Lasers in the Conservation of Artworks: LACONA VI Proceedings, Vienna, Austria, Sept. 21--25, 2005 (Springer Proceedings in Physics)

springer proceedings in physics 116 springer proceedings in physics 96 Electromagnetics in a Complex World Editors: I...

Author: J. Nimmrichter

5 downloads 213 Views 18MB 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

springer proceedings in physics 116

springer proceedings in physics 96 Electromagnetics in a Complex World Editors: I.M. Pinto, V. Galdi, and L.B. Felsen 97 Fields, Networks, Computational Methods and Systems in Modern Electrodynamics A Tribute to Leopold B. Felsen Editors: P. Russer and M. Mongiardo 98 Particle Physics and the Universe Proceedings of the 9th Adriatic Meeting, Sept. 2003, Dubrovnik Editors: J. Trampeti´c and J. Wess 99 Cosmic Explosions On the 10th Anniversary of SN1993J (IAU Colloquium 192) Editors: J. M. Marcaide and K. W. Weiler 100 Lasers in the Conservation of Artworks LACONA V Proceedings, Osnabr¨uck, Germany, Sept. 15–18, 2003 Editors: K. Dickmann, C. Fotakis, and J.F. Asmus 101 Progress in Turbulence Editors: J. Peinke, A. Kittel, S. Barth, and M. Oberlack 102 Adaptive Optics for Industry and Medicine Proceedings of the 4th International Workshop Editor: U. Wittrock 103 Computer Simulation Studies in Condensed-Matter Physics XVII Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 104 Complex Computing-Networks Brain-like and Wave-oriented Electrodynamic Algorithms Editors: I.C. G¨oknar and L. Sevgi 105 Computer Simulation Studies in Condensed-Matter Physics XVIII Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 106 Modern Trends in Geomechanics Editors: W. Wu and H.S. Yu

108 Hadron Collider Physics 2005 Proceedings of the 1st Hadron Collider Physics Symposium, Les Diablerets, Switzerland, July 4–9, 2005 Editors: M. Campanelli, A. Clark, and X. Wu 109 Progress in Turbulence II Proceedings of the iTi Conference in Turbulence 2005 Editors: M. Oberlack, G. Khujadze, S. Guenther, T. Weller, M. Frewer, J. Peinke, S. Barth 110 Nonequilibrium Carrier Dynamics in Semiconductors Proceedings of the 14th International Conference, July 25–29, 2005, Chicago, USA Editors: M. Saraniti, U. Ravaioli 111 Vibration Problems ICOVP 2005 Editors: E. Inan, A. Kiris 112 Experimental Unsaturated Soil Mechanics Editor: T. Schanz 113 Theoretical and Numerical Unsaturated Soil Mechanics Editor: T. Schanz 114 Advances in Medical Engineering Editor: T.M. Buzug 115 X-Ray Lasers 2006 Proceedings of the 10th International Conference, August 20–25, 2006, Berlin, Germany Editors: P.V. Nickles, K.A. Janulewicz 116 Lasers in the Conservation of Artworks LACONA VI Proceedings, Vienna, Austria, Sept. 21–25, 2005 Editors: J. Nimmrichter, W. Kautek, M. Schreiner 117 Advances in Turbulence XI Proceedings of the 11th EUROMECH European Turbulence Conference, June 25–28, 2007, Porto, Portugal Editors: J.M.L.M. Palma and A. Silva Lopes

107 Microscopy of Semiconducting Materials Proceedings of the 14th Conference, April 11–14, 2005, Oxford, UK Editors: A.G. Cullis and J.L. Hutchison

Volumes 70–95 are listed at the end of the book.

J. Nimmrichter W. Kautek M. Schreiner (Eds.)

Lasers in the Conservation of Artworks LACONA VI Proceedings, Vienna, Austria, Sept. 21–25, 2005

With 419 Figures

123

Mag. Johann Nimmrichter Federal Ofﬁce for Care and Protection of Monuments Arsenal 15/4, 1030 Vienna, Austria E-mail: ofﬁ[email protected]

Professor Dr. Wolfgang Kautek University of Vienna, Department of Physical Chemistry W¨ahringer Str. 42, 1090 Vienna, Austria E-mail: [email protected]

Professor Dr. Manfred Schreiner Academy of Fine Arts Vienna, Institute of Science and Technology in Art Schillerplatz 3, 1010 Vienna, Austria E-mail: [email protected]

ISSN 0930-8989 ISBN 978-3-540-72129-1 Springer Berlin Heidelberg New York Library of Congress Control Number: 2007928748 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specif ically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microf ilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer-Verlag. Violations are liable to prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media. springer.com © Springer-Verlag Berlin Heidelberg 2007 The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specif ic statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Prodcution: SPI Publisher Services Cover design: eStudio Calamar Steinen Printed on acid-free paper

SPIN: 12041855

57/3180/SPi

-543210

Preface

Conservation and protection of works of art as well as of rare remnants of natural history has turned more and more into a race against time. Environments all over the world have become increasingly aggressive causing damage or at least deterioration to surfaces meant to be created for eternity. Conventional techniques do a lot against most of these dangers, but new approaches of high technology have to be explored to preserve the heritage of human civilization as well as the precious specimens of former life such as the feathers’ of birds which died out generations ago. Mechanical and chemical methods are involved in traditional conservation treatments. Contactless cleaning by lasers, on the other hand, is a new and prospering ﬁeld of laser materials processing. It allows avoiding mechanical disruption and the disadvantage of cleaning ﬂuids – may they be toxic or just water – which could cause potentially long-term degradation of the substrate or health hazards. Moreover, laser cleaning may have the potential to accelerate conservatory work with high quality and moderate costs, and, thus, may help archives’, museums’ and collections’ strained budgets. Laser cleaning in semiconductor, automotive and aerospace industries has already been motivated by cost-savings, yield enhancement, and environmental concerns so that substantial literature about laser processing and cleaning of technical surfaces has accumulated in scientiﬁc and technological journals in recent years. This wealth of knowledge and experience, however, is usually not accessible to the conservation, museum, and archiving community. Therefore, the series of the “International Conferences on Lasers in the Conservation of Artworks” – LACONA – was initiated by Costas Fotakis organizing LACONA I 1995 in Heraklion, Greece. This was followed by LACONA II 1997 in Liverpool, Great Britain, LACONA III 1999 in Florence, Italy, LACONA IV 2001 in Paris, France, and LACONA V 2003 in Osnabrück, Germany. The success of these unique conferences motivated the LACONA Permanent Scientiﬁc Committee to organize a LACONA VI – this time in the very heart of Europe, in Vienna, Austria.

VI

Preface

The general development in laser conservation has led to the observation that scientiﬁc approaches and diagnostics have been introduced in an extent as never before in conservation. The key issues of the state of the art and future developments of laser cleaning of artefacts turned out to be as sketched in the following. Paradigm Change of Conservation. Laser cleaning applies highly localized deposition of heat by a laser beam in contrast to traditional conservation involving both room-temperature mechanical and chemical methods. Advanced Chemical Analysis and Diagnostics. In addition to the inspection by the conservator’s eye, micromorphological and spectroscopic methods are increasingly employed. Inhomogeneity and Precision. The high-precision deliverance of laser radiation to morphologically and chemically inhomogeneous artefact surfaces allows an unprecedented treatment quality. Integration. Merging laser cleaning with complementary conventional restoration steps may provide unrivalled solutions. Automation. Laser precision processing can be highly automated allowing better precision, safety and cost-eﬀectiveness in the future. The 6th International Conference on Lasers in the Conservation of Artworks (LACONA VI) took place in Vienna, Austria, 21–25 September 2005. It represented the above listed new developments which entered the present proceedings volume. Moreover, LACONA VI ran under the auspices of the United Nations endorsed “World Year of Physics 2005” initiative which started by the European Physical Society to demonstrate that natural sciences provide a significant basis for the development of understanding nature, and that scientiﬁc research and its applications are a major driving force to scientiﬁc and technological development, and remain a vital factor in addressing the challenges of the 21st century. The “World Year of Physics 2005” highlighted the vitality of natural science and its importance in the coming millennium, and will commemorate the pioneering contributions of Albert Einstein in 1905. I want to thank Johann Nimmrichter, Chairman, and Manfred Schreiner, Co-Chairman of LACONA VI, for there unmatched enthusiasm and dedication to make LACONA VI a success. Further there has to be mentioned the invaluable support by the LACONA Permanent Scientiﬁc Committee, the LACONA Local Organizers (public institutions in Vienna), the LACONA Local Congress Committee, and last not least the LACONA Sponsors. Finally, I would like to thank Robert Linke and Ed Teppo for their careful and generous support during the preparation of the proceedings of LACONA VI. Vienna, May 2007

Wolfgang Kautek

Contents

List of Committees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVII List of Sponsors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XXI List of Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .XXIII 1 Serendipity, Punctuated J.F. Asmus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Part I Metal 2 Laser Cleaning of Corroded Steel Surfaces: A Comparison with Mechanical Cleaning Methods Y.S. Koh, J. Powell, A. Kaplan, and J. Carlevi . . . . . . . . . . . . . . . . . . . .

13

3 Laser Cleaning of Gildings M. Panzner, G. Wiedemann, M. Meier, W. Conrad, A. Kempe, and T. Hutsch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21

4 Current Work in Laser Cleaning of the Porta del Paradiso S. Agnoletti, A. Brini, and L. Nicolai . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29

5 Cleaning Historical Metals: Performance of Laser Technology in Monument Preservation A. Gervais, M. Meier, P. Mottner, G. Wiedemann, W. Conrad, and G. Haber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37

6 Laser Cleaning the Abergavenny Hoard: Silver Coins from the Time of William the Conqueror M. Davis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

45

VIII

Contents

Part II Stone 7 The Application of Laser Cleaning in the Conservation of Twelve Limestone Relief Panels on St. George’s Hall M. Cooper and S. Sportun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

8 The Potential Use of Laser Ablation for Selective Cleaning of Indiana Limestone K.C. Normandin, L. Powers, D. Slaton, and M.J. Scheﬄer . . . . . . . . .

65

9 Laser Cleaning of a Renaissance Epitaph with Traces of Azurite J. Nimmrichter and R. Linke . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

75

10 Laser Cleaning of Peristyle in Diocletian Palace in Split (HR) D. Almesberger, A. Rizzo, A. Zanini, and R. Geometrante . . . . . . . . . .

83

11 Phenomenological Characterisation of Stone Cleaning by Diﬀerent Laser Pulse Duration and Wavelength S. Siano, M. Giamello, L. Bartoli, A. Mencaglia, V. Parfenov, and R. Salimbeni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

87

12 The Cleaning of the Parthenon West Frieze by Means of Combined IR- and UV-Radiation K. Frantzikinaki, G. Marakis, A. Panou, C. Vasiliadis, E. Papakonstantinou, P. Pouli, T. Ditsa, V. Zaﬁropulos, and C. Fotakis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

97

13 A Comprehensive Study of the Coloration Eﬀect Associated with Laser Cleaning of Pollution Encrustations from Stonework P. Pouli, G. Totou, V. Zaﬁropulos, C. Fotakis, M. Oujja, E. Rebollar, M. Castillejo, C. Domingo, and A. Laborde . . . . . . . . . . . . . . . . . . . . . . .

105

14 Poultices as a Way to Eliminate the Yellowing Eﬀect Linked to Limestone Laser Cleaning V. Vergès-Belmin and M. Labouré . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

115

15 Experimental Investigations and Removal of Encrustations from Interior Stone Decorations of King Sigismund’s Chapel at Wawel Castle in Cracow A. Koss, J. Marczak, and M. Strzelec . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

125

16 Nd:YAG Laser Cleaning of Red Stone Materials: Evaluation of the Damage C. Colombo, E. Martoni, M. Realini, A. Sansonetti, and G. Valentini

133

Contents

IX

17 Exists a Demand for Nd:YAG Laser Technology in the Restoration of Stone Artworks and Architectural Surfaces? E. Pummer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

143

18 The SALUT Project: Study of Advanced Laser Techniques for the Uncovering of Polychromed Works of Art G. Van der Snickt, A. De Boeck, K. Keutgens, and D. Anthierens . . . .

151

Part III Inorganic Materials 19 Comparison of Wet and Dry Laser Cleaning of Artworks A. Sarzyński, K. Jach, and J. Marczak . . . . . . . . . . . . . . . . . . . . . . . . . . .

161

20 Laser Cleaning of Avian Eggshell L. Cornish, A. Ball, and D. Russell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

169

21 Removal of Strong Sinter Layers on Archaeological Artworks with Nd:YAG Laser J. Hildenhagen, K. Dickmann, and H.-G. Hartke . . . . . . . . . . . . . . . . . . .

177

22 From the Lab to the Scaﬀold: Laser Cleaning of Polychromed Architectonic Elements and Sculptures M. Castillejo, C. Domingo, F. Guerra-Librero, M. Jadraque, M. Martín, M. Oujja, E. Rebollar, and R. Torres . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

185

23 Integration of Laser Ablation Techniques for Cleaning the Wall Paintings of the Sagrestia Vecchia and Cappella del Manto in Santa Maria della Scala, Siena S. Siano, A. Brunetto, A. Mencaglia, G. Guasparri, A. Scala, F. Droghini, and A. Bagnoli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

191

24 Preliminary Results of the Er:YAG Laser Cleaning of Mural Paintings A. Andreotti, M.P. Colombini, A. Felici, A. deCruz, G. Lanterna, M. Lanfranchi, K. Nakahara, and F. Penaglia . . . . . . . . . . . . . . . . . . . . .

203

Part IV Organic Materials 25 Preliminary Results of the Er:YAG Laser Cleaning of Textiles, Paper and Parchment A. Andreotti, M.P. Colombini, S. Conti, A. deCruz, G. Lanterna, L. Nussio, K. Nakahara, and F. Penaglia . . . . . . . . . . . . . . . . . . . . . . . . .

213

X

Contents

26 Simultaneous UV-IR Nd:YAG Laser Cleaning of Leather Artifacts S. Batishche, A. Kouzmouk, H. Tatur, T. Gorovets, U. Pilipenka, V. Ukhau, and W. Kautek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

221

27 An Evaluation of Nd:YAG Laser-Cleaned Basketry in Comparison with Commonly Used Methods A. Elliott, A. Bezúr, and J. Thornton . . . . . . . . . . . . . . . . . . . . . . . . . . . .

229

28 Novel Applications of the Er:YAG Laser Cleaning of Old Paintings A. Andreotti, P. Bracco, M.P. Colombini, A. deCruz, G. Lanterna, K. Nakahara, and F. Penaglia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

239

29 A Final Report on the Oxidation and Composition Gradients of Aged Painting Varnishes Studied with Pulsed UV Laser Ablation C. Theodorakopoulos, V. Zaﬁropulos, and J.J. Boon . . . . . . . . . . . . . . . .

249

30 A New Solution for the Painting Artwork Rear Cleaning and Restoration: The Laser Cleaning S.E. Andriani, I.M. Catalano, A. Brunetto, G. Daurelio, and F. Vona . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

257

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints by Laser Irradiation at IR, VIS and UV Wavelengths M. Westergaard, P. Pouli, C. Theodorakopoulos, V. Zaﬁropulos, J. Bredal-Jørgensen, and U. Staal Dinesen . . . . . . . . . . . . . . . . . . . . . . . .

269

32 Traditional and Laser Cleaning Methods of Historic Picture Post Cards M. Mäder, H. Holle, M. Schreiner, S. Pentzien, J. Krüger, and W. Kautek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

281

33 Femtosecond Laser Cleaning of Painted Artefacts; Is this the Way Forward? P. Pouli, G. Bounos, S. Georgiou, and C. Fotakis . . . . . . . . . . . . . . . . . .

287

34 Laser Cleaning of Polyurethane Foam: An Investigation using Three Variants of Commercial PU Products U. Staal Dinesen and M. Westergaard . . . . . . . . . . . . . . . . . . . . . . . . . . . .

295

35 Excimer Laser Ablation of Egg Tempera Paints and Varnishes P.J. Morais, R. Bordalo, L. dos Santos, S.F. Marques, E. Salgueiredo, and H. Gouveia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

303

Contents

XI

36 Laser Cleaning of Undyed Silk: Indications of Chemical Change K. von Lerber, M. Strlic, J. Kolar, J. Krüger, S. Pentzien, C. Kennedy, T. Wess, M. Sokhan, and W. Kautek . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

313

37 Determination of a Working Range for the Laser Cleaning of Soiled Silk J. Krüger, S. Pentzien, and K. von Lerber . . . . . . . . . . . . . . . . . . . . . . . .

321

38 Laser Versus Conventional Cleaning Methods: Do the Costs Outweigh the Beneﬁts? P. van Dalen, R. Broere, and H.A. Aziz . . . . . . . . . . . . . . . . . . . . . . . . . .

329

Part V Analytical Techniques 39 Raman Spectroscopy: New Light on Ancient Artefacts P. Vandenabeele and L. Moens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

341

40 Pigment Identiﬁcation on “The Ecstasy of St. Theresa” Painting by Raman Microscopy D. Marano, M. Marmontelli, G.E. De Benedetto, I.M. Catalano, L. Sabbatini, and F. Vona . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

349

41 Colorimetry, LIBS and Raman Experiments on Renaissance Green Sandstone Decoration During Laser Cleaning of King Sigismund’s Chapel in Wawel Castle, Cracow, Poland A. Sarzynski, W. Skrzeczanowski, and J. Marczak . . . . . . . . . . . . . . . . . .

355

42 Non-Destructive Observation of the Laser Treatment Eﬀect on Historical Paper via the Laser-Induced Fluorescence Spectra K. Komar and G. Śliwiński . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

361

43 Eﬀects of LIBS Measurement Parameters on Wall Paintings Pigments Alteration and Detection R. Bruder, D. Menut, and V. Detalle . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

367

44 A Parametric Linear Correlation Method for the Analysis of LIBS Spectral Data E. Tzamali and D. Anglos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

377

45 Investigation on Painting Materials in “Madonna col Bambino e S. Giovannino” by Botticelli D. Bersani, P.P. Lottici, A. Casoli, M. Ferrari, S. Lottini, and D. Cauzzi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

383

XII

Contents

46 Laser-Induced Plasma Spectroscopy for the Analysis of Roman Ceramics Terra Sigillata A.J. López, G. Nicolás, M.P. Mateo, V. Piñón, and A. Ramil . . . . . . .

391

47 Laser-Induced Fluorescence Analysis of Protein-Based Binding Media A. Nevin, S. Cather, D. Anglos, and C. Fotakis . . . . . . . . . . . . . . . . . . . .

399

48 Applications of a Compact Portable Raman Spectrometer for the Field Analysis of Pigments in Works of Art S. Bruni and V. Guglielmi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

407

49 Classiﬁcation of Patinas Found on Surfaces of Historical Buildings by Means of Laser-Induced Breakdown Spectroscopy C. Vázquez-Calvo, A. Giakoumaki, D. Anglos, M. Álvarez de Buergo, and R. Fort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

415

50 Laser-Induced Breakdown Spectroscopy of Cinematographic Film M. Oujja, C. Abrusci, S. Gaspard, E. Rebollar, A. del Amo, F. Catalina, and M. Castillejo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

421

51 Online Monitoring of the Laser Cleaning of Marbles by LIBS Sulphur Detection V. Lazic, F. Colao, R. Fantoni, V. Spizzichino, and E. Teppo . . . . . . .

429

52 Low Resolution LIBS for Online-Monitoring During Laser Cleaning Based on Correlation with Reference Spectra M. Lentjes, K. Dickmann, and J. Meijer . . . . . . . . . . . . . . . . . . . . . . . . . .

437

53 Pigment Identiﬁcation on a XIV/XV c. Wooden Cruciﬁx Using Raman and LIBS Techniques M. Sawczak, G. Śliwiński, A. Kaminska, M. Oujja, M. Castillejo, C. Domingo, and M. Klossowska . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

445

54 MOLAB, a Mobile Laboratory for In Situ Non-Invasive Studies in Arts and Archaeology B.G. Brunetti, M. Matteini, C. Miliani, L. Pezzati, and D. Pinna . . . .

453

Part VI Scanning Techniques 55 From 3D Scanning to Analytical Heritage Documentation M. Schaich . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

463

Contents

XIII

56 Cleaning of Painted Surfaces and Examination of Cleaning by 3D-Measurement Technology at the August Deusser Museum, Zurzach P.-B. Eipper and G. Frankowski . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

473

57 Applicability of Optical Coherence Tomography at 1.55 µm to the Examination of Oil Paintings A. Szkulmowska, M. Góra, M. Targowska, B. Rouba, D. Stifter, E. Breuer, and P. Targowski . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

487

58 Varnish Thickness Determination by Spectral Optical Coherence Tomography I. Gorczyńska, M. Wojtkowski, M. Szkulmowski, T. Bajraszewski, B. Rouba, A. Kowalczyk, and P. Targowski . . . . . . . . . . . . . . . . . . . . . . . .

493

59 Multidimensional Data Analysis of Scanning Laser Doppler Vibrometer Measurements: An Application to the Diagnostics of Frescos at the US Capitol J. Vignola, J. Bucaro, J. Tressler, D. Ellingston, A. Kurdila, G. Adams, B. Marchetti, A. Agnani, E. Esposito, and E.P. Tomasini .

499

60 Spectral Domain Optical Coherence Tomography as the Proﬁlometric Tool for Examination of the Environmental Inﬂuence on Paintings on Canvas T. Bajraszewski, I. Gorczyńska, B. Rouba, and P. Targowski . . . . . . . .

507

61 Polish Experience with Advanced Digital Heritage Recording Methodology, including 3D Laser Scanning, CAD, and GIS Application, as the Most Accurate and Flexible Response for Archaeology and Conservation Needs at Jan III Sobieski’s Residence in Wilanów P. Baranowski, K. Czajkowski, M. Gładki, T. Morysiński, R. Szambelan, and A. Rzonca . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

513

62 Evaluation by Laser Micro-Proﬁlometry of Morphological Changes Induced on Stone Materials by Laser Cleaning C. Colombo, C. Daﬀara, R. Fontana, M.Ch. Gambino, M. Mastroianni, E. Pampaloni, M. Realini, and A. Sansonetti . . . . . . .

523

63 A Mobile True Colour Topometric Sensor for Documentation and Analysis in Art Conservation Z. Böröcz, D. Dirksen, G. Bischoﬀ, and G. von Bally . . . . . . . . . . . . . . .

527

64 Reconstruction of the Pegasus Statue on Top of the State Opera House in Vienna using Photogrammetry and Terrestrial and Close-Range Laser Scanning C. Ressl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

535

XIV

Contents

65 Some Experiences in 3D Laser Scanning for Assisting Restoration and Evaluating Damage in Cultural Heritage L.M. Fuentes, J. Finat, J.J. Fernández-Martin, J. Martínez, and J.I. SanJose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

543

66 Monitoring of Deformations Induced by Crystal Growth of Salts in Porous Systems Using Microscopic Speckle Pattern Interferometry G. Gülker, A. El Jarad, K.D. Hinsch, H. Juling, K. Linnow, M. Steiger, St. Brüggerhoﬀ, and D. Kirchner . . . . . . . . . . . . . . . . . . . . . .

553

67 Cultural Heritage Documentation by Combining Near-Range Photogrammetry and Terrestrial Laser Scanning: St. Stephen’s Cathedral, Vienna F. Zehetner and N. Studnicka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

561

68 Laser Engraving Gulf Pearl Shell – Aiding the Reconstruction of the Lyre of Ur C. Rawcliﬀe, M. Aston, A. Lowings, M.C. Sharp, and K.G. Watkins .

573

69 Fluorescence Lidar Multispectral Imaging for Diagnosis of Historical Monuments, Övedskloster: A Swedish Case Study R. Grönlund, J. Hällström, S. Svanberg, and K. Barup . . . . . . . . . . . . . .

583

R Measurement Technology for Use on Surfaces 70 OptoSurf of Historic Buildings and Monuments Cleaned by Laser W.P. Weinhold, A. Wortmann, C. Diegelmann, E. Pummer, N. Pascua, Th. Brennan, R. Burkhardt, and L. Goretzki . . . . . . . . . . . .

593

71 Multi-Tasking Non-Destructive Laser Technology in Conservation Diagnostic Procedures V. Tornari, E. Tsiranidou, Y. Orphanos, C. Falldorf, R. Klattenhof, E. Esposito, A. Agnani, R. Dabu, A. Stratan, A. Anastassopoulos, D. Schipper, J. Hasperhoven, M. Stefanaggi, H. Bonnici, and D. Ursu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

601

72 Time-Dependent Defect Detection by Combination of Holographic Tools E. Tsiranidou, V. Tornari, Y. Orphanos, C. Kalpouzos, and M. Stefanaggi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

611

Part VII Safety and Miscellaneous 73 Health Risks Caused by Particulate Emission During Laser Cleaning R. Ostrowski, St. Barcikowski, J. Marczak, A. Ostendorf, M. Strzelec, and J. Walter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

623

Contents

XV

74 Generation of Nano-Particles During Laser Ablation: Risk Assessment of Non-beam Hazards During Laser Cleaning St. Barcikowski, N. Bärsch, and A. Ostendorf . . . . . . . . . . . . . . . . . . . . .

631

75 A Novel Portable Multi-Wavelength Laser System A. Charlton and B. Dickinson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

641

List of Committees

Permanent Scientiﬁc Committee Prof. Dr. Wolfgang Kautek (President) University of Vienna Department of Physical Chemistry Waehringer Strasse 42 1090 Vienna, Austria E-mail: [email protected] Prof. Dr. John F. Asmus (Honorary President) IPAPS University of California, San Diego UCSD Physics Dept 9500 Gilman Drive La Jolla, CA 92093, USA E-mail: [email protected] Margaret Abraham Los Angeles County Museum of Art 5905 Wilshire Blvd Los Angeles, CA 99036, USA E-mail: [email protected] Prof. Dr. Giorgio Bonsanti Opiﬁcio Delle Pietre Dure di Firenze Via Alafani 78 50121 Firenze, Italy E-mail: [email protected] Dr. Marta Castillejo Consejo Superior de Investigaciones Cientiﬁcas Instituto de Química Física Rocasolano Serrano 119 28006 Madrid, Spain E-mail: [email protected]

XVIII

List of Committees

Dr. Martin Cooper The Conservation Centre Whitechapel Liverpool L1 6HZ, UK E-mail: [email protected] Prof. Dr. Klaus Dickmann Fachhochschule Münster Laserzentrum Stegerwaldstr. 39 48565 Steinfurt, Germany E-mail: [email protected] Prof. Dr. Costas Fotakis Foundation for Research and Technology – Hellas (FO.R.T.H.) Institute of Electronic Structure & Laser Vassilika Vouton, P.O. Box 1527 Heraklion 71110, Crete, Greece E-mail: [email protected] Prof. Dr. Eberhard Koenig Freie Universitaet Berlin Kunsthistorisches Institut Koserstrasse 20 14195 Berlin, Germany E-mail: [email protected] Dr. Mauro Matteini Opiﬁcio delle Pietre Dure di Firenze Laboratorio Scientiﬁco Viale Strozzi 1 50100 Firenze, Italy E-mail: [email protected] Mag. Johann Nimmrichter Federal Oﬃce for Care and Protection of Monuments (Bundesdenkmalamt) Department for Restoration and Conservation (Abteilung für Restaurierung und Konservierung) Arsenal, Objekt 15, Tor 4 1030 Wien, Austria E-mail: [email protected] Dr. Roxana Rãdvan National Institute of Research and Development for Optoelectronics (INOE) Centre for Restoration by Optoelectronical Techniques (CERTO) Platforma Magurele, 1 Atomistilor Str. 76900 Bucharest, Romania E-mail: [email protected]

List of Committees

Dr. Renzo Salimbeni Consiglio Nazionale delle Ricerche Istituto di Elettronica Quantistica Via Panciatichi 56/30 50127 Firenze, Italy E-mail: [email protected] Véronique Vergès-Belmin Laboratoire de Recherche des Monuments Historiques 29 rue de Paris 77420 Champs sur Marne, France E-mail: [email protected] Prof. h.c. Dr. Gert von Bally University of Münster Laboratory of Biophysics, Institute of Experimental Audiology Robert-Koch-Str. 45 48129 Münster, Germany E-mail: [email protected] Prof. Dr. Kenneth Watkins The University of Liverpool Department of Mechanical Engineering Liverpool, L69 3BX, UK E-mail: [email protected] Prof. Dr. Vassilis Zaﬁropulos Superior Technical Educational Institute of Crete Department of Human Nutrition & Dietetics Ioannou Kondylaki 46, 723 00 Sitia, Crete, Greece E-mail: zaﬁ[email protected]

Local Congress Committee Johann Nimmrichter Chairman, Bundesdenkmalamt, Vienna, Austria Manfred Schreiner Co-Chairman, Academy of Fine Arts, Vienna, Austria Wolfgang Kautek Co-Chairman, Dept. of Phys. Chem., Univ. of Vienna, Austria Wolfgang Baatz Academy of Fine Arts, Vienna, Austria Andrea Böhm Bundesdenkmalamt, Vienna, Austria Dimitrios Boulasikis Conservator-Archaelogist, Mödling, Austria

XIX

XX

List of Committees

Giancarlo Calcagno Conservator-Restorer, Bassano del Grappa, Italy Gabriele Gürtler Bundesdenkmalamt, Vienna, Austria Eva Maria Höhle Bundesdenkmalamt, Vienna, Austria Manfred Koller Bundesdenkmalamt, IIC-Austria, Vienna, Austria Gabriele Krist University of Applied Arts, Vienna, Austria Robert Linke Bundesdenkmalamt, Vienna, Austria Erich Pummer Conservator-Restorer, Rossatz, Austria Johannes Riegl RIEGL Laser Measurement Systems GmbH, Horn, Austria Dieter Schuöcker Vienna University of Technology, Vienna, Austria Christopher Weeks Conservator-Restorer, Tring, UK Robert Wimmer Behindscreen, Vienna, Austria Wolfgang Zehetner Dombaumeister, Architect of St. Stephens Cathedral, Vienna, Austria

Local Organizing Institutions Federal Oﬃce for Care and Protection of Monuments Austria (Bundesdenkmalamt) Academy of Fine Arts Vienna (Akademie der bildenden Künste) University of Vienna (Universität Wien) Vienna University of Technology (Technische Universität Wien) Cathedral Masons Lodge of St. Stephens, Vienna (Dombauhütte St. Stephan) International Institute for Conservation (IIC), Austrian Group Austrian Conservator-Restorer Association (Österreichischer Restauratorenverband)

List of Sponsors

The ﬁnancial support of all organisations is gratefully acknowledged. Riegl Laser Measurement Systems GmbH, www.riegl.com Bundesministerium für Bildung, Wissenschaft und Kultur, www.bmbwk.gv.at Bundesdenkmalamt, www.bda.at Akademie der bildenden Künste, www.akbild.ac.at Dr. Michael Häupl, Mayor of Vienna, www.wien.gv.at Casinos Austria, www.casinos.at COST G7 Artwork conservation by laser, http://alpha1.inﬁm.ro/cost Bundeskammer der Architekten und Ingenieurskonsulenten, www.arching.at Linsinger Kulturgutvermessung, Photogrammetrie, 3D Scanning, www.linsinger.at oﬁ – Technologie & Innovation GmbH, Abteilung Bauwesen, www.oﬁ.co.at ELEN GROUP hightech laser, www.elengroup.com Quanta Systems S.p.A. Lasers & Lasersystems, www.quantasystem.com Rest. Felix Mackowitz, [email protected] Rest. Mag. Klaus Wedenig, info@denkmalpﬂegegmbh.at Rest. Mag. Ralph Kerschbaumer, [email protected] Rest. Erich Pummer, www.lasertech-artcons.at Steinmetzﬁrma Wolfgang Ecker, [email protected] Rest. Otto Blassnig, [email protected] Steinmetzﬁrma Rada, www.rada.at Rest. Johann Lindtner, [email protected]

XXII

List of Sponsors

Steinmetzﬁrma Johann Schaden, www.marmorbau-schaden.at Rest. Johannes Schlögl, [email protected] Rest. Mag. Josef Weninger, [email protected] Landesinnung Wien der Steinmetzmeister, [email protected] Rest. Gerhard Zottmann, www.zottmann.at Rest. Werner Campidell, [email protected] Arctron Ausgrabungen & Computerdokumentationen GmbH, www.arctron.de

List of Contributors

Abrusci, C., 421 Adams, G., 499 Agnani, A., 499, 601 Agnoletti, S., 29 Almesberger, D., 83 Álvarez de Buergo, M., 415 Anastassopoulos, A., 601 Andreotti, A., 203, 213, 239 Andriani, S.E., 257 Anglos, D., 377, 399, 415 Anthierens, D., 151 Asmus, J.F., 1 Aston, M., 573 Aziz, H.A., 329 Bagnoli, A., 191 Bajraszewski, T., 493, 507 Ball, A., 169 Bärsch, N., 631 Baranowski, P., 513 Barcikowski, St., 623, 631 Bartoli, L., 87 Barup, K., 583 Batishche, S., 221 Bersani, D., 383 Bezúr, A., 229 Bischoﬀ, G., 527 Böröcz, Z., 527 Bonnici, H., 601 Boon, J.J., 249 Bordalo, R., 303 Bounos, G., 287 Brüggerhoﬀ, St., 553

Bracco, P., 239 Bredal-Jørgensen, J., 269 Brennan, Th., 593 Breuer, E., 487 Brini, A., 29 Broere, R., 329 Bruder, R., 367 Brunetti, B.G., 453 Brunetto, A., 191, 257 Bruni, S., 407 Bucaro, J., 499 Burkhardt, R., 593 Carlevi, J., 13 Casoli, A., 383 Castillejo, M., 105, 185, 421, 445 Catalano, I.M., 257, 349 Catalina, F., 421 Cather, S., 399 Cauzzi, D., 383 Charlton, A., 641 Colao, F., 429 Colombini, M.P., 203, 213, 239 Colombo, C., 133, 523 Conrad, W., 21, 37 Conti, S., 213 Cooper, M., 55 Cornish, L., 169 Czajkowski, K., 513 Dabu, R., 601 Daﬀara. C., 523 Daurelio, G., 257 Davis, M., 45

XXIV

List of Contributors

De Benedetto, G.E., 349 De Boeck, A., 151 deCruz, A., 203, 213, 239 del Amo, A., 421 Detalle, V., 367 Dickinson, B., 641 Dickmann, K., 177, 437 Diegelmann, C., 593 Dirksen, D., 527 Ditsa, T., 97 Domingo, C., 105, 185, 445 dos Santos, L., 303 Droghini, F., 191 Eipper, P.-B., 473 El Jarad, A., 553 Ellingston, D., 499 Elliott, A., 229 Esposito, E., 499, 601 Falldorf, C., 601 Fantoni, R., 429 Felici, A., 203 Fernández-Martin, J.J., 543 Ferrari, M., 383 Finat, J., 543 Fontana, R., 523 Fort, R., 415 Fotakis, C., 97, 105, 287, 399 Frankowski, G., 473 Frantzikinaki, K., 97 Fuentes, L.M., 543 Gambino, M. Ch., 523 Gaspard, S., 421 Geometrante, R., 83 Georgiou, S., 287 Gervais, A., 37 Giakoumaki, A., 415 Giamello, M., 87 Gładki, M., 513 Góra, M., 487 Gorczyńska, I., 493, 507 Goretzki, L., 593 Gorovets, T., 221 Gouveia, H., 303 Grönlund, R., 583 Gülker, G., 553 Guasparri, G., 191

Guerra-Librero, F., 185 Guglielmi, V., 407 Haber, G., 37 Hällström, J., 583 Hartke, H.-G., 177 Hasperhoven, J., 601 Hildenhagen, J., 177 Hinsch, K.D., 553 Holle, H., 281 Hutsch, T., 21 Jach, K., 161 Jadraque, M., 185 Juling, H., 553 Kalpouzos, C., 611 Kaminska, A., 445 Kaplan, A., 13 Kautek, W., 221, 281, 313 Kempe, A., 21 Kennedy, C., 313 Keutgens, K., 151 Kirchner, D., 553 Klattenhof, R., 601 Klossowska, M., 445 Koh, Y.S., 13 Kolar, J., 313 Komar, K., 361 Koss, A., 125 Kouzmouk, A., 221 Kowalczyk, A., 493 Krüger, J., 313, 321 Krüger, J/, 281 Kurdila, A., 499 Laborde, A., 105 Labouré, M., 115 Lanfranchi, M., 203 Lanterna, G., 203, 213, 239 Lazic, V., 429 Lentjes, M., 437 Linke, R., 75 Linnow, K., 553 López, A.J., 391 Lottici, P. P., 383 Lottini, S., 383 Lowings, A., 573 Mäder, M., 281 Marakis, G., 97

List of Contributors Marano, D., 349 Marchetti, B., 499 Marczak, J., 125, 161, 355, 623 Marmontelli, M., 349 Marques, S.F., 303 Martínez, J., 543 Martín, M., 185 Martoni, E., 133 Mastroianni, M., 523 Mateo, M. P., 391 Matteini, M., 453 Meier, M., 21, 37 Meijer, J., 437 Mencaglia, A., 87, 191 Menut, D., 367 Miliani, C., 453 Moens, L., 341 Morais, P.J., 303 Morysiński, T., 513 Mottner, P., 37 Nakahara, K., 203, 213, 239 Nevin, A., 399 Nicolás, G., 391 Nicolai, L., 29 Nimmrichter, J., 75 Normandin, K.C., 65 Nussio, L., 213 Orphanos, Y., 601, 611 Ostendorf, A., 623, 631 Ostrowski, R., 623 Oujja, M., 105, 185, 421, 445 Pampaloni, E., 523 Panou, A., 97 Panzner, M., 21 Papakonstantinou, E., 97 Parfenov, V., 87 Pascua, N., 593 Penaglia, F., 203, 213, 239 Pentzien, S., 281, 313, 321 Pezzati, L., 453 Piñón, V., 391 Pilipenka, U., 221 Pinna, D., 453 Pouli, P., 97, 105, 269, 287 Powell, J., 13

Powers, L., 65 Pummer, E., 143, 593 Ramil, A., 391 Rawcliﬀe, C., 573 Realini, M., 133, 523 Rebollar, E., 105, 185, 421 Ressl, C., 535 Rizzo, A., 83 Rouba, B., 487, 493, 507 Russell, D., 169 Rzonca, A., 513 Śliwiński, G., 361, 445 Sabbatini, L., 349 Salgueiredo, E., 303 Salimbeni, R., 87 SanJose, J.I., 543 Sansonetti, A., 133, 523 Sarzyński, A., 161 Sarzynski, A., 355 Sawczak, M., 445 Scala, A., 191 Schaich, M., 463 Scheﬄer, M.J., 65 Schipper, D., 601 Schreiner, M., 281 Sharp, M.C., 573 Siano, S., 87 Skrzeczanowski, W., 355 Slaton, D., 65 Sokhan, M., 313 Spizzichino, V., 429 Sportun, S., 55 Staal Dinesen, U., 269, 295 Stefanaggi, M., 601, 611 Steiger, M., 553 Stifter, D., 487 Stratan, A., 601 Strlic, M., 313 Strzelec, M., 125, 623 Studnicka, N., 561 Svanberg, S., 583 Szambelan, R., 513 Szkulmowska, A., 487 Szkulmowski, M., 493 Targowska, M., 487 Targowski, P., 487, 493, 507

XXV

XXVI

List of Contributors

Tatur, H., 221 Teppo, E., 429 Theodorakopoulos, C., 249, 269 Thornton, J., 229 Tomasini, E.P., 499 Tornari, V., 601, 611 Torres, R., 185 Totou, G., 105 Tressler, J., 499 Tsiranidou, E., 601, 611 Tzamali, E., 377 Ukhau, V., 221 Ursu, D., 601 Valentini, G., 133 van Dalen, P., 329 Van der Snickt, G., 151 Vandenabeele, P., 341 Vasiliadis, C., 97

Vázquez-Calvo, C., 415 Vergès-Belmin, V., 115 Vignola, J., 499 von Bally, G., 527 von Lerber, K., 313, 321 Vona, F., 257, 349 Walter, J., 623 Watkins, K.G., 573 Weinhold, W.P., 593 Wess, T., 313 Westergaard, M., 269, 295 Wiedemann, G., 21, 37 Wojtkowski, M., 493 Wortmann, A., 593 Zaﬁropulos, V., 97, 105, 249, 269 Zanini, A., 83 Zehetner, F., 561

1 Serendipity, Punctuated J.F. Asmus Institute for Pure and Applied Physical Sciences University of California San Diego 9500 Gilman Dr., La Jolla, CA 92093-0360, USA [email protected] Summary. Laser divestment entered the ﬁeld of art conservation through a nonlinear sequence of positive accidental events (serendipity) that involved the cinema industry, the invention of spread-spectrum and frequency-hopping communications, nuclear space propulsion, and oceanography. The unlikely chain of events began with the invention of a secure military communications system by a Viennese motion picture actress (1942). A ﬁrst evaluation of the novel communications concept took place during a high-altitude nuclear test (TEAK) over the Paciﬁc Ocean in 1958. The secure radio link proved to be a failure; however, analyses of the backscattered electromagnetic radiation contributed to the realization that nuclear-explosion plasmas need not be spherically symmetrical. Nobel Laureate Freeman Dyson exploited this nuclear option to guide in the design and prototype development of the ORION spaceship that was to rendezvous with the planet Saturn in 1970. For this space vehicle the high-speciﬁc-impulse nuclear propulsion was generated by means of superradiant X-ray-beam ablation of the spaceship’s rear surface by the remote detonation of a sequence of asymmetrical bombs projected rearward from the ORION. In the wake of the Nuclear Test Ban Treaty (1962) ORION was canceled. Through a Scripps Institution of Oceanography project in Venice (involving ORION scientists and holographic technology) the nondestructive radiation-ablation process found a resurrection in the ﬁeld of stone conservation (1972). Ironically, the ﬁrst major art-conservation project to employ laser ablation (Porta della Carta of the Palazzo Ducale) was paid for in part by Warner Brothers Motion Picture Studios (1980). Finally, the “Venice Laser Statue Cleaner” followed the Viennese actress (Hedy Lamarr/Hedwig Eva Maria Kiesler) to Hollywood where it was employed to treat the granite veneer of the Warner Center (1981).

1.1 Introduction The ﬁelds of art conservation and laser science merged, formally and ﬁttingly, in the land of Polyclitus and Democritus with a 1995 event now called LACONA I (held at FORTH). However, appropriate that symbolic recognition of the sources of Western cultural heritage may seem, LACONA VI has, in Vienna, returned to the direct technological genesis of lasers in the service

2

J.F. Asmus

of art. The implausible trajectory of “unintended consequences” that led to the introduction of laser technology into art conservation was triggered in 1941– 1942 when Viennese cinema actress Hedy Lamarr invented a novel (jamming proof) concept for the radio transmission of guidance information to naval torpedoes. Subsequent decades witnessed initial evaluations of the Lamarr modulation schemes that helped uncover new avenues in nuclear weapons design as well as in the invention of the nuclear-propelled spaceship (ORION). Subsequently, the holographic plasma diagnostics developed for the engineering design of the spaceship were applied to the in situ archival recording of deteriorating Venetian statuary. This, in turn, led to the improbable realization that the radiation-propulsion mechanism of ORION could provide a means of selflimiting divestment (and conservation) of crumbling marble statues. The series of “connections” and happy accidents that helped in bringing about LACONA VI are summarized in the paragraphs that follow.

1.2 Hedy Lamarr and Her Communications Patent In 1942 Viennese motion picture actress Hedy Lamarr (Figs. 1.1 and 1.2) (Hedwig Eva Maria Kiesler) of MGM was granted US Patent #2,292,387 for a “Secret Communication System” based on her invention of spread-spectrum (Figs. 1.3 and 1.4) and frequency-hopping concepts. Evidently, the idea was a merging of art and science in that it sprang from her knowledge of the military business of her husband, Fritz Mandl, and her understanding of the player piano (gained from her friendship with artist George Antheil). As her discovery formed the basis of cell phone technology, Wi-Fi protocols, and the wireless Internet, she won a US$1/4M infringement claim against Corel Corporation and received the 1997 Electronic Frontier Award. (Upon receiving the award, 55 years after the fact, her response, “It’s about time,” received almost as much notice as her “au naturel” appearance in the 1933 Czech ﬁlm, “Ecstasy.”)

Fig. 1.1. MGM motion picture star Hedy Lamarr

1 Serendipity, Punctuated

3

Fig. 1.2. The ﬁrst page of Hedy Lamarr’s 1942 patent, “A Secret Communication System,” that introduced the frequency-hopping and spread-spectrum concepts to the communications ﬁeld

Fig. 1.3. Spread-spectrum communication link of Project ARGUS during the highaltitude nuclear detonation, TEAK (inset)

Fig. 1.4. The receiver site on the island of Niihau, Hawaii

The ﬁrst evaluation of Hedy Lamar’s approach to secure communications was carried out between Hawaiian Paciﬁc Islands in 1958 during the Johnston Island high-altitude nuclear explosion TEAK (3.8 MT at 77 km altitude). Disappointingly, the experimental radio-wave transmission link was completely blacked out by the bomb’s gamma-ray-induced aurora. However, spectral

4

J.F. Asmus

analyses of the backscattered electromagnetic signal revealed that the H-bomb had, through a performance asymmetry, ejected a plasma jet.

1.3 Orion: Nuclear Spaceship The ARGUS backscatter data together with other theoretical and experimental results predicted that nuclear explosive devices possessed the potential for being redesigned into directed-energy radiation sources. Upon this realization, members of the TEAK team joined with theoretical physicist Freeman Dyson and virtuoso minibomb designer Theodore Taylor to exploit and optimize this phenomenon in order to develop a nuclear-propelled spaceship, ORION, for a mission to the planet Saturn (scheduled for 1970). Following a ﬁrst ORION test ﬂight (1962), the adoption of the Nuclear Test Ban Treaty led to the demise of the program. Figures 1.5–1.7 display a few of the test results of laboratory proof-of-principle ORION technology demonstrations that reveal the impulse delivered by laser ablation.

Fig. 1.5. Deformation of a restrained metallic coin through the impulse delivered by laser ablation pressure at a multigigawatt and kilojoule level

Fig. 1.6. A streak camera record of the laser propulsion of an unrestrained metallic disk to V = 20 km s−1

1 Serendipity, Punctuated

5

Fig. 1.7. Hypervelocity impact crater (and its cross section) produced by the energy released by a laser-propelled projectile

Fig. 1.8. A conceptual portrayal of a nuclear-driven ORION spaceship

Figure 1.8 presents a conceptual rendering of the ORION space vehicle near Mars.

1.4 Laser Divestment in Venice The real-time holographic diagnostics developed for ORION were resurrected for art-conservation purposes in Venice in January 1972. This was a consequence of a collaboration of Scripps Institution of Oceanography geophysicists and ORION project alumni in research directed toward the alleviation of the “acqua alta” problem being experienced by the lagoon. By March 1972 the ruby holographic laser was being employed to clean marble sculpture by means of radiation-induced ablation in accordance with results from the radiationhydrodynamic modeling of the earlier X-ray-beam nuclear-propulsion system (Fig. 1.9). This came about at the suggestion of Lorenzo Lazzarini and Giulia Musumeci of the Venetian Soprintendenza in response to the unacceptable cleaning results on friable stone with conventional air-abrasive and chemical approaches (Fig. 1.10).

6

J.F. Asmus

Fig. 1.9. Arch. Calcagno’s ﬁrst in situ ruby-laser cleaning demonstration on one of the “Ruskin Capitals” of the Palazzo Ducale in Venice)

Fig. 1.10. Unsuccessful conventional cleaning test areas on a Piazza San Marco facade

This preamble to the formation of LACONA (more than a decade later by scientists at FORTH) came full circle with the entrance of an MGM competitor, Warner Brothers Films. At a 1977 Venice Film Festival event (a showing of “Clockwork Orange” in Asolo), Jack Warner, Jr. oﬀered to divert a portion of his proﬁts from “Clockwork Orange” to pay for a laser restoration feasibility project. He, together with his friends and associates, raised US$ 5,000 toward this end. Arch. Giancarlo Calcagno of the Soprintendenza selected the artwork to be the subject of the ﬁrst laser cleaning demonstration. The piece he selected in the Porta della Carta was a marble relief depicting “The Last Supper.” It was approximately 60 cm high and 180 cm wide. After a protracted sequence of laser validation tests in the laboratory, the actual laser demonstration took place in 1980 when an Nd:YAG laser was used to clean the marble relief in support of the overall Porta della Carta conservation eﬀort. As the relief had been laid horizontally on its back for treatment, the laser was mounted on a beam above the artwork. The laser beam was directed vertically downward to impinge on the marble surface. The laser head was attached to the supporting beam with a swivel joint so that the laser beam

1 Serendipity, Punctuated

7

Fig. 1.11. Porta della Carta of the Palazzo Ducale in Venice and a vertically mounted laser cleaning the marble relief of “The Last Supper”

Fig. 1.12. The central, cleaned area of the marble piece

could be manually scanned across the surface. The laser functioned in the normal mode (400 µs) at 1 J per pulse. In most areas a spot size of 3 mm was employed. Figure 1.11 shows the Porta della Carta and the laser pointing downward onto the relief. The initial, centrally cleaned area is shown in Fig. 1.12.

