å, c = 8.916(2) Å, and β = 106.68(3)◦ . It contains one cobalt atom, Co(1), tetrahedrally coordinated and two, i. e. Co(2) and Co(3) octahedrally coordinated (Fig. 22). What it is different from the other metal diphosphonates, and interesting, is the presence of a microporous structure, with channels running along the c-axis. Co(2) and Co(3) octahedra share edges and form zigzag chains along the b-axis and Co(1) share a corner with Co(1) tetrahedron. The sixth position of Co(2) is the water molecule. The crystal structures of a series Ni4 [O3 PCH2 PO3 ]2 (H2 O)n , (n = 3, 2, 0) has been solved ab-initio from X-ray powder data [50]. Ni4 [O3 PCH2 PO3 ]2 (H2 O)3 crystallizes in the monoclinic Cc space group with the unit-cell parameters a = 19.177(3) Å, b = 8.0930(9) Å, c = 9.18248(8) Å, and β = 102.387(9)◦ . The compound is layered and it consists of sheets of trimeric edge-sharing units of Ni(1) and N(2) polyhedra in which the central polyhedron Ni(2) is always an octahedron. These trimers are linked together by one edge to form corrugated chains along the b-axis. The bis-hydrated Ni4 [O3 PCH2 PO3 ]2 (H2 O)2 is still monoclinic, but the space group is C2/c as well the anhydrous nickel methylene-bis(phosphonates). The a and c unit-cell parameters shorten with dehydration process. During the first loss of the water molecule, the migration of the nickel cation toward the tetrahedral site leads to the connection of the layers and renders the bis-hydrate three-dimensional. In the anhydrous compound, the trimers consist of a central octahedron with two edge-sharing square pyramids. In conclusion, the Ni(II) coordination is present as four octahedra in the three-hydrate, as three octahedra and one tetrahedra in the bis-hydrate and, finally, in the anhydrous as one octahedron, two square pyramids

452

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

and one tetrahedron. The latter case represents an unique example, with a presence of six, penta and four coordination in the same unit cell.

13.5.3

Magnetic Properties

Few paramagnetic metal(II) bis(phosphonates) derivatives have been studied from the magnetic point of view. Recent findings on the compounds Cr2 [O3 P(CH2 )2 PO3 ].2H2 O [22], Fe2 [O3 P(CH2 )2 PO3 ].2H2 O [23], the series Ni4 [O3 PCH2 PO3 ]2 (H2 O)n , n = 3, 2, 0 [50], and Co2 [O3 PCH2 PO3 ].H2 O [51] will reported and discussed. The Cr(II) derivative is paramagnetic at high temperatures; below 150 K, antiferromagnetic interactions become dominant, as the large negative value of the Weiss constant, θ, (i. e. −73 K) suggests. The Neel ´ temperature has been located at T = 15 K. The Fe(II) analog is a weak ferromagnet and the magnetic properties are similar to those observed in the Fe(II) mono-alkylphosphonates. The transition to the weak ferromagnetic state is observed below TN = 25 K. The magnetic hysteresis loop taken at T = 20 K is reported in Fig. 23. The magnetization increases rapidly at low applied magnetic fields and then with a linear dependence at larger fields, according to the relation: M = Mnc + χAFM H , where Mnc is the non-compensated magnetic moment and χAFM is the antiferromagnetic susceptibility at that temperature. The fit gave the values of Mnc and χAFM of 1125 cm3 Oe mol−1 and 0.0496 cm3 K mol−1 , respectively. But the most interesting one seems to be nickel(II) methylene-bis(phosphonates), the bis-hydrate form, which shows a ferromagnetic transition at TC = 3.8 K [50]. The compound is therefore the first phosphonate to be ferromagnetic, though at a very low temperature.

Fig. 23. Hysteresis loop for Fe2 [O3 P(CH2 )2 PO3 ].2H2 O, measured at T = 20 K.

13.5 Metal(II) Diphosphonates

453

Fig. 24. ZFC () and FC () plots of M against T under 100 G and, in insert, 1/χ of Ni4 [O3 PCH2 PO3 ]2 (H2 O)3 (a), Ni4 [O3 PCH2 PO3 ]2 (H2 O)2 (b), and Ni4 [O3 PCH2 PO3 ]2 (c). (Reproduced from Ref. 50 by permission of the American Chemical Society).

At high temperatures all the Ni(II) compounds are paramagnetic and the Weiss constant, θ , positive (or weakly negative for the anhydrous one, i. e. −1.7 K) (Figs. 24 and 25). The Weiss constant, θ, indicates predominant ferromagnetic neighbor interactions. The magnetic moment is in agreement with the presence of Ni2+ ions. The

454

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

Fig. 25. Plots of M against H plots for Ni4 [O3 PCH2 PO3 ]2 (H2 O)3 (), Ni4 [O3 PCH2 PO3 ]2 (H2 O)2 () and Ni4 [O3 PCH2 PO3 ]2 (). (Reproduced from Ref. 50 by permission of the American Chemical Society).

three-hydrate and the anhydrous compounds are probably “canted ferromagnets [50]. The magnetic properties of Co2 [O3 PCH2 PO3 ].H2 O have been also studied and it has been found to be paramagnetic until 5 K, the lowest measured temperature [51].

Acknowledgments The author acknowledges the collaboration of Dr. G. Righini in the preparation of the manuscript.

References [1] D.B. Mitzi, Progr. Inorg. Chem. 1999, 48, 1 [2] G. Alberti in Comprehensive Supramolecular Chemistry, Vol. 7, Ed. J.M. Lehn, Pergamon Press, 1996, p. 151. [3] (a) A. Clearfield, Progr. Inorg. Chem. 1998, 47, 371; (b) G. Cao, H. Hong, T. Mallouck, Acc. Chem. Res. 1992, 25, 420. [4] R. Navarro in Magnetic Properties of Layered Transition Metal Compounds, Ed. L.J. De Jongh, Kluwer Academic Publishers, Dordrecht, Holland, 1990. [5] D. O’Hare in Inorganic Materials, Eds D.W. Bruce and D. O’Hare, John Wiley and Sons, New York, 1992, pp.165–228. [6] P. Day, Philos. Trans. Royal Soc. London 1985, A314, 145. [7] C. Bellitto, P. Day in Comprehensive Supramolecular Chemistry, Vol. 7, Ed. J.M. Lehn, Pergamon Press, 1996, p. 293. [8] (a) G. Cao, H. Lee, V.M. Lynch, T.E. Mallouck, Inorg. Chem. 1988, 27, 2781; (b) K. Martin, P.J. Squattrito, A. Clearfield, Inorg. Chim. Acta 1989, 155, 7. [9] D.M. Poojary, B. Zhang, P. Bellinghausen, A. Clearfield,Inorg. Chem. 1996, 35, 5254. [10] S.G. Carling, P. Day, D. Visser, R.K. Kremer, J. Solid State Chem. 1993, 106, 111. [11] (a) G.E. Fanucci, J. Krzystek, M.W. Meisel, L.C. Brunel, D.L. Talham, J. Am. Chem. Soc. 1998, 120, 5469; (b) C.T. Seip, G.E. Granroth, M.W. Meisel, D.L. Talham, J. Am. Chem. Soc. 1997, 119, 7084; (c) G.E. Fanucci, M.A. Petruska, M.W. Meisel, D.L. Talham, J. Solid

References

[12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]

[24]

[25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36]

[37] [38] [39] [40] [41] [42] [43] [44] [45]

455

State Chem. 1999, 145, 443; (d) H. Bird, J.K. Pike, D.R. Talham, Chem. Mater. 1993, 5, 709; (e) H. Bird, J.K. Pike, D.R. Talham, J. Am. Chem. Soc. 1994, 116, 7903. M. Ishaque Khan, J. Zubieta,Progr. Inorg. Chem. 1995, 43, 1. (a) A. Michaelis, R. Kaehne, Chem. Ber. 1898, 31, 1048; (b) A.E. Arbuzov, J. Russ. Phys. Chem. Soc. 1906, 38, 687. J.F. Brody, J.W. Johnson, Inorg. Synth. 1995, 30, 241. G. Huan, A.J. Jacobson, J.W. Johnson, E.W. Corcoran, Chem. Mater. 1990, 2, 91. G. Huan, J.W. Johnson, J.F. Brody, D.P. Goshorn, Mater. Chem. Phys. 1993, 35, 199. J. LeBideau, D. Papoutsakis, J.M. Jackson, D.G. Nocera, J. Am. Chem. Chem. Soc. 1997, 119, 1313. D.J. Sutor, Acta Crystallogr. 1967, 23, 418. B. Bleaney, K.D. Bowers, Proc. Roy. Soc. (London) Ser. A 1952, 214, 451. M.E. Leonowicz, J.W. Johnson, J.F. Brody, H.F. Shannon, J.M. Newsam, J. Solid State Chem. 1985, 56, 370. D. Cunningham, P.J.D. Hennelly, T. Deeney, Inorg. Chim. Acta 1979, 37, 95. (a) C. Bellitto, F. Federici, S.H. Ibrahim, J. Chem. Soc. Chem. Commun. 1996, 759; (b) C. Bellitto, F. Federici, S.A. Ibrahim, Chem. Mater. 1998, 10, 1076. (a) C. Bellitto, F. Federici, S.A. Ibrahim, R.F. Mahmoud, 1998 Fall MRS Meetings Symp. Proceedings 1999, 547, 287; (b) A. Altomare, C. Bellitto, S.A. Ibrahim, R.F. Mahmoud, R. Rizzo, J. Chem. Soc. Dalton Trans. 2000, 3213. (a) B. Bujoli, O. Pena, P. Palvadeau, J. LeBideau, C. Payen, J. Rouxel, Chem. Mater. 1993, 5, 583. (b) J. LeBideau, C. Payen, B. Bujoli, P. Palvadeau, J. Rouxel, J. Magn. Magn. Mater. 1995, 140, 1719. G. Cao, V.M. Lynch, L.N. Yacullo, Chem. Mater. 1993, 5, 1000. G.B. Hix, K.D.M. Harris, J. Mater. Chem. 1998, 8, 579. (a) S.G. Carling, P. Day, D. Visser, Inorg. Chem. 1995, 34, 3917; (b) S.G. Carling, P. Day, D. Visser, J. Deportes, J. Appl. Phys. 1991, 69, 6016. A. Altomare, C. Bellitto, F. Federici, S.A. Ibrahim, R. Rizzi, Inorg. Chem. 2000, 39, 1803. G. Cao, T.E. Mallouck, Inorg. Chem. 1991, 30, 1434. J. LeBideau, A. Jouanneaux, C. Payen, B. Bujoli, J. Mater. Chem. 1994, 4, 1319. P. Rabu, P. Janvier, B. Bujoli, J. Mater. Chem. 1999, 9, 1323. S. Drumel, P. Janvier, P. Barboux, M. Bujoli-Doeuff, B. Bujoli, Inorg. Chem. 1995, 34, 148. Y. Zhang, A. Clearfield, Inorg. Chem. 1992, 31, 2821. J. LeBideau, B. Bujoli, A. Jouanneaux, C. Payen, P. Palvadeau, J. Rouxel, Inorg. Chem. 1993, 32, 4617. J. LeBideau, C. Payen, P. Palvadeau, B. Bujoli,Inorg. Chem. 1994, 33, 4885. W.E. Hatfield, W.E. Estes, W.E. Marsh, M.W. Pickens, L.W. Haar, R.R. Weller in Extended Linear Chain Compounds, Vol. 3, Ed. J. S. Miller, Plenum Press, NY, 1983, p. 43. W.E. Hatfield, J. Appl. Phys. 1981, 52, 1985. M.E. Lines, J. Phys. Chem. Solids 1970, 31, 101. See, for example, R.L. Carlin, Magnetochemistry, Springer, Berlin, 1986. T. Moriya, Phys. Rev. 1960, 117, 635. (a) T. Moriya, Phys. Rev. 1960, 120, 91; (b) I. Dzyaloshinsky, J. Phys. Chem. Solids, 1958, 4, 241. See, for example, O. Kahn, Molecular Magnetism, VCH, NY, 1993, p. 26. V. Laget, C. Hornick, P. Rabu, M. Drillon, R. Ziessel, Coord. Chem. Rev. 1998, 178, 1533. M. Kurmoo, Chem. Mater. 1999, 11, 3370. C. Bellitto et al., unpublished results.