1.5 Return to Hollywood and the Cinema A new Warner Brothers corporate oﬃce complex was constructed in the Los Angeles area while the laser work proceeded in Venice. The following year (1981), with the completion of the central corporate tower of the Warner Center, the general contractor found that rubber cushioning used during shipping had discolored the tower’s South African granite veneer slabs. Chemical treatments that removed the stains also etched the stone and left it with a frosted appearance. Figure 1.13 shows the central tower of the Warner Center complex. The dark vertical stripes, as well as the horizontal bands at the top and bottom, are the South African granite veneers.

8

J.F. Asmus

Fig. 1.13. Warner center tower with the prominent granite veneer bands and vertical stripes

Fig. 1.14. A detail of laser treatment (bleaching) of the stained granite

As a last resort, the “Venetian” laser was sent to the Warner Tower for a cleaning trial. At low ﬂuxes and high ﬂuences, the laser-induced optical damage in the mineral grains of the stone. The resulting cleavages within the mineral grains resembled the normal heterogeneity of granite, yet masked the in-depth chemical blemish. This approach was selected as the most suitable treatment. Consequently, good fortune made one further appearance when the “Venetian” laser repaid a debt to the cinema industry by removing the blemishes from the exterior granite of the corporate center after the failure of chemical cleaning techniques. Figure 1.14 shows the results of the laser irradiation of the blemished granite veneer.

1.6 Conclusions Histories of developments in science and technology are replete with instances of unintended and/or unanticipated consequences. Sometimes such surprises are favorable. Often they bode disaster. All of the earliest pioneers of the laser have expressed bemusement at the laser’s entry into the ﬁeld of art conservation practice (as well as its ubiquitous role in the worlds of the audio

1 Serendipity, Punctuated

9

CD and the video DVD). Most certainly, that occurrence is an “untended consequence” of investigations into the very diverse ﬁelds of spread-spectrum communications, deep-space nuclear propulsion, holographic plasma diagnostics, and archival holographic recording. In retrospect it is clear that laser surface divestment would have found its way into the ﬁeld of art conservation at some point. However, the route that did lead initially to the laser in the arts is a testimonial to the tenacious punctuality with which serendipity invades the circuitry of technological progress. This individual route to innovation, beginning, and then returning to Vienna, is one more example (in a myriad of examples) demonstrating that discovery seldom proceeds in a linear and predictable fashion. Acknowledgments Herbert York, Theodore Taylor, Morris Scharﬀ, Edward Creutz, Keith Boyer, Donald Adrian, and Robert Willis are thanked for their oral histories of the ORION and ARGUS Projects. Charles Townes, Ali Javan, Gordon Gould, William Bridges, and Arthur Schawlow are thanked for their oral histories of the invention of the laser. Lorenzo Lazzarini and Giulia Musumeci are thanked for ﬁrst suggesting the laser cleaning of stone sculpture. The author’s science teacher at Chaﬀey High School (Mr. Sweihardt) is thanked for arranging summer employment with Project ARGUS at the US Naval Ordnance Laboratory. Much of the material in the above discourse was not published in a timely manner as an earlier manuscript (1978) was rejected (without review) by the Editor of Studies in Conservation as he deemed laser-divestment cleaning “too hypothetical to be taken seriously.” Many of the technical, political, and ﬁnancial barriers to progress in the activities outlined above were surmounted through the intervention of the Viennese/American, Walter Munk (Director Emeritus, Institute for Geophysics and Planetary Physics, Scripps Institution of Oceanography at the University of California San Diego).

This page intentionally blank

Part I

Metal

2 Laser Cleaning of Corroded Steel Surfaces: A Comparison with Mechanical Cleaning Methods Y.S. Koh1 , J. Powell2 , A. Kaplan2 , and J. Carlevi2 1

2

Kiruna Center for Conservation of Cultural Property, Arent Grapegatan 20, 98132 Kiruna, Sweden, [email protected] Luleå University of Technology, SE-971 87 Luleå, Sweden

Summary. Conservation often requires the removal of oxide layers from metal artifacts and new cleaning methods are being developed all the time. This paper provides a quantitative comparison of eight cleaning methods, three of which are mechanical (brushing or micro-blasting with Al2 O3 or glass beads) and ﬁve of which are laser dependent (TEA CO2 or Nd:YAG laser, with or without surface water). Surface proﬁlometry and scanning electron microscopy have been used to compare the cleaned surfaces with the original, known, surface geometries.

2.1 Introduction For the purposes of the conservation of metal artifacts it is often necessary to remove surface oxide layers without damaging the metal below. Various mechanical and chemical surface treatments are available for the removal of surface corrosion and other contaminants [1], but these can damage the underlying metal. Consequently, there is a great deal of interest in developing a new cleaning technology which is less aggressive to the metal surface below the oxide layer. Over the past few years the traditional cleaning methods employed by conservators have been augmented by treatments involving the use of lasers [2, 3]. Laser cleaning is a non-toxic, environmentally friendly and non-contact process which can be carefully controlled to minimise damage to the metal beneath a corroded surface. For this reason this process has potential advantages over traditional chemical or mechanical cleaning methods. This paper directly compares the relative eﬀectiveness of eight methods of cleaning a corroded steel surface (listed in Table 2.1).

2.2 Experimental Work To avoid use of actual historic artefacts and also to quantify the relative performance of the cleaning techniques, it was decided to produce a large

14

Y.S. Koh et al. Table 2.1. A list of the cleaning techniques compared in this study Mechanical methods

Laser-based methods

Micro-blasting with glass TEA CO2 laser (10,600 nm) Micro-blasting with Al2 O3 Nd:YAG laser (1,064 nm) on a dry surface Rotating steel brush Nd:YAG laser (1,064 nm) on a wet surface Nd:YAG laser (532 nm) on a dry surface Nd:YAG laser (532 nm) on a wet surface

Fig. 2.1. A typical cross section of a grooved sample

number of similar samples with known surface topology and corrosion levels. The cleaned surfaces could then be compared with each other and with a set of machined but uncorroded reference samples. The material used to produce the samples was a carbon steel (SS 1672) with the chemical formulation Fe 98.7%, C 0.47%, Si 0.25% and Mn 0.60%. The samples used in this work were machined to have a surface covered in parallel grooves with a cross section of the type shown in Fig. 2.1. The depth of the grooves was 0.25, 0.5 or 2.0 mm. The sample size was 30 × 30 × 13 mm. The samples were degreased with acetone before being systematically corroded in a corrosion-chamber. The samples were corroded for 3, 5 or 7 weeks by exposure to 0.1 M NaCl solution, which was sprayed on the samples twice a day.

2.3 Results Figure 2.2 shows the results of the cleaning trials for lightly corroded 0.5 mm grooves (i.e. peak-to-peak distance of 1 mm, peak height of 0.5 mm). The results have been arranged with the most eﬀective cleaning method towards the top of the ﬁgure and the least eﬀective towards the bottom (the uncorroded reference sample is shown ﬁrst and the uncleaned, corroded reference sample is shown at the bottom of the ﬁgure). After reviewing a number of statistical roughness comparisons, it was found that a simple comparison of the range of groove amplitude was the most eﬀective. Although corrosion and cleaning had little eﬀect on the maximum groove amplitude, it had a considerable eﬀect on the minimum amplitude measured where the original groove peak or trough would have been located. This location is easy to identify as the pitch of the grooves is known. This

2 Laser Cleaning of Corroded Steel Surfaces Pictures

Optical profilometry

Reference no corrosion

2mm Microblast, Al2O3 Microblast, glass

Amplitude Range (mm) 0.47 – 0.46 = 0.01 0.45 – 0.42 = 0.03 0.45 – 0.42 = 0.03

Nd:YAG laser 532 nm, wet

0.45 – 0.42 = 0.03

Nd:YAG laser 1064 nm, dry

0.42 – 0.37 = 0.05

Nd:YAG laser 1064 nm, wet

0.45 – 0.42 = 0.03

Rotating stainless steel brush

0.45 – 0.42 = 0.03

TEA CO2 laser 10600 nm

0.45 – 0.35 = 0.10

Nd:YAG laser 532 nm, dry

0.41 – 0.24 = 0.17

Reference corrosion

15

0.46 – 0.22 = 0.24

Fig. 2.2. A comparison of cleaning results for the 0.5 mm deep grooves and light corrosion

reduction in groove amplitude is either due to the corrosion products ﬁlling the bottom of the grooves or the corrosion process removing the peaks of the grooves. The groove amplitude measurements give only a rough guide to the eﬀectiveness of the cleaning method and this is made clear by comparison of the micro-blast Al2 O3 and the stainless steel brush results in Fig. 2.2. Although

16

Y.S. Koh et al.

the groove amplitude measurements are identical for both samples, it is obvious that the original groove proﬁle has been retained only in the case of the micro-blast sample. The steel brushing method has, by its nature, eroded the upper surface of the grooves giving the grooves a sharper peak than the original machined proﬁle. Using this combination of groove amplitude measurements and visual assessment of the proﬁles, it was possible to rank the eﬀectiveness of the eight cleaning methods for the 0.5 and 0.25 mm deep grooves for light, moderate and severe corrosion. The results of this grading procedure are presented in Table 2.2. By allocating points to each process in Table 2.2 depending on their performance in each case, it is possible to give an overall performance ranking and this is presented in Table 2.3. In the case of the moderately corroded samples there is not much difference between the proﬁle for the two least eﬀective cleaning methods and the uncleaned reference sample. Only the two micro-blast methods and the Nd:YAG – wet surface techniques reveal a pattern of regular sharp points. In the case of the heavily corroded samples the situation is even worse. Here, only the micro-blasted specimens give us any useful information about the original surface topology. An appropriate analogy here is that of a signal to noise ratio: The regular cycle of the original, uncorroded surface can be considered a signal and the corrosion process can be assumed to be obscuring this signal with random surface features or ‘noise’. The cleaning process attempts to

Table 2.2. A comparison of eﬀectiveness of the eight cleaning methods for diﬀerent grooves and corrosion conditions Light corrosion 0.5 mm 0.25 mm Groove Groove Best

Worst

A B C D E F G H

A B C D E G F H

Moderate corrosion 0.5 mm 0.25 mm Groove Groove A B E C D H G F

A B C E H G F D

Heavy corrosion 0.5 mm 0.25 mm Groove Groove A B C E F D H G

A B E C D H G F

A: Micro-blast Al2 O3 , (2 bar, Ø 0.050–0.075 mm), B: Micro-blast glass beads (6 bar, Ø 0.075–0.15 mm), C: Nd:YAG laser 532 nm per wet (pulse duration 10 nS, repetition rate 2.5 Hz, maximum pulse energy 300 mJ), D: Nd:YAG laser 1,064 nm per dry (pulse duration 10 nS, repetition rate 2.5 Hz, maximum pulse energy 600 mJ), E: Nd:YAG laser 1,064 nm per wet, F: Stainless steel brush (5,000 revolutions per min), G: TEA CO2 laser 10,600 nm (pulse duration 100 nS, repetition rate of 20 Hz, maximum pulse energy 4 J), H: Nd:YAG laser 532 nm per dry

2 Laser Cleaning of Corroded Steel Surfaces

17

Table 2.3. An overall ranking of the eight cleaning methods Ranking Best

Worst

1st 2nd 3rd 4th 5th 6th 7th 8th

Method Micro-blast – Al2 O3 Micro-blast – Glass Nd:YAG 532 nm – Wet Nd:YAG 1064 nm – Wet Nd:YAG 1064 nm – Dry Nd:YAG 532 nm – Dry Stainless steel brush TEA CO2 laser

remove the noise and re-identify the signal. Figure 2.3 shows that the signal was more rapidly corrupted by the corrosion process in the case of the 0.25 mm deep grooves. The results shown here are for the moderate corrosion samples but it is clear that even the micro-blasting techniques are less eﬀective than they were in the case of the deeper grooves. The reason for this decrease in the signal to noise ratio is simply that the shallower grooves represent a ‘weaker signal’ than the deeper ones. This ‘weaker signal’ is more easily lost under the action of the corrosion process.

2.4 Discussion Reviewing all the results presented so far it is clear that the best cleaning results are obtained by the micro-blasting processes and the next most eﬀective techniques involve the use of an Nd:YAG laser on a wet surface. It is clear from this result that the eﬀective removal of corrosion products from an iron (or steel) surface must involve some mechanical action. In the case of the micro-blasting processes this mechanical action is provided by the momentum transfer from the high velocity Al2 O3 or glass particles. For the Nd:YAG/wet surface technique the mechanical action is provided by the rapid expansion and partial vapourisation of the inﬁltrated water in oxide layer which is heated by the laser energy. Only when the oxide layer is very shallow it is possible to remove it eﬀectively by simple laser irradiation (at the energy densities considered here). In this case the mechanical action is simply one of local, thermally induced expansion leading to fracture and removal of the brittle oxide surface. Figure 2.4 provides some clues to the diﬀerence between the cleaning mechanisms for the dry and wet samples. The electron micro-graphs presented in Fig. 2.4 compare the cleaned surfaces after wet and dry laser cleaning with the 1,064 nm Nd:YAG laser. The samples presented here are those with 2 mm deep grooves (which were intrinsically too rough for examination by the optical proﬁlometry used on the 0.25 and 0.5 mm deep grooves). The dry ‘cleaned’ surface is covered in a rather thick oxide layer which is clearly cracked and porous. It

18

Y.S. Koh et al. Pictures

Optical profilometry

Reference, no corrosion

2mm Microblast, Al2O3 Microblast, glass

Amplitude Range (mm) 0.23 – 0.21 = 0.02 0.18 – 0.15 = 0.03 0.19 – 0.15 = 0.04

Nd:YAG laser 532 nm, wet

0.27 – 0.17 = 0.10

Nd:YAG laser 1064 nm, wet

0.19 – 0.10 = 0.09

Nd:YAG laser 532 nm, dry

0.16 – 0.08 = 0.08

TEA CO2 laser 10600 nm

0.09 – 0.06 = 0.03

Rotating stainless steel brush

0.14 – 0.09 = 0.05

Nd:YAG laser 1064 nm, dry

0.08 – 0.03 = 0.05

Reference, corrosion

0.16 – 0.10 = 0.06

Fig. 2.3. A comparison of cleaning results for 0.25 mm grooves and moderate corrosion

seems likely that, in the case of the wet cleaned surface, the water inﬁltrated these cracks and pores. The incident laser has a wavelength (1,064 nm) which would pass though any surface water without being absorbed by it. The laser energy would be absorbed by the solid oxide layer which would rapidly heat up the water in the cracks and pores. The transfer of the heat from the solid to the surrounding water has been noted by Zapka et al. [4] and Grigoropoulos and Kim [5]. However, these studies do not entirely explain the results of this

2 Laser Cleaning of Corroded Steel Surfaces Nd:YAG laser 1064 nm, wet

× 25

SEM

× 100

SEM

× 500

× 25

SEM

× 100

SEM

× 500

19

4mm

Nd:YAG laser 1064 nm, dry

Fig. 2.4. Scanning electron micro-graphs demonstrating the diﬀerence in the cleaned surface for wet and dry laser cleaning

present investigation because here, we are dealing with an adherent coating of oxide tens or even hundreds of microns thick. A probable explanation for the removal of thick, wet oxide layers is the action of the sudden expansion of the vapourising water on the cracks and pores within the layer. The forces generated by vaporisation would rapidly open the cracks and pores to shatter the brittle oxide coating.

2.5 Conclusions Within the scope of this study, micro-blasting techniques cleaned iron oxides from steel surfaces more eﬀectively than laser methods. It was also demonstrated that wet surface Nd:YAG laser techniques were more eﬀective than dry surface techniques. This increase in eﬀectiveness is probably the result of the break up of the oxide layer by the sudden expansion and vaporisation of trapped liquid in cracks and pores. Steel brushing cleaned the oxide from the surface in some cases but this was accompanied by substrate erosion. Finally, it was shown that the TEA CO2 laser was less eﬀective than the Nd:YAG laser in removing oxide layers. Acknowledgements The authors gratefully acknowledge the Swedish National Heritage Board, who sponsored this study. Also thanks goes to Johnny Grahn and Tore Silver for the technical help, Luleå University of Technology, Sweden.

20

Y.S. Koh et al.

References 1. J. M. Cronyn, The Elements of Archaeological Conservation, Routledge, London, 1992. 2. J. F. Asmus, in Laser Techniques and Systems in Art Conservation, Edited by R. Salimbeni, SPIE Vol. 4402, 1–7, 2001. 3. M. Cooper, Laser Cleaning in Conservation: An Introduction, Oxford, 1998. 4. W. Zapka, W. Ziemlich, and A. C. Tam, in Applied Physics Letters Vol. 58, 2217, 1991. 5. C. P. Grigoropoulos and D. Kim, in Laser Cleaning, Edited by B. Luk’yanchuk, 229, Singapore, 2002.

3 Laser Cleaning of Gildings ∗

M. Panzner1 , G. Wiedemann1 , M. Meier3 , W. Conrad2 , and A. Kempe1 , and T. Hutsch1 1

∗ 2 3

Fraunhofer Institute Material and Beam Technology Winterbergstr. 28, 01277 Dresden, Germany [email protected] Freelance Restorer, Obere Parkstr. 10, 06295 Lutherstadt Eisleben, Germany Lower Saxony Department of Preservation of Ancient Monuments Scharnhorststr. 1, 30175 Hannover, Germany

Summary. Results of laser cleaning experiments on diﬀerent gilding types like leaf gilding and ﬁre gilding are presented in this contribution by means of three tested art objects. The reﬂectivity of gold is advantageously high for the typical laser cleaning wavelength of 1,064 nm. Additionally, to avoid damage like gold loss, the transfer of the absorbed laser pulse energy into the art object by thermal conduction is considered. Fire gilded surfaces are most easily cleaned because of the good heat transfer conditions which imply a high threshold intensity with respect to damage. This is diﬀerent for leaf gilded surfaces but suitable laser cleaning parameters have also been found for this case. The results of laser cleaning experiments are presented by photography, microscopy, SEM and EDX analysis.

3.1 Introduction The surfaces of many objects of art and architecture are gilded by various techniques like ﬁre, leaf or electrochemical gilding. Especially in the case of outdoor objects exposed to rain, dust and environmental pollution usually accumulate as dirt layers on the gilded surface. In many cases, the gold layer below is damaged because of corrosion or other aging processes like the development of cracks (craquelure) in the ground layer of leaf gildings. Mechanical cleaning of such surfaces would cause further loss of gold. So a contactless cleaning method like laser cleaning could be advantageous. The pressure wave induced by the ablation of dirt cause much lower mechanical stress than mechanical cleaning techniques would [1]. Laser cleaning makes use of the fact that, in virtually any particular case, one can ﬁnd a range of exposure parameters where selective removal of dirt is feasible so that damage to the object is precluded. Finding suitable parameters is not a matter of trial and error but can be greatly eased by the knowledge of the optical and thermal properties of the materials involved [2].

22

M. Panzner et al.

Fig. 3.1. Reﬂectivity of gold, copper and silver vs. wavelength [3], with Nd:YAG laser and its harmonics shown

Fig. 3.2. Scheme of the energy deposition ﬂow of heat during and after absorption of laser pulses

Above 700 nm, the reﬂectivity of gold exceeds 95% for the Nd:YAG laser wavelength of 1,064 nm (Fig. 3.1). Obviously it would not make sense to apply the harmonics of the Nd:YAG laser to clean gildings. Thus the selective removal of surface pollution from solid gold by laser cleaning with Nd:YAG laser should be possible without damage. The deposited laser power of <5% diﬀuses very fast into depth because of the high thermal conductivity. A very high threshold power density results from this fact. In the cases of gold or a gold layer in direct contact with a metal substrate, there is little (here the general situation is described in comparison with leave gilding!) problem with temperature since the heat generated by irradiation ﬂows quickly into depth because of high thermal conductivity. This implies a rather high damage threshold. Figure 3.2 shows conditions for three gilding types considered here. The situation is quite diﬀerent for layered systems with a layer of poor thermal

3 Laser Cleaning of Gildings

23

conductivity in between as it is the case with leaf gilded objects where the adhesive layer (or bole) forms a heat barrier. Then the absorbed fraction of the laser pulse energy is essentially conﬁned in the gold layer, resulting in high temperature and a correspondingly low damage threshold. Similar conditions can be predicted for the enwrapped textile threads. Here the heat is accumulated in the silver thread, and thread temperatures above the damage level for the textile part have to be prevented. So a careful investigation of the threshold levels is necessary especially for the leaf gilding and the enwrapped textile threads. Three objects were investigated to study the diﬀerent behaviour of the gildings: A ﬁre gilded copper pipe of the church tower of Graz, a leaf gilded spire from Bronnbach and a festive bonnet (Bad Münder) [4] made of ﬁbers enwrapped by ﬂattened, chemically gilded, silver wire.

3.2 Experimental The pulsed BMI Nd:YAG cleaning laser, NL102, with an average output power of 6 W, was used. The energy density at the test surface was controlled by selecting the laser output energy as well as beam diameter by using the focusing optics at the end of the beam delivery system (an articulated arm). Diﬀerent process gases (He, N2 ) were applied by blowing them across the ablation area to inﬂuence the ablation process. For comparison, test areas were cleaned by micro-precision steam cleaner. The samples were investigated by microscopy and SEM. Cross sections were made to investigate the layered structure of the gilded surfaces. Elemental analysis was made by EDX.

3.3 Results 3.3.1 Leaf Gilding Figure 3.3 shows the layered structure of the leaf gilded spire. A multiple gilding of this object can be deduced from the number and thickness of the

Granular pigmented undercoat Copper corrosion

Fig. 3.3. Cross section of surface layer structure of leaf gilded spire from Bronnbach

24

M. Panzner et al.

yellow oil-based under-paints and inter-layers. A reddish paint has penetrated into the granular pigmented under-coat (no it is not!). Cracks in the top gold leaf appearing as craquelure pattern are related to a corresponding crack pattern below. The aim of laser cleaning in comparison with mechanical techniques is to avoid further gold loss. This could only be realized by careful investigation of the surface changes depending on laser pulse energy density. Figure 3.4 shows a threshold between 0.1 and 0.2 J cm−2 . Destruction of the surface takes place in this energy density range by the curling up of the gold layer in the vicinity of cracks. Here the absorption and accumulation of energy is higher compared with that of the clean gold surface. Resulting thermal expansion, melting and evaporation processes of the adhesive layer are curling up the edges of the leaf gold. Further increase of the energy density even causes gold losses beginning in these regions. This is also demonstrated by SEM images (Fig. 3.5). The yellow oil-based paints are volcano-like emissions by the cracks and spread in the vicinity of the crack edges (→). Cleaning with micro-precision cleaner results inevitably in surface damage by partial loss of gold (Fig. 3.6). Therefore, laser cleaning with specially adjusted exposure parameters could be the technique of choice.

Energy density: 0,1 Jcm-2

Energy density: 0,2 Jcm-2

Fig. 3.4. Leaf gold of the Bronnbach spire, laser cleaned by diﬀerent energy densities

Fig. 3.5. (left) Microscopy, SEM. (centre) Comparison of contaminated and laser cleaned zones. (right) Beginning of thermal inﬂuence at the cracks on leaf gold

3 Laser Cleaning of Gildings

25

Fig. 3.6. Leaf gold surface cleaned by micro-precision cleaner, damaged surface areas indicated by an arrow

54 % O, 18 % Cu, 12 % Fe, 10 % S, 4 % Si, 1 % Al

70 % Au 14 % Hg 15 % Cu 1 % Si

64%Cu 35%O 1%Cl

Fig. 3.7. Cross section of a ﬁre gilded surface microscopy (left) SEM/ EDX (right). At the position shown, corrosion products grow only through a small number of gold layer defects

3.3.2 Fire Gilding Cross sections of the ﬁre gilded surface of the copper roof pipe used here as a model object shows massive copper corrosion (Fig. 3.7). Corrosion products are even found above the gold layer at places with progressed corrosion. The investigation of the cross sections by EDX (Fig. 3.7, right) shows a signiﬁcant concentration of mercury in the gold ﬁlm indicating the use of ﬁre gilding technique. The aim of restoration can only be the removal of dust and other contaminations. A removal of corrosion layers found here would unavoidably lead to further destruction of the original surface by gold losses and the development of holes and grooves. So the laser cleaning process was driven in a way that only the dirt layer was removed. This process partly uncovers the black Cu2 O (cuprite) that had grown through holes in the gold layer (Fig. 3.6, left). But also green copper compounds like CuO (tenorite) or CuSO4 (copper sulphate) were also found below the contamination. Because of the heat ﬂow conditions discussed above, rather high energy densities of 0.3–0.4 J cm−2

26

M. Panzner et al.

Fig. 3.8. SEM image of a laser cleaned ﬁre gilded surface with a small number of defects (that is right but, also for this mechanisms or processes exist threshold values, so the power density remains an important value!!!)

Fig. 3.9. Laser cleaned ﬁre gilded furniture ﬁttings contaminated by dirt in a ﬁre

could be applied without gold loss or damage to the original surface. SEM images of the laser cleaned gilded surface are shown in Fig. 3.8. The laser cleaning process is very eﬀective here because there is no overheating in the case of ﬁre gilded surfaces. The successful cleaning is also seen on ﬁre gilded furniture ﬁttings that were contaminated by grime during a ﬁre (Fig. 3.9). 3.3.3 Enwrapped Textile Thread Black stains on silver and gold had to be removed from a festive bonnet made of silk threads enwrapped by ﬂattened gilded silver wire. Mechanical cleaning would cause damage to the silver strips as well as the silk. Remains of the original gold layer at protected places were rediscovered by laser cleaning. Apparently the larger part of the gold was worn oﬀ by use (Fig. 3.10). These remains, too, would get lost in mechanical cleaning but have been preserved here. In contrast to [5], a damage-free cleaning at 1,064 nm was also possible by blowing nitrogen gas across the surface and by careful investigation of the threshold values. An advantage of the 1,064 nm wavelength is its low absorption by textile threads compared with the UV. Figure 3.10 shows expanded zones with

3 Laser Cleaning of Gildings

27

Gold layer remains Fig. 3.10. Enwrapped silk threads of the bonnet with residual gilding on ﬂattened silver wire rediscovered by laser cleaning

Fig. 3.11. (left) Overview of the bonnet and (right) test pattern with diﬀerent process parameters listed earlier

uncovered threads that cannot be protected from laser beam irradiation. The energy density threshold deduced from laser cleaning experiments is above 0.2 J cm−2 . Figure 3.11 shows the test pattern for damage threshold investigation under diﬀerent exposure conditions. The best result has been reached by the combination of laser cleaning with the process gas nitrogen and subsequent

28

M. Panzner et al.

careful mechanical cleaning by cotton wool. To summarize exposure parameters in the corresponding numbered zones: 42: Dry laser cleaned, 10–15 pulses; 0.2 J cm−2 , 2 Hz, black layer removed but area remains dark 43: Wet cleaned by 20 pulses, 0.2 J cm−2 , 20 Hz, black layer removed, metallic gloss appears 44: 20 pulses, 0.2 J cm−2 , 5 Hz, black layer removed, metallic gloss appears 45: Dry, N2 -ﬂushing, 15 s, 0.2 J cm−2 , 5 Hz, after which a reddish hue appears, mechanically removed by silver cleaning cotton wool was best result 46: Dry, N2 -ﬂushing, 15 s, 0.3 J cm−2 , 5 Hz ablation too strong, increased surface roughness

3.4 Conclusions Gold has a high reﬂectivity for the typical cleaning laser wavelength of 1,064 nm. So laser exposure parameter sets for damage-free cleaning without further gold loss could be found for three diﬀerent gilding types: leaf gilding on copper, ﬁre gilding on copper and also chemical gildings on enwrapped textile threads. The optimal parameter sets strongly depend on the heat ﬂow below the gold layer and thus on details of the layered system. Hence, the damage threshold of leaf gilded surfaces is low because the oil-based underlayers with low thermal conductivity act as a heat ﬂow barrier. In the case of the chemical gildings on enwrapped textile threads, a combined technique of laser cleaning and subsequent soft mechanical cleaning showed the best result. Acknowledgements The studies are part of a 3-year research project in Germany concerning the laser cleaning of metallic artworks, starting in summer 2002. The work is supported by the Deutsche Bundesstiftung Umwelt (German Foundation for the Environment), Osnabrück, Germany.

References 1. K. Neumeister and G. Wiedemann, Einsatzmöglichkeiten und Einsatzgrenzen von Reinigungstechniken zum Abtrag von Schmutzschichten auf Bauwerken und Kunstgütern, IBW Studie, Fh IWS, Weimar/Dresden, Teil 4, (1996). 2. H. Siedel and G. Wiedemann, Laserstrahlreinigen von Naturstein, Stuttgart 2002. 3. E. D. Palik, Handbook of Optical Constants of Solids, Academic Press, 1998. 4. A. Gervais and M. Meier, Reinigung historischer Metalle – Lasertechnologie im Einsatz für die Denkmalpﬂege, Berichte zur Denkmalpﬂege in Niedersachsen 1/2005. 5. J.-M. Lee, J.-E. Yu, and Y.-S. Koh, Journal of Cultural Heritage Vol. 4, 157, 2003.

4 Current Work in Laser Cleaning of the Porta del Paradiso S. Agnoletti1 , A. Brini2 , and L. Nicolai3 1

2 3

Conservator of the Opiﬁcio delle Pietre Dure Via Alfani 78, 50121 Firenze, Italy [email protected] Conservator of the Opiﬁcio delle Pietre Dure, Via Alfani 78, 50121 Firenze, Italy Freelance Conservator, Via dei Pilastri 34, 50121 Firenze, Italy

Summary. This paper summarises the conservation of the Porta del Paradiso, currently underway. It brieﬂy illustrates the two cleaning methods employed and the reasons leading to continue the cleaning of the perimeter panels with laser ablation, describing the various working phases.

The ‘Porta del Paradiso’, the East entrance of the Baptistery in Florence, was crafted between 1425 and 1452 by Lorenzo Ghiberti, who had already cast the ‘Porta Nord’ after having won the 1401 competition (Fig. 4.1). The work consists of two bronze doors, each wing contains ﬁve main panels with stories of the Old Testament and 24 perimeter panels with Prophets and Prophetesses (totalling 58 relief panels). The panels are entirely gilded with the mercury amalgam technique and inserted with perimeter hammering. The overall door measurements are 288 × 520 cm and the central panels measure 80 × 80 cm. The rectangular panels of the frieze measure 14.5 × 83/90 cm and the round panels are 12.5 cm in diameter. A cleaning treatment undertaken in the years 1947–1948 revealed the uncovered gilding but, by the end of the 1970s, the surface presented incrustations and deterioration products (Fig. 4.2). The ﬁrst panel was detached in 1979 (chosen among those which had fallen as a result of the 1966 ﬂood and which had been subsequently remounted mechanically) and studies led to the development of a chemical cleaning method which did not interact with the copper oxides that guaranteed the adhesion of the gold to the underlying bronze. On this occasion the condition, the chemical compounds at the bronze–gold interface as well as the alteration products present on the gold principally constituted by copper salts (sulphates, nitrates and chlorides) in addition to atmospheric deposits and organic substances deriving from previous treatments, were studied [1–4]. At the same time, a room within the conservation laboratory at the Opiﬁcio delle Pietre Dure was equipped with a special mechanism which would enable rotation and support of the two doors during conservation, Fig. 4.3.

30

S. Agnoletti et al.

Fig. 4.1. The ‘Porta del Paradiso’ by Lorenzo Ghiberti, Baptistery, Florence

Fig. 4.2. Detail of the alteration products on the gilding

The doors were brought to the Opiﬁcio in 1990, when a replacement replica was installed on the Baptistery. In subsequent years, all ten of the larger panels were detached, the cleaning of the nongilded areas was completed and work began on the extremely complex task of taking apart the perimeter panels (four rectangular and four round). Dismounting enables cleaning, examination, study and documentation of the parts which are not usually visible, although it is undoubtedly invasive in the life of the original construction of the work. The laser ablation technique provides the cleaning of the gilding

4 Current Work in Laser Cleaning of the Porta del Paradiso

31

Fig. 4.3. Door rotation system

Fig. 4.4. Panel with the ‘Stories of Noah’ comparison of the cleaning tests

with the main advantage of being carried out in situ and appears to guarantee greater stability overtime as opposed to the chemical method, even if the latter might immediately appear to have a greater specular reﬂectivity. Research, preliminary investigations and laser tests on the panel named the ‘Stories of Noah’ (Fig. 4.4) were initiated in 2000, followed by an analytical

32

S. Agnoletti et al.

study aimed at a thorough evaluation of the two diﬀerent cleaning processes (Rochelle salt and laser ablation). The conclusions drawn, which have been reported in previous publications, have led to the decision to proceed with chemical cleaning on the larger panels and the already detached perimeter ones and to employ the laser ablation method in situ for those still attached to the original framework [5, 6]. This has been made possible thanks to the collaboration with IFAC of the CNR in Florence and, in particular, with Salvatore Siano and work is being conducted by the authors under the direction of Annamaria Giusti. The head technician of the laboratory, Fabio Burrini, has undertaken and coordinated the phases of dismantling and the operations relating to the chemical cleaning, still underway, of the detached panels. To this day, nine of the ten large panels and ﬁve of the eight dismounted perimeter panels from the left door have been cleaned with salts. So far 16 panels of the frieze in the left door have been laser cleaned and work has begun on one of the panels from the right door. The mechanical cleaning of the two wings of the door devoid of gold has also been completed. The conservation history of the Porta del Paradiso (from 1948, the year when the ﬁrst restoration in the modern sense was undertaken, to the present day) encompasses a long time span in which many changes have occurred in conservation: the approach to problems, technological know-how, the choice of materials employed, the awareness of the validity of scientiﬁc backup and, above all, the presence of and the collaboration between various specialists. Towards the end of 2001 (October) laser ablation cleaning was begun on the panels of the perimeter frame of the left wing which, in this phase of work, was placed in an upright position with its lower edge parallel to the ﬂoor. The laser instrument was formally devised by IFAC and produced by El.En. s.p.a. It is a so-called ‘long Q-Switched’ Nd:YAG laser with an optimised 70 ns pulse duration attaining the best possible compromise between removal eﬃciency and thermal stress to the gold ﬁlm. The ﬂuence usually employed is about 500 mJ cm−2 and can be slightly increased according to the hardness and position of the incrustation. The nongilded bronze frames are protected with a double layer of paper adhesive tape (a material that does not cause any damage in the short term) so as to prevent them from being accidentally hit during laser ablation as well as clearly deﬁning the area to be treated. Another precaution is the covering of underlying areas with nylon sheeting so as to avoid their wetting by running deionised water, necessary in the cleaning treatment. The restorer undertaking cleaning uses the aspirator to extract all the volatile products of laser ablation, wears a mask and has an adequate eye protection. The area to be treated is moistened with deionised water using an extremely soft brush or an atomiser so as to reduce surface temperature. Water runoﬀ, favoured by the vertical position of the object, removes any deposits which are not vapourised and sucked up by the aspirator. For best results it is necessary to correctly dose the water since an insuﬃcient amount

4 Current Work in Laser Cleaning of the Porta del Paradiso

33

of water would not lower the thermal peak and vapourisation would be far too rapid, whereas an excessive amount of water would render ablation ineﬀective and necessitate repetition of the process. Cleaning should preferably be carried out in such a way so as to enable free ‘scanning’ of the surface, bearing in mind that it is particularly eﬀective when the beam strikes the surface at normal incidence. Since we are faced with an extremely articulate modelled surface, it is necessary to vary the working parameters when the plane of incidence becomes oblique (Fig. 4.5). When working on undercut surfaces, good results may be obtained with the aid of a dentist’s mirror. Especially in this phase of work we have found that the aiming beam is really useful to identify the irradiated point through reﬂection in the mirror. On more than one occasion, as a result of either the extremely uneven undercut surface or the presence of a particular concretion of altered substances, the laser treatment did not succeed in entirely eliminating the incrustation. Cleaning must, therefore, in these cases be integrated and ﬁnished with the application of small poultices of Rochelle salt. Japanese tissue is applied to the area to be treated followed by the application of the poultice made up of ﬁne cellulose pulp impregnated in a 35% solution of Rochelle salt. Contact times vary (from 15 to 30 min) according to the tenacity of the incrustations and the results are checked by the conservator who decides whether the poultice needs to be repeated or not. In a few cases a slightly basic solution is adopted with reduced contact times. Considering, however,

Fig. 4.5. During the laser cleaning treatment

34

S. Agnoletti et al.

the more aggressive nature of such a solution, it is rarely employed. Once the poultice has been removed, the surface is rinsed by means of tamping with a small cotton swab and deionised water followed by the direct application of deionised water, Japanese tissue and a poultice of cellulose pulp and deionised water to complete the rinsing action. Where the concretions prove to be particularly resistant or in the undercut surfaces for example, thin needles are utilised to fragment them with the aid of the microscope or binocular magnifying lenses (Fig. 4.6). In the areas devoid of gilding on the bronze surfaces there is a more or less compact layer of cuprite. If laser ablation is ineﬀective then these areas are cleaned by mechanical means. Once the cleaning is complete, the entire treated area is rinsed with deionised water, followed by a dehydration treatment with denatured ethyl alcohol and acetone and ﬁnally dried with a jet of warm air (Fig. 4.7). The intervention protocol for the conservation of the panels does not anticipate any surface protective treatment, given the precarious condition of the surviving gold, but it does envisage exposure in a controlled microclimatic environment. After conservation, the panels are inserted into special showcases which are ﬁlled with nitrogen (substituting oxygen) and silica gel employed for dehumidiﬁcation. In the case of laser ablation cleaning in situ,

Fig. 4.6. Mechanical ﬁnishing

4 Current Work in Laser Cleaning of the Porta del Paradiso

35

Fig. 4.7. Final drying phase

the microclimate is controlled by directly applying a suitable double ﬁlm4 chamber along the nongilded frame of the door by means of a special adhesive tape which leaves no residues on the bronze, specially tried and tested by the Scientiﬁc Laboratory of the Opiﬁcio delle Pietre Dure. The supply of inert gas is guaranteed by a continual ﬂow from a nitrogen-producing source. Acknowledgements The authors wish to thank the staﬀ of the Scientiﬁc Laboratory of the OPD for the continuous analytical support to the present conservation intervention.

References 1. G. Alessandrini, G. Dassu, P. Pedeferri, and G. Re, in Studies in Conservation, Vol. 24, 108, 1979. 2. P. Fiorentino, M. Marabelli, M. Matteini, and A. Moles, in Studies in Conservation, Vol. 27, 145, 1982. 4

Flexible, multilayer, thermoformable, high barrier ﬁlm based on polyamide evoh/polyethylene structure.

36

S. Agnoletti et al.

3. “Lorenzo Ghiberti, Storie di Giuseppe e di Beniamino, Storie di Adamo ed Eva: bassorilievi in bronzo della Porta del Paradiso” in Metodo e Scienza operatività e ricerca nel restauro, edited by Baldini U., Sansoni Editore Firenze (1983) 168–206. 4. E. Mello, Chimica e Restauro, 95, 1984. 5. M. Matteini, C. Lalli, I. Tosini, A. Giusti, and S. Siano, in Lacona IV, 77, 2001. 6. M. Matteini, C. Lalli, I. Tosini, A. Giusti, and S. Siano, in Journal of Cultural Heritage, Vol. 4, 147, 2003.

5 Cleaning Historical Metals: Performance of Laser Technology in Monument Preservation A. Gervais1 , M. Meier2 , P. Mottner3 , G. Wiedemann4 , W. Conrad5 , and G. Haber6 1

2

3

4

5 6

Norddeutsches Zentrum für Materialkunde von Kulturgut e.V. (North German Centre for the Material Science of Cultural Assets), Scharnhorststr. 1, 30175 Hannover, Germany, [email protected] Niedersächsisches Landesamt für Denkmalpﬂege (Lower Saxony Department of Preservation of Ancient Monuments), Scharnhorststr. 1, 30175 Hannover, Germany Fraunhofer Institute for Silicate Research (ISC) Competence Team ‘Environmental Monitoring and Conservation Research’ Bronnbach Branch, 97877 Wertheim-Bronnbach, Germany Fraunhofer Institute for Material and Beam Technology (IWS), Winterbergstr. 28, 01277 Dresden, Germany Freelance Restorer, Obere Parkstraße 10, 06295 Lutherstadt Eisleben, Germany Haber & Brandner GmbH, metal conservation, Lichtenfelser Str. 3, 93057 Regensburg, Germany

Summary. In practical restoration – depending on the object in question and the regional attitude to monument restoration – widely diﬀering techniques and restoration philosophies have been, and still are, applied to the exposure of metal. Depending on the type of metal, this results in diﬀering working materials as well as highly diverging deﬁnitions of the required degree of exposure as far as removing corrosive deposits is concerned. Therefore, particularly where metallic or heat sensitive cultural assets are concerned, the applicability of an eﬃcient cleaning procedure using contact-free laser beam technology, which is also gentle on the material, should be examined.

5.1 Introduction As far as frequency is concerned, monuments and works of art made of metal make up the second largest material group for cultural assets after stone. Within this material group there is a very wide assortment of diﬀerent metals and their alloys. These range from corrosion-resistant precious metals to copper-based alloys used for representative outdoor sculptures – to materials such as zinc or iron, which are susceptible to weathering. Monuments made of metal, museum art and cultural goods which are exposed to the elements, often have surface deposits (corrosion, layers of dirt)

38

A. Gervais et al.

which are the result of environmental inﬂuence. Usually this sedimentation, or parts of it, are damaging to the object and, in addition, not attractive for its physical appearance. Apart from structural reasons or the risk of material decay, disturbing or corrosive surface deposits are a major reason for introducing restoration or conservation measures [1]. Uncovering or cleaning up the surface of cultural assets, especially where metal surfaces (exposed to the elements) are involved, basically represents a relevant and indispensable element of total restoration. The degree of uncovering in particular deﬁnes the future aesthetic appearance of a monument after restoration has been completed and also determines in eﬀect which conservation measures should follow cleaning. In practical restoration – depending on the object in question and the regional attitude to monument restoration – widely diﬀering techniques and restoration philosophies have been, and still are, applied to the exposure of metal. Depending on the type of metal, this practice results in diﬀerent working materials as well as highly diverging deﬁnitions of the required degree of exposure as far as removing corrosive deposits is concerned. Various mechanical and also chemical resources are used such as scalpels, air abrasives, ultrasonic chisels, rotating or static brushes, dental grinders and chemical resources such as solvents, paint strippers, acid/alkaline reagents and complexing agents. The restorer must avoid changing, let alone damaging or harming, the original surface at all costs. With conventional cleaning methods it is often possible to produce cleaning results which can be considered pleasing both from the monument restoration point of view and as far as the materials are concerned. However, with regard to selectivity when working on individual layers, sporadic deﬁcits occur if removal cannot be carried out under laboratory conditions (technoscope). Therefore, particularly where metallic or heat sensitive cultural assets are concerned, the applicability of an eﬃcient cleaning procedure using contact-free laser beam technology, which is also gentle on the material, should be examined. Cleaning the surface of signiﬁcant objects, e.g. made out of bronze, by means of a scalpel and then (wax) conservation, which is now standard knowledge and procedure in Germany, involves very high labour costs. Using a laser on cultural assets made of metal not only looks promising for its conservation potential but also for economic competitiveness. Possibilities and limitations for cleaning historical metallic surfaces by using contact-free laser beam technology are being tested systematically in cooperation between the Lower Saxony Department of Preservation of Ancient Monuments (NLD), the North German Centre for the Material Science of Cultural Assets (ZMK) in Hanover, the Fraunhofer Institutes for Material and Beam Technology (IWS) in Dresden and for Silicate Research (ISC) in Würzburg, with its branch for the protection of cultural assets in Bronnbach (project leader for Nd:YAG laser) as well as the Laser Centre Hanover (LZH, project leader for femtosecond laser) and freelance restorers. At the IWS Dresden, a hand-guided, optically pumped Q-switched Nd:YAG cleaning laser,

5 Cleaning Historical Metals

39

type Artlight NL 102, is being used. This laser is already being used successfully in other areas of restoration. At LZH in Hanover, a new Ti–Sa femtosecond laser system was used for the ﬁrst time. This relatively new type of technology, which has not been used in restoration work before, has the advantage of causing only slight warming on the surfaces. However, so far, the nonportable apparatus can only be used under laboratory conditions, whereas the Nd:YAG laser can be used at construction sites and moreover can be worked manually.

5.2 Experimental Methods with Nd:YAG Laser at 1,064 nm and Ti–Sa Femtosecond Laser at 780 nm The restoration objective in cleaning metallic surfaces (focusing on copper, bronze, brass, silver), is uncovering the top (last) mineral surface belonging to the original, which becomes visible only after coats of paint, dirt and corrosion layers (e.g. copper sulphides and copper sulphates) not belonging to the original are removed. In doing so, as far as possible, the original surface of the work of art should not be altered by abrasion or chemical or physical processes, resulting in colour changes, loss of historical traces of work and polishing marks which accompany them. Examinations were carried out in detail on metallic surfaces which varied considerably in degree of corrosion and surface structure. The materials to be considered were chosen so as to include the most relevant metal or corrosion contents frequently occurring in works of art and cultural assets. These included various corroded copper and silver alloys as well as minerals and metal compounds which were also common in paint pigments. The resulting evaluation was from a monument preservation point of view. Based on parameter examinations, a corresponding processing of the original remaining patina was carried out on the basis of fundamental principles. 5.2.1 Indoor Objects Silver Coin from Eisenberg, Archaeological Find To carry out a cleaning test on a silver object found in the ground, a coin from a hoard found in 1694 was available. The object was freed from clinging corrosion products using a femtosecond laser. Neither the test surface on the reverse side nor the front, which was treated all over, showed any changes to the silver surface after the exposure work. Part of the embossing lustre has remained under the corrosion. After just one repetition of the abrasion process, most of the corrosion products have been removed. According to the analytical examination by LZH, only sealing wax residue, interspersed with corrosion, which had by chance found its way onto the coin during its usage period, remained. No discolouration of the surfaces resulted from the laser treatment, and therefore the result can be judged positive overall.

40

A. Gervais et al.

45

44

46

43

42

Fig. 5.1. Three-piece silver bonnet from Bad Münder. The best result was obtained in ﬁeld N◦ 45. It was laser cleaned dry and, when ﬂushed with liquid nitrogen, dull red annealing colour residues could be removed using silver cleaning cotton wool. (The Fraunhofer Institute for Material and Beam Technology (IWS), Dresden, 2004)

Silver Bonnet from Bad Münder A three-piece bonnet with atmospheric corrosion from a museum collection in Bad Münder was available for cleaning silver. The bonnet is made up of an elaborate brocade ribbon whose strings are braided with metal; to achieve this, thin, twisted, textile ﬁbres were wrapped with ﬁnest sheet silver braids galvanised in gold. Even in historical times, this gold was already more or less rubbed oﬀ during cleaning, resulting in a thick deposit of black silver sulphide on the surface. Therefore the noble metals appeared totally black. With the Nd:YAG laser, the brocade ribbons were dry-cleaned using a nitrogen ﬂush (Fig. 5.1). Minimal remaining tempering colours were removed using silver cleaning cotton wool. The cleaning result is optimal. The diﬃculties which usually occur with conventional cleaning of silver in combination with textiles were eliminated. Bronze in the Mausoleum in Stadthagen Whilst examining a group of ﬁgures by Adrian de Vries in the mausoleum of the counts of Schaumburg in Stadthagen (1617–1620), attempts could be undertaken to expose the bronze surfaces which had interior corrosion phenomena (Fig. 5.2). Here, underneath a blackish green corrosion crust, the surface revealed a predominantly brown coloured, enamel type patina consisting mainly of copper oxides. Shiny golden surfaces occur; however these are somewhat rare in Stadthagen. They are mainly to be found above platforms, in

5 Cleaning Historical Metals

41

Fig. 5.2. Bronze cherub by Adrian de Vries (1617–1620) from the mausoleum in Stadthagen. (Lower Saxony Department of Preservation of Ancient Monuments (NLD), Hannover, 2003)

some weathered grooves and in the grooves of the ﬁgure relief, which, owing to their position, have been protected from dust and other impurities. The laser is highly suitable for exposing these surfaces which are rather inaccessible for restoration. As an example a small area of the back of a baroque bronze cherub from this group of ﬁgures was exposed using femtosecond laser and Nd:YAG laser and the pedestal using only Nd:YAG laser (Fig. 5.3). None of the trial exposures using the femtosecond laser fulﬁls restoration requirements. During removal, laser use follows the surface proﬁle of the bronze, in other words, the ablation depth is constant. A self-limiting process related to the ﬁne patina below does not occur, so the various corrosion products are removed in line with the surface proﬁle. This removal resulted in fragments of the corrosion crust being left in individual exposed ﬁelds, whereas the ﬁne patina on the other hand has, in many places, been destroyed down to the bronze. More deep-seated centres of corrosion have often been cleared away in the past, leaving holes. Therefore for work on the ﬁgures from the mausoleum in Stadthagen, the Ti:Sa laser cannot be used with the process strategy applied here.