456

13 Transition Metal Ion Phosphonates as Hybrid Organic–Inorganic Magnets

[46] P. Palvadeau, M. Queignec, J.P. Venien, B. Bujoli, J. Villieras, Mater. Res. Bull. 1988, 23, 1561. [47] D.M. Poojary, B. Zhang, A. Clearfield, Chem. Mater. 1999, 11, 421. [48] D.M. Poojary, B. Zhang, A. Clearfield, J. Am. Chem. Soc. 1997, 119, 12550. [49] D.L. Lohse, S.C. Sevov, Angew. Chem. Int. Ed. Engl. 1997, 36, 1619. [50] Q. Gao, N. Guillou, M. Nogues, A.K. Cheetham, G. Ferey, Chem. Mater. 1999, 11, 2937. [51] A. Distler, D.L. Lohse, S.C. Sevov, J. Chem. Soc. Dalton, 1999, 1805.

Magnetism: Molecules to Materials II: Molecule-Based Materials. Edited by Joel S. Miller and Marc Drillon c 2002 Wiley-VCH Verlag GmbH & Co. KGaA Copyright ISBNs: 3-527-30301-4 (Hardback); 3-527-60059-0 (Electronic)

14

Magnetic Langmuir-Blodgett Films Christophe Mingotaud, Pierre Delhaes, Mark W. Meisel, and Daniel R. Talham

14.1

Introduction

Recent years have seen significant interest in the magnetic and electronic properties of “soft media,” organized but non-crystalline materials that have not normally been considered as a source of traditionally solid-state phenomena. Many experimental approaches to the controlled organization of “soft media” have been demonstrated, and it is now clear that the assembly of molecular units can give rise to interesting magnetic properties. It is not unusual for these procedures to result in low-dimensional materials, such as lamellar structures (2D), linear chains (1D), or even nanoparticles (0D). As is discussed in several chapters in this series, magnetic properties often depend on the dimensionality of the system, and low-dimensional materials can give rise to new phenomena not observed in isotropic solids. This chapter focuses on intrinsically two-dimensional structures, formed as mixed organic and inorganic monolayer and multilayer thin films using the Langmuir–Blodgett (LB) technique. Some related thin film systems are also discussed. The Langmuir–Blodgett technique is an extremely elegant way of preparing organic monolayer and multilayer films, one layer at a time [1]. Normally, amphiphilic molecules are organized on a water surface to form a Langmuir monolayer that can then be transferred to a solid support. Deposited films are called Langmuir–Blodgett films and can be limited to a monolayer, or with continued deposition cycles can produce multilayered films. Details of the technique are described in the next section. Promising benefits of the method are that it is an aqueous chemistry procedure, normally carried out at or near ambient temperature, and it affords exquisite control for arranging molecules into organized assemblies. LB films are normally based on organic molecules or polymers, and they have been used in investigations of non-linear optics, energy transfer or electron transfer in controlled media, organic conductors, molecular rectifiers and ferroelectrics [2]. However, to date, in those cases where magnetic ordering phenomena or magnetic state switching have been observed, routes to incorporate inorganic species in the films, either as coordination complexes, clusters, or as continuous lattice monolayers, have been developed. The nature of a Langmuir–Blodgett film can be compared to other types of solidstate materials that are studied as two-dimensional magnetic systems. The adsorption of metal ions on a solid surface using vapor deposition methods can lead to magnetic monolayers or overlayers [3]. These studies have subsequently lead to thin film heterostructures and the subject of giant magnetoresistance [4]. The nature of the magnetic interactions in metallic monolayers is different than the exchange between localized moments that is most often seen in LB films. Also, the structure of va-

458

14 Magnetic Langmuir-Blodgett Films

por deposited monolayers, and often their magnetic properties, are directed by the interface, which is a feature that contrasts with most of the LB films described in this chapter. Layered solid-state compounds and intercalation hosts have also been studied from the point of view of two-dimensional magnetic materials [5]. Sometimes very large interlayer separations (up to 50 Å), comparable to those seen in LB films, have been achieved, and some examples of these materials are discussed later in this chapter. They differ from LB films in that the magnetic layers of solid-state compounds are always part of a three dimensional crystalline solid. We will discuss some examples of LB films that are, in fact, single layer analogs of lamellar solid-state compounds. Many of the materials to be discussed in this chapter derive their magnetic properties from molecular phenomena, while other films contain two-dimensional continuous lattice structures. The theoretical basis for understanding magnetism in layered systems has been extensively reviewed [5]. We will only summarize some of the fundamental considerations here. The magnetic continuous lattice LB films are comprised of layers of exchange coupled paramagnetic metal ions whose magnetic response is described by the spin-Hamiltonian: H = −J Si S j + DSi2

(1)

where the exchange interaction, J , between nearest neighbors bearing spin Si and S j is defined by its sign, which is positive for ferromagnetic and negative for antiferromagnetic alignment of nearest-neighbor moments, and D is the single-ion anisotropy. Eq. (1) says nothing about the mechanism of exchange, but in most LB films a superexchange mechanism operates, where magnetic interactions within a layer are modulated by an active ligand that bridges adjacent metal ions. Other possible mechanisms in insulators include direct exchange, as is sometimes seen in transition metals, and dipolar exchange, which is normally small, but can be active over very large distances [6]. Eq. (1) relates only to short-range or nearest neighbor interactions, but does not describe long-range magnetic order. Layered systems can undergo magnetic ordering transitions from a paramagnetic state to either a ferromagnetic ground state below the Curie temperature (TC ), an antiferromagnetic ground state below the Neel temperature (TN ) or even exhibit ferrimagnetic behavior when two sublattices do not compensate. Historically, two-dimensional systems have been of great interest from the point of view of the theory behind magnetic ordering. From the pioneering work of Bloch [7] and the famous proof of Mermin and Wagner [8], it is known that an isotropic, finite range interactions between spins in one- or two-dimensional lattices cannot lead to long range order at a finite temperature [9]. This result is often quoted, but in practice, an ideal isotropic two-dimensional system is difficult to realize, and most real 2D systems order [10]. Magnetic ordering can result if there is any anisotropy in the system, such as non-equivalent spin components (Si x = Si y = Si z ), single ion anisotropy (D = 0), or if there is a small contribution to interlayer magnetic exchange (J⊥ = 0 and J|| = J⊥ where J|| is the in-plane term). It has been suggested that Langmuir–Blodgett films, as nearly free standing magnetic layers, might be used to investigate some of the issues related to magnetic ordering

14.2 The Langmuir–Blodgett Technique

459

in two dimensions [10, 11]. While long-range order has now been observed in several different Langmuir–Blodgett film systems, it has thus far proven difficult to identify the specific interactions responsible for the magnetic ordering. This point is discussed further, below, when describing individual materials. Space does not allow review of all magnetic measurements on LB films. In particular, there have been numerous examples of EPR analyses of radicals used as spin labels in LB films for the purpose of understanding structure or dynamics. We will limit our discussions, here, to systems where magnetic properties were the objective behind the studies. The remainder of this chapter is organized as follows: Section 2 briefly describes the LB technique, showing how it can be used to control the organization of molecular assemblies. We then highlight several recent advances in the area of magnetic LB films, starting in Section 3 with molecular systems, based on magnetic clusters and spin-crossover molecules, and continuing in Section 4 with extended systems and the associated cooperative effects. We provide a brief comparison of LB films to other soft or lamellar materials in Section 5 and conclude with an assessment of future directions in Section 6.