42

A. Gervais et al.

Fig. 5.3. Test cleaning on the back of the bronze cherub. Large test surfaces, top left on cherub, exposed manually using scalpel and stripping; small surfaces, top right: cleaned with femtosecond laser. Lower ﬁelds with red surround: test cleaning using Nd:YAG laser, ﬂushed with nitrogen as inert protective gas. (Lower Saxony Department of Preservation of Ancient Monuments (NLD), Hannover, 2004)

The exposure attempts with Nd:YAG laser on the cherub’s back and on the head show similar results to the test surfaces where femtosecond laser was used. In places, there are corrosion products on the stable ﬁne patina. Sometimes the stable patina has been worn down so that the bronze is exposed or corrosion holes have been removed. In the former case, using a laser on these surfaces must also be ruled out. In those surface areas which are more or less glistening gold beneath the dark corrosion layer, exposure attempts using Nd:YAG laser sometimes produce very good results. Discolouration on diﬀerent test surfaces indicates changes to the surface during exposure. Using nitrogen as an inert protective gas during ablation prevents discolouration. 5.2.2 Outdoor Object Brass on Garden Sculpture in Hanover-Herrenhausen A brass cherub riding on a turtle from the so-called “Neptune group” in the Herrenhäuser Gardens dates from the second half of the seventeenth century. On the inside, there is an intact water conduit system which suggests it was originally used as a fountain ﬁgure. The surface of the ﬁgure is, to a large extent, covered with a thin, light green corrosion layer consisting mainly of alkaline copper sulphates. In many places, black spots can be clearly seen. Many years of erosion whilst in use in the fountain, as well as from rainwater, have left typical water overﬂow staining. Rimstone and soot plaster crusts have only formed sporadically and in thin layers. Greyish and yellowish impurities on the surface are mainly to be found on the parts which are sheltered from weathering, on the water overﬂow stains and on the pedestal. Possibly the

5 Cleaning Historical Metals

43

Fig. 5.4. (left) Riding brass cherub from the so-called “Neptune group” from Hanover-Herrenhausen in its ﬁnal state after complete restoration which led to positive results using Nd:YAG laser in combination with manual cleaning (with scalpel). Afterwards protective treatment with wax was carried out. (right) Example for the deﬁned cleaning aim: The natural patina surface is complete and not disturbed. (Wolfgang Conrad, Eisleben and Jessica Ulrich, Berlin, 2004)

surface was already cleaned once during an old restoration. On the head, the shoulders, the right arm and the left knee, traces of bird excrement can be identiﬁed, which, owing to its aggressive chemical content, has lead to further corrosion. The ﬁgure’s surface shows black marks in diﬀerent places (neck, upper arms, beginning of the wings) resulting from welding done during previous restoration work. The restoration objectives, to remove the dark grey layers whilst retaining the light green patina, were both achieved using a Nd:YAG laser (Fig. 5.4).

5.3 Results and Discussion Two types of laser were used to test their capabilities and limitations in cleaning historical metals: Type A, the femtosecond laser with a wavelength of 780 nm, which makes “cold” exposure of corrosion possible and Type B, the Nd:YAG laser with wavelengths of 1,064 nm, 532 nm, 355 nm and 266 nm. The size of the femtosecond laser test ﬁelds was 1 × 1 mm or 5 × 5 mm and, with the Nd:YAG, 1 × 1 cm. Apart from the examples described above involving silver, bronze and brass, further test cleaning using both types of laser tests was carried out with varying degrees of success on patinated copper plate from the years 1962 and 1966, on copper rooﬁng material from the seventeenth and eighteenth centuries, on parts of a wrought iron fence dating from around

44

A. Gervais et al.

1900 as well as on painted sheet iron from 1920. The ﬁnal assessment of the iron objects tested is not available at the moment. With objects made of bronze or copper alloys, the femtosecond laser shows pleasing restoration results only for removing layers of rust and other impurities. Judging from its past performance, used exclusively under complicated laboratory conditions, its eﬀectiveness and results are inferior to those of the Nd:YAG laser which showed acceptable results in some of the work carried out on the two cherubs. Both laser types proved to have “good” restoration results for cleaning silver objects and should, in many cases, be chosen rather than traditional cleaning methods.

5.4 Conclusions Whether it makes sense to use a laser and the decision as to which type of laser to use should be examined after considering the value of the object and the speciﬁc tasks to be achieved. The question of costs should be solved in a way which makes economic sense by combining contact-free laser beam technology with conventional restoration processes for the preservation of cultural assets. Acknowledgements A project sponsored by the Federal Foundation for the Environment (Deutsche Bundesstiftung Umwelt, DBU), Osnabrueck, Germany.

Reference 1. P. Mottner, G. Wiedemann, G. Haber, W. Conrad, and A. Gervais, in Proceedings LACONA V (2005).

6 Laser Cleaning the Abergavenny Hoard: Silver Coins from the Time of William the Conqueror M. Davis Department of Archaeology & Numismatics, National Museum of Wales, Cathays Park Cardiﬀ CF10 3NP, Wales, UK [email protected] Summary. The Abergavenny Hoard is a recently discovered collection of late Saxon and Norman silver coins, the details of which were badly obscured by corrosion products and iron concretions. A Q-switched Nd:YAG laser, using near infrared radiation at 1,064 nm, was employed to clean the hoard. Detail retained on the surface of the coins after laser cleaning included “rough-out” marks and polishing marks from the original die, as well as the legend. From this evidence surface damage appears to have been minimal, and the treatment very successful.

6.1 Introduction In April 2002 a scattered hoard of 199 silver pennies was found by metaldetectorists in a ﬁeld near Abergavenny, Monmouthshire in Wales. It is not clear whether they had been deliberately hidden or lost. When it was discovered, the hoard was heavily encrusted with iron deposits; within these were preserved traces of fabric suggesting that the coins had originally been kept in a cloth bag (Fig. 6.1). The total value of the hoard came to 16 shillings and seven pence (16s 7d, or £0.83p) which would have represented several months’ wages for most of the population. The hoard was declared as “Treasure” under current legislation for England and Wales, and has been acquired by the National Museum of Wales (NMW). The hoard includes coins of the Anglo-Saxon king Edward the Confessor (1042–1066) and the Norman king William the Conqueror (1066–1087). AngloSaxon and Norman coins form a unique historical source: each names its place of minting and the moneyer responsible. People had access to a network of mints across England (there were none in Wales) and every few years existing money was called in to be reminted with a new design. The King took a cut on each occasion.

46

M. Davis

Fig. 6.1. Some of the coin hoard before treatment. (Average diameter of coins was 19 mm)

6.2 Condition and Preliminary Cleaning When the coins were discovered they were covered in iron concretions which were obscuring most of the detail, and were much harder than the underlying silver. It was important to remove the concretions to obtain valuable numismatic information, but mechanical cleaning methods such as using a scalpel would have damaged the underlying silver. Chemical cleaning was also tried; tests were carried out using 10% citric acid and 10% oxalic acid. The oxalic acid was much more eﬀective and succeeded in removing some of the corrosion and in separating many of the coins but still failed to shift the majority of iron incrustations (Fig. 6.2) from the coins’ surface and much of the detail remained impossible to decipher. Several of the coins were also cracked or broken; physical pressure or chemical cleaning would have caused further damage. Although laser cleaning has been successfully used on metal sculptures made of aluminium [1] and lead [2], cleaning metals, and especially those from archaeological contexts has been approached with some caution; the complexity and variations of surface dirt and corrosion over altered metal cores means that the eﬀects of laser treatment need to be evaluated for each situation. However, tests carried out by Pini et al. [3] did give some indications that laser cleaning could be successful on these coins. In addition, some experimental tests had been carried out at NMW in 2000 on a variety of materials using a Q-switched Nd:YAG laser. One successful treatment had been the removal of green corrosion products from the surface of a very delicate piece of gilded bronze [4]. Laser cleaning in this case looked like a possible successful treatment for several reasons: The coins showed a signiﬁcant contrast between the dull red, black and green corrosion products on their surface and the underlying white metal. This suggested that the removal of the corrosion by the laser could be self-limiting when the overlying incrustations and oxides were ablated and the

6 Laser Cleaning the Abergavenny Hoard

47

Fig. 6.2. (left) Microphotograph (×13) of the woven fabric preserved within the iron incrustations, possibly the remains of a bag or purse and (right) SEM micrograph of the constituent bast ﬁbres (scale bar = 100 µm)

white metal exposed, as had been seen for the cleaning of both archaeological tin and silver by Pini et al. [3]. Another feature which could contribute to the success of laser cleaning in this instance was the fact that the incrustations on the majority of the coins were not a result of the corrosion of the metal itself but of ﬁrmly adhered contamination.

6.3 Observations on Cleaning Two coins were taken to the Conservation Centre at the National Museums Liverpool, where some preliminary tests using a Q-switched Nd:YAG laser with a green and infrared (IR) laser were undertaken. The short pulse duration (5–10 ns) was important to avoid thermal damage, as the “laser interaction with the metals has to be short enough to produce a fast removal of encrustation, avoiding heat conduction into the substrate” [1]. The green laser appeared to have a very damaging eﬀect on the silver, removing the concretions but altering the appearance of the silver surface which looked more porous, less shiny and much less compact; it also destroyed some of the detail. The IR laser was much less damaging, but the cleaning still appeared a bit too harsh. NMW hired a Q-switched Nd:YAG laser to experiment further with cleaning the coins. In order to reduce the energy levels emitted, an aperture was made which kept this at 33 mJ – one-third of the normal lowest energy emitted by the machine. The working distances were mostly between 20 and 25 cm. This was varied: For softer reddish brown iron corrosion products the laser handpiece was moved further away; too close and it tended to reduce the brown to a harder black substance. For very hard, thick black incrustations the laser handpiece was moved closer: This would make the beam smaller and so it could be more accurately directed, and it did help to quicken the

48

M. Davis

Fig. 6.3. A coin prior to laser cleaning (William I, London mint)

dislodging of larger lumps. Water was applied to the surface and this appeared to help considerably in the removal of both types of corrosion, small lumps sometimes detaching themselves from the surface. Moving the object round and cleaning it from a slightly diﬀerent angle also helped dislodge corrosion from detailed areas and where more stubborn incrustations were present. The sound made by the interaction of the laser with the surface became quite a good indicator of how drastic the treatment was, and helped deﬁne the best working distance on particular types of surface. Figure 6.3 shows a typical coin prior to laser cleaning. The surface of the silver itself was superﬁcially altered by the use of the laser, whether or not water was used. It acquired a whitish coloured (Fig. 6.4) bloom, and this in turn also obscured the detail. This bloom was quite loosely attached to the surface and could be rubbed away. In practice, this layer was removed from the coin in the lab with a soft glass bristle brush under a low-powered optical microscope. Cotton wool swabs and other mild abrasives worked – but the glass bristle brush required minimal use and was able to clean around the detailed legend of the coin in very little time while leaving no discernible scratches to the surface under lowpowered magniﬁcation. At this point any tiny remaining incrustations which the laser cleaning had failed to remove could often be ﬂicked oﬀ with a scalpel. When this layer had been removed the silver was very compact and shiny. The detail retained on the surface of the coins was excellent and included “rough-out” marks (Fig. 6.5) and polishing marks on the original die as well as the legend. From these results surface damage appeared to have been minimal, and considerably less than would have occurred with mechanical cleaning or chemical treatments. It was also relatively time-eﬃcient.

6 Laser Cleaning the Abergavenny Hoard

49

Fig. 6.4. The same coin (as Fig. 6.3) after laser cleaning showing (left) the cream coloured “bloom” on the surface and (right) the cleaned coin after the bloom had been removed

Fig. 6.5. Two cleaned coins; (left) (Edward, London mint) with remnant silver sulphide corrosion on the surface, (right) (Edward, Taunton mint) showing “roughout” marks from when the die was made which are still visible on the struck coin

The laser was most successful at removing iron concretions adhering directly to the silver surface. It also removed green copper corrosion products from the coins where this had occurred. It did not remove silver sulphide so well. This was only present on a few of the specimens, but as it had occurred as a direct result of corrosion of the silver itself, the metal in these areas had a slightly rougher, pitted surface. The laser (Fig. 6.6) was not able to remove all the darker corrosion from indents and pores, and a greyish sheen was left on some of the coins. The laser was a particularly good cleaning method when it came to dealing with cracked and broken coins. Because there was no physical pressure, broken

50

M. Davis

Fig. 6.6. (left) A cleaned coin with a crack running down the centre (Edward, London mint) and (right) a corroded, broken coin during laser cleaning (William, irregular issue)

Fig. 6.7. SEM micrographs of a coin with half the oxide bloom still present after laser cleaning: (left) low magniﬁcation (scale bar = 1 mm) and (right) high magniﬁcation (scale bar = 20 µm)

edges and cracked areas could be cleaned conﬁdently and maximum detail retained. Pressure applied by mechanical cleaning or the eﬀects of chemical cleaning could easily have resulted in loss or damage to these much weakened specimens. The surface of the coins was examined under a scanning electron microscope with an energy dispersive X-ray spectrometer (SEM–EDS) to look in more detail at the bloom and the cleaned surfaces. SEM–EDS (Fig. 6.7) showed no signiﬁcant chemical alteration to the silver and the bloom is likely to have been an oxide; under very high magniﬁcation it was composed of a layer of loosely formed crystals and abrasive marks were visible where these had been brushed away. However, under lower magniﬁcations there was very

6 Laser Cleaning the Abergavenny Hoard

51

little diﬀerence between the areas with the bloom and those which were fully cleaned – except that the detail was clearer.

6.4 Conclusions Hoards from western Britain are rare, and the Abergavenny Hoard has produced many previously unrecorded combinations of mint, moneyer and issue. It includes products of 36 identiﬁable mints, as well as some irregular issues which cannot at present be located. The numismatist felt that this was a very successful treatment for this particular hoard, where a large amount of information was gained from cleaning the coins. Ethically, it did less damage than many chemical methods would have done, and was deemed an appropriate method of conservation for these types of artefact. Acknowledgements Martin Cooper and Sam Sportun (The Conservation Centre, National Museums Liverpool) oﬀered help, advice and undertook initial cleaning experiments. Andy Charlton (Lynton Lasers) gave much support and advice for the project and supplied the “home-made” aperture. Edward Besly (Numismatist, NMW) encouraged the work and supplied numismatic information and Bob Child (Head of Conservation, NMW) both encouraged the work and obtained funding for the laser treatment.

References 1. M. Cooper, in Laser Cleaning in Conservation, 74, Oxford 1998. 2. A. Naylor, in Journal of Cultural Heritage, Proceedings of the International Conference LACONA III, Lasers in the Conservation of Artworks III, Edited by R. Salimbeni and G. Bonsanti, 145, 2000. 3. R. Pini, S. Siano, R. Salimbeni, M. Pasquinucci, and M. Miccio, in Journal of Cultural Heritage, Proceedings of the International Conference LACONA III, Lasers in the Conservation of Artworks III, Edited by R. Salimbeni and G. Bonsanti, 129, 2000. 4. Y. Ever-Hadani, Unpublished MA Project, University of Durham, 2000.

Part II

Stone

This page intentionally blank

7 The Application of Laser Cleaning in the Conservation of Twelve Limestone Relief Panels on St. George’s Hall ∗

M. Cooper and S. Sportun Conservation Technologies, National Museums Liverpool, The Conservation Centre, Whitechapel, Liverpool L1 6HZ, United Kingdom ∗ [email protected] Summary. The use of laser cleaning within the conservation ﬁeld in the United Kingdom has tended to be restricted to indoor work within the studios of a small number of national museums and private conservation companies. This paper describes one of the ﬁrst large-scale projects to be carried out on a public building in the United Kingdom.

7.1 Introduction During the past 15 years, laser cleaning of stone has been employed in a number of high proﬁle, prestigious large-scale conservation projects across Europe including the cathedrals of Amiens, Chartres, and Paris in France [1,2], the Church of the Maddalena in Venice [3], the Cathedral of Oviedo in Spain, the Monastery Jeronimos in Lisbon, St. Stephen’s Cathedral in Vienna [4] and the west frieze of the Parthenon in Athens [5]. The highly selective nature of laser cleaning has allowed skilled conservators to remove unwanted surface accretions from sometimes fragile crumbling stone surfaces in an extremely controlled and sensitive manner. In this way, it has been possible to preserve important patina and ﬁne surface detail on carved stonework. During the same period, large-scale laser cleaning projects in the United Kingdom have been far less commonplace. Laser cleaning work has been concentrated in a small number of national museums and a few private conservation companies. With the exception of a small number of projects [6, 7], work has tended to concentrate on smaller-scale indoor projects, where the working environment is more controllable and conducive to working with lasers. This paper describes one of the ﬁrst large-scale outdoor conservation projects to be carried out in the United Kingdom where laser cleaning has played such an important role. The neo-classical St. George’s Hall (Fig. 7.1) is arguably Liverpool’s ﬁnest public building. Designed by the architect Harvey Lonsdale Elmes, St. George’s Hall was completed in the middle of

56

M. Cooper and S. Sportun

Fig. 7.1. St. George’s Hall, Liverpool (photo reproduced by kind permission of National Museums Liverpool)

the nineteenth-century and provided the city with a grandiose concert hall and law courts. The sculptural programme of the exterior of the building (completed more than 50 years after the death of Elmes) is believed to symbolise the role of Liverpool as the chief commercial port of England at that time. In 1882, a competition was held to design 12 relief panels to ﬁll the blank panel blocks that Elmes had left between the pilasters on either side of the portico on the east façade [8]. The chosen themes, ‘The Progress of Justice’ and ‘National Prosperity’ represented the legal function of the building and the wealth of the city. The competition was won by the sculptor Thomas Stirling Lee. The six panels representing ‘The Progress of Justice’ (on the left of the portico) were unveiled in 1894. In 1890 Alfred Gilbert, the leading British sculptor of his generation, described the ﬁrst two panels as ‘the best things of their kind in England’. The six panels representing ‘National Prosperity’, located to the right of the portico (Fig. 7.2), were designed by three sculptors (Lee and two local men, Charles Allen and Conrad Dressler) with Lee ensuring overall unity in design. These were not unveiled until 1901. Each relief panel measures approximately 1.8 m × 1.5 m and is carved from Istrian limestone. Istrian limestone was chosen instead of Carrara marble as it was felt it would be more resistant to the industrial atmosphere of nineteenth-century Liverpool and would blend in better with the yellowish

7 The Application of Laser Cleaning

57

Fig. 7.2. Six relief panels on the right side of the portico, representing ‘national prosperity’ (photo reproduced by kind permission of National Museums Liverpool)

tone of the Darley Dale sandstone, of which St. George’s Hall is largely built. Even so, it was still felt necessary to apply a coating of paraﬃn wax to the stone panels to further tone down their surface [8]. Remnants of this coating remain. Over one hundred years of weathering and exposure to pollution in Liverpool’s maritime climate has led to the formation of black encrustations in crevices and undersurfaces (Fig. 7.3). Exposed areas of the surface have developed a dirty buﬀ yellow colour in that time, while some of the highest relief appears weathered and bleached. Parts of the surface have blistered. The scope of the conservation work included cleaning the stone panels and consolidating cracks and blisters in the surface. Once loose dust had been vacuumed away, cleaning tests were undertaken to establish the most suitable technique for removing the black crust material. Steam cleaning was found to be largely ineﬀective at pressures that did not lead to loss of surface. Laser cleaning at 1,064 nm was found to be extremely selective in removing the black crust material without loss of stone surface and original coating. This project had three months for completion and was carried out between January and March 2005. This paper describes the work that was undertaken at St. George’s Hall and discusses some of the factors that had to be considered when undertaking such a laser cleaning project outside on scaﬀold in the middle of winter.

58

M. Cooper and S. Sportun

Fig. 7.3. One of the Istrian limestone panels, ‘Liverpool, a municipality, employs labour and encourages art’, (1.8 m×1.5 m) showing extent of pollution encrustations (photo reproduced by kind permission of National Museums Liverpool)

7.2 Work Site Considerations Most laser cleaning systems are designed primarily for work in the relatively clean environment of an indoor conservation studio. An outdoor scaﬀolded site oﬀers a much more demanding environment. Outdoor sites tend to be very dirty, cold in winter (hot in summer) and generally present more complicated versions of the issues that have to be resolved when setting up work in a studio, including safety, access (for equipment and people) and power (Fig. 7.4). The St. George’s Hall site was enclosed in scaﬀolding with a single lift approximately 5 m above ground level. The six panels to the left of the portico were conserved ﬁrst. Once completed, the scaﬀolding was taken down and re-installed for the remaining panels on the right side. Power was supplied from a 30 kV A generator, which was installed before the scaﬀolding was erected around it. Tin hoarding was ﬁxed to the scaﬀold and used as a roof over the working area. This provided a dry and enclosed environment, within which laser cleaning and the other conservation work could proceed safely. Electrical sockets were installed at regular intervals along the 20 m length of the work area, providing outlets for a laser, lighting and extraction equipment. Access to the working area was through two doors. The interior door was interlocked to the laser so that unauthorised entry to the work area immediately stopped

7 The Application of Laser Cleaning

59

Fig. 7.4. Scaﬀold and tin hoarding erected around panels on left side of portico (photo reproduced by kind permission of National Museums Liverpool)

laser cleaning. A separate emergency exit was also provided. The decision was taken to remove the laser cleaning system from site and store it inside St. George’s Hall every night. A ramp and winch were installed so that one person was able to safely move the laser system (weighing nearly 200 kg) onto and oﬀ the scaﬀold. Pneumatic tyres allowed the laser system to be manoeuvred relatively easily around the work area. The laser cleaning system used was a Class IV laser, which means that both the direct beam and diﬀusely reﬂected beam present an ocular hazard. The nominal ocular hazard distance (the distance from the laser at which the beam can be considered safe) for the laser used is in excess of 100 m, which means that the beam must be contained within a controlled area (in this case the work area) so that members of the public are not put at risk. Once the site had been set up, the laser cleaning system was installed on-site by an engineer from the laser manufacturer. Occasionally, the voltage supplied to a laser can be aﬀected by the use of other equipment running oﬀ the same generator, which can aﬀect the performance of the laser. Fine adjustments to the electronics inside the power supply were made, whilst a number of other pieces of conservation equipment (lights and extractor unit) were used at the same time. The articulated arm of the laser cleaning system was encased in polythene to prevent dust ingress into the joints of the arm, which would eventually cause damage to the mirrors (the joints nearest the work area, usually the handpiece, are most susceptible to damage).

60

M. Cooper and S. Sportun

Local site rules were established to ensure safe use of the laser on-site. These included using a safety curtain to divide the work area into two areas: a laser work area and a non-laser work area, and the wearing of laser safety eyewear by everyone on-site. The safety eyewear chosen transmits visible wavelengths of light eﬀectively and does not aﬀect the visibility of the object being conserved. Particulate masks were also worn during laser cleaning and an extraction unit was used to remove the dirt as it was ejected from the stone surface. Regarding use of the laser, it was important to position the unit so that the air intake pointed away from the panel being cleaned (otherwise it acts as an extractor unit drawing dust into the system). To prevent unauthorised laser use, the key used to activate the system was removed whenever conservators were not on-site. The time of year and threat of freezing temperatures meant that it was also important to leave the laser in ‘stand-by’ mode when on-site but not in use; this keeps the cooling water circulating even though the laser is not operating and prevents freezing. If water inside the laser head freezes, expansion as it turns to ice can lead to damage to optical components that is very expensive to repair. The laser cannot be operated when the ‘stand-by’ mode is activated.

7.3 Conservation of the Twelve Panel Reliefs Loose dirt and dust were vacuumed away initially. A Q-switched Nd:YAG laser (Lynton Lasers Zenith) operating at 1,064 nm (5–10 ns pulse duration) with 15 W average power was then used to selectively remove the black pollution encrustations from the Istrian stone surface. Both articulated arm and ﬁbre optic delivery systems were employed. The vast majority of the surface of each panel was cleaned using an articulated arm delivery system. Cleaning was carried out at an average ﬂuence of approximately 0.5 J cm−2 and a repetition rate of 30 Hz. De-ionised water was applied to the area of surface being worked on as a ﬁne mist spray immediately prior to laser cleaning. Vulnerable areas of surface (cracks, blisters, etc.) were cleaned at a reduced ﬂuence and repetition rate to minimise the risk of damage. With the laser operating at an average power of 10.8 W (after a few weeks on-site, the output power of the laser dropped slightly), cleaning rates of approximately 0.1 m2 h−1 were achieved for relatively thick encrustation (up to 1 mm) and 1.0 m2 h−1 for thin dirt layers where only one pulse was required to remove the layer. Removal of the dirt layers was achieved in such a way that the patina of the stone surface, tool markings and remnants of surface coating were preserved. The underside of the bottom ledge of each panel could not be reached using an articulated arm. In order to clean these areas (which are visible from ground level) it was necessary to replace the articulated arm with a more ﬂexible ﬁbre optic attachment. A bundle of four ﬁbres was used to deliver approximately 5 W average power to the less accessible areas of the surface.

7 The Application of Laser Cleaning

61

Once laser cleaning had been completed, paper poultices were used to draw out some of the yellowish staining (resulting from pollution) from areas of the surface, particularly those that had been more heavily polluted. It is important to note that the staining of the stone surface could be seen before any laser cleaning had been undertaken. In his book ‘Public Sculpture of Liverpool’ Terry Cavanagh describes the exposed middle grounds and backgrounds of the panels as a ‘dirty buﬀ-yellow colour’. It is also worth reiterating at this point that paraﬃn wax was applied to each panel shortly after installation to ‘tone down’ the freshly carved surface. Unfortunately, detailed analysis of the surface was not within the scope of the works on this occasion. After laser cleaning, a paper poultice (using de-ionised water) was applied to the surface and left for 24 h. This reduced the staining within the surface of the stone and also removed potentially damaging salts from within the stone. Cracks and blisters on the carved surface were then locally consolidated and signiﬁcant losses ﬁlled with a mixture of ground stone dust (Bath stone, Guiting stone dust, Oolitic limestone), lime putty and de-ionised water. Failing cement was removed from around each panel and the holes ﬁlled with a colour-matched ﬁll. Finally, a coat of polyvinyl alcohol was sprayed onto the surface of each panel to act as a protective coating. On average, the conservation of each panel took two conservators 1 week (35 h), with approximately 10–15 h of laser cleaning. As soon as laser cleaning of one panel had been completed the rest of the conservation work could proceed – in the meantime, laser cleaning on the next panel could begin. A comparison between Figs. 7.5 and 7.6 shows the result of cleaning. Figure 7.7 shows two panels after conservation.

Fig. 7.5. Detail of panel before conservation (photo reproduced by kind permission of National Museums Liverpool)

62

M. Cooper and S. Sportun

Fig. 7.6. Detail of panel after conservation (photo reproduced by kind permission of National Museums Liverpool)

Fig. 7.7. Two panels representing ‘The Progress of Justice’ after conservation (photo reproduced by kind permission of National Museums Liverpool)

7.4 3D Recording of ‘Liverpool, a Municipality, Employs Labour and Encourages Art’ Once conservation of the 12 panels had been completed, the opportunity was taken (with the scaﬀold still in place) to accurately record in three dimensions the surface of one of the panels: ‘Liverpool, a municipality, employs labour and encourages art’ (Fig. 7.3). The panel was laser scanned (3D Scanners

7 The Application of Laser Cleaning

63

Modelmaker X system) in one day (Fig. 7.8). The equipment used is speciﬁed to an accuracy of 0.1 mm in ideal conditions, but the location of scanning in this case (on scaﬀold) meant that the panel was recorded to an accuracy of nearer 0.3 mm. Even so, processing of the data has provided an extremely accurate 3D record of the panel surface, which provides a very useful addition to its conservation record (Fig. 7.9).

Fig. 7.8. Laser scanning of relief panel ‘Liverpool, a municipality, employs labour and encourages art’ (photo reproduced by kind permission of National Museums Liverpool)

Fig. 7.9. Screenshot of 3D digital model of relief panel ‘Liverpool, a municipality, employs labour and encourages art’ (photo reproduced by kind permission of National Museums Liverpool)

64

M. Cooper and S. Sportun

7.5 Conclusions Laser cleaning has been successfully used in an important large-scale outdoor stone conservation project in Liverpool (and possibly England) for the ﬁrst time. Use of a laser has allowed conservators to sensitively remove black pollution encrustations without disrupting the underlying surface; patina, surface detail and remnants of what is believed to be the original surface coating have been preserved. It is important to note that laser cleaning was not the sole cleaning technique employed and that it was necessary to use poultices to reduce the staining in the stone surface also. The sensitive nature of laser cleaning allowed the panels to be cleaned before any consolidation work was carried out. Careful planning allowed this work to be undertaken during the cold and wet months of winter without disruption to the work programme. In addition, a highly accurate 3D record of the surface of one of the panels now exists, alongside the detailed 2D records captured by ‘conventional’ photography. Acknowledgements The authors are grateful to the Oﬃce of Science and Technology’s PSRE fund for funding of the laser cleaning equipment used in this project and to all of the conservators of Conservation Technologies for their dedication and hard work.

References 1. C. Weeks, in Studies in Conservation, Vol. 43, 101, 1998. 2. P. Bromblet, M. Laboure, and G. Orial, in Journal of Cultural Heritage, Vol. 4 (1), 17, 2003. 3. E. Armani, G. Calcagno, C. Menichelli, and M. Rossetti, in Journal of Cultural Heritage, Vol. 1(1), 99, 2000. 4. G. Calcagno, E. Pummer, and M. Koller, in Journal of Cultural Heritage, Vol. 1 (1), 111, 2000. 5. P. Pouli et al., in Lasers in the Conservation of Artworks, LACONA V, Edited by K. Dickmann, C. Fotakis, and J. Asmus, Berlin, 333, 2005. 6. K. Beadman and J. Scarrow, in Journal of Architectural Conservation, Vol. 4(2), 39, 1998. 7. S. Chapman, in Journal of Cultural Heritage, Vol. 1(1), 75, 2000. 8. T. Cavanagh, in Public Sculpture in Liverpool, Liverpool, 1997.

8 The Potential Use of Laser Ablation for Selective Cleaning of Indiana Limestone K.C. Normandin1 , L. Powers2 , D. Slaton2 , and M.J. Scheﬄer2 1

2

Wiss, Janney, Elstner, Associates Inc. (WJE), 1350 Broadway, New York, NY, USA, [email protected] Wiss, Janney, Elstner, Associates Inc. (WJE), 330 Pﬁngston Road, Northbrook, IL, USA

Summary. The aim of this investigation and conservation study was to examine and evaluate the laser ablation method as a practical technique for cleaning of Indiana limestone, a calcite-cemented stone widely used in historic structures throughout the United States. To this goal, a thorough petrographic characterization of the samples was performed prior to and following laser cleaning tests by Q-switched and short free running Nd:YAG lasers. The main optimization problem was the amber-gray appearance associated with the laser ablation by Q-switching lasers. Following the evaluation of such a cleaning result, two practicable solutions based on suitable pulse duration or wavelength selections were successfully demonstrated and then compared with diﬀerent intervention protocols proposed. This chapter will show that through this case study, an understanding of eﬀective uses of cleaning highly weathered Indiana limestone through the use of three types of Q-switched and short free running Nd:YAG lasers can be most eﬀective in the removal from limestone of surface soiling and thick built-up carbon deposits ranging from 0.5 to 1 mm in thickness. Case study evaluation methods included petrographic examination of composition, texture, and microstructure using optical microscopy and scanning electron microscopy performed on thin and polished sections of limestone sampled from six areas before and after cleaning. The microscopy studies were supplemented with energy-dispersive X-ray spectroscopy to characterize crystalline phases and track changes in chemistry.

8.1 Introduction In the United States, buildings and monuments have been constructed of Indiana limestone for several centuries. The calcite-cemented stone formed of shells and shell fragments is found in massive deposits located almost entirely in Lawrence, Monroe, and Owen counties in Indiana. Limestone is selected for building facades because of its natural beauty, availability, and low cost. Its use ranges from simple panels on 1920s skyscrapers to the elaborate carved features found on early twentieth century monumental Beaux Arts facades of many museums and civic structures. In this case study, fragile limestone

66

K.C. Normandin et al.

samples of a decorative cheneau on an historic structure in a major urban center in the northeastern United States were selected for a conservation cleaning study. The subject building was located with exposure to a moderately severe oceanic climate. In a previous test program conducted by the authors in September 2003 [1–7], samples of limestone commonly used in building construction in the United States were cleaned by selected microabrasive systems. For each group of limestone sample, one sample was retained as a control without cleaning and additional samples were cleaned by Rotec, Facade Gommage, and the Nd:YAG laser cleaning system. After petrographic examination and study of the samples, it was determined that optimum results were obtained through use of the Nd:YAG laser system in combination with targeted use of water in selected areas. This preliminary test program was limited in scope, and additional studies using laser cleaning on controlled limestone samples from a single source were recommended. The following paper provides the ﬁndings of the additional studies performed using laser cleaning methods.

8.2 Laser Cleaning and Study Protocol The application of lasers for cleaning ornamental or fragile architectural building features is becoming widely accepted. The successful use of lasers for cleaning limestone objects and sculpture has suggested the potential eﬃcacy of this cleaning technique for extension to an increasing number of building facades. The present study evaluated the use of Q-switched and short free running Nd:YAG lasers on selected Indiana limestone samples removed from decorative architectural elements. Q-switched lasers generate short duration, high-peak power pulses to vaporize the surface deposits without damage to the substrate. Nd:YAG laser is a handheld device with a wide range of parameters utilizing a pulse rate of 1–25 Hz, with an output control allowing reduction down to a single shot. Three lasers were tested on a sample of the exterior Indiana limestone: A handheld laser (normal mode, type B), Palladio laser (Q-switched, articulate arm), and Michelangelo laser (Q-switched, articulate arm). The cleaning eﬃciency was reported to be lowest for the HHL, which had the smallest spot size (3 mm), and highest for the Michelangelo Laser, which had the largest spot size (12 mm). In the test areas, the repetitive rate and energy (mJ) of each laser was varied. Laser characteristics are as follows: Laser

Rep. Rate

Pulse Width

Wavelength

HHL laser Palladio Michelangelo

1–15 Hz 1–15 Hz 1–25 Hz

100 µs 9 ns 9 ns

1,064 nm 1,064 nm 1,064 + 532 nm

8 The Potential Use of Laser Ablation

67

The cleaned samples were examined at magniﬁcations ranging from ×5 to ×50 using a stereomicroscope equipped with a high-intensity, ﬁber optic light source that could be positioned to illuminate at angles ranging from lowangle, glancing illumination to high-angle illumination normal to the surface studied. Low-angle illumination is ideal for studying surface topography. Portions of the sample that had been masked during the cleaning provided an opportunity to study the nature of the surface soiling. These areas were studied with the stereomicroscope using the viewing conditions described above. A small portion of the soiling crust was gently scraped from the uncleaned surface and examined with a scanning electron microscope (SEM). Portions of the stone were also studied by thin-section microscopy. Findings are described below. 8.2.1 Petrographic Description of the Stone The study sample is a typical example of Indiana limestone used as dimension stone in building construction, and is a Mississippian-age limestone from the Salem Limestone Formation in Indiana. Chemically, this gray and buﬀ stone is approximately 97% calcium carbonate and 1.2% calcium–magnesium carbonate (dolomite). The balance largely consists of iron oxides and hydroxides, traces of silicon dioxide, and other phases. Petrographically, the limestone is a grainstone of fairly uniform texture. It has no preferred direction of splitting, but displays weak bedding. The bedding planes are deﬁned by the orientation and sorting of the fossil constituents. The limestone consists of marine microfossils and pellets cemented by ﬁnegrained calcite and by coarse-grained, sparry calcite. The shells and shell fragments are usually 0.5–1 mm long. The pellets typically have diameters of 0.25 mm. The porosity is relatively high, as can be seen in Fig. 8.1 where epoxy containing blue dye ﬁlls the large pores between fossils and the small pores between calcite crystals. Iron-bearing minerals (hematite and goethite are

Fig. 8.1. Thin-section micrographs, with cross-polarized light, show the texture of the fossiliferous Indiana limestone. Pores have been ﬁlled with epoxy containing blue dye. The ﬁeld width of the photo on the left is 1.2 mm; on the right, it is 2.4 mm

68

K.C. Normandin et al.

Fig. 8.2. Diagonal strip of original black soiling in a location masked during the cleaning trials. Note cleaned areas on each side of black soiling A closer, oblique view of the soiled stone shows that the heavy soiling is conﬁned to the surface region and penetrated surface pores (arrow)

suggested by the SEM analysis) are ubiquitous in small amounts interstitially and as inclusions in the calcite. Small, angular quartz grains are also observed interstitially and as inclusions. SEM analysis detected traces of calcium phosphate (collophane) and barium sulfate (barite). Collophane generally occurs within the spheroidal pellets. 8.2.2 Surface Soiling Prior to the cleaning tests, the surface of the stone was heavily encrusted with black particulates (Fig. 8.2). Patches of green and greenish black biological growth were also observed. The biological growth was typically conﬁned to protected areas such as pores in the stone and in carved recesses. These particulates and biological growth were frequently lodged in the pores and crevices exposed on the surface of the stone but were not “cemented” or bound in any discernable fashion. A gypsum crust was not observed on any portion of the sample. 8.2.3 Visual Observations Prior to cleaning, each side of the sample was marked into test blocks and labeled to show the test parameters. After cleaning, the test areas were studied at magniﬁcations ranging from approximately ×5 to ×50 using a stereomicroscope. 8.2.4 Stereomicroscope Study Typical areas viewed with the stereomicroscope are shown in Figs. 8.3 and 8.4.

8 The Potential Use of Laser Ablation

69

Fig. 8.3. A representative surface having high porosity is shown at magniﬁcations of ×10 and ×25 after the cleaning operations in test area 5. The ﬁelds of view are approximately 12 and 5 mm, respectively. The surface appears clean at the lower magniﬁcation. Higher magniﬁcation reveals soiling and greenish biological growth in crevices

Fig. 8.4. A representative surface having lower porosity is shown at magniﬁcations of ×10 and ×25 after the cleaning operations in test area 2. The ﬁelds of view are approximately 12 and 5 mm, respectively. The surface appears relatively clean at the lower magniﬁcation. Higher magniﬁcation reveals only small amounts of soiling in the deepest surface pores (arrows)

8.2.5 Thin-Section Microscopy Six locations of cleaned sample were selected for thin sections based on the stereomicroscope observations, as described in the table below. The sections were prepared from 19 mm diameter cores drilled in the locations selected. Before cutting parallel to the core axis, the stone was impregnated with epoxy resin containing blue dye to facilitate studies of microfracturing and porosity.

70

K.C. Normandin et al.

Thin section

Laser

Parameters

TS TS TS TS TS TS

HHL Michelangelo Palladio Michelangelo Michelangelo Palladio

15 Hz, 12 Hz, 10 Hz, 25 Hz, 12 Hz, 10 Hz,

1 2 3 4 5 6

800 mJ, 350 mJ, 250 mJ, 400 mJ, 350 mJ, 300 mJ,

Test area 1,064 nm 1,064 + 532 nm 1,064 nm FT 1,064 + 532 nm 1,064 + 532 nm 1,064 nm

6 4 5 5 5 4

TS-1 – Isolated black particles are adhered to calcite crystals and remain lodged in pores on portions of the cleaned surface. Occasionally, short microcracks oriented parallel to the outside surface of the stone occur in the outer 200 µm. A few calcite crystals exhibit ragged exterior surfaces. TS-2 and TS-5 – These thin sections represent the same parameters of the Michelangelo laser but used on areas having somewhat diﬀerent textures. The cleaned surface represented by TS-2 is parallel to the bedding plane deﬁned by the orientation of the larger, elongate shelly fossils. This surface has a few isolated dark particles adhered to the surface, but no crust is present. Several surface-parallel microcracks occur within 1 mm of the surface. These microcracks are black (opaque) and appear to be ﬁlled with deposits similar to the soiling on the surface. Isolated dark particles are locally adhered to the surface of TS-5 but microcracks are less common relative to TS-2. TS-4 – This thin section represents test area 5 using the Michelangelo laser at a frequency of 25 Hz. The cleaned surface has almost no adhered particulates. No surface-parallel microcracks were observed. TS-3 and TS-6 – These thin sections represent slightly diﬀerent settings of the Palladio laser. The power setting was 250 mJ in the area from which TS-3 was taken, and 300 mJ in the area from which TS-6 was taken. The surface textures were also somewhat diﬀerent. The cleaned surface of TS-6 is parallel to the bedding plane in the limestone, while the surface of TS-3 was perpendicular to the bedding plane. No diﬀerences between the cleaned surfaces were noted. Both surfaces have a few minor microcracks oriented subparallel to the surface. The microcracks are ﬁlled with dark deposits.

8.3 Discussion and Conclusions For the samples tested, the Michelangelo at 12, 20, and 25 Hz rep. rate and 350, 400, and 450 mJ per pulse energy produced cleaner surfaces with less residual soiling. The lower rep. rate of the Palladio (10 Hz) at lower energy settings (250 and 300 mJ) often produced similar results. The HHL laser appeared less eﬀective in cleaning the surface in this trial. For the most part, the soiling on this stone before cleaning consists of darkcolored particulates that are weakly adhered to the surface of the stone and are

8 The Potential Use of Laser Ablation

71

wedged into abundant pores and crevices. The particulates are not bound in a hard crust. The limestone also contains iron-bearing mineral inclusions that are similar in size and appearance to the particulates that comprise the bulk of the soiling. These iron-bearing inclusions occur within the fossils, within the calcite matrix, and within pores, and are thought to be the source of the yellowish brown ocherous deposits noted on some of the cleaned surfaces. Surﬁcial accumulations of yellowish brown ocher deposits possibly result from the alteration of reddish brown, iron-bearing minerals disseminated throughout the body of the limestone. This alteration is probably a product of normal environmental exposure and would not be an expected result of the laser cleaning. The yellowish deposits are not abundant and are not readily visible unless the stone is viewed with high-intensity lighting and magniﬁcation. The thin sections prepared from areas cleaned with the Michelangelo and Palladio lasers contain a few parallel, relatively long microcracks just below the cleaned surface. These microcracks are generally ﬁlled with dark-colored deposits (Fig. 8.5). A few shorter but similar microcracks are also present in the thin section representing the area cleaned with the HHL laser. The cause of these microcracks is not clear. However, because they contain deposits similar to those comprising the surface soiling, it is probable that these microcracks existed before the cleaning trials. Such features may have been caused by previous cleaning eﬀorts. Features that were present at least to some extent on cleaned and masked (uncleaned) surfaces included fracturing of the delicate microfossils on the surface of the stone and fracturing of exposed surfaces of the coarse-grained calcite matrix (Fig. 8.6). Calcite is a relatively soft mineral (Mohs hardness of 3) and is easily fractured along rhombohedral cleavage planes producing a saw-tooth surface. Etching of calcite by acid rain or acid-cleaning media generally results in pits and scalloped edges that are visible microscopically. Such rounded surfaces are not prevalent in the thin sections studied. This type of damage may also have been caused by previous cleaning eﬀorts.

Fig. 8.5. Parallel fractures in the outer 200 µm of the limestone are ﬁlled with dark deposits

72

K.C. Normandin et al.

Fig. 8.6. Thin section image of the cleaned surface showing etched and/or fractured calcite crystals. The width of the ﬁeld of view is approximately 600 µm

The results of this test program indicate that the laser cleaning methods studied are eﬀective in removing surface soiling from Indiana limestone without damage to the stone. In addition, laser cleaning warrants further study to compare results obtained with Indiana limestone using other methods typically applied to cleaning of ornamental features, primarily low pressure water misting and rinsing. Additional study is also recommended to provide further comparison between cleaned and uncleaned Indiana limestone from a variety of building facades and to represent a broad range of environmental conditions experienced in service. Acknowledgments The study was supported by Wiss, Janney, Elstner Associates, Inc. (WJE), Northbrook, Illinois, USA. Petrographic examination was performed by Laura Powers of WJE, Northbrook, Illinois, with consultation by L. Brad Shotwell of WJE, Cleveland, Ohio, USA. Laser cleaning samples were performed by architectural conservator Giancarlo Calgano of Altech srl, Applied Laser Technology.

References 1. K. C. Normandin and D. Slaton, in Conference Proceedings. Verband der Restauratoren, September 2003. 2. G. Allessandrini, A. Sansonetti, and A. Pasetti, in IV International symposium on the conservation of monuments in the Mediterranean basin, Rhodes, 1997. 3. M. Cooper, Laser Cleaning in Conservation: an Introduction, Oxford 1998.

8 The Potential Use of Laser Ablation

73

4. K. Dickman, C. Fotakis, and J. F. Asmus, in Lasers in the Conservation of Artworks, LACONA V Proceedings, Osnabrueck, Germany, September 15–18, 2003. 5. S. Siano, R. Pini, and R. Salimbeni, in Integration of Laser with Conventional Techniques in Marble Restoration. 9th International Congress on Deterioration and Conservation of Stone, Venice 19–24 (2000). 6. H. Siedel and K. Hubrich, in Results of Laser Cleaning on Encrusted Oolithic Limestone of Angel Sculptures from the Cologne Cathedral. 9th International Congress on Deterioration and Conservation of Stone, Venice 19–24 (2000). 7. C. Weeks, in Studies in Conservation (1998).

9 Laser Cleaning of a Renaissance Epitaph with Traces of Azurite ∗

J. Nimmrichter and R. Linke Austrian Federal Oﬃce for Care and Protection of Monuments, Department for Conservation and Restoration, Arsenal 15/4, 1030 Vienna, Austria ∗ [email protected] Summary. In Steyr, Austria, a Renaissance epitaph was cleaned by an Nd:YAG laser. Compared to cleaning tests, carried out with microsandblasting, (NH4 )2 CO3 and water compresses, the laser cleaning was much more sensible and faster. Satisfying results were realized for the calcareous sandstone of the frame as well as for the red and compact limestone of the relief. One of the big advantages was the detection of several traces of pigments, which were observed during the layered cleaning process. It can be assumed that other cleaning methods could not hold these last spots of polychrome surfaces. Care had to be taken of the energy. The application of higher energies on the red marble leads to colour changes into yellowish red.

9.1 Introduction In Steyr, an old important city in Upper Austria, an epitaph from the sixteenth century was restored in 2004 on behalf of the Austrian federal oﬃce for care and protection of monuments (Bundesdenkmalamt). The epitaph is mounted on the north wall of the “Margarethenkapelle” of the parish church of Steyr and represents the most important gravestone of this church [1]. The inscription on the epitaph shows the date 1538. For the two reliefs, red limestone (marble from the area of Adnet, Salzburg) was used. The ornamented frame putti and fantasy animals consist of 29 parts of calcareous sandstone and soft limestone. The stones are framed with mural paintings on a limewash. The two pillars, which house the epitaph, were made from quartz and calcareous sandstone and are therefore from diﬀerent periods. A detailed report on the construction and the materials used can be found at the restoration report [2]. Figure 9.1 shows the epitaph before restoration. Originally, the monument was covered all over with mural painting. Nowadays, mural paintings can be found in the more protected areas. Most paintings have survived beneath the roof of the housing. From a report from 1876, it can be determined that the rate of decay increased enormously within the last 100 years [3]. Contrary to yellow, red and gold colours, which were removed

76

J. Nimmrichter and R. Linke

Fig. 9.1. Epitaph before restoration

nearly totally from the surface by previous cleaning treatments, mostly the blue colour was well preserved. The epitaph shows heavy damage, caused not only by environmental inﬂuences but also by former restoration treatments and vandalism. According to the state of preservation of the calcareous sandstone, a former cleaning treatment with acid can be concluded. This results to the fact that only tiny traces of paint could be found on the white sandstone.

9.2 Experimental Methods Scientiﬁc investigations were carried out by scanning electron microscopy with energy dispersive X-ray detection (SEM/EDX) in order to characterize the painting technique, the pigments as well as the corrosion products. Originally, the background of the monument was painted with azurite on a layer of whitewash of lime and lime as binding medium. Today, areas which were covered with mural paintings have mostly been washed oﬀ or have been covered with a crust of dust and gypsum.