14.2 14.2.1

The Langmuir–Blodgett Technique Fabrication of Langmuir–Blodgett Films

Langmuir–Blodgett films are monolayers or multilayers deposited on to a solid substrate [1]. Their fabrication is based on a two-step process. The first stage is the formation of a monolayer at the gas–water interface (called a Langmuir monolayer) by carefully spreading an organic solution on a subphase (see Fig. 1A). The tensio-active molecules that are dissolved in the organic solution organize along the interface. The polar groups are attracted to the water while their hydrophobic tails limit dissolution into the aqueous subphase. On average, the molecules are poorly organized at this stage because the area per molecule is large. To increase the molecular density and organization, a slow compression of the monolayer is then performed (see Fig. 1B). When the molecular packing is sufficiently dense, Langmuir–Blodgett deposition may be attempted by dipping a solid substrate into the subphase resulting in transfer of a monolayer (see Fig. 1C). During the up-stroke, transfer of the next layer occurs leading to a modified substrate on which two monolayers have been deposited (see Fig. 1D). Multilayers can be built-up in this way, and normally, the orientation of successive layers alternates, resulting in head-to-head bilayers (this arrangement is called a “Y-type” LB film). Not all molecules behave ideally and the realization of a perfectly well defined LB film is not always possible. However, when successful, LB deposition results in multilayers with a precise thickness and supramolecular architecture. Usually, amphiphilic molecules (i. e., compounds containing both a polar head and a hydrophobic tail) are used to design LB films. In what is sometimes called the “amphiphilic strategy,” the active molecules are neutral, or if charged the counter

460

14 Magnetic Langmuir-Blodgett Films

Fig. 1. Schematic diagram of an LB through and multilayered deposition (Y-type).

ions play no particular role (Fig. 2A). In the “semi-amphiphilic strategy” (Fig. 2B), charged molecules form the monolayer and charge-compensating ions are adsorbed at the interface. The counter-ions, either metal ions or molecular ions, can also be incorporated into the transferred films as a result of the electrostatic interactions with the template amphiphile. In a number of the examples discussed below, the magnetic properties of the films arise from transition metal ions and complexes that are incorporated as counter-ions in semi-amphiphilic LB films.

14.2 The Langmuir–Blodgett Technique

461

Fig. 2. Schematic representation of Langmuir and Langmuir–Blodgett films: A, amphiphilic molecules; B, semi-amphiphilic system.

As films are compressed at the gas–water interface and the surface pressure rises, molecules become closely packed and orient more or less vertically (see Fig. 2A) to balance the forces imposed by the interactions of the molecules with the aqueous subphase and those between molecules. When films are transferred on to a solid substrate, the stress on these molecules is strongly modified as interactions with the aqueous subphase give way to interactions with the substrate (for a monolayer) or those between adjacent layers (for multilayers). Because of the changes in those interactions, the molecular organization inside each deposited layer differs from that on water. Indeed, tilting of the molecules or part of the molecule (such as alkyl chains) is usually observed in transferred films, as the van der Waals or π stacking forces between molecules optimize within the constraints imposed by the organization of the polar or ionic parts. Crystallization of the layer during the transfer may even occur. For example, it is well known that simple molecules such as fatty acids or alcohols are polymorphic and their packing depends highly on the transfer parameters, including the choice of substrate [12].

14.2.2

Structural and Physical Characterization of Langmuir and LB Films

To understand the molecular organization within transferred films, studies of the precursor monolayers at the gas-water interface are required to verify the film-forming qualities of the amphiphiles and begin to characterize intermolecular interactions. The state of the monolayer can be monitored via its compression isotherm (surface pressure against mean molecular area) and through spectroscopic and optical techniques including fluorescence microscopy, ellipsometry and Brewster angle microscopy (BAM). Sometimes, in situ structural information can be obtained with X-ray reflectivity and grazing angle X-ray diffraction. Structural characterization of built-up LB films is then achieved by combining a variety of complementary techniques [1]. Since no single analysis is sufficient to completely analyze the structure of a film, a number of techniques must normally be used. It should also be remembered that the structure of a film can depend on the solid support on to which it is deposited, or even as the number of transferred

462

14 Magnetic Langmuir-Blodgett Films

layers changes, so care must be exercised in comparing results from different techniques that require different substrates. The layered structure of transferred films can be quantified with ellipsometry, X-ray diffraction, and even neutron diffraction. Electron, optical, scanning probe and new near field microscopies can all contribute structural information about LB films, and also provide evaluations of defects and film roughness. UV–Vis and FTIR spectroscopies are normally sensitive enough to study monolayer or multilayer films, and optical or infrared dichroism can lead to information on the orientation of individual chromophores and even the organization of assemblies of molecules within LB films. True in-plane structure characterization requires grazing incidence X-ray diffraction, and is usually performed at a synchrotron source to overcome the small amount of material that is available in an LB film [2]. Magnetic characterizations are performed by electron paramagnetic resonance (EPR), or by SQUID magnetometry. EPR is highly sensitive, and for some organic radicals, where the linewidth is a few gauss or less, a single monolayer can be studied. For inorganic complexes, however, the linewidth is often hundreds of gauss peak-topeak, and tens to hundreds of deposited layers are needed for study. SQUID magnetometry requires more material than EPR, and samples are multilayers deposited on to diamagnetic polymer substrates that can be stacked or rolled to concentrate the sample [13].

14.3 14.3.1

Magnetic Systems: A Molecular Approach Hybrid LB Films with Magnetic Clusters

Films of magnetic clusters have been prepared using the semi-amphiphilic approach where anionic cluster molecules or transition metal complexes are electrostatically bound to the positively charged monolayer formed from an amphiphilic cation, such as a quaternary ammonium ion. Association of the anions with the monolayer at the air-water interface can generally be detected in pressure-area isotherms and in BAM experiments. Key parameters that affect this interaction include ion concentration and subphase pH, allowing the surface density of both anions and cations to be controlled [14]. These films can be transferred, leading to several new layered organic or inorganic LB film systems that will be presented below. 14.3.1.1

Magnetic Properties of Hybrid LB Films Based on Metal Complexes

It has been shown that metal oxalate complexes (FeIII (C2 O4 )3+ S = 5/2, 3 CrIII (C2 O4 )3+ S = 3/2) could be associated with charged surfactants such as 3 dimethyldioctadecylammonium bromide (DODA Br, 1).

14.3 Magnetic Systems: A Molecular Approach

463

Lamellar materials, based on these monometallic oxalate complexes, were characterized by X-ray diffraction, IR spectroscopy, EPR, and SQUID magnetometry. It was shown that the metal oxalate complexes are organized within the LB film as monolayers trapped in between organic bilayers [13]. The complexes are non-interacting and the magnetic behavior of the films is simply described by the Curie law. Heterobimetallic complexes such as a CrIII –FeII –CrIII polyoxalate (i. e., [CrIII (C2 O4 )32 FeII (H2 O)2 ]4+ ) were synthesized, and similarly deposited between DODA layers. For these materials, preliminary magnetization measurements showed a deviation from the Curie law at low temperature, indicating intramolecular magnetic interactions. A more complete study has been devoted to polyoxometalates, which constitute another interesting class of inorganic compounds. Their structures can be described as molecular fragments of close-packed metal oxides with the general formula Xa Mb On− c (M = Mo, W, V, and X = P, Si, B, Co) [15]. The best known of these are the so-called Keggin ions, based on a central heteroatom surrounded by four M3 O13 groups (M = Mo, W). The first to be studied in an LB film was the cluster (CoW12 O40 )6− that was transferred with DODA. X-ray diffraction demonstrated that only one layer of the heteropolyanion is included between DODA layers. Furthermore, IR dichroism shows that the Keggin polyanions are not randomly oriented within the film, but rather adopt one particular orientation or structural distortion. Magnetic measurements exhibit a decrease in susceptibility at low temperature that originates from the zero field splitting of CoII in a tetrahedral environment [16]. This first case has lead to the study of other polyoxometalates in LB films. For example, a monosubstituted Keggin salt, in which a tungsten atom in the periphery has been exchanged by a MnII (S = 5/2) ion, has been transferred with DODA. In another study, a tetrameric cluster, Co4 O16 , encapsulated in between two diamagnetic anions, (PW9 O34 )9− , was incorporated in an LB film. For the cobalt tetramer, ferromagnetic exchange interactions occur within the cluster leading to a total spin S = 6 [17]. The magnetic properties of the LB film (shown in Fig. 3) deviate from the Curie law, indicating intramolecular ferromagnetic interactions. A decrease of the product (χ T ) at very low temperature suggests the presence of a magnetic ground state with a smaller spin multiplicity.

Fig. 3. Plot of χ T (normalized value) against temperature for a LB film (353 layers) based on DODA and K10 [Co4 (H2 O)2 (PW9 O34 )2 ]. Adapted from Ref. [17].

464 14.3.1.2

14 Magnetic Langmuir-Blodgett Films

LB films of Single-molecule Nanomagnets

The discovery that individual “high-spin” molecules can act as nanomagnets has lead to active research in this field [18, 19]. Current interest in high-spin magnetic clusters is due to the presence of a large magnetic hysteresis, comparable to that observed in hard magnets, providing the possibility of molecular bi-stability and information storage at the molecular level. In addition, these nanomagnets are unique examples for observing quantum magnetization and electron tunneling through a potential barrier from one molecular state to another. The most thoroughly studied single molecule magnets are the mixed valent manganese clusters based on the Mn12 O12 core. They are formed by an internal tetrahedron of four MnIV ions (S = 3/2) surrounded by eight MnIII (S = 2) ions. Exchange interactions within the cluster results in a ground state with a large spin (S = 10) that encounters a thermal barrier for reversal of the direction of magnetization along the uniaxial magnetic axis (D = 0, see Eq. 1). In the crystalline state these neutral clusters, such as Mn12 O12 (acetate)18 , exhibit a stable hysteresis loop with a coercive field as large as 1.5 T at 2 K [20]. Because the clusters are not charged, they are not included in LB films through ionic association. Rather, the neutral clusters are incorporated into Langmuir monolayers by mixing them within a matrix of fatty acid. The monolayers are stable over time if the acid/cluster ratio is equal to or larger than five, and Langmuir films can be transferred. However, the degree of organization of the Mn12 O12 clusters within the LB film depends strongly on the lipid/cluster ratio, as shown in Fig. 4. For ratios near 5:1, a lamellar structure with monolayers of closed packed clusters is obtained. Plots of magnetization against applied magnetic field (M against H ) for films containing Mn12 O12 (acetate)18 contain a hysteresis loop at 2 K with a coercitive field of 0.06 T. The response is anisotropic with a softer hysteresis loop when H is parallel to the plane of the magnetic monolayer than when H is perpendicular. This behavior reflects the preferential mean orientation of these anisotropic clusters within the film [21].