This page intentionally blank

9 Laser Cleaning of a Renaissance Epitaph with Traces of Azurite

77

Fig. 9.2. SEM image (BE mode) of the cross section of the azurite area before laser cleaning. (1) Compact crust of gypsum and dust, (2) two layers of azurite (bright crystals) with lime binding medium; the top layer shows a larger grain size, (3) layer of whitewash and local gypsum crystals, (4) calcareous sandstone

The pre-cleaning of the epitaph was carried out mechanically by microsandblasting (aluminum silicate). Partially, compresses of distilled water and (NH4 )2 CO3 were used in order to reduce soluble salts and gypsum. Scientiﬁc investigations in the SEM revealed that the gypsum crust was very compact, and so the dissolution of (NH4 )2 CO3 was not able to penetrate the layer or convert the gypsum. Figure 9.2 shows a SEM image in the backscattered electron (BE) mode of a cross section of the azurite layer before cleaning. It is remarkable that the azurite was applied in two layers, which can be determined from two diﬀerent grain sizes used. Additionally, investigations by FTIR-microscopy showed a former acrylic treatment of some parts. In order to remove the layers of gypsum and dust without harming the residues of azurite layers, an Nd:YAG laser was used for cleaning the most important parts of the epitaph. While nowadays the cleaning of limestone by Nd:YAG lasers is well established, the cleaning of painted surfaces has not been investigated so well. In dependence of the chemical composition of the paint layer and the laser parameters, the interactions between pigments, binding media, varnish material and the laser activity may produce negative results. A Quanta Systems Palladio Nd:YAG laser at 1,064 nm was used for cleaning both reliefs made of red limestone, the framing bead made of calcareous sandstone and the curved relief on the top, which is made of soft limestone. Additionally, the laser was used locally on the plaster surface in order to crack very compact crusts of corrosion products. The energy varied between 400 and 600 mJ with a pulse width of 6 ns. Areas covered with

78

J. Nimmrichter and R. Linke

azurite were cleaned at 250–300 mJ. The working distance varied between 10 and 25 cm. Overall, the laser was used for approx. 160 hs which yielded approx. 3,500,000 shots.

9.3 Results In order to control the quality of dust removal, accompanying investigations were carried out in the microscope as well as in the SEM. The results reveal excellent cleaning results. No material loss or colour changes were observed at the azurite or the calcareous sandstone. As one can see in Fig. 9.3, the patina was removed without any damage of the azurite layer [4]. Contrary to other cleaning methods, no material loss or other negative cleaning reactions could be observed. Care has to be taken of the binding medium. Comparative applications on diﬀerent objects with azurite and casein showed a darkening of the paint layer. Figure 9.4 shows a comparison of the cleaned (left side) and the original surface (right side) of the red limestone in the SEM by secondary electron (SE) mode. Analyses carried out by SEM/EDX reveal a gentle removal of the grown gypsum layer. As the formation of gypsum crystals also aﬀected the bulk material down to approx. 50 µm, a complete removal of the gypsum by laser treatment was not possible without loss of original material. On the other hand, the reduction of the patina enabled the application of desalination compresses and a consolidation by silicone ester (Wacker OH) and Paraloid B72. Figure 9.5 shows cross sections of samples of the red limestone relief before and after cleaning. As one can see in the left image, the corrosion of the stone and the formation of gypsum lead to cracks parallel to the surface. By applying the laser only a minimal material loss could be observed of these loose parts. Compared with other mechanical methods such as microsandblasting or the scalpel, in this case laser cleaning represents the gentlest technique.

Fig. 9.3. Cross section of the red marble before (left) and after (right) laser treatment

9 Laser Cleaning of a Renaissance Epitaph with Traces of Azurite

79

Fig. 9.4. SEM image (SE mode) of the surface of the red limestone

Fig. 9.5. Cross section of the red limestone before (left) and after (right) laser cleaning

In order to control the quality of the patina removal, additional test were carried out as follows. A piece of red limestone (Adnet marble) was treated with a Palladio laser. Laser parameters were used as mentioned above. The samples were treated with 800 units (approx. 530 mJ) and 950 units (approx. 630 mJ). Investigations carried out on cross sections in the microscope reveal a slight fading of the red marble, which was treated at 630 mJ, while no visible change was observed at lower energy. In the SEM one can see that the laser impact did not harm the marble at all. As can be seen in Fig. 9.6, even old cracks, approx. 20 µm below the surface, did not yield a ﬂaking oﬀ of the stone. The recrystallisation zone at the edge of the crack proves that this crack was already formed before the laser treatment and is not a result from the shock wave. More than this, the laser did not remove these brittle ﬂakes close to the surface.

80

J. Nimmrichter and R. Linke

Fig. 9.6. SEM image (BE mode) of a laser treated sample of red limestone

Referring to the economic aspect, laser cleaning was also faster compared to the application of compresses. Another big advantage was the detection of several traces of pigments, which were observed during the layered cleaning process. It can be assumed that other cleaning methods could not hold these last spots of polychrome paintings. On the other hand, the application of higher energies leads to colour changes into yellowish red. After cleaning, the white limestone was partially consolidated with silicone ester (Wacker OH in ethanol) while the red marble was partially treated with Paraloid B72. For conservation, the red marble reliefs were covered with a mixture of acrylic resin and silicon ester. Additionally, a layer of microcrystalline wax was applied. The white calcareous and the soft limestone were covered with three thin layers of limewash.

9.4 Conclusions The application of an Nd:YAG laser at 1,064 nm showed excellent results for cleaning the red marble as well as the calcareous sandstone. Investigations carried out by SEM/EDX and light microscopy reveal no material loss or colour changes at the azurite or the calcareous sandstone. The red limestone showed some fading but, contrary to other cleaning methods, no material loss or other negative cleaning reactions could be observed. Compared to cleaning tests, carried out with (NH4 )2 CO3 and water compresses, the laser cleaning was much more sensible and faster. However, care has to be taken of the energy due to the eﬀect of bleaching.

9 Laser Cleaning of a Renaissance Epitaph with Traces of Azurite

81

References 1. R. Koch and B. Prokisch (eds.): Stadtpfarrkirche Steyr, 262, Steyr 1993. 2. J. Nimmrichter and M. Koller, in Schlussbericht zur Restaurierung des Forsterepitaphs, Bundesdenkmalamt 2004, unpublished report. 3. J. Wussin, in Die Grabdenkmale der Stadtpfarrkirche in Stadt-Steyr und ihres ehemaligen Friedhofes, Steyr 1876. 4. J. Nimmrichter and T. Krarup, in Restauro, Vol. 6, 1998.

10 Laser Cleaning of Peristyle in Diocletian Palace in Split (HR) D. Almesberger1 , A. Rizzo1 , A. Zanini2 , and R. Geometrante3 1

2 3

SER.CO.TEC. S.r.l., Loc. Dolina 547/2, 34018 San Dorligo della Valle, Trieste, Italy Electronic Engineering S.p.A., Via Baldanzese 17, 50041 Calenzano (FI), Italy Dipartimento di Progettazione Architettonica e Urbana, Università degli Studi di Trieste, Trieste, Italy, [email protected]

Summary. Before starting the cleaning program of the peristyle of Diocletian Palace in Split, a series of tests have been performed on it. First of all, the state of conservation of columns and capitals has been assessed applying non-destructive techniques such as thermography, magnetoscopy and superﬁcial ultrasonic tests. All the areas with black crusts, exfoliation and stone cracks have been determined. In this stage, parameters such as water absorption and colour have been estimated in order to compare them with those measured after the cleaning operation. Then, more than 3-month period of tests have been performed to set up all the parameters concerning the application of the laser cleaning techniques. In this chapter, the results of these preliminary investigations are presented.

10.1 Introduction Around 293 A.D., the Roman Emperor Diocletian started the construction of a rich palace, where he retired after abdication (Fig. 10.1). The peristyle (Fig. 10.2) was located in the central part of the palace and had been built using the calcareous stone extracted from Brazza Island nearby. The 16 columns have been realised with diﬀerent kind of materials; in each side of the peristyle, there are four red granite columns, of surely Egyptian origin, and two of Cipollino marble, likely Egyptian too. The two terminal columns are part of the masonry and are constituted of the same calcareous stone used for capitals and arches. Four columns are integrated inside the Diocletian Palace pediment and are constituted of red granite with the same origin as the others [1, 2].

10.2 Laser Cleaning Tests Most of the calcareous stones constituting the peristyle are covered by quite thick black crust (Fig. 10.3). This black crust does not cover the entire peristyle

84

D. Almesberger et al.

Fig. 10.1. Original Diocletian Palace

Fig. 10.2. Peristyle of the Diocletian Palace

surface as rain running oﬀ has whitening most of the exposed areas. A black layer up to 3 mm thick covers most of the capitals and, in particular, those protected by arches. In some areas, black crust up to 10 mm thick, formed during the 1700-year history of the monument, have been localised [3]. Crust pieces were extracted to identify their composition. Both Zagreb Restoration Institute and “Opiﬁcio delle Pietre Dure” of Florence chemically tested these samples. Deposits of gypsum, carbon from wood smoke have been identiﬁed inside crust samples. Then numerous trials were conducted to determine the speciﬁc correlation between crust thickness and Nd:YAG laser energy (λ = 1, 064 nm). For this

This page intentionally blank

10 Laser Cleaning of Peristyle in Diocletian Palace in Split (HR)

85

Fig. 10.3. Black crust on peristyle surface Table 10.1. Results Crust thickness (mm) 0.5 2

Spot diam. (mm)

Freq. (Hz)

Energy (mJ)

Fluence (J cm−2 )

2 3

5 5

400 1600

12.74 22.65

speciﬁc experimental part, an SFR (short free running) laser system has been used. This system operates with a pulse duration of 50–100 µs, located between that of the lasers with Q-switching short-pulse (2–10 ns) mode and those with normal free running long pulse 200–1,000 µs). More in detail, the speciﬁc pulse duration applied during these trials is of 80 µs. Results can be summarized as following (see Table 10.1): for crust thicknesses of about 0.5 mm, 20% of maximum laser energy is the optimum, whereas for crust of 2 mm thickness, the level applied has to be increased up to 80% of maximum energy. Moreover, two or more successive treatments of 0.8 mm ablation depth each have to be applied in order to avoid stone overheating which could have been caused by one powerful treatment. All these treatments have been performed on surfaces previously got wet. Quality control of ablated surfaces has been performed following the UNESCO RILEM recommendations concerning water absorption test. This trial necessitates the use of special probes which are put in contact with the material under investigation through a surface S(ø = 40 mm, S = 12.56 cm−2 ) that allowed the passage of the water. Water absorption is evaluated over time. The diﬀerence between water absorption, after 30 min. [A(30 )] and after 5 min. [A(5 )], divided by the surface S gives a dimensional parameter [A], which represents water absorbed over a certain period for surface unit . All these data are correlated as following: A = [A(30 ) − A(5 )]/S[ml cm−2 ]

86

D. Almesberger et al.

Water absorbed before laser treatment and in presence of black crust is less than that measured after that treatment. So, to lower water absorption of the laser cleaned surface, a siloxane treatment could be suitable.

10.3 Conclusions The anamnesis and diagnosis of the state of conservation of Diocletian Palace’s peristyle have been achieved by applying non-destructive techniques. The results of this preliminary research have allowed a complete understanding of the nature of the materials constituting the peristyle and the localisation and the importance of all the defects and damage present on it. From an accurate study of all the data collected, the laser technique has been considered to be the most appropriate one for peristyle cleaning. Before starting the laser cleaning program, a series of tests have been performed on sample areas in order to deﬁne all the variables involved in the cleaning process. The cleaning procedure, established applying the parameters obtained by all the previous tests, is now successfully adopted. The ﬁnal quality control consists in the evaluation and comparison of water absorption, colorimetric tests and microscopic investigations.

References 1. Tech. Report “Dioklecijanova palača u Splitu istražni radovi na kostrukciji peristila – Preliminarna ispitivanja sa prijedlogom sanacije i zaštite”, Ser.Co.Tec. 2. Tech. Report “Dioklecijanova palača u Splitu istražni radovi na kostrukciji peristila – Druga faza ispitivanja sa prijedlogom sanacije i zaštite”, Ser.Co.Tec.S. 3. A. Zanini, KONKAM 2004: “LASER cleaning of stone architectural surfaces and objects”. 4. D. Krstić, KONKAM 2004: “Istraživanje stanja peristila Dioklecijanove palače u Splitu”.

11 Phenomenological Characterisation of Stone Cleaning by Diﬀerent Laser Pulse Duration and Wavelength ∗

S. Siano1 , M. Giamello2 , L. Bartoli1 , A. Mencaglia1 , V. Parfenov3 , and R. Salimbeni1 1

∗ 2 3

Istituto di Fisica Applicata, Consiglio Nazionale delle Ricerche, Sesto Fiorentino, Italy [email protected] Dipartimento di Scienze Ambientali, Università degli Studi di Siena, Italy Research Centre Vavilov State Optical Institute, St Petersburg, Russia

Summary. The present work focuses on the main phenomenological features of stone cleaning by lasers. They are the removal rate, degree of cleaning, and chromatic appearance of the treated surface associated with diﬀerent conservation problems and laser parameters. A set of three diﬀerent outdoor stone conservation problems were investigated here. The measurement of the ablation rates were carried out on encrusted stone artefacts and two sets of prepared samples in order to derive general behaviour through repeatable measurements. The analysis of the irradiation tests provided a quantiﬁcation of the diﬀerent eﬃciencies, degree of cleaning, and chromatic appearances associated with the fundamental harmonic of Q-switching, long Q-switching, and short free running Nd:YAG lasers (1,064 nm), as well as with the second harmonic of Q-switching sources (532 nm).

11.1 Introduction Laser cleaning of stones was mostly carried out by short pulse Q-switching Nd:YAG (1,064 nm) laser up to a few years ago both from research and practical use standpoints. At the same time the interdisciplinary debate on the limits associated with the use of this technology grew along with the increasing dissemination of the laser approach in conservation. Several research groups documented and analysed problems related with the aggressiveness [1, 2] and the yellow appearance after QS lasers cleaning of several lithotypes [3–5]. Understanding and control of this latter problem represents a crucial point for the acceptance of the laser cleaning techniques in the various territorial and cultural contexts. This situation provided room for groups of scientists, which started to propose technological solutions for the mentioned problems. In particular, the

88

S. Siano et al.

use of short free running (SFR) Nd:YAG laser systems was formerly proposed [2] mainly to overcome the lack of selectivity and the aggressiveness of short pulse Q-switching (QS) lasers on fragile and deteriorated stone artefacts, in agreement with previous phenomenological observations by Asmus and co-workers, as well as by other teams. The ﬁrst SFR laser had a pulse duration of 20 µs, the shortest duration provided by solid state electronic drivers, at least ten times shorter than the typical pulse duration of FR lasers and 1,000 times longer than the longest available QS laser at that time. Such a temporal regime selection was aimed at ruling out from the ablation process signiﬁcant non-linear optical absorption, photomechanical, and plasma phenomena, which were responsible for the observed aggressiveness. Beside the high gradualness and the absence of photomechanical damage the novel technology was also eﬀective for solving the problem of the yellow appearance in various situations [5,6]. Such a problem was also addressed by using diﬀerent wavelengths second and third harmonics of Nd:YAG). Successful demonstrations were reported for diﬀerent conservation problems [6, 7]. Finally, a novel class of ﬁbre-coupled long Q-switching (LQS) Nd:YAG lasers, with pulse duration from tens of nanoseconds to some microseconds, was proposed for metal cleaning, which could also be employed in stone cleaning problems after a thorough experimentation [8]. The present picture is hence rather more complex than some years ago. The scientiﬁc community should provide a concrete support to restorers in order to help them select the suitable laser systems for the speciﬁc conservation problem they encounter. Here, we focused on three main concerns, which are removal eﬃciency, degree of cleaning, and chromatic appearance.

11.2 Experimental Methods Three Nd:YAG laser cleaning systems were used in the present study. (1) Fibre-coupled SFR laser (Smart Clean II, El.En.) with a variable pulse duration between 50 and 120 µs while increasing the output energy between 0.2 and 2 J per pulse; (2) Fibre-coupled LQS laser, with pulse durations between 120 and 950 ns, 300 mJ per pulse; (3) Articulated arm multi-wavelength QS laser system (Raﬀaello, Quanta System), 8 ns pulse duration, 450 mJ per pulse at 1,064 nm. Some cleaning tests were also performed with a ﬁbrecoupled QS laser (LaserBlast 50, Quantel), 20 ns pulse duration, 160 mJ per pulse. Two sets of prepared samples simulating black crust on stones were used for comparison purposes. The ﬁrst one was a set of tablets (A) achieved by pressing a homogenised dry mixture of xantine (85%) and black carbon (15%) in a typical press used for preparing tablets for IR spectroscopy. Xantine is a white mineral, which was selected as a good matrix for tablet production. The second set (B) was prepared by spreading a water mixture of gypsum (85 wt%), black carbon (10 wt%), and quartz (5 wt%) onto a sandstone sup-

11 Phenomenological Characterisation of Stone Cleaning

89

port. The ablation tests were mostly performed in water-assisted conditions, apart from a few trials using a LQS, as speciﬁed in the following. The laser spot was maintained constant through the tests while the ﬂuence variation was achieved by suitable mirrors and neutral density ﬁlters. The ablation rate (µm perpulse) was measured by a micro-proﬁlometer. Two Carrara marble and a large plaster fragment from outdoor artefacts presenting a black encrustation were selected for the present phenomenological study. These samples were a tortile column from Florence’s cathedral, a baluster from the external loggia of its dome, and an angel from a monumental cemetery of Florence (nineteenth century), respectively. The tortile column and the baluster were removed from their original location because of the advanced state of sulphation of the marble substrate. Stereomicroscope observation, thin section analysis, XRD and FT-IR were employed to characterise the deterioration typology and the cleaning results.

11.3 Ablation Rates on Standards Figure 11.1 displays the measured ablation rates achieved for the set of standards A using diﬀerent pulse durations in the LQS range. The general behaviour is the typical one of thermal ablation, which is characterised by a threshold ﬂuence, Fth , a linear region, and a saturation ﬂuence, Fs . As it can be seen in Fig. 11.1, in the present case Fth and Fs increase with the pulse length between 0.7–1.4 and 1.5–3 J cm−2 , respectively, while the saturation ablation rate is fairly constant around 5 µm, which gives an estimation of the optical penetration depth in the present black crust. This behaviour demonstrates the photomechanical contribution to the material removal increases at shorter pulse durations. Anyway, such a contribution appears to be conﬁned within the optical penetration depth in all of the cases.

Ablation rate [mm/pulse]

6 125 ns

5

290 ns 440 ns

4

950 ns 3 2 1 0 0

1

2 Fluence [J/cm2]

3

Fig. 11.1. Ablation rate of the standards A by diﬀerent pulse durations (LQS laser)

90

S. Siano et al. LQS 120 ns+H20 LQS 120 ns dry QS 532 nm QS 1064 nm

Ablation rate [mm/pulse]

50 40 30 20 10 0 0

0.5

1

1.5

2

2.5

3

Fluence [J/cm2]

Fig. 11.2. Ablation rates of the standards B by QS (1,064, 532 nm) and LQS (120 ns) lasers

Larger ablation rates were found for the set of standards B on which we tested three diﬀerent pulse durations (8 ns, 120 ns, 50–90 µs) and the Nd:YAG second harmonic (532 nm, 8 ns). The measurements of the ablation rates provided by QS laser were very diﬃcult because of the inhomogeneous spatial proﬁle, which produced irregular holes. The data reported in the plot of Fig. 11.2 correspond to averages of maximum ablation depths within the spot and among diﬀerent spots. They show a further decrease of the ablation threshold with the pulse duration when moving from 120 to 8 ns, which indicates a stronger photomechanical contribution. Furthermore, an ablation rate saturation around 40 µm per pulse was found at 120 ns, which allowed us to deduce a non-negligible inertial conﬁnement contribution at short pulse durations. SFR laser at the maximum pulse duration of 120 µs and in water-assisted conditions provided a fairly linear increase of the rate with the ﬂuence and pulse duration, without exhibiting the typical saturation behaviour within the investigated ﬂuence range. Thus in particular, at the maximum ﬂuence, FL = 30 J cm−2 , we found 60 µm per pulse. Such behaviour documents the occurrence of almost purely thermal regimes where the pulse duration can provide a non-negligible contribution to the ablation rate. Ablation rates associated with SFR laser (120 µs) were also measured on the tortile column, where the presence of a rather thick and fairly homogeneous black crust allowed repeatable measurements (see below). As displayed in Fig. 11.3, the derived ablation rates for the tortile column are in good agreement with those of the standards B. The absence of saturation up to 30 J cm−2 is a clear evidence of the mentioned thermal regime (which allows overcoming the ablation limit of the optical penetration depth) and absence of dense plasma formation.

11 Phenomenological Characterisation of Stone Cleaning

91

Ablation rate [mm/pulse]

100 B SFR 50 µsec

80

B SFR 90 µ3 B SFR 90 µ3

60

Column SFR 120 µs

40 20 0 0

5

10

15

20

Fluence

[J/cm2]

25

30

35

Fig. 11.3. Ablation rates provided by SFR lasers for the standards B and black crust of the column

11.4 Characterisation of the Cleaning Problems As mentioned above, we selected three diﬀerent cleaning problems in the present study. Representative stratigraphies are shown in Fig. 11.4. The baluster exhibited the relatively most complex stratigraphy. From the outer to inner layers, black crust, gypsum-binder scialbatura (thin plaster), very sulphated Ca-oxalates ﬁlm, and surface pseudomorphic sulphation can be observed. The scialbatura and the Ca-oxalates ﬁlm include a pigment load constituted by ochres and black carbon. The absence of the scialbatura layer represents the diﬀerence between the stratigraphy of the tortile column with respect to the previous one. Finally the plaster exhibited a pure black crust lying on a very irregular surface (with peaks and valleys) produced by water erosion. Some of the analysed samples also evidenced traces of a lead white layer, which were observable only on some of the peaks of the plaster substrate. All of these stratigraphies were very interesting in order to determine to what extent the presence of yellow-red pigments (ochres) can determine the ﬁnal yellow-orange appearance usually associated with cleaning treatments by the fundamental harmonic of QS Nd:YAG lasers and to investigate the genesis of the chromatic diﬀerences provided by diﬀerent laser systems. Actually, two layers with a pigment load (in the scialbatura and Ca-oxalates ﬁlm) were present within the stratiﬁcation of the baluster, a single layer (Caoxalates ﬁlm) within the one of the column, and none for the pure black crust of the plaster. Conversely, only the stratigraphy of the column was suitable for measuring and comparing ablation rates provided by diﬀerent laser systems.

92

S. Siano et al.

Fig. 11.4. Ultra-thin sections (plane polarised light) showing the stratigraphies of the investigated cleaning problems: baluster (top), tortile column (middle), plaster (bottom). (a) black crust, (b) scialbatura, (c) Ca-oxalates ﬁlm, (d) surface pseudomorphic sulphation, (e) marble substrate, (f ) lead white, (g) plaster substrate. The scale bars are 300 µm

11.5 Chromatic Appearance After Laser Cleaning Diﬀerent chromatic appearances of the laser cleaned surfaces were produced by changing the irradiation parameters. To the naked eye, the largest diﬀerences were found for the baluster, where QS Nd:YAG lasers at the fundamental harmonic provided a relatively strong yellow-orange eﬀect with respect to treatments with second harmonic or long pulse (SFR laser). Conversely, only small colour diﬀerences were found for the tortile column. The result provided by LQS laser on the baluster in dry conditions was very interesting. The cleaning provided the “whitest” surface without producing any relevant side eﬀect. Only QS lasers were extensively tested on the plaster because SFR laser results were not eﬀective and aggressive at the operative ﬂuences. The comparison between ﬁrst and second harmonic shows the latter one was

11 Phenomenological Characterisation of Stone Cleaning

93

Table 11.1. Colour coordinates, chroma, and hue angle of the tested areas after laser cleaning by diﬀerent laser systems and parameters Site/Fluence (J cm−2 ) Baluster A dry, 2 B dry, 2.5 C, 2 D, 7.8 E, 7.8 F, 7.8 G, 7.8 H, 1.2 I, 1.2 L, 0.65 M, 0.70 N, 0.70 O, 0.62 Plaster A, 0.63 B, 0.63 C, 0.52 D, 0.52 Tortile column A dry, 2.0 B dry, 2.5 C, 2.0 D, 0.61 E, 0.45 F, 5.0 G, 6.5

Laser

L∗

a∗

b∗

C∗

h∗ Angle

LQS 120 ns LQS 120 ns LQS 120 ns SFR 50 µs SFR 50 µs SFR 50 µs SFR 50 µs QS 20 ns QS 20 ns QS 8 ns QS 8 ns QS 8 ns QS 8 ns 2nd

65.46 5.63 17.34 68.04 4.87 16.85 62.49 7.34 20.17 70.68 7.33 20.79 74.42 6.26 19.70 70.73 6.95 21.67 69.23 6.19 19.89 62.58 9.08 21.78 73.72 10.47 24.26 56.59 9.68 21.09 69.94 7.74 22.39 71.52 7.80 24.03 75.76 6.32 20.23

18.2 17.5 21.5 22.0 20.7 22.8 20.8 23.6 26.4 23.2 23.7 25.3 21.2

1.26 1.29 1.22 1.23 1.26 1.26 1.27 1.17 1.16 1.14 1.24 1.26 1.27

QS QS QS QS

75.09 79.48 85.55 80.51

17.0 18.0 12.0 13.5

1.21 1.26 1.29 1.30

60.33 9.90 24.98 26.88 62.30 10.11 25.20 27.16 64.88 10.78 25.72 27.89 63.51 9.99 25.59 27.48 65.57 10.82 26.54 28.66 63.41 9.94 22.43 24.54 63.99 9.39 21.19 23.2

1.19 1.19 1.17 1.19 1.18 1.15 1.15

8 ns 8 ns 8 ns 2nd 8 ns 2nd

LQS 120 ns LQS 120 ns LQS 120 ns QS 8 ns 2nd QS 8 ns SFR 50 µs SFR 50 µs

5.98 5.46 3.30 3.55

15.91 17.15 11.55 13.05

eﬀective in reducing a slight yellow colour component associated with the former. All these diﬀerences can be “read” through the colorimetric parameterisation reported in Table 11.1. According to the naked eye observation, the lowest chroma were of the sites A, B, and O of the baluster and of sites C and D for the plaster.

11.6 Degree of Cleaning It is extremely important to associate the observed colour diﬀerences with the corresponding degree of cleaning or level in order to characterise the genesis of the phenomenon.

94

S. Siano et al.

Black crust and scialbatura were completely removed from the baluster by all of the tested laser systems. Whereas, the degree of removal of the Ca-oxalates ﬁlm represented the main stratigraphic diﬀerence of the various cleaning tests. The yellow-orange appearance was relatively pronounced whenever the ﬁlm was not completely removed by laser irradiation. A thicker residual ﬁlm layer was usually associated with a stronger chromatic saturation of the yellow and red components. Thus, for example, Fig. 11.5 shows a comparison between 1,064-nm irradiation of QS and SFR laser. Ca-oxalates ﬁlm residues of diﬀerent thickness and lateral size were observed in the QS case, whereas the ﬁlm was totally removed by SFR laser without any aggression of the surface pseudomorphic sulphation (Fig. 11.5). Even though a similar result was also observed with the tortile column as described above, the chromatic appearance was rather orange at all pulse durations. The ultra-thin section analysis showed that the genesis of this phenomenon is due to the pigmentation of the outer region of the pseudomorphic sulphation layer. As displayed in Fig. 11.6, not even second harmonic laser irradiation was able to remove it.

Fig. 11.5. Stratigraphic details of the ﬁnal cleaning level achieved on the baluster by QS (left) and SFR (right) Nd:YAG lasers (at 1,064 nm). Ca-oxalates ﬁlm residues can be recognised in the former case

Fig. 11.6. Cleaning transition showing the pigmented pseudomorphic sulphation layer after second harmonic Nd:YAG laser cleaning of the tortile column

11 Phenomenological Characterisation of Stone Cleaning

95

Fig. 11.7. Eﬀective and careful cleaning of the plaster provided by 1,064-nm Nd:YAG laser (top), compared to aggressiveness of the second harmonic on the substrate (bottom). (The pulse duration of the 532 nm is always short)

Finally, the 1,064 nm QS laser was very eﬀective in the cleaning of the plaster at a typical average ﬂuence of 0.6 J cm−2 . It provided a rough surface (Fig. 11.7), which represent the best result in the present case, according to the stratigraﬁc description of Fig. 11.4 (bottom). Conversely, the second harmonic and, as mentioned above, the SFR laser were too aggressive on plaster.

11.7 Discussion and Conclusions This work provides a picture of the diﬀerent phenomenologies involved in stone cleaning by a set of laser cleaning systems available on the market. The diﬀerences among ablation rates, degree of cleaning, and chromatic appearances these systems can provide, were characterised here and quantiﬁed by using a set of encrusted stone samples. We have shown that the relatively high eﬃciency of the QS Nd:YAG laser systems is associated with a narrow linear ablation range and yellow appearance for all substrates tested, especially in cases where ochres are present in proximity of the substrate. This latter problem can be solved by second harmonic, SFR, or LQS lasers, which are able to deepen the degree of cleaning up to the complete removal of the sulphated pigmented Ca-oxalates ﬁlm. QS lasers are too aggressive at the operative ﬂuences to achieve such a degree of cleaning. In this concern we also reported an interesting performance of the LQS laser systems, here employed for the ﬁrst time in stone cleaning. In dry irradiation condition their operative ﬂuence can be increased to 2–3 J cm−2 without producing relevant damage to the substrate. In the present tests on the baluster, a good stratigraphic and chromatic result was achieved at 2–2.5 J cm−2

96

S. Siano et al.

in dry conditions. When for some reasons, such as for example the natural diﬀusion, a ﬁnely grained pigment load is also embedded inside the upper region of the pseudomorphic sulphation (tortile column), the yellow appearance after cleaning is almost independent of laser source. Any attempt to remove it would produce serious damages to the artefact. QS lasers also produced a slight yellow appearance while removing the pure black crust on the plaster sample. This interesting feature needs further investigations in order to assess whether it is produced by iron oxides coming from pollution, organic radicals, or by a light scattering eﬀect. The second harmonic provided a white surface but it was unacceptably aggressive. Acknowledgements Thanks to Prof. G. Sabatini for the petrographic interpretation, Dr. L. Speranza, C. Biliotti, and I. Castello (OPD) for the support.

References 1. M. S. D’Urbano, C. Giovannone, P. Governale, A. Pandolﬁ, and U. Santamaria, in Proceeding of the 3 rd Symposium on the Conservation of Monuments in the Mediterranean Basin, V. Fassina, ed., Soprintendenza di Venezia, 1994, 955. 2. S. Siano, F. Margheri, P. Mazzinghi, R. Pini, R. Salimbeni, G. Toci, and M. Vannini, in Proceedings of Laser 95, Charleston, USA, 1995, 441. 3. P. Bromblet, M. Labouré, and G. Orial, in Journal of Cultural Heritage, Vol. 4, 17, 2003. 4. V. Vergès-Belmin and C. Dignard, in Journal of Cultural Heritage, Vol. 4, 238, 2003. 5. S. Siano, A. Casciani, A. Giusti, M. Matteini, R. Pini, S. Porcinai, and R. Salimbeni, in Journal of Cultural Heritage Vol. 4, 123, 2003. 6. S. Siano, A. Giusti, D. Pinna, S. Porcinai, M. Giamello, G. Sabatini, and R. Salimbeni, in Lacona V Proceedings, K. Dickmann ed., Berlin 2005, 171. 7. G. Marakis, P. Pouli, and V. Zaﬁropulos, in Journal of Cultural Heritage, Vol. 4, 83, 2003. 8. R. Salimbeni, R. Pini, and S. Siano, in Journal of Cultural Heritage, Vol. 4, 72, 2003.

12 The Cleaning of the Parthenon West Frieze by Means of Combined IR- and UV-Radiation K. Frantzikinaki1 , G. Marakis1 , A. Panou1 , C. Vasiliadis1 , ∗ E. Papakonstantinou1 , P. Pouli2 , T. Ditsa2 , V. Zaﬁropulos2 , and C. Fotakis2 1

2

∗

The Acropolis Restoration Service (YSMA), 10 Polygnotou str, 10555 Athens, Greece Institute of Electronic Structure & Lasers, Foundation for Research & Technology – Hellas, (IESL – FORTH), P.O. Box 1527, Heraklion, Crete 71110, Greece [email protected]

Summary. This chapter deals with the cleaning of the Parthenon West Frieze by means of an innovative laser cleaning methodology. Following a comparative study of various cleaning methods, laser cleaning was proven to be the most eﬃcient method for the removal of loose deposits and black crusts. The laser system employed is a Q-switched Nd:YAG system emitting at the fundamental and the third harmonic frequencies designed and developed by FORTH-IESL. The system emits in two wavelength beams individually or in combination. This feature, along with possible modiﬁcation of the laser parameters – energy density, number of pulses, the contribution of each beam to the ﬁnal combined beam – for each individual case of encrustation and substrate (marble, monochromatic layers), leads to a safe and controlled cleaning result. The project commenced in 2002 and was completed in January 2005. Since then, the Parthenon West Frieze is on display at the Acropolis Museum in Athens, Greece.

12.1 Introduction The West Frieze blocks were moved from the Parthenon into the Acropolis Museum laboratory in 1993. The systematic interventions commenced in 2000 and were divided into two stages. The ﬁrst stage (2000–2002) included consolidation of the surface, removal of bronze dowels and mortar from earlier treatments, and the reattachment of fragments. The second stage (2002–2005) focused on the aesthetic retrieval that was comprised of the surface cleaning and the sealing of cracks and gaps [1]. In order to achieve the best aesthetic eﬀect and reveal original details of the substrate and historical traces covered by black crust and deposits, the cleaning process commenced. After a comparative study of four cleaning methods, laser cleaning was indicated as the most eﬃcient method for the removal

98

K. Frantzikinaki et al.

of all types of encrustation present on the Parthenon West Frieze. A prototype laser system, emitting both at ultraviolet and infrared wavelengths, was developed under a research collaborative scheme formed between the Acropolis Restoration Service (YSMA) and the Foundation for Research and Technology, Hellas-Institute of Electronic Structure and Laser (IESL-FORTH).

12.2 Condition 12.2.1 Deterioration The deterioration of the Parthenon West Frieze was attributed to a combination of mechanical, physical, and chemical factors along with atmospheric pollution and the microclimate, as well as the microstructure of the Pentelic marble. Disintegration (exfoliation, ﬂaking, and delamination), loss of the marble mass, and formation of cracks are the results of the corrosion agents. Furthermore, mortars based on Meyer cement (composed of magnesium oxide and magnesium chloride) used in previous interventions also contributed to the surface deterioration [2]. 12.2.2 Monochromatic Surface Layers Two monochromatic surface layers are preserved and cover about one-third of the sculptured surface [3]. These layers can be distinguished as follows: (a) An underlying orange-brown layer, very well adhered to the marble surface and with a thickness of around 30–100 µm. Its main components are calcium oxalates, calcium phosphates and iron oxides [4] and it is described as the “epidermis” (skin), which is encountered as “patina” on many Greek and Roman monuments. (b) An outer beige layer that covers the epidermis and is described as the “coating.” It is a thin artiﬁcial layer (about 80–120 µm thick), composed mainly of calcium carbonate. These surface layers preserve original tooling traces and relief details and indicate the original surface. They are thus characterized as historical documents and should not be removed during the cleaning process. 12.2.3 Characterization of Black Crust The presence of sulphur dioxide (in the atmosphere) led to the formation of a gypsum layer, which up to a certain thickness preserves details of the relief and it should not be removed during the cleaning process [5]. Soot deposits uniformly cover the Frieze surface except in areas that were being wetted regularly even after the construction of a protective canopy in 1978. The layers

12 The Cleaning of the Parthenon West Frieze

99

of the deposits vary in thickness and composition, and they are classiﬁed into the following types: Loose deposits. A thin layer (up to 100 µm thick) consisting mainly of gypsum, calcite, organic compounds and traces of other minerals and metals. Homogenous compact crust with good adhesion to the surface, up to 150 µm thick. Dendritic black crust of signiﬁcant thickness consisting of recrystallized and reprecipitated calcium carbonate in dendritic formation and a mixture of gypsum, aluminium silicate compounds, and other atmospheric and mineral particles [6] XRD and SEM-EDAX analyzes have shown that in addition to soot, small amounts of heavy metals, such as copper (Cu), lead (Pb), iron (Fe), zinc (Zn) cobalt (Co), manganese (Mn), and chromium (Cr) are present on the sheltered sections of the marble surface [7].

12.3 Cleaning Process 12.3.1 Evaluation of the Cleaning Methods Following an assessment by the late Professor Th. Skoulikidis [8] of all the currently known cleaning techniques, four were deemed to be the most promising for further study. These were (a) application of absorptive poultices, (b) microblasting, (c) inversion of the gypsum layer into calcite, and (d) laser cleaning. A research project was implemented in order to provide a comparative study of these four techniques. The evaluation of the four cleaning methods was conducted by a gradual approach of the West Frieze’s real condition. Preliminary applications of the cleaning methods were initially performed on newer marble complements (dating to the 1960s) from the West Frieze. The cleaning tests were completed on representative surfaces of sculptures and architectural members of the Acropolis Monuments and ﬁnally on the West Frieze itself. Throughout this research project it was proven that the eﬃciency of the applied cleaning methods depends on the thickness and formation of the crust as well as on the type of substrate. Laser cleaning was selected as the method, which met all the required standards (preservation of the noble patina, the gypsum layer, and the monochromatic surface layers) and was proven to eﬃciently remove, homogeneously, all types of encrustation [6]. 12.3.2 Laser Cleaning Laser cleaning was investigated in collaboration with the IESL-FORTH. To avoid any discoloration (yellowing eﬀect) and structural damage, the FORTHIESL research team suggested combining the action of the two discrete laser ablation mechanisms.

100

K. Frantzikinaki et al.

Fig. 12.1. (a) The laser system employed, (b) West Frieze Block III (detail ), marble substrate during the cleaning process, and (c) West Frieze Block IX half cleaned (the corner, on the bottom right is not a laser cleaned area but an ancient complement recently identiﬁed and adjoined onto the Frieze, where no monochromatic layers were preserved due to its long outdoors’ exposure)

According to this scheme, a prototype laser system was developed (EPO 03386004.0-1253) which in its initial state was tested successfully on the Acropolis site. The laser system developed was a Q-switched Nd:YAG system emitting at the fundamental (1,064 nm) and the third harmonic (355 nm). This hybrid system (Fig. 12.1a) was accomplished by the simultaneous use of two laser beams of diﬀerent wavelengths, whose pulses are temporally and spatially overlapped. The energy densities of the two beams were set to a speciﬁc ratio as determined by the preliminary tests, to respond to every individual case of encrustation and substrate [6, 9, 10]. 12.3.3 Determination of the “Reference Surface Color” The marble appearance after the cleaning process and the deﬁnition of the “Reference Surface” were major aesthetic questions. One acceptable solution for the surface ﬁnal color result was the harmonization with the appearance of the rest of the blocks of the Parthenon Frieze. These are the total of 56 blocks at the British Museum (among them the ﬁrst two blocks of the West Frieze) and the 25 blocks at the Acropolis Museum (16 blocks of the North, 7 of the South and 2 of the East Frieze). Unfortunately, none of these surfaces could be accepted as the “reference surface” because they were exposed to different environmental conditions. The blocks of the Acropolis Museum maintain less monochromatic surface layers and information than that of the West Frieze side, since they remained unsheltered on the monument or buried in the ground after the explosion of 1687. In the case of the British Museum blocks, the extensive cleaning within the years 1936–1937 had irreversibly altered the marble surfaces into lighter colored than the originals. Lightness (blackwhite), hue (red-green), and chroma (yellow-blue) coordinates were measured using a colorimeter based on the CIE L∗ a∗ b∗ system (Table 12.1). Eventually, the color of the in situ architectural members of the Parthenon was selected as the most “acceptable.”

12 The Cleaning of the Parthenon West Frieze

101

Table 12.1. Colorimetric measurements according to the CIE L∗ (lightness), a∗ (hue), and b∗ (chroma) system

West Frieze blocks – washed areas North Frieze blocks South Frieze blocks West Parthenon columns British museum blocks New marble

L∗

a∗

b∗

70.44 70.32 76.50 68.40 77.17 80.80

2.78 4.00 3.60 3.20 1.34 1.10

13.00 17.00 15.60 13.00 9.80 4.90

12.4 Application of Laser Cleaning on the West Frieze 12.4.1 Marble Substrate The combination of the infrared and ultraviolet radiation at relatively low energy densities (FIR = 0.3−0.5 J cm−2 and FUV = 0.08−0.2 J cm−2 ) on loose deposits was successful without any discoloration to the underlying marble surface (Fig. 12.1b). However, in some cases where the substrate was nonhomogeneous due to deterioration, loose deposits remained among the marble crystals. There, infrared radiation alone at higher energy densities about 0.6 − 1 J cm−2 is applied selectively where it is needed. In areas with unstable loose deposits further decrease of the working ﬂuence values (FIR < 0.3 J cm−2 and FUV < 0.08) was considered. For the homogenous compact crust with good adhesion to the marble surface, infrared radiation alone was used at the beginning of the process, aiming to reduce its thickness. Then a combination of infrared and ultraviolet radiation at relatively high energy densities (FIR = 0.4 − 0.8 J cm−2 and FUV = 0.1 − 0.3 J cm−2 ) was successfully used for the removal of the rest of the crust. Application of the combined infrared and ultraviolet laser beams on dendritic black crust had no eﬀect at low energy densities. When the energy density increased signiﬁcantly, the cleaning was non homogeneous and visible fragments of black crust remained on the surface. Most acceptable results were obtained with infrared laser pulses at high energy densities in the range of 1 − 1.8 J cm−2 [10]. It should be mentioned that the cleaning eﬃciency (rate and degree of cleaning) could be enhanced by applying a thin ﬁlm of water saturated with calcium carbonate during cleaning with the use of infrared radiation. The water enhances the ejection of the encrustation material through its rapid heating and vaporization [11]. The presence of calcium carbonate prevents the dissolution of the gypsum layer [10, 12]. In cases where cleaning is achieved through the combination of ultraviolet and infrared radiation, the use of water was avoided since it resulted in uncontrollable and nonhomogeneous cleaning.

102

K. Frantzikinaki et al.

Finally, the combination of infrared and ultraviolet radiation, at densities lower than the ones used for cleaning, was applied to correct possible discoloration of the marble surface. (i.e., yellowing) [13] due to the application of infrared radiation alone. 12.4.2 Monochromatic Surface Layers The use of ultraviolet radiation or any combination of ultraviolet and infrared radiation could cause discoloration (grayish color) to the underlying monochromatic surface layers. The use of infrared radiation alone at 0.5–0.9 J cm−2 resulted macroscopically, in the acceptable cleaning of loose deposits and the homogeneous compact crust. To remove thick dendritic crust it was necessary to signiﬁcantly increase the ﬂuence of the applied radiation (sometimes up to 1.5–1.8 J cm−2 ) [10]. Repeated wetting with water, saturated in calcium carbonate, was also required in all cases to remove the dendritic crust. 12.4.3 Notes on the Cleaning Process During the cleaning process ancient color and tooling traces were revealed. Monochromatic surface layers were found on unexpected areas. Engravings, pins, and mortars from previous interventions not visible before, appeared. All these historical elements were hidden since they were covered by homogeneous compact crust and dendritic black crust of signiﬁcant thickness. It is noteworthy that when cleaning an area, the energy densities should be high at the beginning of the process. Then the cleaning was repeated for second or third time on the same area with lower energy densities until the desired result was reached. In cases where this ﬂuence exposure sequence was not followed, part of the crust could remain on the surface. In areas with thick dendritic crust, the process had to be repeated many times at high energy densities because some of the deposits remained on the surface. When the process met areas with mortars from previous interventions the sound audible during the laser cleaning process changed signiﬁcantly. 12.4.4 Safety and Ergonomy Major importance was given to the safety rules during the cleaning process. Black curtains were used to restrict the working area and block possible beam reﬂections. Conservators wore suitable laser safety eyewear and facemasks. Additionally, a dust extractor was employed. The maximum working time for each user was decided to be 2 h with regular breaks every half an hour. Eye/muscle strain appeared when the user worked either more than 2 h or without regular breaks.

12 The Cleaning of the Parthenon West Frieze

103

12.4.5 Documentation Following a conditional report applied on the Acropolis monuments since 1989 [14], cleaning process is well documented by a series of photographs and overlaid graphic mapping following all the stages of the intervention of individual areas.

12.5 Cleaning Control Quantitative and qualitative tests were performed to evaluate the eﬀects of the cleaning methods, on both the substrata and the encrustation. These included: – Stereomicroscopic observation of the surfaces – Colorimetric measurements following the CIE L∗ a∗ b∗ system for the quantitative assessment of any color variations induced by the cleaning process – Polarizing microscope. Observation of the ultrathin and cross sections (on newer marble samples) to examine the morphology of the deposits and the level of cleaning (in collaboration with the Institute of Geology and Mineral Exploration-IGME) – XRD, FTIR, and SEM-EDAX analyses to assess the composition of the encrustation and the contents of the gypsum both before and after cleaning – Artiﬁcial ageing tests to monitor long-term eﬀects of the applied cleaning methodologies. Irradiated samples originating from the newer marble complements of the West Frieze employed for the preliminary tests were placed in artiﬁcial ageing chambers (salt-spray tests according to ASTM B117 and exposure to UV radiation). No alteration was observed [6].

12.6 Conclusions Laser cleaning is fully controllable, leaves no by-products and oﬀers unique features such as high selectivity and precision, reveals details and historical traces, and yet preserves the sulphated outer surface. In the case of monochromatic surface layers, the use of infrared radiation alone resulted in acceptable cleaning, while the cautious combination of diﬀerent methodologies to adjacent areas did not cause any alteration to the ﬁnal aesthetic result. The combination of infrared and ultraviolet radiation has been shown to remove encrustation homogeneously and in a controllable way with no color change to the original marble surface. For acceptable results there was a unique set of parameters (applied energy density values, ultraviolet to infrared energy density ratios, and number of pulses) for the marble substrate and for thin encrustations. In cases with thick encrustations, infrared radiation alone was necessary at the beginning of the process to reduce the thickness of the crust.

104

K. Frantzikinaki et al.

Acknowledgments The authors would like to thank the Greek Institute of Geology and Mineral Exploration (IGME), the ﬁrst Ephorate of Pre-historical and Classical Antiquities and the Committee for the Preservation of the Acropolis Monuments.

References 1. E. Papakonstantinou, K. Frantzikinaki, A. Panou, P. Pouli, and V. Zaﬁropulos, in Proceedings of the XVI Congress of Classical Archaeology, 23026, Harvard University Art Museum, Boston/Cambridge 2003. 2. P. Koufopoulos, in Study for the Restoration of the Parthenon, Vol. 3a. Athens (1994). 3. A. Galanou and G. Dogani, in Study for the Restoration of the Parthenon. West Frieze, Vol. 3c. Athens (1994). 4. K. Kouzeli, N. Beloyannis, Ch. Tolias, and G. Dogani, in Proceedings of the Int. Symp. on the oxalates ﬁlms: origin and signiﬁcance in the conservation of works of art, Milano, 1989, 327. 5. Th. Skoulikidis, D. Charalambous, E. Papakonstantinou, and N. Beloyannis, in Proceedings of the Int. Congr. on the Deterioration and Preservation of Building Stones, Venice 1979, 439–452. 6. E. Papakonstantinou, K. Frantzikinaki, P. Pouli, and V. Zaﬁropulos, in Study for the Restoration of the Parthenon, Vol. 7. Athens (2002). 7. Th. Skoulikidis and D. Charalambous, in Study for the Restoration of the Parthenon. Heavy metals from the environment on the surface of the Acropolis monuments, Vol. 3c, Athens (1989). 8. Th. Skoulikidis, in Study for the Restoration of the Parthenon, Vol. 3c. Athens (1994). 9. P. Pouli, K. Frantzikinaki, E. Papakonstantinou, V. Zaﬁropolulos, and C. Fotakis, in Springer Proceedings Vol. 100, 2005, 333–340. 10. K. Frantzikinaki, A. Panou, C. Vasiliadis, E. Papakonstantinou, P. Pouli, Th. Ditsa, V. Zaﬁropulos, and C. Fotakis, in Proceedings of the 10th International Congress on Deterioration and Conservation of Stone, Vol. II, 2004, 801–807. 11. M. Cooper, Laser Cleaning in Conservation: an Introduction, Oxford (1998). 12. Th. Skoulikidis and N. Beloyannis, in Studies in Conservation, Vol. 29, 197, 1984. 13. V. Zaﬁropulos, C. Balas, A. Manousaki, Y. Marakis, P. Maravelaki-Kalaitzaki, K. Melesanaki, P. Pouli, T. Stratoudaki, S. Klein, J. Hildenhagen, K. Dickmann, B.-S. Luk’yanchuk, C. Mujat, and A. Dogari, in Journal of Cultural Heritage, Vol. 4, 249, 2003. 14. G. Doganis and A. Moraitou, in Study for the Restoration of the Parthenon, Vol. 2a. Athens (1989).

13 A Comprehensive Study of the Coloration Eﬀect Associated with Laser Cleaning of Pollution Encrustations from Stonework ∗

P. Pouli1 , G. Totou1 , V. Zaﬁropulos1 , C. Fotakis1,2 , M. Oujja3 , E. Rebollar3 , M. Castillejo3 , C. Domingo4 , and A. Laborde5 1

∗ 2 3

4 5

Institute of Electronic Structure and Lasers (IESL), Foundation for Research and Technology-Hellas (FORTH), P.O. Box 1527, Heraklion, Crete, 71110, Greece [email protected] Department of Physics, University of Crete, Heraklion, Greece Institute of Physical Chemistry Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain Institute of Structure of Matter, CSIC, Serrano 123, 28006 Madrid, Spain Instituto del Patrimonio Histórico Español, El Greco 4, 28040 Madrid, Spain

Summary. The application of infrared laser wavelength (at 1064 nm), for the removal of pollution encrustations from stonework is limited in many cases by discoloration eﬀects, which are commonly known as “yellowing”. The understanding of this eﬀect is crucial for the investigation of cleaning methodologies. Several methodologies have been proposed aiming to avoid or rectify such unpleasant colorations. The most promising ones are based on the combination of a cleaning beam at 1064 nm with an ultraviolet beam at 355 nm, either in spatial and temporal overlapping and in variable ﬂuence ratios (wavelength blending approach), or in sequential irradiation conditions (sequential approach). To evaluate the two approaches, a comparative study on two diﬀerent types of encrustations and substrates was carried out. Thick dendritic crust on limestone and homogeneous thin encrustation on marble have been studied on the basis of their composition and origin, together with the conditions that may induce yellowing eﬀects when laser cleaned with infrared wavelength. The analytical methods employed to characterize the crusts and the eﬀects of laser irradiation include colorimetric and optical microscopic observations of surface and cross sections, Confocal Micro-Raman Spectroscopy and FT-Raman. The obtained results have provided a comprehensive approach for understanding the reasons responsible for the discoloration eﬀect, while allowing the establishment of well-deﬁned lasercleaning methodologies that ensure optimum results without any colour or structural alterations of the original stonework surface.

13.1 Introduction In the last years, the number of monuments aﬀected by air and water pollution has greatly increased [1–3]. The most noticeable examples of stone suﬀering

106

P. Pouli et al.

from the ravaging action of urban pollutants are marble and limestone, in general the main materials of many statues and decorative elements in monumental buildings of high artistic and historical value. Air pollution causes black soiling and loss of structural integrity due to attack by atmospheric sulphur in various forms, most noticeable sulphur dioxide and the subsequent formation of gypsum. Sulphur pollution stems from a variety of combustion processes, while exhaust emissions from gasoline and diesel-powered vehicles are one of the main culprits releasing carbonaceous particles into urban atmospheres [4, 5]. For the cleaning of soiled historical buildings and monuments, several cleaning methods have been proposed including micro-sandblasting, chemical agents and laser cleaning [6, 7]. Although spectacular results have been obtained with laser cleaning, this approach is not without its problems. The most common unwanted side eﬀect is the discoloration towards yellow hues [8, 9]. Several interpretations have been introduced for explaining such coloration phenomena associated with laser cleaning at 1,064 nm (Q-switched) on the basis of diﬀerent physical and chemical processes. These include (a) diﬀerential scattering of light on the voids resulted from the selective vaporisation of the dark pollution particles embedded in the crust’s matrix [10], (b) staining of the original surface due to the migration of the yellowish fraction present in the pollution crust due to polar organic compounds [11], this layer due to brownish/yellowish residues of oxidized organic matter below the black crust being resistant to laser irradiation at 1,064 nm but removed at 355 nm [11], (c) uncovering of various pre-existing patinas [8], and (d) yellowing related to the presence of iron-rich spheres of 100–200 nm arising from the transformation of hematite [12, 13]. To avoid or rectify such eﬀects, two laser-cleaning methodologies have been developed and proposed; the wavelength blending approach [10,14] based on irradiation with a beam in which IR and UV radiation is spatially and temporally overlapped in chosen ﬂuence ratios and the sequential approach, which refers to sequential irradiation of IR and UV laser beams [11, 15]. Zaﬁropulos et al. [10] reported on the synchronous use of IR and UV laser pulses for the eﬃcient removal of encrustations from sculptures and monuments. By ﬁxing the ﬂuence levels at a certain ratio, it was possible to remove pollution encrustation without any noticeable coloration. This methodology was applied with success in the Parthenon West Frieze in Athens [14]. This work presents a comparative study for two diﬀerent types of encrustation and substrates, as well as the evaluation of the two diﬀerent laser-cleaning approaches. Homogeneous thin encrustation on marble and thick dendritic crust on limestone have been studied on the basis of their composition and origin, together with the conditions that may induce yellowing eﬀects when laser cleaned with infrared wavelength. Finally the two diﬀerent laser-cleaning strategies have been tested on both types of encrustation and their results have been compared.

13 Coloration Eﬀect with Laser Cleaning of Pollution Encrustations

107

13.2 Experimental Methods 13.2.1 Samples Two types of samples were studied in this work: limestone of the Hontoria variety (Burgos, Spain) and Pentelic marble (Athens, Greece). Three Hontoria limestone samples, named HL1, HL2 and HL3, were taken from the Burgos Museum currently under restoration. Hontoria variety is a high purity, porous, white limestone with well-characterized petrophysical properties [16]. Previous analyses on the monument revealed that the stone has a yellow/brown coloured patina that is the result of intentional application of a layer with protective and aesthetic purposes. The samples used for this study appeared covered with a hard black soiling layer (of non-uniform thickness, but about 0.5 mm thick in some areas) strongly adhered to the substrate. The marble sample (S1) used in this study was taken from a neoclassic monument in the centre of Athens. The marble substrate of Pentelic origin and white colour is covered with a relatively thin layer (∼100 µm) of dark pollution encrustation. The stone surface is sound without signs of weathering or damage. No ancient or intentional patination layers have been identiﬁed. 13.2.2 Laser Irradiation Irradiation of samples HL1 and HL2 was carried out using a Q-switched Nd:YAG laser (Quantel Brilliant B) that delivers pulses of 5 ns (FWHM) at a maximum repetition rate of 10 Hz. Irradiation was performed at the fundamental wavelength at 1,064 nm (IR) and its third harmonic at 355 nm (UV). Various conditions of energy per pulse using the IR or UV wavelength alone or in a sequential (SQ) mode (applying ﬁrst the IR beam) were chosen to irradiate rectangular zones of ca. 1 cm2 (Fig. 13.1, Table 13.1). On the other hand, irradiation of samples HL3 and S1 was carried out by a modiﬁed Q-switched Nd:YAG laser (Spectron, SL850) with pulse duration in the range of 15 ns and maximum repetition rate of 10 Hz. The system was

Fig. 13.1. Visual observation of the areas irradiated on the Hontoria limestone and Pentelic marble: (a) sample HL1, (b) sample HL2, (c) sample HL3 and (d) sample S1. For the irradiation conditions, see Table 13.1

108

P. Pouli et al.

Table 13.1. Irradiation conditions and colorimetric measurements of limestone and marble samples

limestone patina crust IR

UV SQ SM

L∗

a∗

b∗

– – – 0.6–0.9

88.8 70.8 52.8 57–67

2.3 5.3 0.9 2.7–4.1

10.3 14.8 4.8 13.7–19.7

0.3 – 0.6

57–64

0.5–0.8

4.7–8.7

64–65.7

1.8–2.4

10.9–11.8

56.7–70

3.2–4.3

12.5–15

89.6

−0.3

2

Samples and Zones

Range of Fluences (J cm−2 )

– – – HL1 HL2 HL3 HL1 HL2 HL1 HL2 HL3

(2, 4) (1, 3) (1, 3, 5, 8) (3) (4) (1, 5, 6) (2, 5) (2, 4, 6, 7)

–

FIR : 0.7–0.9 FUV : 0.3–0.4 FIR : 0.6–1 FUV : 0.2–0.3 FIR /FUV : 2–4 –

–

–

70–80

1.5–3

12–14

–

–

64.9

2.8

12.2

S1 (1) S1 (2) S1 (3)

0.8 0.2 FIR : 0.8 FUV : 0.2 FIR : 0.8 FUV : 0.2 FIR /FUV : 4

78 73.7 77.6

3.5 1.7 2.2

19.2 10.9 14

80

1.3

12.9

marble bulk marble surface soiling layer IR UV SQ SM

S1 (4)

SQ and SM refer to sequential and synchronous irradiation modes. The irradiation zone is indicated in parentheses

custom made [EPO 03386004.0-1253] in order to emit the fundamental wavelength and its third harmonic simultaneously as well as separately in variable energy outputs. Thus it was possible to control the relative ratio of the two components in the ﬁnal combined beam [14]. Irradiation tests were comparatively performed on both types of substrates/crusts (samples HL3 and S1) in single (IR or UV), sequential (SQ) (ﬁrst IR and then UV) and synchronous (SM) (in various EIR /EUV ratios) modes. The complementary use of a thin layer of water to enhance encrustation removal was also considered. The laser ablation thresholds for both types of crust and substrates were determined in previous studies [17, 18] and are given in Table 13.2.

13 Coloration Eﬀect with Laser Cleaning of Pollution Encrustations

109

Table 13.2. Laser ablation thresholds in J cm−2

1,064 nm 355 nm

Hontoria Limestone

Thick Crust

Pentelic Marble

Thin Crust

1.3 0.5

0.6 0.3

3.5 1.5

0.8 0.35

13.2.3 Instrumental Techniques A Minolta CM 2002 colorimeter served to measure the chromatic properties of the diﬀerent irradiated zones of the samples and to compare with the colour of the unsoiled stone surfaces. The CIE-Lab colour space was used to measure colour shifts expressed in the variables L∗ (luminosity), a∗ (red/green) and b∗ (yellow/blue). Five measurements were averaged to obtain one data point. Two laser-Raman systems were employed. Micro-Raman spectra were recorded with a Renishaw Raman Microscope System RM1000 equipped with a Leica microscope (using magniﬁcation ×50), an electrically refrigerated CCD camera and a diode laser at 785 nm as excitation source, operating at a power level of 3 mW. The spectra reported here have neither smoothing nor baseline treatments. FT-Raman spectra were recorded with an RFS 100/S-G Bruker spectrometer with a cooled Ge detector. The excitation source consisted of a Nd:YAG laser emitting at 1,064 nm. Each data point was the result of the accumulation of 100 scans. The resolution was 8 cm−1 .

13.3 Results and Discussion Visual observation of the areas of the limestone samples (Fig. 13.1) irradiated at 1,064 nm or with a combination of 1,064 and 355 nm, both in the sequential and simultaneous mode, indicate eﬃcient removal of the black soiling. However, uniform removal of the thick soiling layer is hindered by very hard adhesion of the dark crust to the substrate in some zones. The synchronous laser-cleaning mode performed in sample HL3 (Fig. 13.1c) results into a more eﬀective removal of the soiling layer, although still the ﬁnal surface lacks uniformity. Application of the UV laser beam alone could not remove the crust satisfactorily. The most eﬃcient and pleasing cleaning result was obtained with the simultaneous application of IR and UV wavelengths. Irradiation of the marble sample (Fig. 13.1d) with only IR or UV beam causes discoloration, giving rise to yellow and grey hues, respectively. This discoloration is avoided when the two beams are applied in combination. The cleaning eﬀect obtained in sequential use of the two beams lacks uniformity and homogeneity. Nevertheless, the simultaneous use yields very satisfactory results (no discoloration or inhomogeneous cleaning).

110

P. Pouli et al.

13.3.1 Colorimetry Colorimetric data (Table 13.1) indicate that areas of Hontoria samples irradiated with the UV beam only display unacceptable low values of a∗ and b∗ parameters. Irradiation with the IR beam only results into areas with higher b∗ values, indicative of yellowing (also higher a∗ values, more red). Irradiation with a combination of the IR and UV beams in a sequential mode (samples HL1 and HL2) results in colour characterized by lower b∗ (less yellow) and lower a∗ values than IR only. Simultaneous irradiation with the two beams (sample HL3) yields lower b∗ values than the areas irradiated with the IR only. Direct comparison between the sequential and simultaneous irradiation is difﬁcult, as it was not made on the same sample. On the marble sample, the irradiation at 1,064 nm results in a high b∗ value, indicative of unacceptable yellowing. Too low b∗ value is displayed by the UV irradiated area. The comparison between irradiation in the sequential and simultaneous mode with IR and UV laser indicates that the yellowing discoloration produced is corrected in the latter. 13.3.2 Raman Spectroscopy Micro-Raman and FT-Raman spectra (Figs. 13.2–13.4) were employed to extract conclusions on the composition of the samples and possible changes due to laser irradiation under the diﬀerent conditions. (A) For Hontoria limestone samples (HL1, HL2 and HL3), the representative spectra are shown in Figs. 13.2 and 13.3. The bare stone spectra [18] show the bands of CaCO3 as calcite at 1,084, 711 and 279 cm−1 , while spectra taken on remains of yellow/brown patina show additional bands of gypsum

G

G

b) Laser-cleaned HL1

G

G

G

Carbon

G

G

Phosphate

O G

G Cc

G

Cc

Cc

O

Cc

G O G

Cc

Raman signal (a.u.)

a) Black crust on Hontoria limestone

Z2 Z4 Z3

G Z1

800

1200

1600

400

800

1200

1600

Wavenumber/cm−1

Fig. 13.2. Micro-Raman spectra of Hontoria sample (HL1). Cc, G and O stand for calcite, gypsum and oxalate, respectively. Irradiation conditions Z1: SQ FIR = 0.6 J cm−2 and FUV = 0.3 J cm−2 , Z2: FIR = 0.6 J cm−2 , Z3: FUV = 0.3 J cm−2 and Z4: FIR = 0.8 J cm−2

13 Coloration Eﬀect with Laser Cleaning of Pollution Encrustations b) IR only Cc

Cc

Cc Phosphate

C=O

O

Z7

Cc Phosphate

CH, CH3

Cc

G

Cc

Cc

G

G

Cc

Raman signal / a.u.

a) Simultaneous IR+UV

Z6

G G

GG

Z5 Carbon

G O

G

CH, CH3

1000

1500

C=O

O Z1

Z4

500

111

2000

500

1000

1500

2000

-1

Wavenumber / cm

Fig. 13.3. FT-Raman spectra of Hontoria sample (HL3). Cc, G and O stand for calcite, gypsum and oxalate, respectively. Irradiation conditions Z1: FIR = 0.6 J cm−2 , Z4: SM FIR = 0.8 J cm−2 and FUV = 0.3 J cm−2 , Z5: FIR = 0.7 J cm−2 , Z6: SM FIR = 1 J cm−2 and FUV = 0.3 J cm−2 and Z7: SM FIR = 1 J cm−2 and FUV = 0.2 J cm−2

(CaSO4 · 2H2 O) (at 1,136, 1,007, 491 and 412 cm−1 ) and oxalate (C2 O2− 4 ) at 1,464 and 909 cm−1 . The 960 cm−1 band observed in some spectra indicate that the patina layer contains also phosphates, while there is no indication for iron oxides. The black crust (Fig. 13.2a) contains gypsum and carbon, the latter evidenced by the 1,590 and 1, 350 cm−1 graphitic and disordered bands of carbon, normally displayed by soot particles. For irradiation at 1,064 nm, with a ﬂuence of 0.6 J cm−2 (Fig. 13.2b, zone 2), only gypsum is observed in the spectra of the irradiated area, indicating partial removal of crust. Other spectra in the same zone show the presence of calcite; a fact that points out that non-uniform cleaning is performed possibly due to the variable thickness of the crust. Similarly, areas irradiated with the same wavelength but at higher ﬂuences (0.8 J cm−2 , Fig. 13.2b, zone 4) show gypsum and traces of calcite, indicating a more eﬃcient removal of black crust. Appearance of the oxalate bands at 1,464 and 909 cm−1 implies that the patina has been uncovered by laser cleaning. Irradiation at 355 nm (0.3 J cm−2 , Fig. 13.2b, zone 3) gives rise to calcite, gypsum bands and phosphates indicating partial removal of black crust, while areas irradiated by the combination of IR (0.6 J cm−2 ) and the UV (0.3 J cm−2 ) beams in a sequential mode (Fig. 13.2b, zone 1), result into calcite bands (as well as low-intensity gypsum bands) indicative of total crust removal. However, absence of the oxalate bands suggests that the laser has also removed the patina. Figure 13.3 compares the FT-Raman spectra of areas irradiated with the IR beam alone (Fig. 13.3b) and the IR and UV beams simultaneously

112

P. Pouli et al. a) Stone and crust

Bare stone Carbon Cc

Cc

G

G G GG

G R dark soiled

500

1000

1500

2000

c) IR only

Carbon

G

500

1000

1500

2000

Cc

Cc

Cc

CH, CH3

G

1000

Cc

Cc

G

500

Z2

G

d) Simultaneous IR + UV

Cc

Raman units/ a.u.

Cc

Cc

Cc

Cc

Cc

Cc

b) UV only

Z1

C=O G

1500

2000

500

Z4

G

G G GG

G

1000

1500

2000

Wave number/ cm−1

Fig. 13.4. FT-Raman spectra of marble sample (S1). Cc, G and O stand for calcite, gypsum and oxalate, respectively. Irradiation conditions Z1: FIR = 0.8 J cm−2 , Z2: FUV = 0.2 J cm−2 , Z4: SM FIR = 0.8 J cm−2 and FUV = 0.2 J cm−2

(Fig. 13.3a). Bands at 1,433 and 1, 749 cm−1 are assigned to CH, CH3 and CO and associated with organic residues not completely removed by laser irradiation. The spectra indicate a more eﬃcient removal of the soiling layer by simultaneous IR and UV irradiation. (B) In the case of the marble sample, both micro-Raman [18] and FTRaman (Fig. 13.4) results allow us to ascertain the calcite composition of the bare stone, while dark soiled areas contains additional bands assigned to carbon and gypsum. The zone irradiated at 355 nm displays strong carbon bands indicating that the dark soiled area has not been removed. On the other hand, upon irradiation at 1,064 nm, a decrease of the intensity of carbon bands is observed, while gypsum bands are still present. This fact supports the argument that coloration eﬀects in this regime may be due to selective vaporisation of the black particles embedded into the gypsum layer. Irradiation with the combination of the IR and UV beams shows the calcite bands from the substrate and the gypsum bands from the crust. To the limit of Raman detection, no additional chemical compounds, other than calcite and gypsum, are observed.

13 Coloration Eﬀect with Laser Cleaning of Pollution Encrustations

113

13.3.3 Discussion The Raman results indicate that calcite is the main component of the bare stone for both types of samples. The patina of the limestone sample contains oxalates, while the thick black crust, strongly attached to the Hontoria limestone, is composed of carbon, gypsum and phosphates. Black crust on the marble sample is composed of gypsum and carbon. On the limestone and marble samples, the observed bands of CH, CH3 and CO are indicative of organic residues. No signs of iron oxides are present in any spectra of limestone or marble samples. Raman spectra do not reveal any diﬀerence in chemical composition of areas irradiated sequentially or simultaneously in either types of stone. This evidence would support the hypothesis of physical eﬀects based in diﬀerent scattering properties.

13.4 Conclusions Both IR and UV radiation may cause discoloration on limestone and marble to yellow and grey hues, respectively. Nevertheless, these eﬀects can be avoided with the combination of the two beams in sequential and synchronous mode. No evidence of chemical alteration on the areas irradiated with the combined beams in simultaneous or sequential mode could be detected. The hypothesis that discoloration is associated with physical eﬀects based on diﬀerential scattering is supported for these types of samples, although work is underway by using complementary analytical techniques to analyse the possible presence of polar compounds in the crust and the stone matrix. Irradiation in simultaneous mode appears to result in a more satisfactory result both in colour and surface homogeneity. These results have provided a comprehensive approach for understanding the reasons responsible for the discoloration eﬀect and allow the establishment of well-deﬁned laser-cleaning methodologies that ensure optimum results without any colour or structural alterations of the original stonework surface. Acknowledgements Thanks are given to R. Fort (Instituto Geología Económica CSIC) for using the colorimeter. MO and ER acknowledge CSIC (I3P program) for a contract and a fellowship.

References 1. A. Moropoulou and S. Kefalonitou, in Building and Environment, Vol. 7, 1181, 2002. 2. P. Maravelaki-Kalaitzaki and G. Biscontin, in Atmospheric Environment 33, 1699, 1999.

114

P. Pouli et al.

3. A. Moropoulou, K. Bisbikou, K. Torfs, R. Van Grieken, F. Zezza, and F. Macri, in Atmospheric Environment, Vol. 32, 967, 1998. 4. P. Ausset, F. Bannery, M. Del Monte, and R.A. Lefevre, in Atmospheric Environment, Vol. 32, 2859, 1998. 5. C. Saiz-Jimenez, in Atmospheric Environment, Vol. 27B, 77, 1993. 6. M. Cooper, Laser Cleaning in Conservation: An Introduction, Oxford, 1998. 7. C. Rodríguez-Navarro, K. Elert, E. Sebastián, R.M. Esbert, C. María Grossi, A. Rojo, F.J. Alonso, M. Montoto, and J. Ordaz, in Reviews in Conservation, Vol. 4, 65, 2003. 8. V. Vergès-Belmin and C. Dignard, in Journal of Cultural Heritage, Vol. 4, 238, 2003. 9. V. Zaﬁropulos, C. Balas, A. Manousaki, Y. Marakis, P. Maravelaki-Kalaitzaki, K. Melesanaki, P. Pouli, T. Stratoudaki, S. Klein, J. Hildenhagen, K. Dickmann, B.S. Luk’Yanchuk, C. Mujat, and A. Dogariu, in Journal of Cultural Heritage, Vol. 4, 249, 2003. 10. V. Zaﬁropulos, P. Pouli, V. Kylikoglou, P. Maravelaki-Kalaitzaki, B.S. Luk’yanchuk, and A. Dogariu, in LACONA V Proceedings, Springer Proceedings in Physics, Vol. 100, 311, 2005. 11. M. Gaviño, M. Castillejo, V. Vergès-Belmin, W. Nowik, M. Oujja, E. Rebollar, B. Hermosin, and C. Saiz-Jimenez, in Air Pollution and Cultural Heritage, Edited by C. Saiz-Jimenez, A.A. Balkema Publishers, Leiden, 239, 2004. 12. S. Klein, F. Fekrsanati, J. Hildenhagen, K. Dickmann, H. Uphoﬀ, Y. Marakis, and V. Zaﬁropulos, in Applied Surface Science, Vol.171, 242, 2001. 13. M. Gracia, M. Gaviño, V. Vergès-Belmin, B. Hermosin, W. Nowik, and C. Saiz-Jimenez, in LACONA V Proceedings, Springer Proceedings in Physics 100, 341, 2005. 14. P. Pouli, K. Frantzikinaki, E. Papakonstantinou, V. Zaﬁropulos, and C. Fotakis, in LACONA V Proceedings, Springer Proceedings in Physics 100, 333, 2005. 15. S.S. Potgieter-Vermaak, R.H.M. Godoi, R. Van Grieken, J.H. Potgieter, M. Oujja, and M. Castillejo, in Spectrochim. Acta A 61, 2460, 2005. 16. R.M. Esbert, J. Ordaz, F.J. Alonso, and M. Montot, Manual de diagnosis y tratamiento de materiales pétreos y cerámicos, Col.legi d’Aparelladors y Arquitects Tècnics de Barcelona, Barcelona 1997. 17. G. Marakis, P. Pouli, V. Zaﬁropulos, and P. Maravelaki-Kalaitzaki, in Journal of Cultural Heritage, Vol. 4, 83, 2003. 18. P. Pouli, G. Totou, V. Zaﬁropulos, C. Fotakis, M. Oujja, E. Rebollar, M. Castillejo, C. Domingo, P.P. Pérez, and A. Laborde submitted to Spectrochimica Acta A.

14 Poultices as a Way to Eliminate the Yellowing Eﬀect Linked to Limestone Laser Cleaning V. Vergès-Belmin1 and M. Labouré2 1

2

Laboratoire de recherche des monuments historiques, Champs sur Marne, France [email protected] Fondation de l’Oeuvre Notre-Dame, Strasbourg, France

Summary. One of the main drawbacks raised against stone cleaning by Nd:YAG Q-switched lasers (1,064 nm, 6–20 ns) is the yellow aspect left after cleaning. It is well known among stone conservators that one may attenuate this colour using waterbased poultices. We tested four kinds of poultices in two areas on the northern portal of the Saint Denis cathedral, which had been laser cleaned in 1997. The yellow aspect visibly decreased immediately after poultice removal, and the eﬀect remained visible after 12 months. Colour measurements performed before, just after, 3 and 9 months after poulticing conﬁrmed and quantiﬁed observations. Scotch tape tests, optical microscope and SEM-EDS observations and analyses show that this reduction is due at least partly to poultice remnants when poultice contains attapulgite or carboxymethyl-cellulose. Fume silica poulticing also seems to leave relics on the treated surface. The use of cellulose alone leads to a slight de-yellowing without any poultice relics being present. In that case, the de-yellowing could be due to solubilization of yellow products or physical detachment of particles.

14.1 Introduction Nd:YAG Q-switched lasers, working at the fundamental wavelength (1,064 nm), are often held responsible for stone yellowing. The issue has been much debated but the reasons for the yellow appearance are still not clearly established. As a consequence, the technique extensively used on restoration campaigns in France in the 1990s is now somewhat kept away by architects. In the meantime, poultice application on laser-cleaned surfaces was found as a way to reduce the yellow eﬀect. This chapter aims at quantifying the inﬂuence of such poulticing on the stone colour. It was also intended to clarify the reasons why poultices may reduce the yellowing eﬀect. Finally, the durability of this reduction was assessed over a 9-month period. The selected site is the twelfth century portal of the northern transept at the Saint-Denis cathedral, a monument of highest importance for art history, located 10 km away from Paris, France. The portal upper parts were cleaned in 1990 using a combination of techniques: nebulization, Mora poultices and microsandblasting [1].

116

V. Vergès-Belmin and M. Labouré

The lower parts (jambs and pier) were covered with thick black crusts (up to a few millimetres). They were laser cleaned by the Company Quelin-Groux in 1997 [2]. The cleaning was performed with an Nd:YAG Q-switched laser (NL 201 from B.M.I) delivering 1 J at 30 W. Tap water was sprayed on the stone while ﬁring the laser, as usual, in order to increase the cleaning rate. As a result, the portal acquired a particularly contrasted appearance, the lower parts having a yellow-brown hue and the upper parts appearing light-beige. This contrast appeared to be clearly linked to the cleaning techniques [3]. This observation and other similar ones led restorers to think that maybe water would be an active medium to remove, or at least reduce, the laser yellowing eﬀect. As the restorations in St. Denis cathedral were very well documented and the composition of the black crust thoroughly characterized [4], this site appeared to be an appropriate place for testing deyellowing methods.

14.2 Background Knowledge 14.2.1 Diﬀerent Explanations of the Yellowing Eﬀect Many causes of laser yellowing were identiﬁed, possibly occurring simultaneously [5]. First of all, various underlying yellow layers may be uncovered by laser cleaning: a yellowish epigenic gypsum layer is frequently found beneath black deposits on limestone and marble [6, 7]. The yellowish brown hue of the layer has been attributed to impregnation by oxidized organic matter originating from the polluted atmosphere. Gavino et al. [4] present evidence that a yellow fraction composed of polar organic compounds is extracted from St. Denis cathedral black crusts. These compounds are not aﬀected by 1,064 nm Nd:YAG laser irradiation, indicating that, at least, part of the yellowing eﬀect after laser cleaning is linked to organic compounds not eﬀectively removed [8]. Evidence of such a layer has been shown by optical microscopy in cross sections [9, 10]. Orange-brown oxalate patinas, widely described in the literature [11], can also be found on the surface of a sulfated stone substrate, underneath the black crust. Experiments have shown [12] that, during laser cleaning, the oxalate patina can be preserved, without damaging underlying layers. Other underlying yellow layers found on inorganic substrates include pigment washes and coatings such as shellac or oily media [13, 14]. Another explanation raised for a yellow appearance after laser cleaning could be the presence of soiling residues, perhaps transformed by the laser radiation. In the case of stones, this hypothesis is supported by the common observation that individual laser shots on black crusts can result in the creation of a brownish circle of residues around the laser impact spot. Evidence of soiling residues was produced [15]. When laser cleaned at 1,064 nm, artiﬁcially sulphated marble plates, soiled with a mixture of hematite, gypsum and graphite, became measurably yellower (∆b in the order

14 Poultices as a Way to Eliminate the Yellowing Eﬀect

117

of 7–10 units). The authors attributed the yellowing to iron-rich microspheres of 100–200 nm diameter, newly formed over the cleaned surface. They concluded that the yellow appearance was due to iron residues remaining from the soiling. Evidence that yellowing is due to substrate damage was not demonstrated on stone, but rather in the case of paper by Kolar et al. [16]. Soiling in that case plays an important role as it reacts with the radiation leading to substrate damage causing yellowing. Finally, a light scattering model was raised [17], considering that the presence of a thin absorbent layer, changes in the surface roughness and/or lasercreated voids may account for a change in the reﬂectance spectrum towards wavelengths corresponding to a yellow hue. 14.2.2 Everyday Practice to Deal with the Yellow Eﬀect The way that the Paris cathedral western portals were cleaned between 1998 and 2000 illustrates quite well a search for a proper methodology to minimize aesthetical criticisms raised against laser cleaning [2]. Each portal was cleaned in a diﬀerent way: (1)North portal : (a) application by brush of a 5–10% ammonium carbonate + 2–3% CMC solution, (b) application of paper sheets impregnated with the same solution, and (c) rinsing and brushing with water and, last, complete laser cleaning. (2)Central portal : (a) laser cleaning, and (b) cellulose powder + tap water poultices/1 h. (3)South portal: (a) application of moistened paper pulp, and (b) water rinsing and brushing, (c) complete laser cleaning. It is diﬃcult to say today what role each of these diﬀerent procedures had on the slight colour diﬀerences that one can observe on the three portals. De-yellowing poultices were also applied in 2004 by the Quélin-Groux company on the Rue chapel (Somme, France), in order to harmonize the lasercleaned areas of the portal with the micro-sandblasted zones of the façade.

14.3 Materials and Methods 14.3.1 Poultices Taking into account the diﬀerent methods currently used by stone conservators, we tested four kinds of poultices (Table 14.1). The dry powder was mixed in deionized water until getting appropriate properties (plasticity and elasticity) for an application on vertical surfaces. One of the cellulose poultices (No. 4 in Table 14.1) was made with water containing 3 wt% carboxy-methylcellulose (CMC). CMC was added a few days before poultice preparation, in order to give it a better adhesion to the substrate. The poultices were applied on October 17th 2004, and removed after 18 h. drying. The two test areas are located on the lower part of the portal: left side of the pier, which is a nineteenth century restoration, and left jamb, which is a twelfth century part.

118

V. Vergès-Belmin and M. Labouré Table 14.1. Composition of the poultices

Area No. Composition

Trademark

Size

1

clarsol/ hydro-façade

0–50 µm deionized water

CTS-France arbocel arbocel

≈1 µm 200 µm 200 µm

2 3 4 a

20% Attapulgite /80 % glass microspheres (250 µm) fume silica cellulose cellulose

Moisturizing Medium

deionized water deionized water deionized water +3% CMCa

Carboxy-methyl-cellouse

14.3.2 Colour Measurements Naked eye observation and photo recording were realized before, immediately and then 3, 9 and 12 months after the treatments. Colour measurements were performed with a MINOLTA Chromameter CR110 colorimeter ﬁtted with a 5 cm diameter measuring head. Three measurements were performed three times on each 6×6 cm square, before, after poulticing and then 3 and 9 months after the tests. The colour values are expressed in the L∗ a∗ b∗ system. A four or ﬁve unit colour shift for each parameter is usually assumed as the minimum change visible to the human eye. The precision of these measurements is about ±0.1 unit. Total colour variation (treated vs. not treated stone) ∆E ∗ was 2 2 2 calculated according to the formula: ∆E ∗ = (∆L∗ + ∆a∗ + ∆b∗ )1/2 . The evolution of the b parameter was calculated simply by subtracting the b value of the considered area after treatment by the corresponding value before treatment. A de-yellowing thus appears as a negative value of ∆b∗ . As colour variations of stone surfaces exposed to outdoor conditions may vary according to the relative humidity, we performed a series of measurements on a reference zone, not treated, selected as close as possible from the test areas. 14.3.3 Naked Eye, Microscope and SEM-EDS Analyses Nine months after the tests, once all measurements were completed, each zone was observed using a ×20 magnifying lens, and then submitted to a scotch tape test, a technique derived from medical applications. An adhesive tape R (3M Scotch Magic tape 810, size 19 × 60 mm) was applied over the surface, and gently pressed so that any surface deposit with a poor adhesion could be collected. During transportation to the laboratory, the adhesive side of the tape was protected so that no additional particles could soil them. The tapes were observed and analysed using an optical microscope and SEM JEOL JSM T30, ﬁtted to an EDS spectrometer OXFORD and working under low vacuum. The samples were not metallized, and morphological images were collected using a back-scattered electrons (BSE) sensor, accelerated voltage being 15 kV. Qualitative EDS analyses were performed on 300×100–300×300 µm (distance) areas, within the particle clutches being present on the tape surface.

14 Poultices as a Way to Eliminate the Yellowing Eﬀect

119

14.4 Results and Discussion 14.4.1 Naked Eye and ×20 Magnifying Lens Observations One week after the poultices were removed, similar evolutions were recorded on pier and jamb for cellulose-based poultices: the area treated with cellulose alone showed a slight shift towards a lighter hue. The addition of CMC resulted in a much whiter aspect. The mineral poultices, attapulgite/glass beads and fume silica, did not give consistent results on the pier in comparison to the jamb: on the pier, a slightly clearer aspect could be noted as a result of both poultices. On the jamb, fume silica gave a whitish aspect quite similar to the one obtained with cellulose/CMC, while attapulgite/glass beads induced a lightening intermediate between those of fume silica and cellulose alone(Fig. 14.1).

Fig. 14.1. Aspect of the portal and test areas 12 months after poultices removal: (a) pier; (b) jamb

120

V. Vergès-Belmin and M. Labouré

A closer view of the areas, using a ×20 magnifying lens, indicated that there was some kind of ﬁbrous deposit on the areas treated with cellulose/CMC, both on jamb and pier. Twelve months after the test, the evolution of the test areas had kept the same aspect. 14.4.2 Colour Measurements Colour measurements conﬁrmed naked eye observations. The pier and jamb showed a coherent evolution for cellulose-based poulticing: in the case of cellulose alone, a signiﬁcant b value decrease of 4 units was noted, while the addition of CMC in the poultice induced a further b value decrease by 8–11 units. The mineral poultices, attapulgite/glass beads and fume silica, did not give consistent results on the pier and with respect to the jamb: on the pier, the b decrease was the same as the one induced by cellulose alone (about 4 units) while, on the jamb, the b value decrease reached 7 units for clay/glass and 9 units for fume silica (Fig. 14.2). The b value had generally slightly decreased over this time (Fig. 14.3). The dispersion of the b values was signiﬁcantly lower for cellulose poultices.

Delta b* 4 0 reference −4

clay+ glass

fume silica cellulose

cellulose + CMC

−8 −12 Pier

Delta E*

12

Jamb

8 4 0 reference

clay + glass

fume silica

cellulose

cellulose + CMC

Fig. 14.2. ∆b∗ and ∆E ∗ calculated from colour measurements performed on two test areas

14 Poultices as a Way to Eliminate the Yellowing Eﬀect Fume silica

30 25 20 15 10 5

b value

b value

Clay +glass

oct 04 before

oct 04 after

jan-05

30 25 20 15 10 5

jun-05

oct 04 before

oct 04 after

oct 04 after

jan-05

jun-05

Cellulose + CMC

b value

b value

Cellulose 30 25 20 15 10 5 oct 04 before

121

jan-05

30 25 20 15 10 5 oct 04 before

jun-05

oct 04 after

jan-05

jun-05

Reference

b value

30 20 10

oct 04 before

oct 04 after

jan-05

Pier

Jamb

jun-05

Fig. 14.3. Evolution of the b value over time for the poultices tested

14.4.3 Optical Microscope and SEM-BSE/EDS Observations and Analyses Optical microscope and SEM-BSE observations of the adhesive tapes conﬁrmed that the area no. 4 was bearing ﬁbres, suﬃciently loose to have been trapped by the adhesive tape (Fig. 14.4). These ﬁbres appeared to be composed of carbon and oxygen as major elements, the minor elements being Na, Mg, Al, Si, S, Cl and K. We deduced both from morphological and analytical data that these ﬁbres were most probably cellulose ﬁbres. Cellulose is a carbohydrate, it contains C, O both detectable on SEM-EDS, and H, not detectable. It is likely that CMC used on test area no. 4, being a gentle adhesive material, had glued cellulose ﬁbres to the surface. Rinsing and brushing procedures were not suﬃcient to remove these poultice relics. Cellulose powder is a white material. As a consequence, we believe that the strong and permanent colour

122

V. Vergès-Belmin and M. Labouré

Fig. 14.4. Cellulose + CMC treated area on the pier: close view (magnifying lens), transmitted light microscope aspect of the adhesive tape and SEM/BSE image of the tape surface Table 14.2. SEM-EDS qualitative analyses performed on the adhesive tapes; in bold letters: elements systematically present over ten diﬀerent analyses Area no. sample

1 2 3 4

adhesive tape alone ﬁbre on test area no. 4 reference area attapulgite + glass microspheres fume silica cellulose cellulose + CMC

Major elements

Minor elements

C C–O C C–O–Si–Mg C–O C C–O

O Na–Mg–Al–Si–S–Cl–K O–Na–Mg–Al–Si–S– K–Ca–Fe Na–Al–P–S–K–Ca– Ti–Fe Na–Al–Si–S–K–Ca O–Na–Al–Si–S–K–Ca–F Na–Mg–Al–Si–S–Cl– K–Ca–Fe

shift of cellulose + CMC treated areas, on the pier and on the jamb of the portal, is at least partly due to the presence of these remnants. On the other adhesive tapes, 0.2–3 mm rounded or worm-like, very thin clutches of <10–80 µm particles were found. Carbon was systematically detected as a major element (Table 14.2). Most probably, the deposits being present in these clutches were particles in such a low quantity that the spectrometer, on the areas selected for analysis, mainly detected a signal coming from the embedding tape. Clutches present on the tapes from reference area and area no. 3 show only carbon as a major element. This signal is obviously coming from the tape. This means that only a very small quantity of loose deposits is present. The minor elements signature corresponds to airborne particles. Oxygen is found as a major element in all other samples. There, in addition to atmospheric particles, some other phases containing oxygen are present. The tape collected from area no. 1 is the only one showing Mg and Si as other major elements. These elements are likely to originate from attapulgite [whose formula is (Mg, Al)2 Si4 O10 (OH)4 (H2 O)], one of the components of the poultice applied on area no. 1. The colour shift on this area may thus be attributed at least partially to the presence of poultice remnants. The tape collected from area no. 2 does not show any other marking element than

14 Poultices as a Way to Eliminate the Yellowing Eﬀect

123

oxygen. We know that fume silica, i.e. very ﬁne SiO2 material, was applied on this area. We may deduce that there is also the possibility in that case of a contamination of the area by poultice remnants. The colour shift would thus also in that case be at least partly due to these remnants.

14.5 Conclusions Poultice application may reduce, signiﬁcantly and fairly permanently, the laser-induced yellowing eﬀect. Scotch tape tests, optical microscope and SEM-EDS observations and analyses show that this reduction is due at least partly to poultice remnants in the cases where the poultice contains attapulgite or carboxy-methyl-cellulose. Fume silica poulticing also seems to leave relics on the treated surface. The use of cellulose alone leads to a slight de-yellowing without any poultice relics being present. In that case, we are sure that the de-yellowing can be due either to solubilization of yellow products or to physical detachment of particles. New poultice tests are going to be performed at St. Denis on larger lasercleaned areas, in order to go further on the solubilization hypothesis. The poultice will be analysed with the same protocol as the one used for yellow polar organic compounds extracted from St. Denis black crusts [4]. Acknowledgements Authors would like to thank O. Rolland, D. Bouchardon for their help on photo making, M. Guiavarc’h and F. Villard for their assistance in ﬁeld measurements, M. Mengoli, B. Maﬀre, B. Mouton and S. Santos for their constant support and conﬁdence on experiments being held on St. Denis cathedral.

References 1. M. E. Meyohas and V. Picure, in J. Philippon, D. Jeannette, and R.-A. Lefèvre (Eds) La conservation de la pierre monumentale en France, Presses du CNRS: Ministère de la culture et de la communication, Paris, 1992. 2. P. Bromblet, M. Labouré, and G. Orial, in Journal of Cultural Heritage, Vol. 4, suppl. 1, 17, 2003. 3. V. Vergès-Belmin, in La couleur et la pierre, Polychromie des portails gothiques, actes du colloque, Amiens, 12–14 oct. 2000 D. Steyaert & D. Verret (Eds), ARPP et J. Picard publ./Paris et ARP/ Amiens, 151, 2002. 4. M. Gavino, B. Hermosin, V.Vergès-Belmin, and W. Nowik, in Journal of Separation Science, Vol. 27, n◦ 7–8, 513, 2004. 5. V. Vergès-Belmin and C. Dignard, in Journal of Cultural Heritage, Vol. 4, suppl. 1, 238, 2003. 6. V. Vergès-Belmin, in Atmospheric Environment, Vol. 28, 2, 295, 1994.

124

V. Vergès-Belmin and M. Labouré

7. P. Bromblet and V. Vergès-Belmin, Actes des Journées d’étude de la SFIIC, le dessalement des matériaux poreux, Poitiers, 9–10 mai 1996, SFIIC, Champs-surMarne, 55, 1996. 8. M. Gavino, B. Hermosin, M. Castillejo, M. Oujja, E. Rebollar, V. Vergès-Belmin, W. Nowik, and C. Saiz-Jimenez, in Proceedings of the Int. Workshop on Air Pollution & Cultural Heritage, 1–3 Dec. 2003, Seville, Spain, 239, 2004. 9. V. Vergès-Belmin, C. Pichot, and G. Orial in Conservation of Stone and other Materials: Proceedings of the International RILEM/UNESCO Congress, Paris, Jun. 29–Jul. 1, 1993, Editor M.J. Thiel, Vol. 2, 534, 1993. 10. M. Cooper and J. Larson, in The Conservator, Vol. 20, 28, 1996. 11. V. Fassina, in Proceedings of International Symposium “Le pellicole ad ossalati: origine e signiﬁcato nella conservazione delle opere d’arte”, Politecnico di Milano, Milan, 25–26 Oct., 5, 1989. 12. V. Vergès-Belmin: Proceedings of the Workshop Lasers in the Conservation of Artworks, Lacona 1, Heraklion, Crete, Greece, 4–6 Oct 1995, Restauratorenblätter, Sonderband, 17, 1997. 13. M. Cooper (Ed.), Laser Cleaning in Conservation: An Introduction, Oxford, 1998. 14. G. Calcagno, M. Koller, and J. Nimmrichter, in Proceedings of the Workshop Lasers in the Conservation of Artworks, Lacona 1, Heraklion, Crete, Greece, 4–6 Oct. 1995, Restauratorenblätter, Sonderband, 39, 1997. 15. S. Klein, F. Fekrsanati, J. Hildenhagen, K. Dickmann, H. Uphoﬀ, Y. Marakis, and V. Zaﬁropulos, in Applied Surface Science Vol. 171, 242, 2001. 16. J. Kolar, M. Strlic, D. Müller-Hess, et al. in Journal of Cultural Heritage, Vol. 4, suppl. 1, 185, 2003. 17. V. Zaﬁropulos, C. Balas, and A. Manousaki, in Journal of Cultural Heritage, Vol. 4, suppl. 1, 249, 2003.

15 Experimental Investigations and Removal of Encrustations from Interior Stone Decorations of King Sigismund’s Chapel at Wawel Castle in Cracow A. Koss1 , J. Marczak2 , and M. Strzelec2 1

2

Interacademy Institute for Conservation and Restoration of Works of Art, 37 Wybrzeze Kosciuszkowskie, 00–379 Warsaw, Poland, [email protected] Institute of Optoelectronics, Military University of Technology, 2 Kaliskiego Str., 00–908 Warsaw, Poland

Summary. The paper presents selected results of EUREKA E!2542 RENOVA LASER project “Laser renovation of monuments and works of art”, realized in the years 2001–2004, aimed at application of laser technique for conservation of artworks and historic objects in architectural scale. Laser technology has been included into a full program of conservation and restoration of Sigismund’s Chapel at Wawel Castle in Cracow, Poland, covering more than 800 m2 of decorative, sixteenth century sculptured surfaces.

15.1 Introduction The present appearance of the famous Wawel complex in Cracow is due mainly to the royal architect, Bartolomeo Berrecci (ca. 1480–1537). Among several of his inventions, the Sigismund Chapel in the Wawel Cathedral is one of the best known (Fig. 15.1). Financed by King Sigismund I of Poland, the chapel was built between 1519 and 1533. The basic shape Berrecci chose – a cube divided internally by paired pilasters and surmounted by a golden dome, octagonal outside and cylindrical inside, with eight circular windows – represents the Italian ideal. Yet the overall appearance, with its profuse furnishings in a wide range of materials and elaborate surface decoration ﬁlling every available space on the walls, is totally unlike anything to be found in Italy. Sigismund Chapel was hailed by many historians of art as the most beautiful example of Tuscan renaissance north of the Alps and, despite the enormous value of the gilt dome, it has miraculously remained untouched throughout the city’s troubled history. Preliminary work connected with application of pulsed laser radiation for art works encrustation removal started in Poland, at the Institute of

126

A. Koss et al.

Fig. 15.1. Sigismund’s Chapel, Wawel Cathedral

Optoelectronics, Military University of Technology (IOE MUT) at the turn of 1992/1993. In 1998, at the Faculty of Conservation and Restoration of Works of Art, Academy of Fine Arts in Warsaw, laser technology was used for the ﬁrst time to remove encrustation from elements of the Tomb of Unknown Soldier in Warsaw. In the years 2001–2004, IOE MUT and Inter-Academy Institute of Conservation and Restoration of Works of Art coordinated international research project RENOVA LASER “Laser renovation of monuments and works of art”, performed as E!2542 in the frames of EUREKA Initiative of EU. The main aim of the project was to implement laser technique into the conservation of works of art and monuments in an architectural scale. Laser technique has been included into many conservation programs, also into the program of conservation and restoration of internal décor of Sigismund Chapel at Wawel Hill in Cracow, about 800 m2 of decorative sculptured surfaces made of Myslenicki sandstone at the beginning of the sixteenth century. This work should be rated among the largest similar projects in the world, like: (a) Laser cleaning of portals of cathedrals in Amiens, Mantes-la-Jolie, Paris, Chartres and Saint Denis in France [1] (b) Laser renovation of a number of sculptures from the collection of English museums by Liverpool Conservation Centre [2] (c) Laser cleaning of decorated facades and elevations (Rucellai Palace, Cathedral of St. Maria del Fiore in Florence, Cathedrals in Pisa and Sienna) in Italy [3] (d) Renovation of St. Stephen’s Cathedral in Vienna [4] (e) Laser cleaning of classical stone monuments in Athens [5] This chapter is a second one in a series of publications (ﬁrst one [6]) showing the results of our work in Cracow. It justiﬁes the selection of laser technology,

15 Experimental Investigations and Removal of Encrustations

127

evaluates its eﬃciency, describes methodology and workplace and presents selected experimental results, including also supplementary conservation measurements, like thermography of chapel walls. Separate posters, presented during LACONA VI, show the results of colorimetry, LIBS and Raman measurements, research results in the ﬁeld of environment safety during laser cleaning and comparison of wet and dry laser cleaning [7–9].