Fig. 4. Schematic structures of LB films based on behenic acid and Mn12 O12 (acetate)18 for various lipid/cluster ratios. Adapted from Ref. [21].

14.3 Magnetic Systems: A Molecular Approach

14.3.2

465

Spin-transition Systems

In some transition metal complexes, an electronic spin-crossover transition can be observed between states of different spin multiplicity. For example, the ground state of octahedral first-row transition metal complexes, with d-orbital occupancy of d4 to d7 , can correspond to two different configurations. One configuration, called the high spin (HS) state, has a maximum number of unpaired electrons, and the other, called the low spin (LS) state, has a maximum of spin pairing. If the two states lie close in energy, then a small perturbation may be sufficient to cause a transition between them. In solid-state compounds, this transition can be induced thermally, optically, or with pressure [22]. Over the years, a very large number of iron(II) complexes have been investigated, with the majority of them containing nitrogen donor ligands. The ligand field strength can be used to adjust the nature and the characteristic temperature (TS ) of this spin transition. The molecular phenomenon can be coupled to lattice effects, leading to cooperative transitions. Therefore, spin-crossover transitions may be gradual, abrupt, or discontinuous with hysteresis [23]. Theoretical models, based upon a phenomenological thermodynamic approach, have been developed to explain such spin conversions in the simple case of thermal equilibrium, as well as for cooperative cases that may be associated with structural phase transitions. More recently, microscopic models have been developed to describe the nature of the interactions in these spin-crossover solids [22]. Recently the Langmuir–Blodgett technique has been used to study spin-crossover, because of the interest in controlling an electronic spin state interconversion in organized thin films for potential applications. Two main problems need to be overcome in order to prepare LB films of spin-crossover complexes. The first is to introduce amphiphilic character to the complexes without modifying too drastically the spincrossover process. The second issue is to control the molecular surroundings and the associated multilayer organization, which can also influence the spin conversion. Two groups have published on spin-crossover complexes in LB films, with all work involving iron(II) complexes having transition temperatures near room temperature in the bulk. The first attempts were carried out by the group in Saclay (france) with ligands derived from phenanthroline [24] and with an oligomer based on triazole [25]. The main drawback reported by the authors is the chemical instability of the complexes at the gas-water interface, which prevents the formation of physically and chemically well-defined LB films. A second approach has been devoted to two neutral amphiphilic complexes coordinated with bipyridine ligands. These compounds are based on the parent compounds FeII (4,4 -dimethyl-2,2 -bipyridine)2 (NCS)2 (TS ≈ 270 K) and FeII (tmsbpy)2 (NCS)2 where t-msbpy is 4-methyl-4 - trans-styryl-2,2 -bipyridine. This last compound contains the trans–cis photoisomerizable styryl group, which induces a “ligand driven light induced spin change” in the bulk [26, 27]. Two strategies were developed to overcome problems with the chemical and physical stability of these complexes in Langmuir monolayers. The first approach, used in the case of complex 2 (see Fig. 5), was to adapt the Langmuir technique by using

466

14 Magnetic Langmuir-Blodgett Films

an organic subphase based on formamide instead of pure water to reduce hydrolysis of the complex [28].

A second approach to reducing hydrolysis was to modify the chemical structure of the bipyridine ligand, avoiding the need to work with organic solvents. A new complex, 3 (see Fig. 5), with semi-fluorinated chains was synthesized. The semifluorinated hydrophobic groups interact differently than regular alkyl chains and lead to remarkably enhanced stability of the complex on the water subphase [29]. After transfer on to a solid substrate to form well-defined multilayers, the thermal spin-crossover processes could be investigated through both spectroscopic and magnetic measurements. Monitoring the temperature dependence of the IR spectrum of the LB films allows one to check that the chemical integrity of the complexes is maintained during the LB deposition process [28] and to follow the thermally induced spin change within the multilayers. The frequency of the C–N stretch of the NCS ligands depends on the spin state of the inorganic complex. Therefore, two bands can be found near 2050 cm−1 and 2110 cm−1 associated with iron complexes in a high-spin state and low-spin state, respectively. A plot of the ratio of intensities of the two bands indirectly characterizes the spin change, as shown in Fig. 6 for compound 2. During the first thermal cycle, the spin-crossover process is partially reversible and centered around 292 K, as in the pristine powder. However, in contrast to the bulk, the IR peak at ca 2050 cm−1 does not completely disappear at low temperature. This low temperature residue indicates that a large quantity of the iron complex does not

Fig. 5. Compression isotherms of a complex 2 monolayer (open circles) spread on a formamide-KNCS aqueous solution (concentration of KNCS 10−2 M; v/v 75:25) and of a complex 3 monolayer (full circles) spread on a KNCS 10−2 M aqueous solution. Adapted from Ref. [29].

14.3 Magnetic Systems: A Molecular Approach

467

Fig. 6. Ratio AHS /(AHS + ALS ) against temperature. AHS and ALS are the areas of the IR peaks, respectively, in the range 2000–2095 cm−1 and 2095–2135 cm−1 for an LB film of complex 2 (19 layers). Full circles: cooling, open circles: heating. Adapted from Ref. [28].

undergo the spin conversion process within the LB film. Magnetic measurements, performed for the first time on such a system, demonstrate that the molecules corresponding to this residue are in a low-spin state [30]. These results suggest that a large percentage of the iron complexes are partially distorted in the Langmuir film because of the high molecular density imposed by the compression of the monolayer. This distortion may indeed be similar to a pressure effect in the bulk, which stabilizes the low-spin state of the iron complexes. Presumably because of defects in the LB organization, a minority of the molecules within the multilayers experiences little or no distortion and undergo the spin crossover process detected by infrared spectroscopy or magnetic measurements. Therefore, the specific organization induced by the LB film can change or even suppress the spin conversion. If the multilayer is heated above 330 K, then the residue observed at low temperature by IR is much smaller (around 30%), as seen in the second thermal cycle (Fig. 6). Furthermore, in the annealed LB films, the spin conversion from the low-spin (S = 0) to the high-spin (S = 2) state is clearly observed in a plot of the product χM T against T (Fig. 7). Such behavior is in agreement with the hypothesis that the LB packing causes a molecular distortion. Heating the LB films above 330 K may result in melting of the alkyl

Fig. 7. Plots of the χM T product against T for a complex 3 LB film (1280 layers) on Mylar. Thermal cycle: 2 K → 350 K → 2 K. Adapted from Ref. [30].

468

14 Magnetic Langmuir-Blodgett Films

Fig. 8. Relative high spin (HS) molar fraction deduced from magnetic measurements under light irradiation (630–670 nm, 1 mW) as a function of time at 10 K for a powder sample and for a LB film of complex 3. Adapted from Ref. [33].

chains, leading to relaxation of the distortion previously induced by the packing. After annealing, the spin crossover process is then totally recovered. Even if the lamellar structure of the LB film induces a stress on the majority of the complexes, and therefore partially suppresses the spin transition, the thermodynamics of the spin-crossover process in the LB films is the same as that seen in the bulk. Indeed, for both compounds [29] in LB films or in the bulk, the experimentally observed spin conversions can be described as a chemical equilibrium with enthalpic and entropic parameters similar to a classical solid-state case, where the molecules do not interact significantly, and the resulting spin transition is gradual [31]. Therefore, the packing within the multilayer does not increase the cooperativity of the spin crossover but rather controls the number of molecules that undergo spin conversion. Finally, photo induced spin-crossover processes are of interest in these new lamellar materials. Two approaches have been explored so far. The first is a photochemical route involving photoisomerization of a ligand at room temperature (“ligand driven light-induced spin change”), such as the styryl group included in molecule 2 [32]. This effect has not yet been realized in an LB film [27]. The second possibility is the LIESST effect (“light induced excited spin-state trapping”), which is a photophysical phenomenon [22]. A preliminary experiment has been performed on an LB film of complex 3 and the LIESST effect has been observed at 10 K [33] (Fig. 8). This promising result may lead to new magneto-optical devices with high efficiency because of the large available surface area.

14.3.3

Comments on Molecular Magnetism in LB Films

These early experiments demonstrate that various LB approaches can be used to build well-ordered arrays of inorganic entities within thin films, leading to new organized multilayers with useful magnetic properties having molecular origin. For the systems containing anionic clusters that have been discussed so far, electrostatic repulsions are likely to lead to rather large distances between the anions within the LB films. As a result, the anion-anion magnetic interactions are quite weak and no cooperative magnetism is observed. However, cooperative magnetism is not the only route to useful magnetic properties. Spin-crossover has been demonstrated in LB films, and solid-state effects were seen that might ultimately be used to create thermal hysteresis. Purely molecular phenomena that can lead to thermochromic and

14.4 Extended Systems and Cooperative Effects

469

photomagnetic properties, which also hold potential for information storage and display devices, should be pursued in spin-crossover LB films. And finally, genuine single molecule magnetism has been observed in LB films, in the case of the high-spin mixed-valent manganese clusters. Together, the work on molecule-based LB films has shown that several interesting magnetic phenomena, which had previously been known in solution or solid-state studies, can now be imparted to interfaces, where they will ultimately be needed if high density information storage is the goal.