15.2 Selection of a Cleaning Method by Laser Selection of a laser process for cleaning of an architectural object as large as the Sigismund’s Chapel, according to criteria in [4] – over 300 m2 , was based on the following data: 1. Laser cleaning of stone and other substrates, demonstrated in Europe for more than 20 years and over 10-year experience of the authors in Poland, has conﬁrmed the virtues and safety of a method based on selectivity and, so-called, an eﬀect of self-limitation to a laser ablation process. 2. Own construction of Nd:YAG Q-switched lasers have been developed that are characterised with adequate values of beam parameters and possibility of their full in situ control as well as good laser mobility. 3. It has been possible to include E!2542 RENOVA LASER project, connected with implementation of a laser technology into artworks conservation, into the restoration works carried out in the Sigismund’s Chapel. 4. Application of laser cleaning is promising for signiﬁcant (but not total) elimination of chemical and mechanical cleaning of a stone that have stronger disadvantageous inﬂuence on environment than laser cleaning has. 5. For the ﬁrst time in Poland, it was possible to apply laser conservation for such a large object (over 800 m2 ) being a magniﬁcent historical object. This object has been renovated many times in the period of 500 years with the use of various materials (as replacements of original ones), i.e. artiﬁcial stones, plaster, wood, etc. and the chapel interior has been painted many times in order to protect it and unify its colouring. 6. For the ﬁrst time the authors could remove encrustations from green and grey-green tint (dyed by glauconite) sandstone from Myslenice vicinity what has not been reported in the literature yet. Selection of a laser cleaning technology was predicted with estimation of its eﬃciency. A speed of removal of encrustation from artwork surfaces depends on a required energy density, i.e. on a diameter of a laser beam at the object, pulses repetition rate and overlapping level of “spots” at the object. If the spot diameter is ϕ, the overlapping coeﬃcient is s = 1/2 and the laser frequency repetition of a laser is f , then the theoretical speed of cleaning in both directions N ↑ and N → will be (1/2)ϕ what gives in result:

128

A. Koss et al.

N = N↑ · N → · f =

1 1 1 · ϕ · · ϕ · f = · ϕ2 · f. 2 2 4

(15.1)

Substituting the parameters of laser system ϕ = 1 cm and f = 10 Hz, we will obtain the theoretical speed of surface cleaning N = 1/4×1×10 = 2.5 cm2 s−1 . Counted per square metre, a theoretical speed of removal of encrustations is 0.9 m2 h−1 with the overlapping coeﬃcient of beams, s = 1/2. Of course, this speed can be higher or lower depending upon energy density required for encrustation removal. The speed of a cleaning process of 0.9 m2 h−1 was calculated with an assumption that the required energy density is F = 1 J cm−2 . In practice, eﬃciency of removal of complex encrustations in the chapel using ReNOVALaser 2 was from 0.4 to 1.6 m2 per workday.

15.3 Experimental Results 15.3.1 Methodology For selection of a proper procedure of stone cleaning, in an architectural scale, by means of laser radiation, the conditions of optimal cleaning results were veriﬁed by testing and analysis of laser radiation inﬂuence on chosen types of stone. Using chemical–physical analyses, black encrustation was characterised in order to determine its content and possible presence of polychromy and patina in it. The output laser energy E was determined and the laser beam diameter ϕ was matched in order to obtain the required energy density at the object’s surface F = E/ϕ2 . The “method of squares” was used, i.e. the same area was irradiated with the radiation of precisely regulated energy density. Due to assistance of well-experienced conservator, a range of energy densities being safe for substrate and patina was deﬁned. It is illustrated in Fig. 15.2. Using ReNOVALaser 2 system with a pantograph and focusing lens of a focal length of 1,000 mm, the minimum beam diameter can be of 3.0 mm and maximum energy density can reach even 9 J cm−2 . But high energy densities

Fig. 15.2. Illustration of “square” method: mask (left), and result of laser irradiation (right)

15 Experimental Investigations and Removal of Encrustations

129

were not used because they would be dangerous for a substrate and could cause such damages as vitriﬁcation, oxidation, melting, destruction and formation of micro-cracks at the surface. 15.3.2 Description of a Workplace Removal of encrustations from stone, in architectural scale, i.e. at surfaces larger than 300 m2 [4], is signiﬁcantly diﬀerent and is much more complex than laser cleaning of sculptures or objects in a laboratory. Laboratory cleaning is easier because all needed elements are ready for work and easily available as well as there is suﬃcient intensity of light and easy laser service. When the works are carried out at a scaﬀold with nine levels and very narrow passages and the cleaned surface of stone is over 800 m2 , additional equipment is necessary for a given laser system that ensures: – Stable power supply of 230 V and power of 5 kW (three laser systems) with 1% tolerance of voltage value for protection of internal laser subassemblies. – Mobility and adequate spatial arrangement of laser devices in a workplace. Three non-overlapping areas were prepared and all of them were protected against dust and having housings resistant to mechanical disturbances. – Protection against laser radiation emission into outside (i.e. in direction to people working at other levels of a workplace) due to application of absorbing polyethylene housing. Each laser operator had eye-safe glasses of optical density of over OD 8 for a wavelength of 1,064 nm. – Carrying out initial tests based on results of experimental cleaning, selection of laser parameters ensuring optimal cleaning and up-to-dating the “register of works” with current technical and photographical documentation. 15.3.3 Results of Work A program of laser renovation of Sigismund’s Chapel required parallel realization of diﬀerent, specialized investigations of encrustation and original substrates before, during and after cleaning process: – – – – – –

Petrography and stratigraphy Surface and cross-sectional composition of elements (SEM/EDS) Reﬂectometry and colorimetry Spectrometry, including laser induced emission spectroscopy (LIBS) Temperature gradients (thermography) Optimization of laser ﬂuence and number of laser pulses in the same area needed for safe removal of encrustation (shown before)

As it is written at the beginning of the chapter, most of the experimental results can be found elsewhere [6–8]. Particularly, it concerns spectroscopic and SEM/EDS measurements of green Myslenicki sandstone, for the ﬁrst time presented in the literature. Figure 15.3 shows several examples of laser cleaning performed at diﬀerent places of chapel bas reliefs.

130

A. Koss et al.

Fig. 15.3. Fragments of Sigismund’s Chapel decor during laser cleaning: (a)–(d) ornamentations (conference inscriptions made by laser), (e) window pilaster

15.4 Conclusions Due to results obtained from removal of secondary encrustations and overpaintings from the Sigismund Chapel interior with application of pulsed laser radiation, the already known virtues of laser method of objects renovation have been conﬁrmed. These virtues are the following: – Versatility. Various energy densities were used for precise removal of non-coherent encrustations of various thicknesses and various chemical compositions – Selectivity. ReNOVALaser 2 can be regulated in such a way to interact only with deﬁnite substances of various optical absorption coeﬃcients. Moreover, only deﬁnite encrustation thickness can be removed, separating an original substrate’s material – Self-limiting. Direct control of a physical process is ensured by a substrate material that reﬂects laser radiation in a level dependent on dirt – Environmental friendly. Amount of pollution in a process of laser cleaning is signiﬁcantly less than from mechanical and chemical processes. Also the noise is much less intensive

15 Experimental Investigations and Removal of Encrustations

131

– Time of operation. When eﬃciency of laser and conventional processes is comparable, surface preparation for a laser cleaning is minimal. It was observed in many cases that physical and chemical characteristics of encrustations can change from point to point even in one sample. An absorption coeﬃcient of layers depends even on such factors as weather or air humidity. Variable properties of the cleaned samples makes it necessary for the laser operator to change the exposure parameters even during a cleaning process. In such a case he has to use his knowledge, experience or sometimes intuition. From the author’s experience, it results unquestionable that removal of secondary encrustations from stone is signiﬁcantly diﬀerent during various times of a day. Thus, various energy densities should be used for the same encrustation. In the morning, due to water in encrustation pores, even two times lower energy densities can be used. Acknowledgements The work was funded by the Polish Ministry of Science and Information Society Technologies, project E!2542 “Laser renovation of monuments and art works” (grant No 217/E-284/SPUB-M/EUREKA/T-11/DZ 203/2001–2003). The authors wish to thank all the partners in the project for their helpful discussions.

References 1. P. Bromblet, M. Labouré, and G. Orial, in Journal of Cultural Heritage, Vol. 4, 17, 2003. 2. J. Larson, C. Madden, and I. Sutherland, in Journal of Cultural Heritage, Proceedings of LACONA III (Florence, April 1999), Elsevier, (2000), 79. 3. S. Siano, F. Fabiani, R. Pini, R. Salimbeni, M. Giamello, and G. Sabatini, in Journal of Cultural Heritage, Vol. 1, 47, 2000. 4. G. Calcagno, E. Pummer, and M. Kollerc, in Journal of Culural Heritage Vol. 1, 111, 2000. 5. P. Maravelaki-Kalaitzaki, V. Zaﬁropulos, and C. Fotakis, in Applied Surface Science, Vol. 148, 92, 1999. 6. J. Marczak, M. Strzelec, and A. Koss, in Proc. SPIE, Vol 5958, (2005), in print. 7. Sarzynski, W. Skrzeczanowski, and J. Marczak, “Colorimetry, LIBS and Raman experiments on renaissance green sandstone decoration during laser cleaning of King Sigismund’s Chapel in Wawel Castle, Cracow, Poland”, this issue. 8. R. Ostrowski, J. Marczak, M. Strzelec, S. Barcikowski, J. Walter, and A. Ostendorf, “Health risks caused by particulate emission during laser cleaning”, this issue. 9. Sarzynski, J. Marczak, and K. Jach, “Comparison of steam and dry laser cleaning of artworks”, this issue.

16 Nd:YAG Laser Cleaning of Red Stone Materials: Evaluation of the Damage ∗

C. Colombo1 , E. Martoni2 , M. Realini1 , A. Sansonetti1 , and G. Valentini3 1

∗ 2 3

Istituto per la Conservazione e la Valorizzazione dei Beni Culturali, Sezione di Milano “Gino Bozza”, p.zza Leonardo da Vinci 32, 20133 Milano, Italy [email protected] IUAV Istituto Universitario di Architettura di Venezia, Italy Dip. Di Fisica, Politecnico di Milano, p.zza Leonardo da Vinci 32, 20133 Milano, Italy

Summary. Lasers have been tested, during the recent past, as a useful cleaning method in conservation treatments: this is due to selectivity and precision of its performance. Nevertheless some colour changes have been detected using Nd:YAG laser sources, especially on white and red coloured substrates. Colour changes on white marble and other white architectural materials have already been widely surveyed. This chapter focuses on the interaction of laser radiation with two kinds of red materials: red Verona limestone and terracotta. These materials have been chosen because of their large use in northern Italian architecture and in statuary. Red Verona limestone is not homogenous in hue, owing to the presence of calcareous nodules (lighter in colour) and clay veins (dark reddish colour). On the other hand, terracotta is homogeneous in colour in most of the cases. Experimental tests have been carried out on specimens comparing two diﬀerent Quanta System Nd:YAG laser devices: Michelangelo and a SYL 201 lab model equipped by a scanning computer-aided system. In this way a better radiation control has been obtained, both to address laser pulses on surfaces and to count exactly their number per surface unit. Fluences have been chosen on the basis of a previous survey, which indicated the harmfulness threshold ﬂuence on red Verona limestone as 1.5 J cm−2 . To better bring out laser eﬀects, ﬂuences have been increased up to 3 J cm−2 . The interactions between laser radiation and substrates have been inferred comparing the changes of surface both in colour and morphology. Surface colour changes have been measured with a reﬂectance colorimeter (CIE L∗ a∗ b∗ system). Morphology has been examined by diﬀerent microscopic techniques: thin and polished cross sections have been observed by optical microscopes; moreover, a close examination of surfaces has been carried out with the aid of scanning electron microscopy.

16.1 Introduction Many diﬀerent laser devices have been tested, during the recent past, as a cleaning method in conservation treatments, both in laboratory and in ﬁeld; selective and precise performance helped conservators solve cleaning problems

134

C. Colombo et al.

which could hardly be confronted with traditional methods. These are some of the reasons why lasers are, at the present time, common in conservation practice. Nevertheless scientiﬁc research must still play a role in evaluating eﬃcacy and damage and in measuring harmful threshold ﬂuences. Regarding Q-switched Nd:YAG laser sources (λ = 1, 064 nm), some authors observed discoloration of stone after cleaning tests. These harmful eﬀects were particularly evident on white substrates and on red pigments. Discoloration on white marbles and other white stone materials has been already widely surveyed [1–6]. Some other authors observed a speciﬁc sensitivity concerning materials pigmented with iron compounds and darkening of calcareous materials with impurities of iron hydroxide (goethite) after laser irradiation [7–9]. The survey is aimed to study the interaction of laser radiation with red stone materials containing iron compounds; hence red limestone (Rosso Ammonitico Veronese) and a ceramic material (terracotta) have been chosen and, moreover, for their large use in northern Italian architecture and in statuary. In the case of red Verona limestone, the iron compounds constitute a low percentage dispersed in a carbonatic matrix, while in the terracotta they are instead contained in a higher content and are dispersed in a non-carbonatic matrix. This chapter focuses on the correlation between discoloration and morphological changes in consequence of the interaction with the radiation produced by diﬀerent laser devices. A previous work reports mechanical damage of red Verona limestone at 1.5 J cm−2 ﬂuence [10], even if signiﬁcant discolorations were recorded also at lower ﬂuences. For these reasons ﬂuences lower than 1.5 J cm−2 have been used to verify if morphological changes are produced, even if only at a microscopic scale; higher values have been used to stress the phenomena and to better bring out laser eﬀects.

16.2 Experimental 16.2.1 Materials Red Verona limestone belongs to an important Jurassic formation outcropping in many localities near Verona; it includes a large variety of nodular limestone, which is called diﬀerently according to their colour and texture. It is not homogenous in hue, owing to the presence of calcareous nodules (lighter in colour) and clay veins (dark reddish colour). The percentage of Fe2 O3 ranges from 0.11% to 0.30% [11]. On the contrary terracotta is homogeneous in colour and texture in most of the cases. Bibliographic data show that the Fe2 O3 content could reach and exceed 5% [12].

This page intentionally blank

16 Nd:YAG Laser Cleaning of Red Stone Materials

135

The survey on red Verona limestone has been carried out on specimenprepared slabs of 5 × 5 × 1 cm3 size, on which XRF analysis provided an Fe2 O3 content of 0.35%. Terracotta specimens have been prepared by moulding clay and ﬁring them in a muﬄe kiln to 950◦ C. XRF analysis provided an Fe2 O3 content of 5.76%. Both specimen series have been honed with the aim to smooth their surfaces giving them morphological homogeneity. An automatic hone has been used employing a 180 mesh abrasive paper at 5 N for 4 min at the speed of 300 revolutions per minute. 16.2.2 Laser Devices Two Q-switching Nd:YAG laser devices were employed for testing operation (Quanta System Michelangelo and SYL 201 ) at ﬂuences of 1, 1.25, 1.5, 2, 2.5, 3 J cm−2 . Michelangelo is designed for conservation sites and it supplies pulses of 5–10 ns duration. The laser beam was delivered via a jointed articulated arm and through a focusing lens in the handpiece. The SYL 201 is a lab model and it supplies pulses of 9 ns duration. Its laser beam was coupled to a 600-m core quartz ﬁbre. The distal end of the ﬁbre was imaged onto the sample by means of a lens, while the beam was deﬂected by a two-axis galvo scanner, controlled by a computer. This allowed us to replace the quasi-gaussian proﬁle of the original laser beam with a ﬂat top proﬁle onto the sample thanks to the multimode superposition inside the ﬁbre. This setup allowed a very precise control of each laser parameter; moreover, the laser beam was focused in spots having a precise size and delivered to the specimen in a regular pattern through an X–Y galvanometric scanner. The precise control of ﬂuence and repetition density, deﬁned as the number of full scans delivered to the specimen surface, is a very signiﬁcant feature because it is not available in the ﬁeld. In this way a comparison between a commercial and lab laser was performed in the optimal condition. No water was used to wet the surface before laser cleaning. 16.2.3 Analytical Methods Microscopic observations and colour measurements have been carried out on specimens, both before and after laser irradiation. Colour measurement. A Minolta Chroma Meter CR 200 colorimeter has been used. Data have been recorded in CIE L∗ a∗ b∗ system, and then the diﬀerence between before and after laser radiation has been calculated for each parameter (∆L∗ , ∆a∗ , ∆b∗ ). Optical microscopy. Surface observations have been carried out in the same areas before and after laser radiation, using a reference grid, by a Leitz Wild M420 stereomicroscope; polished cross sections have been observed by a Leitz Ortholux with Ultrapack illuminator microscope; thin cross sections have been observed by a Nikon Eclipse E400 Pol. The microscopes were equipped with a digital image capturing system.

136

C. Colombo et al.

Scanning electron microscopy. Secondary electron and backscattered electron images have been collected with a JEOL 5910LV microscope. Morphology has been directly observed on specimen surfaces and on polished cross sections to improve the interpretation of surface proﬁle.

16.3 Results and Discussion Surface colour showed signiﬁcant changes for ﬂuences higher than 1.50 J cm−2 (Table 16.1). In particular, on red Verona limestone, both lasers produced a dramatic increase in lightness (L∗ ) and a decrease in chromaticity with an overall eﬀect of whitening at 3 J cm−2 ; surface morphology is characterised by micro-abrasions diﬀerently distributed. The manual control of Michelangelo caused an irregular pattern of alteration while the scanner control of SYL 201 produced a very regular pattern characterised by well-deﬁned spots (Fig. 16.1). The laser increased the surface roughness highlighting the calcite crystals, whose speciﬁc surface increase is responsible for higher light scattering (Figs. 16.2, and 16.3). As colour measurements have conﬁrmed, this Table 16.1. Chromatic variations in red Verona limestone and terracotta consequent to laser source irradiation (M: Michelangelo – SYL: SYL 201 ) at diﬀerent ﬂuences (Fl) Fl

Red Verona Limestone ∆L∗

J cm−2 1 1.25 1.5 2 2.5 3

M

SYL

∆a∗ M

SYL

Terracotta ∆b∗

M

SYL

∆L∗ M

0.07 −0.82 0.15 −1.63 1.42 −0.67 −0.11 −2.59 −3.88 0.15 −0.50 −0.85 −1.23 0.00 −0.44 4.11 −1.50 −2.54 −0.61 1.40 0.56 2.38 6.86 −3.50 −5.89 −2.28 −4.20 1.49 6.11 12.55 −5.85 −7.88 −4.52 −7.42 1.48 10.47 7.86 −8.14 −7.83 −8.31 −6.65 2.23

SYL 0.31 1.25 1.09 1.06 0.93 1.17

∆a∗ M −1.71 −1.89 −2.25 −5.60 −7.53 −8.32

SYL −2.91 −4.71 −5.59 −5.79 −7.61 −8.79

∆b∗ M −0.78 −0.51 −0.24 −2.21 −3.37 −4.45

SYL −0.73 −1.40 −2.03 −1.80 −3.28 −5.31

Fig. 16.1. Pattern of alteration of red Verona limestone: Michelangelo at 2 J cm−2 (left), SYL 201 at 2 J cm−2 (right). The non-irradiated area is at the top of ﬁgures

16 Nd:YAG Laser Cleaning of Red Stone Materials

137

Fig. 16.2. Surface micro-morphology of red Verona limestone due to laser interaction: Michelangelo at 2 J cm−2 (left), SYL 201 at 2 J cm−2 (right)

Fig. 16.3. Red Verona limestone. On the left: micro-morphology induced by Michelangelo at 2 J cm−2 (particular area of Fig. 16.2). On the right: SYL 201 at 3 J cm−2 . Boundary line between irradiated (B–D) and non-irradiated sample portions (A–C). A and B, clay vein; C and D, calcareous nodule

phenomenon depends on laser devices, and it is proportional to ﬂuence: the higher the energy, the wider the interaction area. This is probably due to mechanical spallation eﬀects on the periphery of the spot area (Fig. 16.4). Moreover, clay veins are more sensitive to laser radiation as compared to calcareous matrix because of their lower compactness, weak cohesion and higher absorption. Here an increase in roughness detected due to the higher removal of clay materials (Fig. 16.5). As concerns terracotta colour, two diﬀerent behaviours as a function of laser devices have been observed: Michelangelo produced important changes for ﬂuences higher than 1.50 J cm−2 except for L∗ parameter which did not change signiﬁcantly, while SYL 201 eﬀects are very signiﬁcant at even the lowest ﬂuence tested, especially on a∗ parameter. Contrary to red limestone, the morphological alterations on terracotta surface are not clear by stereomicroscope observations (Fig. 16.6). The crystalline structure of the irradiated surface is similar to the honed one, so the induced variation of scattering is very low (Fig. 16.7).

138

C. Colombo et al.

Fig. 16.4. Red Verona limestone: SYL 201 at 2, 2.5, 3 J cm−2 (from left to right). The spot size was constant for each test

Fig. 16.5. Detailed view of the preferential loss of material corresponding to a clay vein of the red Verona limestone: polished cross section observed by optical microscopy (left) and by scanning electron microscopy (right)

Fig. 16.6. Pattern of alteration of terracotta: Michelangelo at 2 J cm−2 (top), nonirradiated area (bottom)

Owing to homogeneous micro-crystalline structure of terracotta, the laser eﬀects are homogeneously distributed on the surface and they are well evident only by observing the surface proﬁle, which is characterised by small but frequent losses of material (Fig. 16.8). Important modiﬁcation to the micro-crystalline structure have been observed on sample irradiated at 2.5 and 3 J cm−2 with SYL 201, where the surface has changed from a polycrystalline structure to a vitriﬁed one (Fig. 16.9).

16 Nd:YAG Laser Cleaning of Red Stone Materials

139

Fig. 16.7. Terracotta surface irradiated by SYL 201 at 2 J cm−2 (right) and nonirradiated area (left)

Fig. 16.8. Thin cross section showing terracotta proﬁle before (left) and after (right) Michelangelo irradiation at 1.5 J cm−2 (bar = 100 µm)

Fig. 16.9. Terracotta surface before (left) and after (right) SYL 201 at 3 J cm−2 . Vitrifying phenomena are evident

16.4 Conclusions The two materials tested reacted diﬀerently to laser radiation. At low ﬂuences (1–1.5 J cm−2 ) eﬀects are observed as regards chromaticity on terracotta only; at higher ﬂuences (2–3 J cm−2 ), whitening prevails especially on red Verona

140

C. Colombo et al.

limestone. The diﬀerent colour changes, whitening for limestone and yellowing for terracotta, have to be ascribed to compositional and structural diﬀerences of materials. Laser radiation produces important morphological changes, which have side eﬀects on the colour; both materials undergo yellowing, which is clearly visible on the terracotta only. On red limestone the increase in surface roughness resulted in highlighted calcite crystals. A possible explanation is that the consequent increase in scattering causes a whitening of surface that covers the yellowing. The clay veins are more sensitive than calcite to laser eﬀects, because of crystal dimensions and compactness. For this reason the loss of material on terracotta is homogeneously distributed on the surface, while on red limestone, that is characterised by the mixture of clay and calcite minerals, the loss of material is very irregular, determining a characteristic surface proﬁle. Moreover, clay veins are discontinuity points, which may cause loss of material even in the calcitic nodules. The diﬀerent behaviour between laser devices is to be ascribed to the difﬁculty to obtain a uniform beam proﬁle with Michelangelo, in terms of beam focalisation and of scanning of the surface. The observed damage, characteristic of the beam shape, is in fact very similar for both Michelangelo and SYL 201. So it is possible to note that 1.5 J cm−2 ﬂuence could be considered as a damage threshold, even if this survey has been carried out on non-soiled specimens, to better investigate the interaction eﬀects; possibly the eﬀects obtained when removing soiling may be weaker on the surfaces, allowing the conservator to use the indicated threshold ﬂuence. Acknowledgement Authors would like to acknowledge Anna Brunetto for her valued collaboration in using Quanta System Michelangelo Laser Device.

References 1. G. Alessandrini, A. Sansonetti, and A. Pasetti, in Proc. of 4th Inter. Symposium on the Conservation of Monuments in the Mediterranean, Vol. 3, 19–30, 1997. 2. A. Sansonetti and A. Pasetti, in Proc. of 4th Inter. Symposium on the Conservation of Monuments in the Mediterranean, Vol. 3, 345–354, 1997. 3. A. Aldrovandi, C. Lalli, G. Lanterna, and M. Matteini, in Journal of Cultural Heritage, Vol. 1, 55, 2000. 4. D. Eichert, V. Vergès-Belmin, and O. Kahn, in Journal of Cultural Heritage, Vol. 1, 37, 2000. 5. S. Klein, F. Fekrsanati, J. Hildenhagen, K. Dickmann, H. Uphoﬀ, Y. Marakis, and V. Zaﬁropoulos, in Applied Surface Scienc,e Vol. 171, 242, 2002. 6. V. Vergès-Belmin and C. Dignard, in Journal of Cultural Heritage, Vol. 4, 238, 2003.

16 Nd:YAG Laser Cleaning of Red Stone Materials

141

7. A. Sansonetti and M. Realini, in Journal of Cultural Heritage Vol.1, 189, 2000. 8. M. Chappè, J. Hildenhagen, K. Dickmann, and M. Bredol, in Journal of Cultural Heritage, Vol. 1, 264, 2003. 9. M. Gradia, M. Gavino, V. Vergès-Belmin, B. Hermosin, W. Nowik, and C. Saiz-Jimènez, in Lacona V Proc. – Osnabruck 2003 – Springer 2005. 10. A. Sansonetti, M. Realini, L. Toniolo, and G. Valentini, in Lacona V Proc. – Osnabruck 2003 – Springer 2005. 11. G. Bortolaso, L. Lazzarini, L. Menegazzo Vitturi, and G. Rampazzo, in Proc. of VIth Inter. Congress on Deterioration and Conservation of Stone – Torun – 1988. 12. Z. Goﬀer, Archaeological Chemistry, New York 1980.

17 Exists a Demand for Nd:YAG Laser Technology in the Restoration of Stone Artworks and Architectural Surfaces? E. Pummer Atelier Erich Pummer for Conservation and Restoration, 3602 Rossatz 165, Austria [email protected]

17.1 Introduction The question as to whether there is a call for laser technology in the restoration of stone artwork is not new. It is to be assumed that restorers and conservators of monuments are already very well informed in this ﬁeld, as a result of a number of test series by renowned scientists and institutes, the presentation and publication of their ﬁndings in specialist magazines and books, several “laser forums” in the past years and the discussion of the method at the sixth LACONA Conference since 1995. There is currently no other technology that has been subject to such critical scrutiny from every angle as laser technology. Almost 10 years ago I opted for the use of laser technology in the forthcoming restoration of the great gate, “Riesentor” (project of the Federal Oﬃce for Care and Protection of Monuments Austria) and the “Albertinischer Chor” (Atelier Pummer project) of St. Stephens Cathedral in Vienna (Fig. 17.1). Within 30,000 h, 82 m of wine leaf ornament and parapets, 30 gargoyles and symbolic statues, 17 decorated windows, 14 pillars with ﬂoral decorations were cleaned with a Nd:YAG Laserplast 500/1000 from Quantel, Art Master from EV Laser, NL103 from Thales Laser and Palladio from Quanta Systems. Another important examples of laser cleaning are the Vienna Gothic Minoriten Church, the Gothic chorus of the basilica from Bad Deutsch Altenburg in Lower Austria (12,000 h) and Siena Pallazzo Publico, where parts of sandstone and terracotta were cleaned by seven diﬀerent types of lasers within three months (5 Palladios, 1 BMI NL 102, 1 laserblast 500). A countless number of test series and trial areas to compare mechanical and chemical cleaning methods conﬁrmed the beneﬁts of laser technology and the superiority of the method was then conﬁrmed in its practical application. Some of the examples of applied laser technology cited represent, in my studio alone, what is eﬀectively 40,000 man hours in the last 10 years. The time invested in the experience we have gained gives me a certain degree of

144

E. Pummer

Fig. 17.1. Vienna St. Stephens Cathedral

Fig. 17.2. Cleaning samples/intensity: above line with four test areas cleaned with Palladio laser: 1,300, 2,350, 3,400, 4,450 mJ, beneath one test area cleaned with Laserblast 500 and 400 mJ

conﬁdence in assessing cleaning techniques, in particular in direct comparison, and also in combination with other methods (Fig. 17.2). As well as assessing the mineral surface to be cleaned it is also important to set up guidelines and parameters to protect passer-bys and those working on the object, whether they are the laser users or not, from the laser beam and any emissions. The areas on which work is being carried out must be covered

This page intentionally blank

17 Exists a Demand for Nd:YAG Laser Technology

145

in non-reﬂective sheets or plates, and the ﬁne dust that is emitted when the crust that has been growing for hundreds of years has to be vacuumed away. Eye-protection UVEX/laser-Vision DIR 1.5, wavelength: 1,048–1,064 nm, max. power density: 100 W cm−3 , max. energy density: 0.05 J cm−3 , daylight transmission: 70% respiratory protection 3M mask 6300:compination gas ﬁlter: 3M 6057 ABE 1+particulate ﬁlter 3M, P2 (5,925 coarse) (2,125 ﬁne) for organic and inorganic gas until ×30 limiting value or until 1,000 ppm. Dipl. Ing. Janos Bekesi, working with the “ÖSBS” (Österreichische Staubund Silikosebekämpfungsstelle – Austrian Anti-Dust and Silicone Oﬃce) led a project to assess the danger of dust particles resulting from laser cleaning that can enter the alveoli; the reaction to this project has been the mandatory use of protective ﬁlter masks for users. It goes without saying that protective glasses must always be worn for wavelengths of 1,048–1,064 nm/infrared. Despite the vast range of information and scientiﬁcally proven facts available, the contracting authorities and architects still lack awareness. A standard statement, such as the one shown below, is often used in service agreements concerning the cleaning of valuable stone surfaces: “Severely compromised details must be cleaned with a Nd:YAG laser or equivalent.”

17.2 Experimental The test results presented will bridge any remaining voids in the information available, and should contribute to clarifying the results of laser cleaning. The test was conducted in cooperation with INNOWEP GmbH in Würzburg, Germany: A 40 × 40 cm2 plate of “Kehlheimer”/“Solnhofer” limestone with a highly polished surface was selected as the test plate. This material was selected as, with a Mohs hardness of 3–3.5, it is one of the softest polishable stones. The structure is also even, so the conditions on a 40 × 40 cm2 surface are comparable. Stone specialist know that this limestone is very sensitive to mechanical treatment during polishing, while allowing cleaning methods to be very accurately assessed (Fig. 17.3). The polished plate was coated in a thin layer of black silicate colour (“Keim Granital”), diluted 1:1 with water. A square in the centre of the plate was covered to serve as a reference. Once the coating had dried out, ﬁve diﬀerent laser/blasting techniques were employed to remove this coating as gently as possible. Test areas (1) JOS powder (limestone powder), (2) aluminium oxide 180, (3) Duroplast (combined plastic granulate), (4) nutshell granulate, and (5) Nd:YAG-laser. Four of these techniques involved a microsand blaster, at the lowest setting at which residue could be removed (Figs. 17.4–17.9). The ﬁfth test method involved a pulsed Nd:YAG-laser (Thales NL 103 and Quanta Palladio). Once cleaned, the area surrounding the central reference area was divided into 6 × 6 mm2 squares and analysed using the UST method (Universal Surface Tester) from INNOWEP in Würzburg, Germany. The Universal

146

E. Pummer

Fig. 17.3. “Kehlheimer” limestone plate, ﬁne polished surface, coated with black silicate colour (Mohs 1–1.2), 0 = virgin surface, X = black silicate colour (mixed with water 1:1)

Fig. 17.4. Reference area Rq = 0.162 µm, Rz = 0.489 µm

from

the

pure

limestone:

Ra = 0.116 µm,

R R Surface Tester (UST ) uses a tactile procedure (mistan ), where a material surface is mechanically explored in three steps along a straight and ﬂat surface. Roughness parameters and topographic representations are calculated on the basis of this exploration.

Roughness parameter: Roughness depth: Rz and Rmax (DIN 4788) Average roughness value: Ra and Rq (DIN 4762, DIN 4768, ISO 4287/1)

17 Exists a Demand for Nd:YAG Laser Technology

147

Fig. 17.5. Microblasting with Jos powder: Ra = 0.570 µm, Rq = 0.740 µm, Rz = 1.704 µm

Fig. 17.6. Microblasting with corundum 180: Ra = 1.303 µm, Rq = 1.604 µm, Rz = 3.554 µm

17.3 Discussion

1. 2. 3. 4. 5.

JOS powder Aluminium oxide 180 Duroplast Nut shell granulate Nd:YAG laser

Colour coating removed/limestone Colour coating removed/limestone Colour coating removed/limestone Residues of colour coating Colour coating removed/limestone

abraded abraded not abraded not abraded

These test results meant that the cleaning results achieved with Duroplast granulate and with the Nd:YAG laser can be regarded as virtually comparable.

148

E. Pummer

Fig. 17.7. Microblasting with Duroplast: Ra = 0.156 µm, Rq = 0.226 µm, Rz = 0.526 µm

Fig. 17.8. Microblasting with nutshell: Ra = 0.650 µm, Rq = 0.922 µm, Rz = 1.530 µm

The next step was to compare these two cleaning methods on an original piece of “Adnet Limestone” with a uniform layer of black crust (Fig. 17.10). The test results quite clearly demonstrated that the Duroplast granulate is too soft to remove the black crust from the limestone. A negative side eﬀect was also that the softer areas on the stone (clay pockets) were aﬀected by the granulate, thus giving the surface a very uneven result. Cleaning with the Nd:YAG laser produced a very uniform and attractive result, regardless of whether there was a soft or hard surface. This result also mirrors our practical experience on all sandstones. It is permissible to apply these test results to virtually all sand stone varieties, as the mechanical pressure caused by abrasive

17 Exists a Demand for Nd:YAG Laser Technology

149

Fig. 17.9. Nd:YAG-laser: Ra = 0.369 µm, Rq = 0.559 µm, Rz = 0.850 µm

Fig. 17.10. Above cleaned area: Nd:YAG-laser Thales NL 103/175 mJ/20 Hz/focus 4.5/test 10 min/8 h/m2 , below cleaned area: Microblaster/Duroplast 150 g/4 bar/ test 10 min/8 h m−2

blasting/laser treatment may not be immediately visible or measurable, but repeated use certainly does involve a loss of the original surface.

17.4 Conclusions Apart from the great diﬀerence in the cost of laser cleaning (laser per hour/approx. (80) and the cost of microsand cleaning (microsand blasting per hour/approx. (5), these two methods cannot be regarded as equivalent in terms of the cleaning result. The sentence formulated in stone cleaning

150

E. Pummer

proposals: “Severely compromised details must be cleaned with a Nd:YAGlaser or equivalent is thus not valid. This also answers the question as to whether laser technology has its place in the preservation of monuments and historical buildings. The test results comparing the various cleaning methods would assume that there is a very great demand for laser technology in the preservation of monuments and historical buildings. It is not possible to establish high-quality laser technology in the light of unprofessional and ignorant assessments of completely diﬀerent test results using methods that cannot be compared.

18 The SALUT Project: Study of Advanced Laser Techniques for the Uncovering of Polychromed Works of Art ∗

G. Van der Snickt1 , A. De Boeck1 , K. Keutgens1 , and D. Anthierens2 1

∗ 2

Hogeschool Antwerpen, Conservation Studies, Blindestraat 9, 2000 Antwerp, Belgium [email protected] Hogeschool Antwerpen, Industrial Sciences, Paardemarkt 92, 2000 Antwerp, Belgium

Summary. In order to ﬁnd out whether the existing laser systems can be employed to remove superimposed layers of paint on secco wall paintings in a selective way, laser tests were carried out on three types of prepared samples simulating three stratigraphies that are frequently encountered in practice. OM, EPMA, colorimetry, µRaman, and FT-IR were used to evaluate the results. It was found that Q-switched Nd:YAG lasers emitting at 1,064 nm could be employed to remove unwanted layers of oil paint and limewash, but the treatment of large areas requires implementation of a computer-controlled X–Y–Z station in order to control the parameters. However, the applicability of this technique will remain limited as ablation at the established optimum parameters implied a discoloration of the pigments cinnabar, yellow ochre, and burnt sienna. Moreover, it was observed that no ablation took place when the limewash thickness exceeds 25 µm. Unwanted layers of acrylic could be removed in an eﬃcient way with an excimer laser emitting at 193 nm.

18.1 Introduction A considerable part of the patrimony of wall paintings and other polychrome surfaces in Western Europe is covered by several superimposed layers of paint. The reasons for this can be found in changing fashion trends, outbreaks of iconoclasm, building alterations, etc. Wall paintings or polychromes in Western Europe, and especially in Flanders, were generally applied “a secco,” which means that the pigments were mixed with a medium (e.g., oil, tempera) and placed on a dry substrate (e.g., plaster). Current practice for the removal of superimposed layers involves use of solvents and/or mechanical action using a scalpel (Fig. 18.1). However, solvents should be avoided because of their toxicity and retention. The disadvantage of this method is its high cost as expensive working hours have to be paid. Moreover, this technique always

152

G. Van der Snickt et al.

Fig. 18.1. Wall painting in the Church of Saint Jacob, Antwerp, Belgium, partially uncovered with traditional removal techniques

implies certain damage to the original surface and the result correlates with the experience of the conservator. As a result an important part of our cultural heritage suﬀers damage or is left uncovered. A new uncovering technology, both quick and safe, would allow conservators to treat more murals and other polychrome surfaces in a more qualitative way. Starting from the issues described above, a proposal for the SALUT project was formulated and handed in at the IWT-Flanders (Institute for Science and Technique Flanders). After approval, a commission composed of researchers, conservators, and laser companies was formed to investigate whether laser cleaning could oﬀer a solution.

18.2 Experimental Methods In order to ﬁnd out whether the existing laser systems can be employed to remove superimposed layers of paint on polychromed surfaces, laser tests were carried out in three stages (1) preliminary, (2) advanced, and (3) validating. Throughout the ﬁrst stage, several laser systems were put to test on three types of samples (4 × 4 cm2 ), simulating diﬀerent stratigraphies frequently encountered in practice. The goal of these preliminary tests consisted of determination of the proper wavelength and selection of the most suitable laser

18 The SALUT Project: Study of Advanced Laser Techniques

OVERPAINTING

(c)

PICTORIAL LAYER (b) SUBSTRATE

(a)

DUMMY TYPE 1

DUMMY TYPE 2

DUMMY TYPE 3

OIL PAINT (LEADWHITE) OIL PAINT (CRO)

LIMEWASH TEMPERA (CRO)

ACRYL (TIO) OIL PAINT (CRO)

PLASTER

PLASTER

PLASTER

153

Fig. 18.2. Schematic representation of the three types of samples used during the preliminary laser tests

system for this delicate task. The samples consisted of a three-layered system (a) a substrate of plaster, (b) a pictorial layer of paint, which should not be damaged during treatment, and (c) an unwanted overpainting with an approximate thickness of 15–25 µm (Fig. 18.2). The samples of Type 1 (oilon-oil) simulated a real-life case which often poses particular problems when traditional conservation techniques are applied: An overpainting of oil paint, in this case with the pigment lead white, was placed over a pictorial layer of oil paint with chromium oxide. This last pigment was selected for the preliminary tests as it is easy to distinguish from lead white both visually and with diagnostic techniques. Two superimposed layers of oil paint are hard to separate in a selective way by means of solvents and scalpel, as the adherence between the two layers (b–c) is often superior to the adherence between the pictorial layer and the substrate (a–b). Type 2 (limewash-on-tempera) imitates the situation which is most commonly met in practice: An unwanted layer of limewash is applied over a pictorial layer of tempera, in this case, with chromium oxide. To conclude, an overpainting of acrylic with titanium white was selected for the samples of Type 3 (acrylic-on-oil) as this commercial paint was frequently applied on murals during maintenance works throughout the second half of the twentieth century. Thanks to the cooperation of several laser companies, universities, and scientiﬁc institutes, it was possible to test a multitude of IR- and UV-producing laser systems (193, 248, 355, 1,064 nm) on these three types of samples. The parameters were varied within the limitations of the system and/or available setup. The results were evaluated by means of visual examination and optical microscopy (OM). The second stage of the project concerned the advanced laser tests and aimed at the optimization of the selected parameters and the evaluation of the impact of the laser treatment on the pigments and media of the pictorial layer. Samples were prepared with six historic pigments: yellow ochre, burnt sienna, cinnabar, lamp black, lead white, and green earth. The selection of the pigments was not based on the color but on their sensitivity toward laser light. Also a comparison was made between the results of an uncovering by means of traditional techniques on the one hand and a laser on the other hand. The results were evaluated by means of OM, colorimetry, electron probe microanalysis (EPMA), and micro-Raman.

154

G. Van der Snickt et al.

Finally, a limited number of tests on a historical wall painting investigated the correlation of these tests on samples with reality.

18.3 Results and Discussion 18.3.1 Type 1: Oil-on-Oil The results of the laser tests proved that the UV wavelength of 248 nm shows potential for the selective removal of superimposed layers of oil paint. However, in view of the dimensions and weight of the current excimer lasers, it is not feasible to employ these systems in situ on yards with scaﬀolding and limited working space. The results of the tests at 193 and 355 nm were heterogeneous, since it was found impossible to free a deﬁned area of all overpaint. Moreover, the ablation of the unwanted oil paint implied considerable damage to the pictorial layer, to that extent that the substrate was locally revealed. Several commercial laser systems were tested to investigate the potential of IR-laser light (1,064 nm) for this delicate task. Though these lasers were intended for self-limiting purposes (i.e., the cleaning of stone), most results were promising (Fig. 18.3). The Q-switched Nd:YAG laser was able to eliminate the overpaint completely, while the interaction of the laser beam with the pictorial layer appeared minimal. Research on the surface by means of OM. (×200) demonstrated that the texture of the pictorial layer modiﬁed from relatively smooth to porous and granular as a result of direct contact with the laser beam. Yet, the study of cross sections (OM ×200) revealed that the damage was restricted to the upper microns and no substantial thinning was established if the ablation was carried out with optimum parame-

Fig. 18.3. Detail of a Type 1 sample, partially uncovered. The oil paint with lead white was ablated by a Q-switched Nd:YAG laser (1,064 nm)

18 The SALUT Project: Study of Advanced Laser Techniques

155

ABLATION RATE AT 165 mJ/cm2,1Hz

ABLATION DEPTH (µm)

160 140 120 100 80 60 40 20 0 1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 # PULSES

Fig. 18.4. Graph showing the ablation rate at 165 mJ cm−2 and 1 Hz

ters. Also, the variation of the texture did not aﬀect the visual appearance of the color. In spite of these good results, the treatment of larger areas is expected to be problematic as the manual operation of the handpiece involves an unknown beam overlap and uncontrollable laser exposure parameters. The implementation of a computer-controlled X–Y–Z work station, as used for the excimer lasers mentioned above, could enhance the selectivity of the ablation and homogeneity of the result. In view of these promising results, advanced laser tests were carried out with the Q-switched Nd:YAG lasers on the Type 1 samples. The established optimum ﬂuence varied between 106 and 165 mJ cm−2 . Using OM and EPMA, the ablation rate was determined, as the graph in Fig. 18.4 illustrates. Colorimetry measurements (CIE L∗ a∗ b∗ ) revealed that the ablation of the overpaint at the established optimum parameters (F =100–160 mJ cm−2 ), also involved a considerable change in color of the yellow ochre (Fig. 18.5), cinnabar, and burnt sienna pigments (Table 18.1). Raman analysis was carried out by Dr. Peter Vandenabeele of the University of Ghent. The results conﬁrmed the degradation of the cinnabar, burnt sienna, and yellow ochre pigments, as illustrated in Fig. 18.6. Consequently, an uncovering by means of laser cannot be considered when these pigments are involved. 18.3.2 Type 2: Limewash-on-Tempera The number of tests with the excimer lasers remained limited as it soon became clear that no ablation of the limewash took place during irradiation with UV laser light (193, 248, and 355 nm).

156

G. Van der Snickt et al.

Fig. 18.5. Sample with yellow ochre, partially uncovered. Colorimetry measurements showed that the pigment was discolored as a result of the laser treatment Table 18.1. Results of the colorimetry measurements L burnt sienna lead white green earth lamp black cinnaber yellow ochre

Reference sample a b

33.41 91.5 36.78 22.67 49.88 61.39

13.69 −0.64 −7.47 0.28 34.22 16.15

10.86 4.15 11.24 0.22 16.07 44.37

Sample uncovered ∆L ∆a ∆b 7.93 −3.19 −0.67 −0.37 −15.36 −6.64

−1.97 0.38 0.09 −0.94 −26.5 −4.65

−3.73 1.77 −1.79 −1.5 −12.65 −13.58

∆E 8.98 3.66 1.92 1.81 33.14 15.82

Fig. 18.6. Type 1 sample: “oil-on-oil.” Uncovering by laser (left) vs. uncovering by scalpel (right). OM ×50

18 The SALUT Project: Study of Advanced Laser Techniques

157

Better results were obtained with IR-radiation at 1,064 nm. During the preliminary laser tests, it was possible to remove the layer of limewash almost completely at a ﬂuence of 100–150 mJ cm−2 . Although visually no alteration of the uncovered pictorial layer was established, inspection of the surface with OM (×200) revealed that the tempera was aﬀected by direct contact with the laser beam. Analogous with the samples of Type 1, the texture of the pictorial layer modiﬁed from smooth to granular as a result of the interaction with the laser beam. However, examination of cross sections made it clear that no signiﬁcant thinning of the pictorial layer took place if the ablation was carried out with optimal exposure parameters. After ablation, a minor residue of limewash remained on the pictorial layer in the form of small clusters scattered over the surface. The remains of limewash were mainly located in small depressions, where the overpaint was thicker. In order to assess the impact of the thickness on the result, a thicker layer (>25 µm) was applied on the samples which were tested during the advanced layer tests. The parameters were varied within the capabilities of the laser system, but no ablation of the limewash took place. Unfortunately, wetting the surface did not improve the results substantially. Other factors, such as the carbonation rate of the limewash were considered, but it was concluded that the lack of ablation was caused by the increased thickness of the unwanted layer. 18.3.3 Type 3: Acrylic-on-Oil Laser tests showed that the wavelengths of 1,064 nm (IR), 248 nm, and 355 nm (UV) are less suitable for this task as the removal of the acrylic implied, in all cases, important damage to the pictorial layer. Better results were achieved with an excimer laser emitting at 193 nm. The acrylic paint was completely ablated at a ﬂuence of 460 mJ cm−2 and 200 pulses per site. As for the tests mentioned above, the pictorial layer was slightly ablated without causing substantial thinning, discoloration, or other visual modiﬁcations.

18.4 Conclusions Evaluation of the test results proved that Q-switched Nd:YAG laser systems emitting at 1,064 nm show potential for the removal of unwanted, superimposed layers of oil paint or limewash on secco wall paintings, especially when the results are compared with uncovering by means of scalpel. Although these laser systems were designed for self-limiting applications, the overpaint on the samples could be removed without substantial thinning of or damage to the pictorial layer. However, as a result of direct interaction with the laser beam, the texture of the pictorial layer was modiﬁed. Advanced laser tests, using the established optimum parameters (F =100–165 mJ cm−2 ) were carried out

158

G. Van der Snickt et al.

Fig. 18.7. The pictorial layer after ablation of the overpainting, OM ×200. Type 1 and 2 were uncovered with a Q-switched Nd:YAG laser emitting at 1,064 nm; Type 3 was treated with an excimer laser emitting at 193 nm

on samples with historic pigments. No signiﬁcant change in color was determined by means of colorimetry except for cinnabar, yellow ochre, and burnt sienna. Consequently the uncovering of wall paintings cannot be considered when these quite common pigments are present in the pictorial layer. Moreover, no ablation took place when the limewash layer was thicker than 25 µm. As a result the application of this technique will be limited. Also, the treatment of larger areas will require the implementation of a computer-controlled X–Y–Z station in order to maintain constant beam exposure parameters (overlap, ﬂuence, etc.). An acrylic overpaint was successfully removed by an excimer laser emitting at 193 nm (460 mJ cm−2 , 200 pulses per site) but a low ablation rate. Figure 18.7 gives a visual summary of the overall results. Acknowledgments The SALUT project was funded by the IWT-Flanders. We would like to thank all participating members and other institutes or companies that lent their support to our project. Unfortunately, the list is too long to mention them all. Special thanks to the researchers at IESL-FORTH and also to Dr. B. Lievens for reviewing this article.

Part III

Inorganic Materials

19 Comparison of Wet and Dry Laser Cleaning of Artworks ∗

A. Sarzyński , K. Jach, and J. Marczak Institute of Optoelectronics, Military University of Technology, 2 Kaliskiego Str., 00-908 Warsaw, Poland ∗ [email protected] Summary. This paper presents results of numerical calculations of graphite (model medium for black encrustation) removal from aluminium (model medium for artwork substrate) using laser ablation method. Theoretical comparison of cleaning rate of graphite as a solid encrustation as well as porous graphite ﬁlled with air or water, using pulsed laser radiation, are presented. Moreover, results of numerical modeling of pressure pulse, generated by laser pulse in graphite ﬁlled with water, are shown.

19.1 Introduction Laser surface cleaning has become a mature technological process, widely applied in electronic, air, nuclear industry as well as artworks conservation [1–4]. The eﬃciency of the cleaning process depends on many factors such as characteristics of laser radiation, optical, mechanical and thermal properties of material under treatment. An important role in the process is played by the presence of liquid, e.g., water, at the cleaned surface [5] which substantially increases, for diﬀerent substrates, the eﬃciency of particle and encrustation removal. Usually porous structure of encrustation facilitates absorption of atmospheric gases and water. A large surface area of pores and, most of all, the presence of gases and water, change the physical properties of encrustation and beam propagation conditions: – Often increase average absorption coeﬃcient – Laser radiation repeatedly refracts and reﬂects, causing interference and local increase of power density – Gas in pores favour ionization, causing additional increase of radiation absorption This paper presents the results of numerical modeling of solid and porous graphite (encrustation imitation) removal from homogeneous substrate. It has

This page intentionally blank

162

A. Sarzyński et al.

been assumed that pores are ﬁlled with air (dry cleaning) or water (wet cleaning). The inﬂuence of water as an inert layer on the amplitude of the pressure pulse in irradiated sample is also presented.