14.4 14.4.1

Extended Systems and Cooperative Effects “Literally Two-Dimensional Magnets”

The earliest extensive investigation of magnetism in LB films was a study of manganese stearate. Careful EPR measurements demonstrated antiferromagnetic exchange within the monolayer and provided evidence for a magnetically ordered state at low temperature. This pioneering work was performed by Melvin Pomerantz and coworkers in the late 1970s at the IBM T.J. Watson Research Center [10]. Their studies of magnetism in the limit of a monolayer had two motivations. First, one could imagine that a magnetic monolayer might be considered the ultimate magnetic memory storage device. And second, a single layer of magnetic ions provided an opportunity to investigate theoretical predictions about magnetic order in a truly two-dimensional system. Pomerantz and co-authors called the manganese stearate LB films “Literally Two-Dimensional Magnets.” Using the semi-amphiphilic approach for preparing LB films, Mn2+ ions were confined to a single layer between two stearate layers deposited head-to-head. The manganese stearate bilayers were deposited on to substrates made hydrophobic with a monolayer of the non-magnetic cadmium stearate. Magnetic exchange in two-dimensions was demonstrated using EPR. Experiments were performed on a stacked sample of 50 slides, each containing a manganese stearate bilayer. The EPR linewidth and resonance field were each shown to exhibit anisotropy consistent with exchange in a two-dimensional lattice. The dipolar broadening of the EPR signal depends on spin diffusion and in 2D follows the form H = A + B(3 cos2 − 1)2

(2)

where is the angle between the external field and the film normal and A and B contain contributions from dipolar broadening. The manganese stearate data were shown to conform to this model, demonstrating the expected behavior for antiferromagnetic exchange in a 2D array and showing that there was not a significant amount of Mn2+ in 3D environments (Fig. 9). A large increase in the EPR linewidth below 10 K for the manganese stearate film signaled a divergence of the magnetic correlation length and a dramatic shift in the resonance field below 2 K provided evidence of an ordered state. The anisotropy of

470

14 Magnetic Langmuir-Blodgett Films

Fig. 9. Plot of the X-band EPR linewidth against θ, the angle between H0 and the film normal, of a manganese stearate LB film at 80 K. The solid curve is a fit to Eq. (2) with A = 220 Oe and B = 65 Oe. The plot is adapted from Ref. [10].

the resonance field measured at 1.4 K is different than that seen at high temperature, which signals the development of an internal field. The authors concluded that the manganese stearate monolayers order in a canted antiferromagnetic state [10]. The low critical temperature makes experiments difficult, and a complete characterization of the ordered state has never been performed. Haseda et al. [34]. found an even lower ordering temperature of 0.3 K. However, the work did indicate that magnetic exchange and magnetic order could be observed in monolayer films and it has inspired other approaches to magnetism in LB films.

14.4.2

Metal Phosphonate LB Films

A more recently developed class of LB films is the metal salts of alkylphosphonic acids (Fig. 10). For example, octadecylphosphonic acid forms LB films with a variety of divalent, trivalent or tetravalent metal ions resulting in layers of the metal ions sandwiched within bilayers of the organophosphonate [35–37]. An interesting and useful feature of these systems is that solid-state metal phosphonates form layered continuous lattice structures, and it has been possible to learn about the in-plane structure of the LB films by comparing them to the known solids. In nearly all cases, it has been shown that the LB films possess the same in-plane bonding as their solidstate analogs. Along with forming the continuous lattice solid-state structures in LB films, it is possible to introduce typically solid-state properties, such as magnetism, into the films through the inorganic network [35]. The most extensive studies have been on manganese phosphonate LB films. In the solid-state, manganese alkylphosphonates crystallize in the orthorhombic space group Pmn21 [38]. Sheets of manganese ions, in a distorted square array, are bonded on top and bottom by layers of the organophosphonate (Fig. 11). Within a layer, each phosphonate group bridges four metal ions, and each metal ion is coordinated by five oxygen atoms from four different phosphonate groups. The distorted octahedral coordination of the metal is completed by a water molecule of hydration. Magnetic exchange within the layers is antiferromagnetic, and the alkylphosphonates (Cn H2n+1 PO3 )Mn.H2 O (n = 2–6) each undergo magnetic ordering transitions between 13 K and 15 K. A weak moment develops below the ordering temperature and the solid-state phosphonates are known to be canted antiferromagnets [39, 40].

14.4 Extended Systems and Cooperative Effects

471

Fig. 10. Schematic comparison of the similar layered structures of a Y-type LB film, left, and the layered metal alkylphosphonates exemplified by Ca(O3 PC6 H12 )2 , right. The calcium phosphonate structure was generated from data in Ref. [38b].

The manganese phosphonate LB films were studied with X-band EPR and with SQUID magnetometry [35, 41]. At high temperatures, the EPR behavior is similar to that of manganese stearate [10]. The EPR signal is dipolar broadened, with line widths that range from 218 to 300 G depending on the orientation of the film in the magnetic field, and the anisotropy of the line width is consistent with antiferromagnetic exchange within the 2D sheets [41]. The temperature dependence of the EPR intensity was used to estimate the magnitude of the nearest-neighbor exchange, J, by fitting the data to a numerical expression for the susceptibility of a quadratic layer Heisenberg antiferromagnet. The value of J/kB = −2.8 K, obtained from the fit, is nearly identical to the values of exchange observed for the solid-state manganese phosphonates. However, no transition to long range magnetic order is seen in X-band EPR. At temperatures below 50 K, the line width begins to increase and becomes too broad to measure below 17 K. This behavior is consistent with antiferromagnetic fluctuations, a precursor to magnetic ordering, but no direct observation of the ordered state can be seen at 9.3 GHz [41]. An ordering transition is observed in static magnetization measurements, obtained with a commercial SQUID magnetometer [35]. The temperature dependent magnetization (Fig. 12) was measured upon warming the film from 5 K for the cases where the sample was cooled from room temperature in zero applied field (ZFC)

472

14 Magnetic Langmuir-Blodgett Films

Fig. 11. The layered nature of Mn(O3 PC6 H5 ).H2 O is shown in a cross-section at left, and the manganese ion plane is viewed perpendicular to the layer at right, where the phenyl groups have been omitted for clarity. Each manganese ion is coordinated by five phosphonate oxygens and one water molecule. Key: oxygen, small open circles, manganese; crosshatched circles; phosphorus, diagonal hatched circles. Phosphorus atoms above and below the metal ion plane are distinguished by hatch marks with different directions. Figures were generated with crystallographic data from Ref. [38a].

and where the sample was cooled from room temperature in a magnetic field of 0.1 T (FC). The ZFC data show the signature of an ordering transition at 13.5 K. The FC data also show the ordering transition, and the increased magnetization below the ordering temperature (TN ) is evidence for canted antiferromagnetism in the film. The spontaneous magnetization arising from the canting moment is shown more clearly when the difference between the FC and ZFC magnetization ( M) is plotted (Fig. 12). Since the mass of the sample is small compared to the sample holder and Mylar substrate, the difference plot subtracts the signal due to the sample support and allows the film magnetization to be quantified. The magnetization at 5 K is greatest with the applied field oriented parallel to the film, indicating that the canting moment is in-plane. Further evidence for this magnetic structure comes from field dependent magnetization in the ordered state. For the case where the applied field is oriented perpendicular to the plane of the film, there is evidence for a spin-flop transition at Hsf = 2.5 T (Fig. 13). The spin-flop transition in this orientation indicates that the axis of antiferromagnetic alignment is perpendicular to the plane of the film, and therefore perpendicular to the manganese phosphonate layers. Canting of the moments away from the easy axis gives rise to the observed in-plane magnetization. The LB film also exhibits magnetic memory below TN . Hysteresis is observed in the vicinity of zero field for positive and negative scans of the applied field as it is cycled between ±5 T [35]. Hysteresis is the signature of magnetic memory, and a small coercive field of 20 mT and remnant magnetization of 10 memu are seen [35] Although the effect is small, the result nevertheless demonstrates that the LB method can be used to produce soft thin magnetic films.

14.4 Extended Systems and Cooperative Effects

473

Fig. 12. Plot of magnetization against temperature for an 81-bilayer manganese octadecylphosphonate LB film with the measuring field applied parallel to the plane of the film. At top, the fieldcooled (FC) and zero field-cooled (ZFC) data are compared. A difference plot ( M) clearly shows the spontaneous magnetization below TN = 13.5 K. The spontaneous magnetization arises from spin canting as illustrated in the scheme. Adapted from Ref. [35].

Fig. 13. Plot of magnetization against applied field at 2 K for a manganese octadecylphosphonate LB film. Data are shown for two orientations, parallel to the plane of the sample (open circles) and perpendicular to the plane (filled diamonds), and are normalized to the value at 5 T. The spin-flop transition is seen in the perpendicular orientation at 2.5 T, indicating that the easy axis is perpendicular to the sample plane. Adapted from Ref. [35].

14.4.3

Organic and Inorganic “Dual Network” Films

Metal phosphonate LB films are not limited to octadecylphosphonic acid in the organic network. Amphiphilic phosphonic acids containing aromatic groups, azobenzenes, metalloporphyrins and tetrathiafulvalene derivatives have all been prepared and studied as metal phosphonate LB films [42–46]. Such systems open up the possibility of preparing LB films with composite properties. For example, the organic net-

474

14 Magnetic Langmuir-Blodgett Films

work might be conducting, mesomorphic, photochemically switchable, redox active, or contain non-linear optical chromophores, and the inorganic network can provide typically inorganic solid-state properties such as conductivity, semiconductivity, or as discussed in this chapter, magnetism. Within the context of metal phosphonate LB films, some preliminary questions needed to be answered. Can these substituted phosphonic acids form organized LB films within the structural constraints of the inorganic metal phosphonate lattice? Likewise, as the organic substituents get larger, is there a limit where the inorganic continuous lattice no longer forms? However, several combinations of organic and inorganic networks have now been found that form “dual network” metal phosphonate LB films, and some are magnetic. Manganese phosphonate LB films have been prepared with the substituted phosphonic acids 4 and 5 and both films order as canted antiferromagnets near 13 K [45, 46], just as the manganese octadecylphosphonate film does.

An interesting feature of the films with amphiphile 5 is that the azobenzene groups aggregate, and the packing within the organic and inorganic networks is likely incommensurate. The four-carbon linker between the azobenzene chromophore and the phosphonate head-group is flexible and takes up the strain between the inorganic and organic networks with differing periodicity [43]. Tetrathiafulvalene (TTF) and its derivatives such as bis(ethylenedithio)tetrathiafulvalene (BEDT-TTF) are often the basis molecules of conducting, semiconducting, and superconducting molecular solids. There are now several examples of semiconducting and conducting LB films based on TTF derivatives [47, 48]. With the metal phosphonate approach, it may be possible to form LB films with a conducting organic network and a magnetic inorganic network. Compounds such as 6 and 7 have been prepared to explore this possibility. Thus far, only semiconducting examples have been produced, but the manganese phosphonate films of 6 and 7 both show magnetic ordering of the inorganic network also [46].

14.4 Extended Systems and Cooperative Effects

475

These early examples show that it is possible to form “dual network” LB films where the organic and inorganic components contribute complementary properties. The next steps will be to tune the materials to find systems where the networks act in synergy, leading to effects such as magnetoresistance or photochemical switching of the magnetization.

14.4.4

Bimetallic Compounds

New classes of molecular ferromagnets or ferrimagnets have recently been developed which are heterobimetallic compounds with bridging oxalate and cyanide ligands [49]. Metal cyanide compounds with Prussian blue-like structures were then considered as possible routes to LB films with cooperative magnetic phenomena. These cubic inorganic solids have the general formula Ak [B(CN)6 ].nH2 O where A and B can be divalent or trivalent transition metals, and k depends on the relative charges of the different metal ions and the number of vacancies in the structure. The CN ligands are good mediators of magnetic exchange, and examples of these mixed valence bimetallic compounds show ferrimagnetic or ferromagnetic ordering at critical temperatures that range from liquid helium temperatures to above room temperature [50]. Several metal cyanide extended solids have now been studied as hybrid organic and inorganic LB films. LB films of Prussian blue related compounds have been prepared using routes similar to the previous studies on polyoxometalates (see Section 3.2). A classical semiamphiphilic approach is used where a positively charged lipid such as DODA is spread on a very dilute colloidal dispersion of the inorganic compound as the subphase. Thus far, this route has lead to films containing the classical Prussian blue, Fe4 [Fe(CN)6 ]3 .12H2 O (TC = 5.5 K), and the mixed metal analog Cu3 [Fe(CN)6 ]2 .12H2 O (TC = 21 K). In both cases a stable charged monolayer can be obtained and then transferred on to different solid substrates, trapping the inorganic entities within the amphiphilic layers (Y-type transfer). The resulting hybrid films have been characterized with IR dichroism and X-ray diffraction [51]. For both materials, the CN stretching frequencies are consistent with bridging cyanides, and the IR dichroism shows that at least a part of the CN bridging groups are oriented parallel to the substrate plane. In addition, an interesting feature is seen in low angle X-ray diffraction experiments on the Prussian blue analog. As shown in the Fig. 14 both Kiessig fringes and a Bragg (001) peak are recorded. This result demonstrates that a well-defined lamellar structure is present, at least partially, inside the LB film (one should not rule out the possible adsorption of inorganic nanoparticles along the interface and then trapped in the multilayer). The calculated periodicity is about 42–44 Å. By accounting for the length and the tilt angle of the alkyl chains, one can estimate a thickness of about 5 Å for the inorganic layer. In that case, this structure could correspond to a single layer of cyanide (bridged or not) iron ions. The situation is not so clear for the mixed copper-iron compound where a sharp Bragg (001) peak has not been detected so far [52]. Even if the physical and chemical equilibria inside the colloidal subphase or

476

14 Magnetic Langmuir-Blodgett Films

Fig. 14. X-ray diffractogram of a LB film (19 layers) based on DODA and Prussian Blue. The arrow designates the (001) Bragg refection. Adapted from Ref. [51].

along the interface are difficult to characterize, these experiments demonstrate that hybrid systems based on such compounds can be prepared by the LB technique. It is interesting to compare the structure of the metal cyanide containing LB films to previously described mesoscopic inorganic structures formed within lamellar organic templates. Two-dimensional cyanide derivatives based on NiII (CN)2− 4 bridged III , have been prepared within cast films of by CuII , or FeII (CN)4− bridged with Fe 6 bilayers of charged lipids, using a stepwise deposition technique [53, 54]. X-ray diffraction on the iron cyanide phase suggests an inorganic layer thickness of 5.1 Å, which is in agreement with the observation made on the Prussian-blue containing LB film. Both d.c. and a.c. magnetization and the associated magnetic susceptibilities of these hybrid LB films have been investigated. Their values are expressed per unit of surface and number of deposited layers. Classic Curie–Weiss law is observed at higher temperatures for the Prussian blue film, with a positive Weiss constant indicating the presence of ferromagnetic interactions. A ferromagnetic ground state is evidenced at low temperature by the onset of a spontaneous magnetization below TC . Fig. 15 shows the zero field-cooled (ZFC) then a field-cooled (FC) magnetization plot for a Prussian blue LB film (300 layers). A Curie temperature is defined (TC = 5.7 ± 0.1 K), which corresponds to long range ferromagnetic ordering. This temperature is slightly higher

Fig. 15. Plot of field-cooled (FC) and zero field-cooled (ZFC) magnetization against temperature for a LB film (300 layers) based on DODA and Prussian Blue. Adapted from Ref. [51].

14.4 Extended Systems and Cooperative Effects

477

Fig. 16. Angular dependence of the EPR linewidth and g-factor on the angle between the magnetic field and the LB film plane at 4.3 K. Adapted from Ref. [55].

than that determined under the same experimental conditions for the commercial Prussian blue powder used in this work as precursor (TC = 5.1 ± 0.1 K). An a.c. magnetization measurement confirms this result, where the in-phase component exhibits a peak around TC , which is characteristic of magnetic losses in a ferromagnet [55]. Furthermore, EPR experiments at X-band furnish further insight about the magnetic anisotropy of this lamellar system. Strong anisotropy in the line position (gfactor) and the linewidth ( H in gauss) near TC are presented in Fig. 16. As with other LB film systems (see Sections 4.1 and 4.2), the angular dependence of the linewidth exhibits behavior consistent with a two-dimensional magnetic system. Moreover, the temperature dependence of the g-factor and linewidth change at T < TC because of the onset of internal fields that lead to the progressive appearance of ferromagnetic resonance. A similar magnetic study has been carried out on the mixed metal compound. Plots of field-cooled magnetization (H = 10 gauss) against temperature are different for the pristine powder (TC = 21 K) and the corresponding LB film (TC = 25 K). While the surface magnetization increases linearly with the number of layers, the critical temperature for the LB film is independent of the total number of deposited layers.

478

14 Magnetic Langmuir-Blodgett Films

Fig. 17. Hysteresis loop at 3 K in the plot of magnetization against field for an LB film based on DODA and Cu3 [Fe(CN)6 ]2 . Adapted from Ref. [52].

This result indicates the layer by layer deposition process is constant, even if the final structure is not so well defined by X-ray diffraction. Because the compound is EPR silent, there is yet no direct proof of the 2D (or lack of 2D) character of the magnetic properties. Finally, a plot of the isothermal magnetization against applied field gives a hysteresis loop characteristic of the presence of magnetic domains inside a soft ferromagnetic state (see Fig. 17). A similar hysteresis loop is found for the Prussian blue LB film [51]. These hybrid layers appear to be 2D magnetic systems with Curie temperatures as high as in the analogous 3D crystalline compounds. A similar approach should be applicable to other bimetallic derivatives, many of which exhibit even higher transition temperatures. Recently an LB film containing a NiII and CrIII mixedmetal cyanide as the inorganic part and DODA as the amphiphile was prepared that exhibits a Curie temperature TC = 46 K. The supramolecular architecture of these multilayers depends on a number of factors, including the adsorption and reaction of the inorganic compounds along the interface [56]. Studies on those interfacial processes should be made using other surfactants in order to better control the organization . Ultimately, they should lead to new well-defined hybrid nanomaterials based on such inorganic derivatives. Potential applications of magnetic LB films are evident if a ferromagnetic state can be stabilized near room temperature. However, some other interesting studies of these multilayers will likely involve coupling of complementary properties within the hybrid materials. Examples might include films with both electroactive and magnetic properties (in situ electrochemistry is easily performed on a thin film, but more difficult on a powdered solid [57]) or films that are both magnetic and photoactive.

14.5

Comparison with Other Lamellar and Colloidal Systems

LB films comprise just one class of mixed organic and inorganic soft media, and it is worthwhile to offer some comparisons with related materials that have been explored for their magnetic properties. The transition metal phosphonate and Prussian blue

14.5 Comparison with Other Lamellar and Colloidal Systems

479

systems, discussed in Section 4, are LB analogs of known solid-state structures, and parallels can be drawn with traditional layered solids that have been studied for the purpose of exploring low-dimensional magnetism. Also, LB films are normally based on amphiphilic building blocks and can be compared with colloidal organic and inorganic materials, where the self-organizing surfactants either template or complement the formation of the inorganic network.