19.2 Numerical Model of Laser Ablation Using hydrodynamic code described earlier [6, 7], radiation absorption as well as pressure, temperature, and density of expanding plasma, has been simulated. The system of partial diﬀerential equations describing an evolution of material heated by laser radiation, expresses the laws of conservation of mass, momentum, and energy. Phase transitions like vaporization and melting are not taken into consideration. For one-dimensional case, the form of equations is as follow: Law of conservation of mass (conservation of continuity) ∂ρ ∂υ ∂ρ +υ +ρ =0 ∂t ∂z ∂z Law of momentum conservation ∂υ ∂υ ∂p ρ +υ =− ∂t ∂z ∂z Law of conservation of energy dε ∂υ ∂ ∂q − ∂T ∂q + ρ = −p + + − ρKre − Qs χ − dt ∂z ∂z ∂z ∂z ∂z where

(19.1)

(19.2)

(19.3)

d ∂ ∂ = +υ (19.4) dt ∂t ∂z is the so-called substantial derivative. Other denotations are as follows: ρ, density; p, pressure; ε, speciﬁc internal energy; t, time; z, spatial coordinate; T , temperature; χ, coeﬃcient of thermal conductivity; Qs , ionization energy losses; q + , q − , incident and reﬂected laser radiation ﬂux (reﬂection takes place for so-called critical electron concentration); Kre , coeﬃcient describing the interaction of medium thermal radiation with matter. Thermodynamic, mechanical, and optical properties of materials are described using a few dozen constants. These data are usually unattainable and that was the reason for substitution of encrustation by graphite and substrate by aluminium, materials with known parameters. Such a simulation is suﬃcient for qualitative calculations. Grüneisen’s equation of state [8–10] has been applied for graphite, while the Thomas–Fermi model [9] served for determination of carbon ionization level. Equations (19.1)–(19.4) have been solved numerically using ﬁnite diﬀerences, applying explicit diﬀerential scheme.

19 Comparison of Wet and Dry Laser Cleaning of Artworks

163

19.3 Results of Numerical Calculations Presented numerical model, after minor modiﬁcations of absorbing medium, allows us to evaluate the inﬂuence of medium porosity on laser ablation. Laser ﬂuency has been varied from 1 to 200 J cm−2 for constant laser pulse duration of 5 ns. Moreover, relative pore volume varied from 10% to 60% (Fig. 19.1). Figures 19.2 and 19.3 present the results of numerical modeling of vaporized layer thickness d as a function of pulse energy in the case of graphite pores ﬁlled with air or water. For a graphite layer thickness of 3 µm and pores relative volume of 60 %, energy density of only 100 mJ cm−2 is enough to remove the whole surface layer. Thus, further calculations have been performed for a graphite thickness of 8 µm. The presence of pores increases the encrustation removal rate. In the wet medium, ablation rate is even higher (as in our experiment and literature). Laser pulse

Laser pulse

Laser pulse

Encrustation Substrate-aluminium

Substrate-aluminium

a)

b)

Substrate-aluminium c)

Fig. 19.1. Geometry of medium illumination: (a) graphite–aluminium, (b) porous graphite ﬁlled with water–aluminium, and (c) porous graphite ﬁlled with water– aluminium 6.0 60% 5.0 60%

d [µm]

4.0 3.0

40% 20% 10%

2.0 1.0 0%

0.0 0.001

0.010

0.100

1.000

10.000 100.000 1000.000

E[J/cm2]

Fig. 19.2. Thickness of graphite layer vaporized from aluminium substrate by laser pulse. Numbers denote graphite pores volume in percentages (ﬁlled with air). Laser pulse duration –5 ns

164

A. Sarzyński et al.

Fig. 19.3. Inﬂuence of pore contents on the thickness of graphite layer vaporization. Numbers denote graphite pores volume in percentages. Laser pulse duration – 5 ns Local thickness of liquid layer

Graphite

Inert layer

L ~ several mm L

L Substrate-aluminium

Substrate-aluminium

Fig. 19.4. Geometry of laser illumination of a real medium and its numerical equivalent

Physics of interaction of laser radiation with porous medium still hides many ambiguities and mysteries thus, models of such interactions are still not accurate. Another important situation occurs when water is not penetrating the porous encrustation structure, but forms a certain liquid layer at the surface. In this case, water creates inert layer with density comparable to density of encrustation, signiﬁcantly increasing the amplitude of pulse pressure. It is illustrated in Fig. 19.4. In calculations of inﬂuence of inert layer on the amplitude of pressure pulse, it was assumed that absorption (encrustation) layer is represented by solid graphite, treated substrate is aluminium and the eﬀect of inert layer is reduced to the boundary pressure condition. External pressure at the edge of absorption layer has been introduced as: Pext = ρext vb2

(19.5)

19 Comparison of Wet and Dry Laser Cleaning of Artworks

165

Fig. 19.5. Inﬂuence of an inert layer on the shape of pressure pulse, propagating in the aluminium sample at 25, 50, 75, and 100 ns from the start of laser pulse (τ = 10 ns) for two laser energy densities, 1 and 10 J cm−2 . Lowest characteristic curve – lack of inert layer; middle curve – water inert layer; upper curve – solid wall

166

A. Sarzyński et al.

where vb is the velocity of external surface of absorption layer and ρext is the density of inert layer sticking to the absorption layer (air, water, or heavy glass): ρext = 0.001 g cm−3 (for air), 1 g cm−3 (for water), and ∞ (for solid). All calculations have been performed for triangular 10 ns laser pulse, 5 µm thickness of graphite absorption layer and 600 µm thickness of aluminium sheet. For calculation time of 100 ns used, pressure pulse did not reach the backside of sheet. Figure 19.5 shows comparison of pressure distribution in aluminium foil for two laser pulse energy densities: 1 J cm−2 and 10 J cm−2 , at 25, 50, 75, and 100 ns after the beginning of laser pulse. It can be seen that absorption and inert layer increase the pressure pulse amplitude from three to ten times. In principle, pulse propagation velocity is independent on its amplitude, which results from the fact that shock waves under consideration are weak. Density variations are no greater than 10%. Strong shock waves occur in solids only for pressures above 105 –106 bars (10–100 GPa). In the cases considered above, pressure amplitudes do not exceed 5 GPa.

19.4 Conclusions Amplitude of pressure pulse depends, ﬁrst of all, on the parameters of absorption and inert layer. The substrate inﬂuence on the pressure pulse shape could be observed only in a short time, which is not presented there. In principle, pressure pulse amplitude is independent of type of substrate. However, a distinct dependence of pressure on mechanical properties of absorption layer has been observed. As it follows from numerical calculations, water can substantially increase the eﬀectiveness of porous encrustation removal. However, a thicker water layer can act an as inert layer. In the latter case, pressure pulse amplitude in the cleaned substrate will signiﬁcantly increase, leading to the object damage. Application of water as an inert layer could be favorable in laser micromachining of metals. Water as an inert material permits decrease of laser beam parameters (e.g., energy density), thus machining costs, but excludes the use of short-wavelength radiation. The higher pressures have been obtained for laser radiation at 1,064 and 532 nm (second harmonic). It is also impossible to apply higher pulse repetition frequencies, because it is necessary to recreate the homogeneous water layer before the next pulse in series. Finally, peculiar precautions should be taken during water-assisted ablative removal of art works encrustation. The generation of a strong, local pressure pulse is possible, which can sometimes be ten times higher than in the case of dry laser cleaning. Acknowledgments The work was funded by the Polish Ministry of Science and Information Society Technologies, project E!3483 “Advanced Laser Renovation of Old

19 Comparison of Wet and Dry Laser Cleaning of Artworks

167

Paintings, Paper, Parchment And Metal Objects” (grant No 120/E-410/SPB/ EUREKA/KG/DWM 97/2005-2007).

References 1. J. F. Asmus, C. G. Murphy, and W. H. Munk, in Proceedings of SPIE, Vol. 41, 19, 1973. 2. J. Asmus, in Laser Focus, Vol. 12, 56, 1976. 3. M. Cooper: Lasers in Conservation. An Introduction, Oxford 1999. 4. J. Marczak, in Proceedings SPIE, 2001, Vol. 4402, 202–209. 5. B. S. Luk’yanchuk, (Ed.), Laser Cleaning, 2002. 6. K. Jach, A. Morka, M. Mroczkowski, R. Panowicz, A. Sarzyński, W. St¸epniewski, R. Świerczynski, and J. Tyl, in Wydawnictwo Naukowe PWN, Warszawa, 2000. 7. J. Marczak, in Wydawnictwo Bel Studio, Warszawa, 2004. 8. G. I. Kerley and L. Chhabildas, in Sandia Report, SAND2002-2619, Sept. 2001. 9. M. von Allmen and A. Blatter, Laser-Beam Interactions with Materials, Springer, 1995. 10. S. Eliezer, A. Ghatak, H. Hora, and E. Teller, An Introduction to Equations of State – Theory and Applications, Cambridge, 1986.

20 Laser Cleaning of Avian Eggshell ∗

L. Cornish , A. Ball, and D. Russell The Natural History Museum, Cromwell Road, South Kensington SW7 5BD UK ∗ [email protected].

Summary. A low vacuum SEM was used to evaluate the eﬀect of using an Nd:YAG laser as a non-contact technique for cleaning avian eggshells. The technique shows potential, since there are no obvious deleterious eﬀects from cleaning, but further study is required to understand how the laser is interacting with the sample surface.

20.1 Introduction Surface cleaning of museum objects is a frequently performed task in conservation. Since the 1970s interest has developed in the use of alternative non-contact methods of treatment [1]. This is primarily in response to the fragility of some surfaces and the invasive nature of more traditional methods, such as mechanical and chemical cleaning. In chemical cleaning for example there is often surface penetration and the chemical action can continue long after the chemical has apparently been removed [2]. Laser cleaning is one such non-contact method and it is fast becoming an established cleaning method for certain types of museum object, e.g. historic buildings, monuments and sculpture. The use of lasers to clean natural history material is a more recent innovation with comparatively few publications and has yet to become fully recognized as an established cleaning technique.

20.2 Background A disadvantage for many conservators is the high costs involved in purchasing and maintaining a laser. In 1999 a Joint Research Equipment Initiative was set up between the Natural History Museum, Imperial College of Science, Technology and Medicine, the Victoria and Albert Museum, The Royal College of Art and the Tate Gallery. A grant application was submitted to the Engineering and Physical Sciences Research Council for funds to aid the purchase of a portable Q-switched Nd:YAG dual wavelength laser (1,064/532 nm) cleaning

170

L. Cornish et al.

system. The application was successful and the participating institutions share equal time on the laser [3]. The Q-switched Nd:YAG dual wavelength laser was chosen by the group, as it is the most commonly used laser for cleaning in conservation, emitting the most appropriate wavelength and energy of radiation for selective cleaning of a wide variety of surfaces [4]. The Q-switched laser is found to be very eﬀective for cleaning natural history objects [5]. The Q-switch acts as an extremely high-speed shutter and shortens pulse length of the laser. This results in an extremely intense pulse of energy with very short pulse duration (5–10 ns). The short pulse length ensures little or no temperature rise in the underlying surface and, therefore little risk of thermal damage.

20.3 The Avian Eggshell Project The Walter Rothschild Zoological Museum in Tring, Hertfordshire is home to the ornithological research collections and library of The Natural History Museum (NHM). The ornithological collections are amongst the largest and most comprehensive in the world. There are approximately 1,000,000 eggs, with an additional 2,000 nests. The collection is consulted by researchers throughout the world. The cleaning of eggshell is a diﬃcult task due to fragility of the surface. Older collections are considered to be more vulnerable due to the eﬀects of aging and therefore the current policy is that eggs are not cleaned. The evaluation of a non-contact cleaning method such as the laser was therefore considered to be an important progression in improving the standards of the egg collections. The eggs loaned from the NHM for laser evaluation came in-part from the collection of the English Portuguese naturalists, William Chester Tait and Alfred Welby Tait. The collection is important as it also includes extensive documentation which records the development of Tait’s interest in Portuguese birds. Tait’s egg collection is of considerable importance both historically and scientiﬁcally as very little oological material from Portugal is known. The collection is unfortunately very dirty having been neglected for many years prior to its arrival at the NHM. Two of the Scops Owl (Otus scops scops) eggs, collected in 1878 were selected for use in the evaluation. Many eggs also contain pigmentation on their surface and it was decided to extend the study by testing such areas on a Guillemot egg (Uria aalge) to see if the laser aﬀected the pigmented areas more than the unpigmented areas. Eggshell is not a homogeneous structure. It is composed of an organic matrix, or framework, of delicate interwoven ﬁbres, and an interstitial substance composed of a mixture of inorganic salts. The proportions of these constituents vary in the eggshells of diﬀerent species of birds. The organic matrix is a collagen-like protein and the minerals are mainly carbonates and phosphates of calcium and magnesium, of which calcium carbonate is the most plentiful. There are also minute pores within the eggshell which are ﬁlled with

This page intentionally blank

20 Laser Cleaning of Avian Eggshell

171

Fig. 20.1. (a,b) SEM views of Scops owl eggshell surface using BSE detector. The surface is divided into irregular shell units overlaid by a thin cuticle which appears transparent to the BSE detector. The cuticle in this sample appears to have been invaded by fungal hyphae. Chamber conditions: 23 Pa, 20 kV, 14 mm WD. Scale bar: (a) 200 µm, (b) 50 µm

protein ﬁbres. On the external surface of most bird eggs there is an extremely thin transparent coating of protein – the cuticle (Fig. 20.1). Cuticular pigments are deposited as granules in the deeper portion of the cuticle. The intensity of pigmentation depends upon the thickness of this deposit [6]. An important criterion for both conservator and curator was agreeing what part of the eggshell to clean. Eggs were commonly labelled by writing directly onto the surface in ink. It was agreed that only areas without writing should be cleaned to avoid potential loss of data. In order to evaluate the eﬀect of laser cleaning of the eggshell surface a low vacuum scanning electron microscope (SEM) was utilized. Low vacuum SEM allows samples to be imaged without the need to apply conductive carbon or metallic coatings. This non-invasive procedure allows items to be removed from museum collections, photographed and returned with no alteration or surface treatments. If samples are introduced into a conventional high vacuum SEM without a conductive surface coating, surplus electrons from the primary electron beam can accumulate on the sample surface where they deﬂect the beam and interfere with the image forming process. This phenomenon is called “charging”. In a low vacuum SEM, the electron gun and column operate at high vacuum, whilst the chamber is maintained at a low vacuum. Typically a high vacuum SEM operates at a chamber pressure of 3 × 10−4 Pa, whereas a low vacuum SEM will achieve images from 5 Pa to 2–3 kPa (atmospheric pressure is 4.2 kPa). The atmosphere within the sample chamber is typically air at low pressures and water vapour at higher chamber pressures. In either case, the “charge” on the sample is absorbed by gas molecules in the chamber. The vacuum system pumps these charged gas molecules away and a computercontrolled leak valve maintains the gas pressure at a constant pre-set level. The result is charge-free imaging on virtually any sample. The low vacuum conditions also allow samples that may be damaged by high vacuum conditions to be examined with some degree of safety.

172

L. Cornish et al.

High vacuum SEM’s are normally ﬁtted with either secondary (SE) or backscattered electron (BSE) detectors. BSEs are high energy electrons scattered from relatively deep inside the sample (typically 2–3 µm) and the BSE detector can give some information about the chemical composition of the sample. Secondary electrons (SE) are produced close to the sample surface (less than 10 Å). This signal provides more information about the topography of the sample than the BSE signal. The standard BSE detector can be used in a low vacuum SEM, but the detection of an SE signal in the gaseous atmosphere of the chamber requires a dedicated low vacuum SE detector.

20.4 Methods Specimens were examined visually and photographed. They were then examined using a LEO 1455VP SEM (Fig. 20.2). Charge-free imaging was obtained at an accelerating voltage of 15–20 kV and a chamber pressure of ca. 20–25 Pa. Electron detection was via a four quadrant diode-type BSE detector, with one quadrant switched oﬀ to give some topographical contrast, or a variable pressure secondary electron detector (VPSE). Working distances ranged from 15 to 40 mm. Specimens were cleaned using the Q-switched Nd:YAG laser at 1,064 nm. Due to the delicate nature of the eggshell, the laser handpiece had an aperture inserted to give a reduced energy output. The aperture reduced the energy to 25 % of the monitor reading – hence 100 mJ on monitor = 25 mJ actually used. Cleaning parameters are given in Table 20.1.

Fig. 20.2. Internal views of SEM specimen chamber showing a Guillemot egg in position on the stage. The various detectors are labelled. The TV cameras are positioned on the back of the chamber door (left image) and the left wall of the specimen chamber (right image)

20 Laser Cleaning of Avian Eggshell

173

Table 20.1. Laser cleaning parameters used Specimen egg of

Fluence (J cm−2 )

Repetition rate (Hz)

Energy (mJ)

Wavelength (nm)

Scops owl Guillemot

<0.3 <0.3

1 1

25 25

1,064 1,064

Fig. 20.3. Scops owl egg. Examination of the transition from cleaned to uncleaned regions. The lines mark the approximate boundary between the cleaned area (top) and uncleaned area (bottom). (a) BSE image showing sub-cuticular structures, (b) VPSE image showing detail at level of the cuticle. Chamber conditions: 19 Pa, 20 kV, 24 mm WD. Scale bar, 1 mm. (c) BSE image at higher magniﬁcation. Chamber conditions: 24 Pa, 20 kV, 14 mm WD. Scale bar, 400 µm

20.5 Results Initial examinations of the owl egg where part of the shell had already been test cleaned were problematic. Although the cleaned portion of the shell was clearly visible to the naked eye (Fig. 20.5), there was no detectable diﬀerence between the cleaned area and the uncleaned areas under the electron beam. To ensure that the cleaned region was identiﬁable in the SEM, the region was marked out with conservation grade putty (Fig. 20.3). Whilst the marked out regions could be clearly identiﬁed, it still proved impossible to recognize diﬀerences between cleaned and uncleaned portions of eggshell (Fig. 20.3). It was decided to try a more accurate way of marking the cleaned areas using a conservation grade marker pen. Tests were carried out on an eggshell fragment from a common hen to check that the ink could be detected under SEM (Fig. 20.4) and subsequently removed by laser. This test proved successful and an area was marked out for cleaning and imaged under SEM. After cleaning, the area was compared under SEM with the adjoining uncleaned area. The same method was used in evaluating the Guillemot egg (Fig. 20.5). Once again, examination under SEM revealed no recognizable diﬀerences between the cleaned and uncleaned areas (Fig. 20.6). However, in this case, visual examination of the egg also revealed that the pigmented areas appeared unaﬀected by laser treatment.

174

L. Cornish et al.

Fig. 20.4. Hen egg. Tests using curatorial grade pen and diﬀerent electron detector settings. (a) BSE detector. The pen mark is almost impossible to detect using this technique. (b) VPSE detector. The pen mark is highly visible using this technique, however cracks in the shell appear brighter than the background, an artefact of VPSE imaging. (c) A mix of VPSE and BSE signals provides a good compromise between detection of surface and underlying features. Chamber conditions: 23 Pa, 20 kV, 20 mm WD. Scale bar, 1 mm

Fig. 20.5. Optical eﬀects of cleaning. (a) Scops owl egg. The left side is uncleaned. This is the same sample shown in Fig. 20.3, arrowheads mark the position of the conservation grade putty. Scale bar, 10 mm. (b) Guillemot egg. This image shows the marked-out square visible in Fig. 20.6 after cleaning. Arrows 1 and 2 refer to images in Fig. 20.6. Laser cleaning does not aﬀect the pigmentation. Scale bar, 2.5 mm

In Uria, the eggshell has a thick (8–10 µm) polysaccharide cuticle [6]. Since the VPSE detector shows only the cuticle, the pen marks are clearly visible (Fig. 20.6). The BSE signal comes from the underlying calcitic layer below the pen marks. Despite considerable experimentation with accelerating voltages and chamber pressures to optimize the images, no setting was found where the pen marks were visible under BSE. This is signiﬁcant since many low vacuum SEMs will only be equipped with a BSE detector. The pigmentation in Guillemot eggs lies at varying depths within the shell [6]. This may explain why the laser does not aﬀect pigmentation, and why it is visible optically, but not under the SEM, which is essentially a tool for examining surface features.

20 Laser Cleaning of Avian Eggshell

175

Fig. 20.6. Guillemot egg before and after laser cleaning. (a) VPSE and (b) BSE images of the ﬁrst area arrowed in Fig. 20.5. Chamber conditions: 19 Pa, 20 kV, 18 mm WD. Scale bar, 100 µm. (c) VPSE and (d) BSE images of the second area arrowed in Fig. 20.5 after cleaning. Chamber conditions: 25 Pa, 20 kV, 19 mm WD. Scale bar, 500 µm

20.6 Conclusions These initial results show that the laser has considerable potential for this type of cleaning, but underlie the importance of understanding, in detail, the nature of the material to be cleaned. Future research will be directed towards detecting variations in light elements, perhaps using Secondary Ion Mass Spectrometry [7]. This may reveal changes in the eggshell too subtle for SEM examination to reveal. In addition a detailed investigation of eggshell cross-sections will be carried out to check the impact of laser cleaning in proﬁle. Acknowledgements The NHM photographers (Fig. 20.5). Ed Teppo for funding and Errol Fuller for loan of a Guillemot egg.

References 1. J. F. Asmus, C. G. Murphy, and W. H. Munk, in Proceedings of SPIE, Vol. 41, 19, 1973. 2. A. Burnstock and T. A. Kielich, in ICOM-CC 10th Tirennial Meeting Preprints, J. Bridgland (Ed). London: James and James, 253–62, 1997.

176

L. Cornish et al.

3. M. Sokhan, P. Gaspar, D. McPhail, L. Cornish, C. Hubbard, and D. Pullen, in Lasers in Conservation of Artworks, LACONA IV. ICOMOS France. 303, 2001. 4. M. Cooper, Laser Cleaning in Conservation: an Introduction, Oxford, 98, 1998. 5. L. Cornish and C. Jones, in Conservation Science 2002, 101–106. Archetype, 270, 2002. 6. K. E. Mikhailov, in Avian eggshells: an atlas of scanning electron micrographs. British Ornithologists’ Club Occasional Publications. No. 3. 88, 1997. 7. D. S. McPhail, (in press) Some applications of SIMS in conservation science, archaeometry and cosmochemistry.

21 Removal of Strong Sinter Layers on Archaeological Artworks with Nd:YAG Laser ∗

J. Hildenhagen1 , K. Dickmann1 , and H.-G. Hartke2 1

∗ 2

Laser Center Fachhochschule Münster (LFM), FB Physikalische Technik, University of Applied Sciences, Stegerwaldstr. 39, 48565 Steinfurt, Germany [email protected] Rheinisches Landesmuseum Bonn, Colmantstr. 14–16, 53115 Bonn, Germany

Summary. For numerous archaeological objects covered with strong sinter layers, restoration using traditional cleaning methods often fails. Within this study the potential of laser cleaning applied to archaeological pottery and glass artwork is presented. In this case the application of various Nd:YAG laser wavelengths leads to very satisfactory results, experimentally proven by SEM and EDX analyses.

21.1 Introduction Archaeological artworks being exposed over hundreds of years to soil elements in solid and dissolved form show diﬀerent types of destruction and modiﬁcation. One common type is an overlaid sinter layer of lime mixed with other elements. Ground water is enriched with carbon dioxide and carbonic acid. As a result water is able to dissolve lime from soil and stones. Subsequent changes of temperature, pressure and pH value cause a separation of lime. Archaeological objects located in the ground are enhancing by this eﬀect and thus they are oﬀering an agglomeration area for limy sinter layers. Due to storage conditions like soil humidity and pressure, as well as the artwork composition, there are diﬀerent types of sinter layers: weak ones can be cleaned by brushes, harder ones by scalpel, acid or complexing agents. In combination with sensitive original surfaces, these sinter layers are a challenge for laser restoration. The Rheinisches Landesmuseum (Bonn/Germany) is responsible for restoration of a wide variety of archaeological ceramic, pottery and glass objects from Roman and medieval times. Many of these objects are covered with strong sinter layers as described above. In co-operation with the Laser Center FH Münster, a Q-switched Nd:YAG laser (Thales SAGA 220/10) was used for cleaning experiments for a gentle and satisfactory removal.

178

J. Hildenhagen et al.

21.2 Initial Situation Based on a close co-operation between Rheinisches Landesmuseum and Laser Center FH Münster, a wide variety of archaeological samples have been made available for laser cleaning studies. All samples derive from excavations around the region of Cologne. Most of the objects are fragments of Roman military camps as well as medieval pieces. For these samples, strong covering with sinter layers is not common. However, selected pieces represent very nicely this widespread problem that is well known from the Mediterranean region. In many cases of traditional restoration, the removal of strong sinter layers is not trivial due to the sensitivity of the original artwork surface. Nowadays restorers apply mechanical removal techniques such as scalpel or micro-sandblasting or they use chemical solvents or complexing agents. In cases where the sinter layer shows a greater hardness or increased chemical resistance compared to the original artwork surface, the removal is risky or even impossible. Within this study the authors investigated the potential of Nd:YAG laser cleaning in order to solve these problems. Special attention was paid to ﬁnd the optimal wavelength amongst the four applied wavelengths (266, 355, 532, 1,064 nm). A state-of-the-art cleaning laser (SAGA 220/10) including articulating arm and handpiece was used. The maximum pulse energy was 1.5 J at the fundamental wavelength and decreased with shorter wavelength (e.g. 0.17 J at 266 nm) [1]. For the analysis of the removal results, optical microscopy, SEM and EDX analyses were used.

21.3 Laser Cleaning of Archaeological Pottery The medieval pottery mug shown in Fig. 21.1 is characterised by typical circular grooves covered by a thin but strong and brittle sinter layer. Mechanical cleaning techniques have shown a removal process in terms of a spallation of tiny ﬂakes also accompanied by a spallation of the original surface underneath. In contrast, laser cleaning at 1,064 nm yielded a very promising result. We have determined a threshold for the sinter layer of 1.3 J cm−2 ; for laser cleaning of extended areas, a ﬂuence of 2.1 J cm−2 was used. Moistening of the surface enabled a reduction of the ﬂuence (1.5 J cm−2 ), produced a better cleaning result and increased the cleaning rate (by approx. 2×). This eﬀect is due to water absorption of lime inside the sinter layer and subsequent laser induced vaporisation, resulting in an enhanced removal mechanism. In combination with water moistening, we obtained a cleaning rate of several cm2 per minute for the pottery shown in Fig. 21.1. It should be mentioned that traditional mechanical cleaning methods totally failed in this case. Using light optical microscopy and colour measuring (L∗ a∗ b∗ ) analysis, no laser induced side eﬀects on the original surface could be detected.

21 Removal of Strong Sinter Layers on Archaeological Artworks

179

Fig. 21.1. Medieval pottery mug; partially laser cleaned at 1,064 nm

However, these promising results could not be transferred to a Roman period pot with a glazed surface covered with lime sinter layer. The shiny eﬀect of the glaze was lost after laser irradiation with a ﬂuence of 1.1 J cm−2 at 1,064 nm. Further cleaning studies were carried out on pottery fragments of a Roman bowl covered by sinter layers (300–400 AD). The high iron content is responsible for the red colouring of the sinter layer. As already mentioned above, the removal of sinter layers at 266 nm (Ep,max = 170 mJ cm−2 ) and 355 nm (Ep,max = 320 mJ cm−2 ) was not suﬃcient. (In principle, ﬂuences above the ablation threshold can also be obtained for these wavelengths. However, corresponding spot sizes are too small in order to achieve suﬃcient cleaning rates). A delicate removal of sinter layers from the glazed surface was possible by use of 1,064 nm (1.5 J cm−2 ) and 532 nm (0.7 J cm−2 ), respectively. Unfortunately, laser cleaning was accompanied by a slight discolouration of the original pottery surface as depicted in Fig. 21.2. Use of excessive ﬂuence at 532 nm led to bluish discolouration whilst irradiation at 1,064 nm results in greenish discolouration. In this case these side eﬀects were irreversible and revealed a signiﬁcant drawback of the cleaning process and moistening of the surface did not show any enhancement. For further archaeological pottery samples, we obtained an inﬂuence of quartz grains being embedded within the sinter layer. Laser removal becomes critical since quartz shows high transparency over a wide wavelength range (VIS, NIR). This problem is well known from laser cleaning of archaeological ironwork covered with strong corrosion crusts [2]. An improvement is possible by moistening the surface. The generated vaporisation supports the removal

180

J. Hildenhagen et al.

Fig. 21.2. Fragment of pottery (300–400 AD) with laser cleaned areas: (a) 532 nm, 100 mJ cm−2 ; (b) 532 nm, 200 mJ cm−2 ; (c) 532 nm, 300 mJ cm−2 ; (d) 1, 064 nm, 140 mJ cm−2 and (e) 1, 064 nm, 120 mJ cm−2

of grains but also increases the risk of damaging the original surface due to spallation. This occurs particularly for sinter layers showing a strong porosity.

21.4 Laser Cleaning of Archaeological Glass As mentioned above, due to unsuﬃcient output energy at 266 and 355 nm (for the applied laser system), the cleaning of archaeological glass was only carried out at 532 and 1,064 nm. Sinter layers from a small medieval bottle neck could be successfully removed without any damaging of an underlying sensitive corrosion layer (Fig. 21.3). This layer is known as the “gel” or “iris” layer and is very fragile. In particular cases characterised by sinter layers being close connected to the gel layer, a local damaging of the gel layer could not always be prevented. Particularly this eﬀect has been observed during laser cleaning of a medieval glass window fragment. The strong and thick black sinter crust was better inter-connected to the gel layer compared to the connection of the gel layer to the original glass surface. In this speciﬁc case there was only a complete removal at the crust/gel-compound possible (Fig. 21.4). There was no diﬀerence in using either 532 or 1,064 nm and no side eﬀects in the bulk glass were detected. Further cleaning studies have been carried out on a fragment of a thinwalled Roman glass bottle (200–300 AD). This sample is covered by an irregular, strong sinter layer. Laser cleaning at 1,064 and 532 nm was successful as depicted in Fig. 21.5. The corresponding SEM-analysis revealed a satisfying removal of the sinter layer without observable side eﬀects on the gel layer underneath. Additionally, EDX analysis proved this result.

21 Removal of Strong Sinter Layers on Archaeological Artworks

181

Fig. 21.3. Fragment of a small medieval bottle neck. Left, before; right, after laser cleaning

Fig. 21.4. Glass fragment (approx. 400 AD) with strong cinter crust. Top: laser cleaning at various wavelengths. Bottom: glass cross section

182

J. Hildenhagen et al.

Si O 30

Cleaned

20

10

Al Mg Na

Ca

0 2

4

6 Energy (keV)

Fig. 21.5. Top left: Fragment of a Roman glass (300 AD) with strong sinter layer, partially cleaned. Top right: SEM analysis before and after laser cleaning. Bottom: EDX spectra before and after laser cleaning at 532 nm

21.5 Conclusions For the removal of strong sinter layers on archaeological pottery and glass, ﬂuences in the range of 100–500 mJ cm−2 are necessary. The ablation threshold may even vary over the sample surface due to inhomogeneities of adhesion

21 Removal of Strong Sinter Layers on Archaeological Artworks

183

and surface roughness of the sinter layer. For suﬃcient cleaning with the applied laser system, only 1,064 and 532 nm were useful. Concerning the cleaning result, there was no observable diﬀerence between the two wavelengths. Cleaning rates up to several cm2 per minute were obtained. It has turned out that laser cleaning of a sinter layer covering pottery becomes critical in those cases where quartz grains are imbedded within the gel layer. Furthermore, a discolouration of the original surface cannot always be avoided. However, very satisfying results may be obtained for individual restoration problems as has been demonstrated exemplarily for a medieval pottery bowl. In contrast, laser cleaning of archaeological glass was much more satisfactory. Only in cases of very close inter-connections between the sinter layer and the gel layer, a removal of both layers at the same time could not always be prevented. For practical applications it should be mentioned that partially transmitted laser light might also cause interactions on the back side of the glass. Acknowledgements This project is funded by the EUREGIO (INTERREG III Program) within the “EUREGIO CENTER for Art Restoration Technology” (ECEACT).

References 1. M. Chappé, J. Hildenhagen, K. Dickmann, and M. Bredol, in Journal of Cultural Heritage, Vol. 4, Sup.1, 152, 2003. 2. K. Dickmann, J. Hildenhagen, J. Studer, and E. Müsch, in LACONA 5, Berlin, 71.

22 From the Lab to the Scaﬀold: Laser Cleaning of Polychromed Architectonic Elements and Sculptures ∗

M. Castillejo1 , C. Domingo2 , F. Guerra-Librero3 , M. Jadraque1 , M. Martín1 , M. Oujja1 , E. Rebollar1 , and R. Torres1 1

∗ 2 3

Institute of Physical Chemistry Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain [email protected] Institute of Structure of Matter, CSIC, Serrano 121, 28006 Madrid, Spain Coresal, Donoso Cortés 90, 28015 Madrid, Spain

Summary. This work presents the results of laboratory tests aiming at the characterization of painting materials by LIB and FT-Raman spectroscopies and at identiﬁcation of the best laser cleaning conditions of polychromes of Spanish Heritage: polychromes on gypsum mortar of the Church-Fortress of Santa Tecla of Cervera de la Cañada, Zaragoza, ﬁfteenth century, and appliqué relief brocades on wooden sculptures of the Chapel of San Miguel, Cathedral of Jaca, Huesca, sixteenth century.

22.1 Introduction Whilst laser cleaning of architectural façades and sculptures is a widely accepted restoration procedure [1, 2], its application to polychromed surfaces is a subject of concern due to the sensitivity of painting materials to light, including pigments and binders. Discoloration and degradation caused by laser irradiation constitute the object of ongoing research [3,4]. Application of lasers for the elimination of dirt and contamination layers from polychromes requires a preliminary and systematic study of the adequate conditions to guarantee the preservation of the artwork. Here we present the results of a collaborative eﬀort in which a preliminary study in the laboratory allowed the identiﬁcation of both painting materials and the optimum laser parameters for the restoration of two types of polychrome surfaces. First we considered the mural paintings on mortar in the Church-Fortress of Santa Tecla of Cervera de la Cañada, Zaragoza, ﬁfteenth century, a ﬁne example of Mudéjar architecture of Aragon and a Spanish World Heritage site. Before restoration, these polychromes, based in chalk white and tempera applied to a gypsum plaster, were covered by a contamination layer consisting of humidity stains, soot and greasy deposits. Some

186

M. Castillejo et al.

parts were also covered by overpaintings of porous tempera. We also studied appliqué relief brocades on wooden sculptures as found in the Chapel of San Miguel, Cathedral of Jaca, Huesca, sixteenth century. Appliqué relief brocades are one of the most sophisticated styles of polychromes developed to imitate the ceremonial clothes of the high rank religious dignitaries and nobility. Brocades are complex multilayer systems consisting of overimposed tin and gold foils applied on a substrate, with the indentations on the outermost surface usually ﬁlled with red lacquer. Restoration with mechanical methods or using chemical solvents is not possible in this type of artwork due to the poor adhesion between layers and the fragility of their structure. After laboratory tests, restoration of most of the polychromed surfaces in the two-mentioned sites was ﬁnally carried out with a Q-switched Nd:YAG laser at the wavelength of 1,064 nm.

22.2 Experimental Methods Identiﬁcation of the composition of painting and contaminating layers and study of the eﬀects induced upon laser irradiation was carried out by LIB and FT-Raman spectroscopies. Cleaning tests were conducted in the laboratory with a Q-switched Nd:YAG laser (Quantel, Brilliant B, 6 ns pulses, 10 Hz) operating at the fundamental frequency and up to its third harmonic and, in the case of brocades, also with a free running Nd:YAG laser with ﬁbre delivery (Smart Clean, EL.EN SpA, 1,064 nm, 50–120 µs pulses). 22.2.1 Samples Tests were performed on a series of painting samples of white and red colours taken from the ceiling of the chorus of the Church-Fortress of Santa Tecla. On the other hand, laser cleaning tests were made in an appliqué relief brocade panel (50 × 15 cm2 ) from Chapel of San Miguel, Cathedral of Jaca (Fig. 22.1). Samples were irradiated as received without any pretreatment (e.g. wetting with water). 22.2.2 LIB and Raman Spectroscopies LIB and Raman spectroscopies were employed to analyse and identify the composition of the painting and dirt layers and to assess the modiﬁcations induced by laser irradiation [5]. LIB spectra were taken by focusing the Nd:YAG laser on the surface of the sample by a 15 cm focal length lens (to ﬂuences up to 3 J cm−2 ). The plume emission was collected at right angle to the laser beam and analysed with a 0.30 m spectrograph (TMc300 Bentham, grating of 1,200 grooves mm−1 , blazed at 500 nm) coupled to an intensiﬁed charge coupled device (ICCD 2151 Andor Technologies). The spectra were recorded with zero time delay and a temporal window of 100 µs.

This page intentionally blank

22 Laser Cleaning of Polychromed Architectonic Elements and Sculptures

187

Fig. 22.1. Laser cleaned painted ceiling on the chorus of Santa Tecla showing (a) an uncleaned square patch statue from Chapel of San Miguel decorated with appliqué relief brocades, (b) before and (c) after laser cleaning

FT-Raman spectra were recorded with an RFS 100/S-G Bruker spectrometer with a cooled Ge detector and a 1,064 nm Nd:YAG laser as the excitation source. Laser power output (40 mW) was proved not to induce any damage to the samples. The light scattered from a surface area of ∼0.01 cm2 was collected in backscattering geometry. Each data point was the result of the accumulation of 1,000 scans. The resolution was 4 cm−1 .

22.3 Results and Discussion 22.3.1 Mural Paintings of Santa Tecla of Cervera de la Cañada LIB and FT-Raman analysis and laser cleaning tests were performed on a series of painting samples of white and red colours taken from the chorus of the church. Figure 22.2 shows LIB spectra taken in a sample from a whitepainted area. The spectrum from a well-preserved surface (Fig. 22.2a) shows the presence of Ca, Na and Pb lines. In the spectrum taken in a degraded part of the same sample (Fig. 22.2b), additional C2 Swan bands are indicative of the contamination layer. Correspondingly the FT-Raman spectra indicate that calcite (as chalk white), gypsum and lead white are the compounds used for the white paint (Fig. 22.2c) and carbon is the main component of the contaminating layer (Fig. 22.2d). On the other hand, LIB and Raman analysis carried out in red painted samples allowed the identiﬁcation of two types of pigments based in vermillion (HgS) and iron oxides (spectra not shown). Laser cleaning tests were carried out on a sample covered with a grey overpainting. This layer could be eﬃciently removed with 10–20 pulses of the Q-switched Nd:YAG laser at 1,064 nm (350 mJ per pulse, 1.5 J cm−2 ). Irradiation under these conditions resulted in the uncovering of the original white paint. Figure 22.3 shows the FT-Raman spectra of the overpainting with strong carbon bands from contamination, of the original white paint showing gypsum (CaSO4 · 2H2 O) as main component, and of laser-treated

500

Lead white

1500

400

Carbon

Raman (d) C2

300

1000

C2

500

600

500

Wavelength /nm

1000 Wavenumber / cm−1

Carbon

LIBS (b)

Gypsum

Na (I)

Ca (I)

Ca (I)

Calcite

Raman (c)

Ca (I) Ca (I) Ca (I)

Ca (II)

Ca (II)

LIBS (a)

Ca (II)

M. Castillejo et al.

Pb (I)

188

1500

500

750

1000

(a)

(b) Anhydrite

Gypsum

Gypsum

Carbon

Carbon

Fig. 22.2. LIB and Raman spectra of (a) a white-painted area and (c) with good degree of conservation; (b) and (d) degraded

(c)

1250

1500

1750

Wavenumber / cm−1

Fig. 22.3. FT-Raman spectra of a grey overpainting with (a) strong carbon bands from contamination; (b) the original white paint showing gypsum band; (c) lasertreated overpainting revealing the original gypsum and a strong anhydrite band

overpainting. Whilst removal of the dirt layer from the surface proceeds without apparent discoloration or damage to the materials, the spectra taken after laser treatment reveal the presence of an additional strong anhydrite (CaSO4 ) band. The appearance of anhydrite indicates a partial dehydration of gypsum as a result of local heating induced by laser irradiation at 1,064 nm [5]. Due to the reversibility of this transformation in presence of water, the need for continuously monitoring the state of the restored surfaces is stressed. Darkening was observed when irradiating polychrome areas painted with vermillion at 1,064 nm; this eﬀect was absent upon irradiation of red iron oxide painted areas.

22 Laser Cleaning of Polychromed Architectonic Elements and Sculptures

189

22.3.2 Appliqué Relief Brocade Panel Laser cleaning tests showed discoloration and loss of adhesion between layers upon irradiation with the free running Nd:YAG laser. Long pulses delivered with this laser system induced a strong thermal eﬀect by virtue of the high thermal conductivity of the metallic layers of the substrate. Good cleaning results were observed with the Q-switched system (fundamental and up to third harmonic) with most satisfactory results, in terms of ﬁnal surface appearance, obtained by operating at 1,064 nm and ﬂuences around 0.16 J cm−2 .

22.4 Conclusions LIB and Raman spectroscopies allowed identiﬁcation of materials and chemical modiﬁcations (i.e. gypsum to anhydrite). Preliminary study in the laboratory led to the identiﬁcation of optimum laser parameters for the restoration of polychrome surfaces: mural paintings on mortar (Church of Cervera de la Cañada) and appliqué relief brocades on wooden sculptures (Chapel of San Miguel, Cathedral of Jaca). Restoration of most parts of polychromes was ﬁnally carried out with a Q-switched Nd:YAG laser at 1,064 nm, although according with cleaning tests performed on samples of mural paintings, great care was taken to avoid laser cleaning of the polychrome areas containing vermillion. For brocades, laser cleaning was the only possible cleaning method, as due to the porosity and fragility of the substrate the use of more conventional methods could result in damage of the artwork. Acknowledgements Funding by Projects MCYT BQU2003-08531-C02-01 and 06/0134/2003 (Comunidad de Madrid ), Spain. MO and ER thank CSIC I3P program for contract and fellowship, respectively.

References 1. M. Cooper Ed. (1998), Lasers in Conservation: an Introduction. Oxford 1998. 2. C. Rodríguez-Navarro, K. Elert, E. Sebastián, R.M. Esbert, C. María Grossi, A. Rojo, F.J. Alonso, M. Montoto, and J. Ordaz, Rev. in Conser. 4, 65M (2003). 3. A. Athanassiou, A.E. Hill, T. Fourrier, L. Burgio, and R.J.H. Clark, J. Cultural Heritage 1, 209 (2000). 4. M. Castillejo, M. Martín, M. Oujja, D. Silva, R. Torres, V. Zaﬁropulos, O.F. van den Brink, R.M.A. Heeren, R. Teule, and A. Silva, Analyt. Chem. 74, 4662 (2002). 5. M. Oujja, E. Rebollar, M. Castillejo, C. Domingo, C. Cirujano, and F. Guerra– Librero, J. Cultural Heritage 6, 321 (2005).

23 Integration of Laser Ablation Techniques for Cleaning the Wall Paintings of the Sagrestia Vecchia and Cappella del Manto in Santa Maria della Scala, Siena ∗

S. Siano1 , A. Brunetto2 , A. Mencaglia1 , G. Guasparri3 , A. Scala3 , F. Droghini3 , and A. Bagnoli4 1 ∗ 2 3 4

Istituto di Fisica Applicata – CNR, Sesto Fiorentino, Italy [email protected] Restoration Firm, Vicenza, Italy Dipartimento di Scienze Ambientali, Università degli Studi di Siena, Italy Soprintendenza per il Patrimonio Storico Artistico di Siena, Siena, Italy

Summary. We report the results of a successful experimentation of the laser cleaning technique on wall paintings using intermediate pulse duration Nd:YAG lasers at the fundamental wavelength (1,064 nm).

23.1 Introduction Painted surfaces represent the main challenge for the laser cleaning techniques. Despite several works were reported in the last decade, including systematic investigations on the characterisation of laser interaction eﬀects induced on pigments, binders, and varnishes [1–3], as well as several case studies [3, 4, 8], the laser approach is still far from a wide acceptance in the conservation community. Various laser systems were tested, such as excimer lasers (308, 248, 193 nm) [1–3, 7, 8], Free running (FR) Er:YAG laser (2.94 µm) [3, 9], and Q-Switching (QS) Nd:YAG laser at the fundamental (1,064 nm) and higher order harmonics (532, 355, 266 nm) [4–6, 9–12]. The use of this latter was mostly investigated in laboratory tests aimed at assessing the stability of pigments [6,9,10] and paint layers under irradiation conditions [12,13], a part from a few reported case studies of wall paintings [4]. Despite the laser–material interaction mechanisms associated with UV radiation emitted by excimer (or high order harmonics of QS-Nd:YAG) lasers are very diﬀerent from the ones of the speciﬁc near-IR wavelength emitted by Er:YAG laser, their use in the cleaning of painted surfaces has a common feature. In both of the cases the basic idea is to exploit the very short optical penetration depth (∼1 µm), which is signiﬁcantly smaller than the thickness of the layer to be usually removed (aged varnishes, deposits, over-painting, etc.).

192

S. Siano et al.

As evidenced by various authors in thorough parametrisation studies, the direct irradiation of the paint layer by UV radiation is armful because of the possible discoloration and other side eﬀects occurring at the operative ﬂuences of the laser cleaning [2, 7, 8]. In speciﬁc case studies reported up to now, the laser treatment was used only for partial removal of aged varnishes, thus avoiding the direct irradiation of the paint layer, or to remove incoherent deposits by relatively low ﬂuences [8]. A careful parametrisation of the ablation phenomenology on paint layer standards and artworks associated with the use of FR Er:YAG laser is not available yet. Anyway, also in this case the experimentation pointed out some limitations due to thermal side eﬀects [8]. The material removal process is not thermally conﬁned because of the long pulse duration (about 250 µs [3]), whose corresponding thermal diﬀusion length is at least one order of magnitude larger than the optical penetration depth of the typical materials of interest. In principle, alternative processes to the fast phase explosion produced by excimer laser and the conduction thermal regime of the Er:YAG lasers are provided by spallation, pressure conﬁnement and slow vaporisation regimes [13], which can be produced in absorbing layers by suitable laser pulse duration at “penetrating” wavelengths, such as the ﬁrst harmonic of the Nd:YAG laser systems. This idea comes from the observation of the relative high stability of pigments and paint layer under irradiation at 1,064 nm, already reported in the literature [10, 12–15]. Moreover, it should be taken into account that the published damage thresholds were probably underestimated because of the very inhomogeneous spatial proﬁle of the QS Nd:YAG lasers used in tests. In the present contribution we demonstrate the practicability of the novel alternative approaches based on spallation and slow vaporisation regimes through the presentation of successful applications of ﬁbre-coupled long QSwitching (LQS) and short free running (SFR) Nd:YAG lasers to the conservation problems of a set of wall paintings decorating the Sagrestia Vecchia (Old Sacristy) and Cappella del Manto (Chapel of the Mantle) in Santa Maria della Scala, Siena.

23.2 Artworks The Sagrestia Vecchia and the Cappella del Manto are two painted rooms of the large complex of Santa Maria della Scala, located just in front of the Cathedral of Siena. As well known, it is one of the oldest hospital of Europe, which was instituted by canons of the cathedral about 1,000 year ago. The hospital terminated its activity during the seventies and it was afterward gradually transformed in a museum pole. The walls and vault of the Sagrestia Vecchia were painted with themes of the Old and New Testament by Lorenzo Vecchietta between 1446 and 1449. The artworks were partially rediscovered around 1937 under various layers of lime scialbaturas (whitewashes) applied in the past. One more documented

This page intentionally blank

23 Integration of Laser Ablation Techniques

193

Fig. 23.1. The Sagrestia Vecchia: the wall paintings of the central area of the composite cross vault depicting the Cristo Benedicente (in the middle), the evangelists, and prophets

restoration intervention was carried out in the eighties, which involved a further cleaning and protection by Paraloid. The Cappella del Manto, used in the past as ﬁrst aid room, is composed by three spans, which are subdivided into cross vaults: the ﬁrst two outwards were frescoed in 1370 by Cristoforo di Bindoccio and Meo di Pero with a Madonna in throne and some saints on the walls and archivolts. While the third span contains a valuable frescoed lunette painted in 1513 by Domenico Beccafumi. Also in this case the artworks were covered in the past by lime scialbaturas and varnish. One of the main problems the present restoration intervention is encountering is actually the removal of the inner lime scialbatura layer on the wall paintings of the second span, which led to experiment the laser approach (Fig. 23.1).