14.5.1

Hybrid Lamellar Systems

Metal organophosphonates in the solid state frequently form layered structures, and several different structure types are known. The binding within the metal phosphonate layer varies as the metal ions and the organic groups change, but the layered nature persists for many structures. The magnetic properties of several families of metal phosphonates have been studied. The canted antiferromagnetic state in manganese alkylphosphonates was demonstrated through a combination of magnetization and neutron scattering measurements, and antiferromagnetic resonance. Manganese compounds are not the only layered phosphonates to show magnetic order. Layered phases based on CrII , FeII , NiII , CuII , and CoII are known to form weak ferromagnets at temperatures that range from 2 K to 35 K [58]. A similar class of materials are the alkylammonium metal halide layered perovskites, (RNH3 )2 MX4 where R is an organic group, X a halide, and M = CrII , MnII , FeII , CdII , or CuII ) [59, 60]. The MnII and FeII analogs are layered antiferromagnets, while the CuII and CrII cases are ferromagnets, as a result of a cooperative Jahn–Teller distortion that renders the magnetic orbitals on adjacent metal sites orthogonal. It was recognized long ago that such materials provided an opportunity to combine typically organic and inorganic phenomena in a “molecular composite.” In one example, the alkylammonium ions in the compositions above were replaced with aminodiacetylene cations, which could then be photopolymerized. Another interesting series of organic and inorganic layered compounds are the ferrimagnetic mixed valence oxalates, A[MM (C2 O4 )3 ], where A is a quaternary ammonium ion, and, for example, M = FeIII and M = MoII , FeII , CoII , NiII , or CuII [61-63]. The structures consist of planar honeycomb networks separated by the organoammonium ions. Some of them exhibit an interesting crossover from positive to negative magnetization under a weak applied magnetic field, and in some cases TN increases with interlayer separation. Transition metal hydroxides with the formula Cu2 (OH)3 X.nH2 O can form organic and inorganic analogs if X is an alkyl carboxylate or sulfonate. For the carboxylates, both antiferromagnetic and ferromagnetic ground states are observed, depending on the separation between layers [64–66]. A similar cobalt series with formula Co5 (OH)8 X.nH2 O has been explored, where here, X is a dicarboxylic acid. The magnetic data show all these compounds behave as two-sublattice ferrimagnets, due to the presence of both octahedral and tetrahedral sites [66]. The organic and inorganic copper hydroxides, at least, can be prepared either directly or by ion exchange intercalation. Another outstanding example of magnetism in organic and inorganic layered intercalation compounds is stilbazolium MPS3 (with M = MoII and

480

14 Magnetic Langmuir-Blodgett Films

P = CdII ). The inorganic network gives rise to permanent magnetization below 40 K, whereas the organic network provides a large second-order optical non-linearity [67]. The systems mentioned above all possess structural features that could make them amenable to fabrication as LB films. In addition, several have been previously studied in the context of two-dimensional magnetism. It is interesting, in the cases of the layered perovskites and copper hydroxides, especially, that magnetic order is seen in materials where the interlayer spacing is as large as 40 Å. Discussion continues about the precise mechanism for magnetic ordering in systems where interlayer coupling should be very small. Current ideas suggest that even very weak dipolar coupling between layers could be sufficient to lead to 3D ordering [68]. LB films, where interlayer distances can expanded even further, offer promise for investigation of the factors that lead to long-range order in nominally two-dimensional materials. However, so far, the small amount of material present in monolayer or even multilayer LB films have hindered detailed studies of these issues.

14.5.2

Self-organized Media

There are other approaches for forming organic or inorganic composite films with magnetic properties. For example, the alternating deposition of polyelectrolytes provides a powerful way to achieve alternating layer thin films [69]. Related to the polyoxometalate containing LB films discussed in Section 3.1 are hybrid multilayered films prepared by alternate adsorption of octamolybdate (Mo8 O26 )4− and linear polycations [70]. Moreover surfactant encapsulated clusters of very large polyanions, based on a Mo57 or Mo132 core having a diameter of a few nanometers with a well defined supramolecular architecture, have been realized by combining surfactant chemistry with ligand exchange process and the Langmuir technique [71]. Prussian blue has also been incorporated into other types of composite films. The Prussian blue framework was condensed within the polar regions of a cast film of charged lipids [53, 54]. By employing a photoisomerizable vesicle that includes an azobenzene chromophore, these authors are able to control the low temperature magnetic properties of the hybrid material by photoillumination [72]. An approach to thin films that does not involve surfactants concerns the formation of functionalizable sol-gel thin films. For example, ion exchange with further complexation in solution gave rise to a sol-gel composite including aggregates of Prussian blue [73]. Returning to surfactants, classical colloid processes have been successfully employed to build mesoscopic inorganic and organic architectures. In the past, surfactant-based colloids have been used to stabilize a variety of semiconductor and precious metal nanoparticles [74]. More recently, this approach has been extended to magnetic transition metals. The composite materials have even been used as ferrofluids. Ferrite nanoparticles have been synthesized in reverse micelles [75], and the photomagnetic characteristics of polymer-nanoferrite composites have been investigated [76]. The composite particles can also self-organize, and a 2D array magnetic nanosized cobalt particles was produced and magnetically studied [77, 78].

14.6 Conclusion

481

These examples show that organic and inorganic frameworks offer versatility in the design and control of magnetic materials. Control over the size and organization of the magnetic entities is easily combined with issues related to processing. Furthermore, the mixed organic and inorganic make-up allows the possibility of engineering materials with composite properties. LB films are an important subset of these self-organizing soft media. While new materials are being developed, LB films can be used for detailed studies of the interface between the organic and inorganic networks that are central to all of these examples.

14.6

Conclusion

Organized thin films, including LB films, constitute a novel branch of magnetic materials that combines aspects of molecular magnetism with some classical solid-state chemistry. Attractive features of molecular studies are retained, such as processing and the ability to control function through molecular synthesis, and in some examples, they are combined with traditional, but still useful, solid-state phenomena like cooperative magnetism. Furthermore, the promise of designing molecule based systems that afford a positive contribution to properties from both organic and inorganic networks, giving rise to synergistic properties, seems very real. Certainly, progress in magnetic LB films parallels other efforts to prepare mixed organic and inorganic magnetic materials, such as those that take advantage of other deposition processes to form thin films, and spontaneous assembly to form colloidal networks. While they clearly represent their own class of materials, LB films and the associated methods also provide well-defined and controllable systems for studying the organic and inorganic interface that is common to all of these “soft” magnetic materials. With respect to the magnetic properties of LB films, we can emphasize three main points based on work to date. First, cooperative magnetism has been observed in continuous lattice systems made up of individual monolayers. The physics behind the long-range magnetic order is not yet clear, but in each case, either the existence of magnetic anisotropy, or weak dipolar coupling between well-spaced layers, appear to give rise to the observed cooperative phenomena. Clearly, it is possible to prepare magnetic monolayers from wet-chemical methods. Second, important molecule-based phenomena have also been observed in multilayered films. Whether it is the quantum effects of the molecular clusters and nanoparticles, or molecular spin-crossover, these phenomena have now been observed in those lamellar systems. To achieve the limits in the density of magnetic information storage that molecular processes promise, these materials will have to be deposited on to surfaces. The work of thin organized films demonstrates the feasibility of confining molecular magnetic materials to surfaces. Finally, their organic and inorganic nature, combined with the inherent extremely high level of control, make LB films an ideal medium to pursue the design of molecular materials with composite properties. Combinations that include magnetism should be possible.

482

14 Magnetic Langmuir-Blodgett Films

Acknowledgments DRT and MWM thank the US National Science Foundation and the Petroleum Research Fund, administered by the ACS, for partial support of some of the work described. DRT also thanks the CNRS for funding his visit to CRPP as a chercheur associe, ´ during which time part of this chapter was prepared. CM and PD thank the European Network TOSS for partial financial support.

References [1] A. Ulman. An Introduction to Ultra-thin Organic Films from Langmuir–Blodgett to Selfassembly, 1st edn; Academic Press, Boston, 1991. [2] Langmuir–Blodgett films (Ed. G. Roberts), Plenum Press, Boston, 1990. [3] J.M.D. Coey, M. Viret, S. von Molnar. Adv. Phys. 1999, 48, 167–293. G.A. Prinz. Science, 1998, 282, 1660–1663. [4] (a) S.S.P. Parkin, N. More, K.P. Roche. Phys. Rev. Lett. 1990, 64, 2304–8; (b) M.N. Baibich, J.M. Broto, A. Fert, F.N. Van Dau, F. Petroff, P. Etienne, G. Greuzet, A. Friedrich and J. Chazelas. Phys. Rev. Lett. 1988, 61, 2472–5. [5] J. de Jongh. Magnetic Properties of Layered Transition Metal Compounds, Kluwer, Dordrecht, 1990. [6] F. Palacio in Molecular Magnetism: From Molecular Assemblies to Devices (Eds.: E. Coronado, P. Delhaes, ` D. Gatteschi, J.S. Miller) NATO-ASI series, Kluwer Acad. Publ., 1996, p 5–63. [7] F. Bloch. Z. Phys. 1930, 61, 206. [8] N.D. Mermin and H. Wagner. Phys. Rev. Lett. 1966, 17, 1133–6. [9] L.D. Landau and E.M. Lifshitz in Statistical Physics (Part 1), Pergamon Press, London, 1980, p 537. [10] (a) M. Pomerantz in Phase Transitions in Surface Films (Eds.: J.G. Dash, J. Ruvalds) NATO-ASI series, Plenum: New York, 1980, p 317–346; (b) M. Pomerantz, F. H. Dacol, A. Segmuller. ¨ Phys. Rev. Lett. 1978, 40, 246–249; (c) M. Pomerantz. Surf. Sci. 1984, 142, 556-70. [11] P. Day. Chem. Br. 1983, 306–14. [12] A. Bonnerot, P.A. Chollet, H. Frisby and M. Hoclet. Chem. Phys. 1985, 97, 365–77. [13] M. Aiai, J. Ramos, C. Mingotaud, J. Amiell, P. Delhaes, A. Jaiswal, R.A. Singh, B. Singh, and B.P. Singh. Chem. Mat. 1998, 10, 728–33. [14] M. Clemente-Leon, B. Agricole, C. Mingotaud, C.J. Gomez-Garcia, E. Coronado, and P. Delhaes. Langmuir 1997, 13, 2340–7. [15] E. Coronado and C.J. Gomez-Garcia. Comments Inorg. Chem. 1995, 17, 255–81. [16] M. Clemente-Leon, C. Mingotaud, B. Agricole, C.J. Gomez-Garcia, E. Coronado and P. Delhaes. Angew. Chem. Int. Ed. Engl. 1997, 36, 1114–6. [17] M. Clemente-Leon, C. Mingotaud, C.J. Gomez-Garcia, E. Coronado and P. Delhaes. Thin Solid films 1998, 327-9, 439–42. [18] D. Gatteschi, A. Caneschi, L. Pardi and R. Sessoli. Science 1994, 265, 1054–8. [19] S.H.J. Aubin, M.W. Wemple, D.M. Adams, H.L. Tsai, G. Gristou and D.N. Hendrickson. J. Am. Chem. Soc. 1996, 118, 7746–54.

References

483

[20] L. Thomas, F. Lionti, R. Ballou, D. Gatteschi, R. Sessoli, B. Barbara. Nature 1996, 383, 145–7. [21] M. Clemente-Leon, H. Soyer, E. Coronado, C. Mingotaud, C.J. Gomez-Garcia and P. Delhaes. Angew. Chem. Int. Ed. Engl. 1998, 37, 2842–5. [22] P. Gutlich, ¨ A. Hauser, H. Spiering. Angew. Chem. Int. Ed. Engl.1994, 33, 2024–54. [23] P. Gutlich, ¨ J. Jung, H.A. Goodwin in Molecular magnetism from molecular assemblies to the devices (Eds.: E. Coronado, P. Delhaes, D. Gatteschi and J.S. Miller) NATO-ASI series, Kluwer Acad. Publ., 1996, pp 327–78. [24] A. Ruaudel-Teixier, A. Barraud, P. Coronel, O. Khan. Thin solid films 1988, 160, 107–9. [25] F. Armand, C. Badoux, P. Bonville, A. Ruaudel-Teixier, O. Kahn. Langmuir 1995, 11, 3467–72. [26] M.L. Boillot, H. Soyer New. J. Chem. 1997, 21, 889–92. [27] M.L. Boillot. S. Sour, P; Delhaes, C. Mingotaud, H. Soyer. Coord. Chem. Rev. 1999, 190-2, 47–59. [28] H. Soyer, C. Mingotaud, M.L. Boillot, P. Delhaes. Langmuir 1998, 14, 5890–5. [29] H. Soyer, E. Dupart, C. Mingotaud, P. Delhaes Colloid and Surfaces A, 2000, 171, 275–82 [30] H. Soyer, E. Dupart, C.J. Gomez-Garcia, C. Mingotaud, P. Delhaes. Adv. Mater. 1999, 11, 382–4. [31] O. Khan. Molecular magnetism, C.V.H. Editions, New York, 1993. [32] M.L. Boillot. S. Chantraine, J. Zarembowitch, J.Y. Lallemand, J. Prunet. New J. Chem. 1999, 23, 179–83. [33] J.F. Letard, O. Nguyen, H. Soyer, C. Mingotaud, P. Delhaes, O. Khan. Inorg. Chem. 1999, 38, 3020–1. [34] T. Haseda, H. Yamakawa, M. Ishizuka, Y. Okuda, T. Kubota, M. Hata, K. Amaya Solid State Commun. 1977, 24, 599–602. [35] C. T. Seip, G. E. Granroth, M. W. Meisel, D. R. Talham. J. Am. Chem. Soc. 1997, 119, 7084–94. [36] (a) G. E. Fanucci, C. T. Seip, M. A. Petruska, S. Ravaine, C. M. Nixon, D. R. Talham. Thin Solid Films 1998, 327–329, 331–5; (b) G. E. Fanucci, D. R. Talham. Langmuir 1999, 15, 3289–95. [37] C. T. Seip, D. R. Talham Mat. Res. Bull. 1999, 34, 437–45. [38] (a) G. Cao, H. Lee, V. M. Lynch, T. E. Mallouk. Inorg. Chem. 1988, 27, 2781–5. (b) G. Cao, V. M. Lynch, J. S. Swinnea, T. E. Mallouk. Inorg. Chem. 1990, 29, 2112–7. [39] (a) S. G. Carling, P. Day, D. Visser, R. K. Kremer. J. Solid State Chem. 1993, 106, 111–9, (b) S. G. Carling, P. Day, D. Visser. J. Phys.: Condens. Matter. 1995, 7, L109–13, (c) S. G. Carling, P. Day, D. Visser. Inorg. Chem. 1995, 34, 3917–27, (d) D. Visser, S. G. Carling, P. Day, J. Deportes. J. Appl. Phys. 1991, 69, 6016–8. [40] G. E. Fanucci, J. Krzystek, M. W. Meisel, L.-C. Brunel, D. R. Talham. J. Am. Chem. Soc. 1998, 120, 5469–79. [41] C. T. Seip, H. Byrd, D. R. Talham. Inorg. Chem. 1996, 35, 3479–83. [42] M. A. Petruska, G. E. Fanucci, D. R. Talham. Chem. Mater. 1998, 10, 177–89. [43] M. A. Petruska, D. R. Talham. Chem. Mater. 1998, 10, 3673–82. [44] D. R. Talham, C. T. Seip, S. Whipps, G. E. Fanucci, M. A. Petruska, H. Byrd. Comments Inorg. Chem. 1997, 19, 133–51. [45] G. E. Fanucci, M. A. Petruska, M. W. Meisel, D. R. Talham. J. Solid State Chem. 1999, 145, 443–51. [46] M. Petruska, Ph.D. Thesis, University of Florida, 2000. [47] C. Dourthe, M. Izumi, C. Garrigou-Lagrange, T. Buffeteau, B. Desbat, P. Delhaes. J. Phys. Chem. 1992, 96, 2812–20. [48] T. Bjornholm, T. Hassenkam, N. Reittel. J. Mater. Chem. 1999, 9, 1975–90.

484

14 Magnetic Langmuir-Blodgett Films

[49] O. Kahn. Molecular magnetism : From molecular assembly to the devices (Eds. E. Coronado, P. Delhaes, D. Gatteschi, J.S. Miller) NATO. ASI. Series, Kluwer Acad. Publ., 1995, p 243–88. [50] S. Ferlay, T . Mallah, R. Ovahes, P. Veillet, M. Verdaguer. Nature 1995, 378, 701–3. [51] C. Mingotaud, C. Lafuente, J. Amiell, P. Delhaes. Langmuir 1999, 15, 289–92. [52] C. Lafuente, C. Mingotaud, P. Delhaes. Chem. Phys. Lett. 1999, 302, 523–7. [53] N. Kimizuka, T. Handa, I. Ichinose, T. Kunitake. Angew. Chem. Int. Ed. Engl. 1994, 33, 2843–5. [54] N. Kimizuka, T. Handa, T. Kunitake. Mol. Cryst. Liq. Cryst. 1996, 277, 139. [55] C. Mingotaud, C. Lafuente, C. Gomez-Garcia, S. Ravaine, P. Delhaes. Mol. Cryst. Liq. Cryst. 1999, 335, 349–58. [56] G. Romualdo-Torres, C. Mingotaud, P. Delhaes to be published. [57] R. Saliba, B. Agricole, C. Mingotaud, S. Ravaine. J. Phys. Chem. B 1999, 103, 9712–6. [58] (a) J. Le Bideau, C. Payen, P. Palvadeau, B. Bujoli. Inorg. Chem. 1994, 33, 4885–90; (b) J. LE Bideau, C. Payen, B. Bujoli, P. Palvadeau, J. Rouxel. J. Magn. Magn. Mater. 1995, 140–144, 1719–20; (c) B. Bujoli, O. Pena, P. Palvadeau, J. Le Bideau, C. Payen, J. Rouxel Chem. Mater. 1993, 5, 583–7. [59] M.J. Stead, P. Day. J. Chem. Soc. Dalton Trans. 1982, 1081–4. [60] P. Day. Phil. Trans. R. Soc. London 1985, A 314, 145–8. [61] H. Tanaki, Z.-J. Zhong, N. Matsumoto, S. Kida, M. Koikawa, N. Achiwa, Y. Hashimoto, H. Okawa. J. Am. Chem. Soc. 1992, 124, 6974–9. [62] C. Mathoniere, C.-J. Nuttall, S.G. Carling, P. Day. Inorg. Chem. 1996, 35, 1201–6. [63] S. Decurtis, R. Pellaux, G. Antorrena, F. Palacio. Mol Cryst. Liq. Cryst. 1999, 335, 173–81. [64] P. Rabu, S. Rouba, V. Laget, C. Hornick, M. Drillon. Chem. Com. 1996, 1107–8. [65] V. Laget, C. Hornick, P. Rabu, M. Drillon, P. Turek, R. Ziessel, Adv. Mater. 1998, 10, 1024–8. [66] M. Kurmoo. Chem. Mater. 1999, 11, 3370–8. [67] P.G. Lacroix, R. Clement, K. Nakatani, J. Zyss, J. Ledoux. Science 1994, 263, 658–60. [68] M. Drillon, P. Panissod. J. Magn. Magn. Mater 1988, 188, 93–8. [69] G. Decher. Science 1997, 277, 1232–7. [70] I. Ichinose, H. Tagawa, S. Mizuki, Y. Luov, T. Kunitake. Langmuir 1998, 14, 187–92. [71] (a) D.G. Kurth, P. Lehmann, D. Volkmer, H. Colfen, M.J. Koop, A. Muller, A. du Chesne. Chem. Eur. J. 2000, 6, 385–93; (b) D. Volkmer, A. du Chesne, D. G. Kurth, H. Schnablegger, P. Lehman, M.J. Koop, A. Muller. J. Am. Chem. Soc. 2000, 122, 1995–8. [72] Y. Einaga, O. Sato, T. Iyoda, A. Fujishima, K. Hashimoto. J. Am. Chem. Soc. 1999, 121, 3745–50. [73] Y. Guo, A.R. Guadalupe, O. Resto, L.F. Fonseca, S.P. Weisz. Chem. Mater. 1999, 11, 35–40. [74] J. H. Fendler, F. C. Meldrum. Adv. Mat. 1995, 7, 6007. [75] C.T. Seip, E.E. Carpenter, C.J. O’Connor. V.T. John, S. Li. IEEE transactions on Magnetics 1998, 34, 1111–3. [76] C.J. O’Connor, S. Li, M. Evain, J. Gareth, V.T. John. Mol Cryst. Liq. Cryst. 1999, 335, 95–108. [77] C. Petit, A. Taleb, M.P. Pileni. Adv. Mater. 1998, 10, 259–61. [78] C. Petit, A. Taleb, M.P. Pileni. J. Phys. Chem. B 1999, 103, 1805–10.