23.3 Conservation Problems As it was some years ago for stone cleaning, also in the present cases the possible integration of laser cleaning technique was considered only after the evaluation of chemical and mechanical techniques, which did not provide satisfactory results in some areas. For the Sagrestia Vecchia they were the ﬁgures of the vault where the residues of scialbatura, not completely removed by previous restoration

194

S. Siano et al.

Fig. 23.2. The darkest face in the left picture represents the situation of the Sagrestia Vecchia before laser cleaning, the white areas in the right picture the one of the Cappella del Manto

interventions, and the presence of yellowish-brown aged Paraloid were relatively abundant. Whereas, the most of the areas including the azurite backgrounds and the vaulting-ribs were chemically treated with binary, ternary or quaternary micellar solutions speciﬁcally devised for the present conservation problems (private communication by Luigi Dei), along with acetone and alcohol washings. In the case of the Cappella del Manto the laser ablation was tested for removing the inner scialbatura layer on the decorating frieze of the second span vault after a preliminary lightening by mechanical removal of the outer two stratiﬁcation layers. The removal of the inner scialbatura resulted indeed very harmful because of its relatively strong adhesion to the paint layers. Examples of initial appearances before laser cleaning treatments are shown in Fig. 23.2, where they are evidenced by comparison with adjacent areas cleaned by combined mechanical and chemical approaches. Painting techniques and associated conservation problems of the areas under study, as well as results of the cleaning tests were investigated by means of stratigraphic, mineralogical and chemical analyses using grazing light observation of the surfaces, thin section optical analysis, SEM–EDX, FT-IR, XRD and colorimetry (not reported here). Furthermore, the reﬂectance at 1,064 nm of the surfaces before and after laser cleaning were also measured in order to estimate the energy distribution during laser irradiation (Fig. 23.3).

23 Integration of Laser Ablation Techniques

195

Fig. 23.3. Two representative stratigraphies of the Sagrestia Vecchia (left) and Cappella del Manto (right). Left: (a) lime plaster with sandy aggregate; (b) discontinuous layer (10 µm average thickness) of red ochre by fresco; (c) thin layer (5 µm average thickness) of a red lake (likely Kermes lake) by organic binder; (d) very thin blackish layer formed by deposits; (e) lime scialbatura (10 µm average thickness) with aged Paraloid. Right: (a) lime plaster with sandy aggregate; (b) lime layer (40 µm average thickness) with coal carbon by dry application; (c) yellow ochre and black carbon layer (50 µm average thickness) by fresco on the underlying layer; and (d) lime scialbatura

23.4 Laser Cleaning Tests Two Nd:YAG laser systems were used through extensive cleaning tests. (1) Fibre-coupled SFR laser (EOS 1000, El.En. S.p.A.) with a variable pulse duration between 50 and 120 µs while increasing the pulse energy between 0.1 and 1 J per pulse. (2) Fibre-coupled LQS laser, with pulse durations of 120 ns, 100 mJ per pulse. We also carried out a few trials with an articulated arm multi-wavelength QS laser system, but the scarce homogeneity of the spot did not allow a practicable control of the ﬂuence. Several areas were selected for laser cleaning test in the vault of the Sagrestia Vecchia. Among them, the red mantle and the book of Gesù Benedicente, the face of the angel shown in Fig. 23.2 (left), the book of San Giovanni the evangelist, the vestment of a prophet. Conversely, no particular preliminary selections were performed in the Cappella del Manto since the frescos were mostly completely covered by the lime scialbatura. Self-terminated ablation processes were achieved by LQS laser on diﬀerent red–yellow hue areas in the ﬁgures of the Sagrestia Vecchia where the paint layers include ochre, calcite, and a red lake (likely Kermes lake by EDX analyses) as main pigments. The operative ﬂuences were between 0.3 and 0.6 J cm−2 , whereas apparent ablation damages (no darkening) started to occur above 1 J cm−2 . Conversely, unsatisfactory results were provided on these areas by SFR laser, whose cleaning threshold was higher (1.5–2 J cm−2 ). The main problems encountered were the incomplete removal of the scialbatura residues,

196

S. Siano et al.

Fig. 23.4. Comparative tests on the red mantle (ochre and red lake) of the Gesù Benedicente: water assisted LQS laser provided the best result to the naked eye

Fig. 23.5. Grazing light views of areas cleaned by SFR (left) and LQS (right) lasers in water assisted conditions

aggression of the organic binder red lake of the mantle, likely by overheating, and greenish appearance in some zones. Water assisted LQS laser ablation provided the best aesthetical result. Thus for example on the red mantle of the Gesù Benedicente it did not exhibit the slightly whitish appearance associated with other treatments (Fig. 23.4), which was essentially due to incomplete cleaning. The stratigraphic analyses conﬁrmed the evaluation to the naked eye. They demonstrated the presence of scialbatura and Paraloid residues in zones cleaned by SFR and QS lasers, which was also relatively aggressive because of the inhomogeneous laser spot. Figure 23.5 displays a macro-view by grazing illumination of the surfaces cleaned by SFR and LQS lasers where the incomplete cleaning by the former can be recognised, as compared with the smooth surface texture associate with the eﬀective cleaning by LQS laser. The perfect preservation of the ﬁnishing organic binder red lake of the mantle allowed by LQS laser and the corresponding possible aggressiveness of the SFR are shown in Fig. 23.6.

23 Integration of Laser Ablation Techniques

197

Fig. 23.6. Representative stratigraphies of the cleaning tests by LQS (left) and SFR (right) lasers on the red mantle of the Cristo Benedicente (apart from the diﬀerent thickness, a–c are the same as in Fig. 23.3 left). The former allowed a careful discrimination of the organic binder red lake layer, (c, 30 µm maximum thickness), which was sometime damaged or removed by SFR laser

Fig. 23.7. Face of the angel of Fig. 23.2 during the laser cleaning by LQS Nd:YAG laser

Conversely, the long pulse laser (SFR) provided the best result on the white-grey zones of the books. Actually, the QS and LQS laser produced a too deep cleaning and a slightly yellow appearance of the surface. Only the SFR laser allowed discriminating the greyish shadings of the pages and preserve the original white colour. Following these and other positive testing of the laser approach extensive cleanings and corresponding analytical validations have been carried out in the Sagrestia Vecchia. As an example, Fig. 23.7 displays the partial laser cleaning of the selected angel (Fig. 23.2).

198

S. Siano et al.

Fig. 23.8. Testing of diﬀerent laser systems on the vault of the Cappella del Manto

The laser cleaning tests carried out in the Cappella del Manto provided better results with respect to mechanical means, which did not allow a satisfactory discrimination of the paint layers. At the same time, the operative ﬂuence margins were rather narrower than the previous case. The best was achieved by a combination of the two present laser systems, which exhibited diﬀerent interaction intensity with the various paint layers of the frieze under cleaning. Thus for examples azurite and green earth paint layers were better treated by LQS whereas grey hues by SFR laser, according to the results of the Sagrestia Vecchia described above. Both of them were satisfactory on yellow ochre layers and both aggressive on red motives where the pigment was minium. The repetition of the motives of the frieze was exploited in order to select the suitable irradiation parameters during the cleaning treatments of relatively large areas (Fig. 23.8).

23.5 Brief Discussion and Conclusions In this work we demonstrated the practical applicability of the intermediate pulse duration laser sources in the cleaning of wall paintings. The analyses described through the text to support this thesis are only a small part of the thorough investigation we are carrying out on the important masterpieces of the Sagrestia Vecchia and the decoration of the Cappella del Manto in Siena. From a physical standpoint we proved the eﬀectiveness of two novel approaches to conservation problems of wall paintings based on spallation and slow vaporisation ablation regimes using LQS and SFR Nd:YAG lasers. As it can be seen in Table 23.1, in the case of Paraloid and scialbatura residues the

23 Integration of Laser Ablation Techniques

199

Table 23.1. Vault of Sagrestia Vecchia: reﬂectance measurements Area Red mantle Deep red Light red Face of the angel Rosy forehead Rosy cheek Light band in the air Chin Shadow of the nose Eyebrows Book White zone Shading Azurite backgrounds Various Vestment of a prophet Light yellow band

R before (%) 20.8 29.2

27.7

R after (%) LQS 54.3 60.1 LQS 69.7, 74.2, 68.5 53.9 54.5

37.1, 36.0

25.8, 31.3

7.4, 10.9 42.6, 42.0, 47.2,

56.2 34.8, 32.6 SFR 62.1 46.7 15.9, 12.5, 18.8 LQS 69.3, 64.8, 60.9

Fig. 23.9. Cleaning transition achieved by LQS (left) and SFR (right) lasers. Perfect selectivity in the former case and slight aggression in the latter one. Left: (a) lime plaster with sandy aggregate; (b) lime layer (50 µm average thickness); (c) pigmented layer with lead–tin yellow and green earth by fresco on the underlying layer (50 µm average thickness); (d) blackish deposit ﬁlm; (e) lime scialbatura (100 µm maximum thickness). Right: (a) lime plaster with sandy aggregate; (b) pigmented layer with lead–tin yellow (20 µm average thickness), likely by fresco; (c) organic binder azurite layer (60 µm average thickness); (d) lime scialbatura (100 µm maximum thickness)

former photomechanical regime was favoured by direct absorption of the stratiﬁcation and the relatively high reﬂectance of the paint layers, as measured by an integrating sphere. Diﬀerently, the spallation dynamics producing the removal of the coherent white scialbatura of the Cappella del Manto is mediated by water vaporisation driven by laser heating of absorbing deposits laying on the paint layer (Fig. 23.9). In both of these cases the stratiﬁed material was

200

S. Siano et al.

removed in form of ﬂakes produced by detachment and fragmentation of the stratiﬁcation within the irradiated spot. The removal of the Paraloid and scialbatura residues of the Sagrestia Vecchia by LQS laser ablation was often characterised by a single shot cleaning eﬀect. Slow vaporisation is instead the ablation regime characterising the cleaning of the white-grey areas in the Sagrestia Vecchia according to the very gradual material removal experimentally observed. These qualitative descriptions are supported by detailed physical investigations we will report elsewhere. In conclusion, this work demonstrates the Nd:YAG (1,064 nm) laser systems, which are well-established in stone and metal conservation, can be also used for solving complex problems of wall painting cleaning. Among these systems the ones with intermediate pulse durations provide a higher control and other practical advantages with respect to the typical short pulses emitted by QS Nd:YAG lasers for cleaning wall paintings. They actually allow selfterminated ablation processes, such as “soft” photomechanical spallation and slow vaporisation, without relevant non-linear optical absorption phenomena and formation of micro-plasmas, usually responsible for aggressive eﬀects of QS lasers. These regimes can allow eﬀective and safe cleaning of wall paintings. The good result is also favoured by the high beam homogeneity of the present class of laser systems. Further works will be carried out on wall paintings aimed at extending the ﬁeld of application of the ideas reported in the present paper. Acknowledgements The authors wish to thank Dr. Enrico Toti, Conservatore of Santa Maria della Scala, Institution of Siena’s Municipality, for having entrusted them with the present conservation problems, Dr. Giorgio Bonsanti for having stimulated the participation to LACONA VI, and the restorer Massimo Gavazzi for the helpful discussions and logistic support.

References 1. I. Zergiotti, A. Petralis, V Zaﬁropulos, C. Fotakis, A. Fostiridou, and M. Doulgeridis, in: Proceedings of LACONA I (Crete 1995), Restauratorenblatter, 57, 1997. 2. R. Salimbeni, P. Mazzinghi, R. Pini, S. Siano, M. Vannini, M. Matteini, and A. Aldrovandi, in Proceedings of the 1st International Congress on: Science and Technology for the Safeguard of Cultural Heritage in the Mediterranean Basin, A. Guarino ed., Consiglio Nazionale delle Ricerche, Roma, 1998, 811. 3. A. de Cruz, M. L. Wolbarsht, and S. A. Hauger, in Journal of Cultural Heritage, Vol. 1, 173, 2000. 4. M. C. Gaetani and U. Santamaria, in Journal of Cultural Heritage, Vol. 1, 199, 2000. 5. A. Sansonetti and M. Realini, in Journal of Cultural Heritage, Vol. 1, 189, 2000.

23 Integration of Laser Ablation Techniques

201

6. P. Pouli and D. C. Emmony, in Journal of Cultural Heritage, Vol. 4, 181, 2000. 7. A. Athanassiou, A.E. Hill, T. Fourrier, L. Burgio, and R. J. H. Clark, in Journal of Cultural Heritage, Vol. 1, 209, 2000. 8. R. Teule, H. Scholten, O. F. van den Brink, R. M.A. Heeren, V. Zaﬁropulos, R. Hesterman, M. Castillejo, M. Martin, U. Ullenius, I. Larsson, F. GuerraLibrero, A. Silva, H. Gouveia, and M.B. Albuquerque, in Journal of Cultural Heritage, Vol. 4, 209, 2003. 9. P. Bracco, G. Lanterna, M. Matteini, K. Nakahara, O. Sartiani, A. De Cruz, M. Wolbarsht, E. Adankiewicz, and M. P. Colombini, in Journal of Cultural Heritage, Vol. 4, 202, 2003. 10. M. Chappé, J. Hildenhagen, K. Dickmann, and M. Bredol, in Journal of Cultural Heritage, Vol. 4, 264, 2003. 11. P. Pouli, D.C. Emmony, C.E. Madden, and I. Sutherland, in Journal of Cultural Heritage, Vol. 4, 271, 2003. 12. J. Hildenhagen, M. Chappé, and K. Dickmann, in: LACONA V Proceedings, Springer-Verlag, Berlin, 297, 2005. 13. A. Schnell, L. Goretzki, and Ch. Caps, in: LACONA V Proceedings, SpringerVerlag, Berlin, 291, 2005. 14. R. Torres, M. Jadraque, M. Castillejo, and M. Martin, in: LACONA V Proceedings, Springer-Verlag, Berlin, 285, 2005. 15. S. Siano, R. Pini, and R. Salimbeni, in ALT 99, International Conference on Advanced Laser Technologies, Edited by Pustovoy and V.I. Konov, SPIE Vol. 4070, 27, WA USA, 2000.

24 Preliminary Results of the Er:YAG Laser Cleaning of Mural Paintings ∗

A. Andreotti1 , M.P. Colombini1 , A. Felici2 , A. deCruz3 , G. Lanterna2 , M. Lanfranchi2 , K. Nakahara2 , and F. Penaglia1 1

∗ 2

3

Dipartimento di Chimica e Chimica Industriale, Università di Pisa, via Risorgimento 35, 56126 Pisa, Italy [email protected] Opiﬁcio delle Pietre Dure, Fortezza da Basso, Viale Filippo Strozzi 1, Firenze, Italy Duke University, Department of Chemistry, Durham, NC 27708 USA

Summary. The conservation of mural paintings requires a deep knowledge of the alterations caused by natural ageing, environmental agents and previous restoration treatments. All the operations concerning cleaning and consolidation of wall paintings must assure the safety of the paint layers. This is especially true the more fragile the painting technique. For example, “a secco” paintings, executed with organic binders such as tempera, oil, glue, when altered and damaged, present a very weak adhesion to the mortar underneath, and provoke detachment of paint fragments. In this circumstance it is necessary to ﬁnd a feasible alternative to the usual cleaning methods (wet and mechanical ones) and a valid way to operate. Moreover, the removal of scialbo layers (a thick, pure lime layer applied on the wall painting) presents diﬃculties in order to preserve the integrity of the painting layers. Previous experiments carried out in Opiﬁcio with Er:YAG laser on easel painting cleaning, lead us to extend the experiments on the cleaning of mural paintings. This program is intended to establish parameters, such as emitted wavelength, ﬂuency and associated energy, applied on mural paintings. The ﬁrst step was to verify the upper limits of energy for a large set of laboratory samples, prepared following the traditional techniques of mural paintings. Then experiments were carried out on the following mural paintings: a Giotto school fresco, located in the Bargello Palace in Florence, which presents areas with scialbo and secco painting, and a sixteenth century mural painting in the Boiardo’s fortress in Scandiano (RE). The analyses were performed by using optical microscopy, SEM, FTIR, and GC-MS techniques. This paper reports and documents the most signiﬁcant results obtained during the experiments.

24.1 Introduction The cleaning of mural paintings includes the removal of many types of materials. The surfaces can be aﬀected by various damage mechanisms, including altered retouches and old organic ﬁxatives, biological infestations and covered

204

A. Andreotti et al.

by lime-made “scialbature” (limewash). Er:YAG laser irradiation experiments followed those of previous studies carried out in Opiﬁcio delle Pietre Dure of Florence on panel and canvas paintings [1]. The methodology applied demonstrates the gradual and controlled ablation of organic materials from the painted surfaces, as reported in the literature [2]. The Er:YAG laser emits pulses at 2.94 µm, in the mid-IR spectral region where the absorption of energy is maximum for the O–H bond; thus the surfaces dampened with an aqueous liquid or a solvent containing hydroxyl group, eﬀectively increases the yield of old encrustation removal without inducing unwanted chemical or physical changes on the original painted surfaces of wall paintings. Also the wall painting techniques are important for the evaluation of the laser eﬃciency and safeguard. There are three main mural painting techniques: “buon fresco,” when wet pigment is applied on the ﬁne fresh mortar surface. After the mortar sets, the pigment is strongly included on the wall; “limewash painting,” when the pigment is dispersed in limewash and applied on a dried mortar. After this layer sets, a painted lime layer is superimposed on the mortar; “a secco,” when the pigment is mixed with a medium and applied on a prepared dry wall by mean of organic primers. Depending on the painting technique, there are diﬀerent degrees of adhesion and cohesion between each layer. These experiments point out the cases where Er:YAG laser ablation is eﬃcient for various surface materials removal and positively supported by the inner layers. The aim is to provide restorers with a comprehensive knowledge on laser eﬃciency, ﬂuence, threshold limits on each pigment type, possible interactions between binders and pigments and the action of auxiliary liquids. The study of the surfaces, before and after removal of the top layers by Er:YAG laser, was performed by optical microscopy, SEM, FT-IR and GCMS techniques. The combined use of these techniques provides information on the morphology and chemical composition of surfaces exposed to Er:YAG radiation.

24.2 Experimental Methods The equipment used is a portable unit (Light Scalpel ) manufactured by Mona Laser, Inc. (USA) and was used both in the laboratory and on the scaﬀolds during the restorations. The laser pulses were delivered on the surface by the shield of (and through) a microscope slide (cover-slip); ablated materials were collected according to procedure previously published [3]. The ﬁxed spot size was 1 mm. The GC-MS method is based on published procedures [3] for the analysis of amino acids, fatty acids, and terpenoid compounds. Optical microscopy (OM) with VIS and UV sources (Zeiss) were used to observe the painted specimen surfaces and their modiﬁcation after the laser pulses. SEM (Leica Cambridge) and EDS (Link-Oxford) were used to study the complex morphology

This page intentionally blank

24 Preliminary Results of the Er:YAG Laser Cleaning

205

changes of the surfaces, both in ﬂat samples and in cross sections. FTIR (Thermo-Quest) was used to compare the materials left on the surfaces and the ablated materials collected on the glass cover-slip. 24.2.1 Experiments on Laboratory Models Experiments of Er:YAG laser cleaning tests have been executed initially on the laboratory wall painting models prepared in 1999, which simulate “buon fresco” and “a secco” technique. The aim is to study the response of the paint layers behavior and any color changes in laboratory models of each technique, when exposed to increased laser energy and determine the maximum acceptable threshold. The selection of pigments was based on their frequent use and also sensitivity to Er:YAG laser, that is, high content of –OH bonds or its crystal transition phase. Pigments tested belong to iron silicates (green earth, burnt sienna, yellow ochre), metal oxides (red lead, manganese black), basic carbonates (malachite, lead white), cobalt glass (smalt), carbon, and mercury sulphide (vermilion). All these pigments were applied in “buon fresco” and “a secco” techniques, following the above described. For “a secco” models, the binder was whole egg. Figure 24.1 shows the paintings sample tile. The ablation procedures followed previous experiments on easel paintings [4]; all the laser tests have been executed under the control of stereo microscope

Fig. 24.1. Mural paintings sample tile. The pair of pigments was applied with (left) egg “buon fresco” and (right) tempera “a secco” technique. They are (from top-left to lower-right): lime white, yellow ochre, burnt sienna, green earth; umber, smalt, vine black, manganese black; lead white, minium, vermillion, malachite, azurite

206

A. Andreotti et al.

by mural painting conservators and laser testing was executed by application of progressively increasing energy: ﬁrst, using dry methods and, secondly, with wet methods. The range of energy used was approximately 3 up to 50 mJ with increments of about 3 mJ. The auxiliary wetting liquids were applied with small cotton swab. The hydroxyl liquids applied were distilled water, ethanol, water–ethanol mixture (1:1 v/v) and hydrocarbon mixtures mixed with O–H containing substances (circa 12% ethylene glycols). Laser cleaning tests were designed to study removal of various scialbo layers from lab models. Three diﬀerent types of scialbo layers were prepared: pure lime, lime with milk, lime with olive oil. These three scialbo layers were applied on secco paint models of diﬀerent colors in the medium of tempera (English red, blue smalt, malachite green, and gold leaf adhered by vinylic adhesive). After preliminary laser tests, an energy of about 45 mJ on the surface dampened with distilled water/ethanol (1:1) was most eﬃcient to weaken the scialbo and facilitate the subsequent removal with mechanical methods (scalpel, brushes), (Fig. 24.2). It was found that a pretreatment with a poultice of distilled water/ethanol (1:1) for a few minutes before laser ablation permitted penetration of auxiliary liquid through the thick scialbo layers, allowing for more eﬀective laser

Fig. 24.2. Scialbo removal from an a secco painting (laboratory tests)

24 Preliminary Results of the Er:YAG Laser Cleaning

207

ablation. Because the resistance of the secco substrate is often weaker than scialbo one, it may be necessary to apply several passes of laser pulses at lower energy. 24.2.2 Experiments on Wall Painting After the preliminary experiments on laboratory paint models, the laser tests proceeded on two original wall paintings which had an extremely diﬃcult problem in the removal of scialbo layers or degraded materials. This condition threatened the original substrates. The ﬁrst case is a Giotto school fresco painting in the Magdalen Chapel located in the Bargello Palace in Florence, where there are areas with secco paint which were covered by scialbo layers, removed partially in 1840. Another example is a sixteenth century secco painting in the “camerino del Paradiso,” located in the Boiardo fortress in Scandiano (RE). These wall paintings are covered by various layers of scialbo applied during diﬀerent periods and, at present, it remains completely on the surface; in this case the original substrate has a very weak adhesion to the mortar. The Er:YAG laser tests, dry and with liquids, did not show any evident improvement, but a slight thinning of scialbo was observed applying 10–30 mJ, wetting with the white spirit/glycol. The substrate is unusually fragile, and any mechanical intervention must be avoided. For this reason a preliminary consolidation is necessary to strengthen both scialbo and paint layer, which may facilitate the safe removal of the scialbo. At this time further experiments will be carried out to resolve this problem.

24.3 Results and Discussion 24.3.1 Evaluation of Experiments on Laboratory Samples The results on the laboratory models are representative of those colors which are present in real case history. They are summarized as follows: Green earth. For the secco painting, it was observed that, in dry method, it was mechanically resistant up to a certain energy (12 mJ), but provoked chromatic alteration according to an energy increase; with auxiliary liquids, the resistance increased by only a few energy. In the fresco sample, chromatic alteration is not observed but there arises a mechanical damage of the surface planarity; wetting lowers the threshold (Fig. 24.3). Burnt sienna. In fresco, the wetting lowered the threshold (about 12 mJ). All the wetting agents provoked the same eﬀect. Yellow ochre. In fresco, 6 mJ provokes a partial glossy aspect of the surface; the overcoming of threshold energy induced a red-brown color change. The wetting agents avoid the color change, but not

208

A. Andreotti et al.

Fig. 24.3. Sample mural painting tile. Green earth painted surface detail with laser treated areas seen as small squares: (center ) fresco painting, (right) secco painting

the glossy aspect. In secco painting, over 12 mJ energy begins the microscales detachment; wetting with liquids enhanced that phenomenon. Minium. In secco, wetting with liquids increased the threshold until 12 mJ. Vermillion In secco, wetting with liquids increased the threshold until 12 mJ. Water seems to be the better liquid for treatment. Malachite. In secco, wetting with liquids increased the threshold until 6 mJ. Water seems to be the better liquid for treatment. Smalt. In secco, technique presents an exceptionally high threshold (>40 mJ). The test with liquids is lacking. Manganese black. In secco, it had an exceptionally high threshold (>40 mJ). The test with liquids is lacking. Lead white. In secco, wetting with liquids increased the threshold (21 mJ). Pure, distilled water must be avoided because it provoked a slight yellowing and mechanical damage. The energy threshold for buon fresco and secco painting is, on average, 12 and 18 mJ, respectively, using dry methods. For the malachite-secco layer, the threshold is just 6 mJ in dry method (this, however, might be due to the bad condition of malachite-model compared to other models) and for yellow ochre buon fresco-layer which resists only to 3 mJ, at which point the reaction

24 Preliminary Results of the Er:YAG Laser Cleaning

209

provoked a slight lucidity. It was noticed that some colors such as smalt (threshold over 40 mJ) and manganese black (threshold over 30 mJ) are more resistant to laser ablation. It was also observed that energy over threshold can provoke some negative eﬀects such as blanching of surface, chromatic changes, micro- and macroﬂaking, up to carbonization of organic materials. It was also observed that when –OH bearing liquids are suspended on the paint surface, the thresholds increase by 3–5 mJ. However, in a laboratory sample (burnt sienna, buon fresco and secco) when the –OH bearing liquids have saturated the paint layer, the threshold decreased signiﬁcantly. 24.3.2 Evaluation of Experiments on Site The mural paintings tested with Er:YAG laser demonstrated a better resistance than the laboratory model. Generally the energy supported is 20% greater compared to laboratory models. Surely, this is due to the better adherence of the wall painting to the mortar. This is true with exception of scialbo, which is more diﬃcult to remove when aged. The Er:YAG laser action works better if applied on a moistened surface. The eﬀectiveness is best for water, then for water/ethanol 1:1 (v/v) and ﬁnally for the dilute glycol solution in Ligroine. The mechanism of ablation of aged scialbo seems to be due to a phase transition of liquid present near the surface. The transition provoked by laser pulse shatters the scialbo lattice forming a dusty material that ﬂies away from the irradiated area. For the Scandiano painting, a mixture of Ligroine was used as wetting agent because of the fragility of the paint layer below; its greater volatility avoids a deep penetration of ﬂuid which was maintained near the surface, thus the removal is limited in depth, gradual and progressive. Several laser passes are necessary to thin the scialbo layer, but after the laser action also the delicate action of a scalpel on the surface also become more eﬀective, probably because of the weakening of the scialbo layer (Fig. 24.4). Test carried out also on surfaces aﬀected by biodeterioration (algae, moulds, fungi) gave remarkable results.

24.4 Conclusions Laser cleaning is an eﬀective procedure when helped by wetting the surface. On scialbo samples and original layers, it is easier to achieve a removal combining a previous thinning with mechanical tools, followed by several passes of Er:YAG laser energy. With fragile substrates, where mechanical intervention must be avoided, a preliminary consolidation is necessary to strengthen both scialbo and paint layer. On this hardened surface, it may be possible to thin the residual scialbo without endangering the original paint layer. At this time further experiments will be carried out to thin the residual scialbo layer after a preconsolidation with polymers.

210

A. Andreotti et al.

Fig. 24.4. Scialbo removal by laser ablation. Giotto school fresco painting (Bargello Palace, Florence)

Acknowledgments Ente Cassa di Risparmio of Florence is gratefully acknowledged for the large contribution given for this research. Thank to the Municipality of Scandiano (RE), the Soprintendenza per Beni Artistici, Storici ed Etnoantropologici di Modena e Reggio Emilia, the Bargello Museum Management, the Management of Polo Museale of Florence for the agreement on Er:YAG laser experiments.

References 1. E. Adamkiewicz, P. Bracco, M. P. Colombini, A. De Cruz, G. Lanterna, M. Matteini, K. Nakahara, O. Sartiani, and M. L. Wolbarsht in Journal of Cultural Heritage, Vol. 4, 202, 2003. 2. A. De Cruz, S. Hauger, and M. L. Wolbarsht, in Opt. Photon. News Vol. 10, 36, 1999. 3. A. Andreotti, M. P. Colombini, G. Lanterna, and M. Rizzi. in Journal of Cultural Heritage, Vol. 4, 355, 2003. 4. P. Bracco, G. Lanterna, M. Matteini, K. Nakahara, and O. Sartiani, in OPD restauro, Vol. 13, 192, 2001.

Part IV

Organic Materials

25 Preliminary Results of the Er:YAG Laser Cleaning of Textiles, Paper and Parchment ∗

A. Andreotti1 , M.P. Colombini1 , S. Conti2 , A. deCruz3 , G. Lanterna2 , L. Nussio2 , K. Nakahara2 , and F. Penaglia1 1

2

∗ 3

Dipartimento di Chimica e Chimica Industriale, Università di Pisa, via Risorgimento 35, 56126 Pisa, Italy Opiﬁcio delle Pietre Dure, Fortezza da Basso, Viale Filippo Strozzi 1, Firenze, Italy [email protected] Duke University, Department of Chemistry, Durham, NC 27708 USA

Summary. The main constituents of textiles, paper and parchment are proteins and polysaccharides. These organic matters are particularly sensitive to damage such as spotting, dimensional deformations, depolymerisation, and oﬀer a surface particularly suitable towards the deposition of various materials. In the case of the paper and parchment cleaning, traditional techniques are well known, as well as the risk to make halos and shades when using the mechanical action of a scalpel or solvent mixtures to thin the spots. For textile artefacts the need to remove dust, spots, and mud residues is a duty though this action is quite diﬃcult with traditional methods because the dense weft and warp structure does not allow a complete cleaning, without a minor damage of the artwork. The authors set up a research program to verify the possibility and the results of an Er:YAG laser equipment, emitting at 2,940 nm, following the results achieved previously in OPD on the treatment of organic materials (LACONA IV proceedings). The wavelength and the uniform ﬂuency distribution were chosen in order to respect the delicate organic material. The ﬁrst step for both paper and textiles was to determine the threshold of safety for the constituting materials, afterwards was tested the eﬃciency of the simple laser pulse or the pulse absorbed by wetting agents (such as water, alcohols, and other organic solvent mixtures). Particularly, the study wanted to determine the cleaning eﬃciency, controlling the laser-induced changes in the morphology, and chemical composition of the laser-exposed surfaces. The study of these eﬀects on the surfaces, before and after the Er:YAG laser treatment, was performed by using optical microscopy, SEM, FT-IR and µ-FTIR, and GC–MS techniques. The results obtained show that the materials can absorb a small amount of energy which is suﬃcient to remove dust and spots without change or damage. For textiles, due to the complex structures of ﬁbres and yarns, a multi-step laser cleaning which exposed the surface several times to a very low energy gave the best results without any damage.

This page intentionally blank

214

A. Andreotti et al.

25.1 Introduction Experiments on textiles were carried out to verify both the maintenance of the superﬁcial morphology and the chemical/physical characteristic of ﬁbres, yarns, dyes, and mordants. The restoration tools may change the structure and the properties of the textile surface, so that reﬂectance decreases or the colour fades. In fact pulse energy could thermally transform the surface of the ﬁbres and yarns producing a kind of matted eﬀect, and this must be avoided. For this reason, before proceeding with further evaluations, the study of the threshold of various ﬁbre types to the Er:YAG laser pulse was carried out for in-depth examinations. The main components of textiles, parchment, and paper are proteins and polysaccharides. These organic matters are particularly sensitive to damage such as spotting, dimensional deformations, depolymerisation, and oﬀer a large surface particularly sensitive to deposits of various materials. Especially textile materials, having a weft and warp structure, are very receptive to stains, dust, dirt, etc. Particular care must be given to the micro-aspect of the surface: pressed ﬁbres (paper), thinned leather (parchment), and fabric (textiles) have a particular surface morphology and/or texture that provides characteristic optical and chromatic eﬀects in terms of reﬂectance to their decorations (such as patterns, drawings, gilding, and prints). The loss or modiﬁcation of these features can provoke unwanted optical diﬀerences between adjacent areas. This is one of the most important secondary eﬀects to be avoided in the restoration of paper, parchment, and textiles.

25.2 Experimental Methods The materials chosen for exposure to the Er:YAG laser pulse were standard recent materials, aged materials, and some fragments from sixteenth and seventeenth century. Tests on paper were studied for various cellulose types, texture and colour of paper (following old recipes), and also paper from books and engravings (around 70–100 years old). Residual adhesive tapes, stains, and spots of organic origin were also tested. New parchment scraps and old fragments with yellowing spots and ink traces were studied. Textiles tested were recent standard yarns rolled up in small, dense skeins; pieces of aged cloth in raw wool, pure wool, and silk, with organic dyes in the form of mordant lacquer, were also tested. As an original case study, an old silk furnishing heavily aﬀected by 1966 Florence ﬂood damage was tested for cleaning. These tests were carried out with a portable Er:YAG laser (light scalpel) manufactured by Mona Laser, Inc. (USA). The operating parameters were:

25 Preliminary Results of the Er:YAG Laser Cleaning Emission wavelength Pulse rate Energy range pulse Fluence range Delivery system

215

2.94 µm (MIR region) 5 and 10 Hz 2–30 mJ 100–1, 500 mJ cm−2 a hollow glass ﬁbre, 1/16 in. i.d., silver coated

a

For purposes of this article, to convert energy to ﬂuency it is necessary to multiply the energy per pulse by 50 cm−2

The laser pulse was delivered to the surface through a shield of a microscope coverslip; ablated materials were collected according to previously published procedure [1]. The auxiliary wetting liquids, applied with small cotton swab, were –OH containing substances (distilled water, ethanol, water–ethanol mixture (1:1 v/v), distilled water with 1% surfactant [Tween 20 ]); non-OH containing liquid (light aliphatic hydrocarbons [Ligroine]); non-OH containing liquid added with –OH containing substances [White Spirit with 15 % di-ethyleneglycol]). For the textile materials only a low-pressure steam generator was used to gently moisten the ﬁbres. GC–MS method was based on published procedures [2] for the analysis of amino acids, fatty acids, and terpenoid compounds. OM (Zeiss), with halogen and UV sources, was used to observe the painted specimen surfaces and their modiﬁcation after the laser pulses. SEM (Leica Cambridge) and EDS (LinkOxford) were used to study the complex morphology changes of the surfaces both in ﬂat samples and in cross sections. FTIR (Thermo-Quest) was used to compare the materials left on the surfaces and the ablated materials collect on glass coverslip. 25.2.1 Experiments on Textiles Experiments were carried out on raw materials such as yarns to test the behaviour of the micro-ﬁbres to the laser eﬀects. For this purpose small skeins of Japanese puriﬁed silk, Tussah raw silk, and British wool were prepared. These three kinds of yarns are characterised by a very ﬁne structure. All the tests were executed both in dry mode and in wet mode. The aim of adding water or other hydroxylated liquids is to increase the Er:YAG laser pulse eﬃciency on the surface. But in the case of textiles the manufacture and nature of natural ﬁbres oﬀer a great chemical and structural aﬃnity with water (hygroscopicity, capillarity), and that could result as an excessive amount of water into the textile, with uncontrolled eﬀects on the laser absorption. To avoid that situation, a low-pressure steam generator was employed to gently moisten the surface. In this way the Er:YAG laser action was limited to the surface. Moreover a low pulse frequency (5–10 Hz) was used because this was considered more controllable by restorer.

216

A. Andreotti et al.

Fig. 25.1. Fibre resistance test to the laser energy carried out on three yarn samples: (on top, clockwise): Japanese silk, Tussah silk, British wool. Each sample was pulsed from left to right and from top to bottom, with increased energy (3 mJ step) until the evident damaging, both in dry (D) and wet (W) method. Some thermal modiﬁcation arise, but it is evident that the resistance increases with wet method

The laser energy was applied in increasing succession on each skein: ﬁrst ﬁbres were irradiated in dry mode with 3 mJ and followed with increased energy of 2 mJ per step. Then the area of each skein was moistened with steam. At over 7–9 mJ in dry method, a superﬁcial modiﬁcation was observed, due to change of surface reﬂection in the silk, and increasing thermal eﬀect for both silk and wool. The results show that the threshold energy rose around 40% with introduction of steam moisture (Fig. 25.1). Because the energy employed was below the threshold, the dyed fabric samples were not aﬀected by any signiﬁcant fading or colour changes. The case study reported here concerns the cleaning of a soft ﬁnish red silk heavily damaged by a thick deposition of mud. This sample belongs to a huge series of artworks damaged by the 1966 ﬂood in Florence. This kind of dirt is really a complex mix of slime, oils, solublem, and insoluble organic materials. It is very hard to remove with mechanical tools because of the scarce residual tensile strength of fabric, and hard to remove chemically because

25 Preliminary Results of the Er:YAG Laser Cleaning

217

Fig. 25.2. Er:YAG laser mud removal from a 1966 damaged silk cloth. The detail shows in the right lower corner the laser cleaned fabric (steam moisted pre-treated surface, 5 Hz pulse frequency, 6 mJ pulse energy)

of the complexity of materials present on the surface. A washing procedure usually adopted for textiles did not reach an acceptable cleaning result. Steam pre-moistening of the surface followed by repeated laser passes around 5 mJ achieved the aim of complete removal of thick dirt in a safe way without detectable changes to the ﬁbres surface (Fig. 25.2). 25.2.2 Experiment with Paper and Parchment Also for paper and parchment it is necessary to consider the extreme sensitivity of the surfaces to moisture. The synergy between Er:YAG laser pulse and the –OH containing liquids, in this case, is opposed to the hygroscopic nature of polysaccharides and polypeptides. The tests were conducted both in dry and wet methods to compare the diﬀerent surface behaviour. The appearance of the tested surface shows the threshold energy limits for each type of white paper and for the parchment are around 20 mJ pulse energy. Coloured papers have a lower resistance due to the transformation of dye or pigment (indigo, ochre), normally 10 mJ is the threshold (Fig. 25.3). An interesting behaviour was found in cleaning ancient paper with wet method. An appreciable swelling is visible. The eﬀect increases with higher pulse energy. In this case the strong hygroscopic nature of paper absorbs and

218

A. Andreotti et al.

Fig. 25.3. Energy tests on diﬀerent type of paper and parchment (wet method) carried out increasing the energy; a 13 × 13 mm mask was used to delimitate the areas. The colour change indicate the threshold energy: The working parameters must be maintained within each threshold

keeps hydroxyl liquid in pores. The laser pulse absorption provokes an immediate phase change, with a swelling eﬀect on the surface. This phenomenon can probably be corrected with standard paper restoration techniques (stretching, leveling). Observation with both microscope and SEM shows a modiﬁcation on the micro-superﬁcial structure; in all paper types, even if pulsed with low energies, a slight lightening is visible in the areas treated with laser (Fig. 25.4). In UV light a soft yellowish ﬂuorescence is perceptible, due to the photooxidation of –OH functional group of the glucose units. Increasing the energy causes the surface morphology to change with a series of micro-bores, probably due to the thermal eﬀect. In wet method, the surface is more compact and the bores less deep and less frequent. The analysis of the proteinaceous material was performed by means of the GC–MS and the amino acids percentage content of the paper samples were compared to a database of reference samples (egg, animal glue, and casein) using the principal component analysis (PCA; the resulting loading plot and score plot are reported in Fig. 25.5b, c, respectively). The brittle stain on the ancient paper was mainly constituted by animal glue (the chromatogramm of sample 4.1.A and the amino acids percentage contents of the two sample are reported in Fig. 25.5a), as it can be ascertained on the basis of an high content of glycine, proline and the occurrence of hydroxyproline, together with the results of the PCA (Fig. 25.5c). The high content of glutammic acid of sample 4.1.B, is responsible of the shift of the second component in the

25 Preliminary Results of the Er:YAG Laser Cleaning

219

Fig. 25.4. SEM morphology study comparison of pulsed and untreated paper. Above: untreated paper (left ×100 right ×500 magniﬁcation); below : 15 mJ dry pulsed (left ×100, right ×500); SEM detector secondary electrons, acceleration 15 kV, samples were gold sputtered TIC:160104_3.D

Abundance 1.4e+07

4.1. Ancient paper “vergellata” with glue stains

1.2e+07 1e+07 8000000 6000000 4000000 2000000 Time--> 0

a) 15.00

20.00

ala gly val 4.1.A 10.7 24.8 2.8 4.1.B 6.3 14.5 2.7

leu 4.0 3.6

25.00

ile 1.9 1.8

30.00

met 0.0 0.0

ser 3.4 5.3

35.00

40.00

pro phe asp glu lys 5.1 1.2 12.6 23.0 4.5 11.6 2.6 14.1 32.8 0.7

hyp tyr 6.1 0.0 3.9 0.0

b)

c)

3 2

colla −10

−8

uovo database

1

−6

−4

−2

0 0

2

4

4.1.A

−1 −2 −3

casein a

4.1.B

−4

Fig. 25.5. Chromatogram of sample 4.1.A and amino acids percentage content of two samples from a stain on the ancient paper (a); and principal component analysis loading plot (b) and score plot (c)

220

A. Andreotti et al.

score plot towards the casein cluster. The laser ablation of the material from the paper caused a decrease of the proline content. The amino acids proﬁle of parchment was similar to that of animal glue, but in a few samples after laser ablation, a decrease of glutamic acid, and proline was observed. Materials deposited on the paper surface are easier to ablate than materials which penetrated into the structure. This is not due to the laser energy absorption, but to the separation between laser spots and inner paper structure. When the materials impregnate the ﬁbres it is more diﬃcult to control the laser action avoiding secondary eﬀects on the substrate. Future experimentation will take into account the Er:YAG laser combination with traditional restoration procedures to verify if many of these eﬀects can be reduced.

25.3 Conclusions The cleaning of paper is more diﬃcult than cleaning of other materials because of the very broad parameters encountered: Diﬀerent stains on the same substrate can achieve an excellent result or a bad one. The same stain on a diﬀerent substrate could in one case be diﬃcult while, on another, an easy and complete stain removal is possible. The thin layer of the superﬁcial materials, the strong adhesion among these materials, the ﬁbrous structure of substrates, the wide variety of paper, and their state of conservation can be complex with many variables which complicate the approach to the problems with use of the Er:YAG laser and require further studies and applications. The laser cleaning is an eﬀective procedure when helped by a wet cleaning. Acknowledgement Ente Cassa di Risparmio of Florence is gratefully acknowledged for the large contribution given for this research.

References 1. E. Adamkiewitz, P. Bracco, M. P. Colombini, A. De Cruz, G. Lanterna, M. Matteini, K. Nakahara, O. Sartiani, and M. L. Wolbarsht, in Journal of Cultural Heritage, Vol. 4(1001), 202, 2003. 2. A. Andreotti. M. P. Colombini, G. Lanterna, and M. Rizzi, in Journal of Cultural Heritage, Vol. 4(1001), 355, 2003.

26 Simultaneous UV–IR Nd:YAG Laser Cleaning of Leather Artifacts ∗

S. Batishche1 , A. Kouzmouk1 , H. Tatur1 , T. Gorovets2 , U. Pilipenka3 , V. Ukhau3 , and W. Kautek4 1

∗ 2 3

4

National Academy of Sciences of Republic of Belarus, Institute of Physics, F. Scorina Ave. 68, 220012 Minsk, Belarus [email protected] National Art Museum of Belarus, Minsk, Belarus Research Technological Enterprise “Belmicrosystems” of “Integral” Amalgamation, Minsk, Belarus University of Vienna, Department of Physical Chemistry, Waehringer Str. 42 1090 Vienna, Austria

Summary. Ancient leather samples from original upholstered furniture were treated with nanosecond Nd:YAG laser radiation with wavelengths of 1,064, 532, and 266 nm. The novel approach was the simultaneous application of these wavelengths. It opened new approaches for laser cleaning leather. Extensive diagnostics such as absorbance of diﬀerent layers of leather, chemical composition and microscopic inspection studies before and after cleaning were conducted. Advantageous results with simultaneous UV–IR (266 nm + 1, 064 nm) radiation are presented and discussed.

26.1 Introduction Laser cleaning of biogenetic artifacts has concentrated mainly on painting varnish, paper, and parchment in recent years [1–5]. In this context, best results for the removal of dark contaminants from, for example, cellulose and collagen materials were observed with visible laser radiation (532 nm) allowing maximum contrast between the light absorption of foreign materials and the substrate [4, 5]. Leather processing such as laser engraving and marking, in contrast to cleaning, has been developed to an industrial process relying on the in-depth ablation of the substrate. In this chapter, results optimized multiwavelength conditions of laser cleaning dark contaminants on a leather surface with a Nd:YAG laser are presented.

222

S. Batishche et al.

26.2 Experimental Methods The prototype Nd:YAG laser cleaning system [6, 7] employed in this study allows the spatial and temporal overlap of the fundamental beam (1,064 nm, up to 300–500 mJ) with its second (532 nm, up to 800 mJ), and fourth (266 nm, up to 300 mJ) harmonic frequencies. The absorbance spectra measurements of leather samples were performed by a Varian Cary 500 spectrometer. The cleaning results were evaluated with an optical microscope (Stemi 2000-C) and a scanning electron microscope (JEOL 840, Stereoscan-360 with EDXSpectrometer AN-10000). The colorimetric investigations were conducted following the standards indicated by a microspectrophotometer (MPV-SP, Leitz). In these investigations three light sources (A – incandescent lamp, C – daylight lamp, D65 – luminescent lamp) were used. The study of the elemental composition of diﬀerent layers of leather samples before and after cleaning was conducted by the methods of local X-ray spectral analysis on a raster-type electronic microscope (S-360 with an AN-10000 analyzer) and by secondary-ion mass spectroscopy (IMS-4F, CAMECA). The following parameters were employed: primary ions O2 + , Cs+ ; element range from H to U; depth resolution of 5–30 nm.

26.3 Results and Discussion Ancient leather samples from original upholstered furniture were laser treated (Fig. 26.1). A schematic structure of leather samples is shown in Fig. 26.2. The

Fig. 26.1. Original leather samples from an upholstered furniture. Austrian Museum of Applied Arts/Contemporary Art (MAK), Vienna

26 Simultaneous UV–IR Nd:YAG Laser Cleaning of Leather Artifacts

223

Fig. 26.2. Schematic layer structure of a contaminated leather surface

surface was covered by a thin dark layer of pollutants located mostly in cracks and grooves. A number of areas may be chosen corresponding to particular conditions of the contaminant layer (1) an area on the right side of the sample that had been subjected only to minimal light and minimal contacts, referred to as “original,” (2) an area on the right side of the sample that was subjected to intense light but minimal contacts, “faded,” (3) an area on the right side of the sample that was subjected to intense light and strong contacts, “polluted,” (4) a transition layer between the leather itself and the dye layer, “intermediate,” (5) an area on the right side of the sample from which the layers of pollutants, varnish, dye, and intermediate were scraped oﬀ, “base,” and (6) an area on the backside of the sample, “backside.” Figure 26.3 shows approximate (including absorbance and scattering of light) absorbance spectra of various parts of the sample. The spectra were taken from KBr pellets (20 mm in diameter, 1 mm thickness, mass 1 g) with powdered material from various parts of the sample surface (5 mg, thin layers were scraped oﬀ the surface). The absorbance spectra of pure KBr was subtracted. The densities of contaminant, leather, and intermediate phases were found to be 0.706, 1.276, and 0.987 g cm−3 , respectively. From Fig. 26.3, it is seen that the absorbance for all samples show localized peaks in the ranges of 500, 360, and 260 nm. For 266-nm radiation, however, a maximum is observed around 260 nm. Use of 266-nm radiation obviously allows removing pollutants layer by layer to a certain extent. Irradiation of the backside with 35 and 120 mJ cm−2 gives analogous results – a minor rise in absorbance in the vicinity of 355 nm in comparison with the nonirradiated backside. That suggests that 266-nm irradiation causes little photochemical changes of a collagen structure of leather.

224

S. Batishche et al.

4000

266 nm

3000

K, cm−1

355 nm

2000

KB01 KB11 KB21 KB31 KB41 KB51 KB71 KB91 KB61 KB81

532 nm 1000 1064 nm

0 300

400

500

1100

λ, nm Fig. 26.3. Typical absorbance spectra of various parts of a contaminated leather (see Fig. 26.1): Original – clean part kept in the dark, Faded – clean part that was exposed to sunlight, Polluted – polluted part, Intermediate – a layer between the top side and the base, Base – a layer of the base just below the Intermediate, Cleaned – polluted parts of the top layer cleaned by 266-nm laser radiation at 35 mJ cm−2 , Cleaned01 – faded parts of the top layer cleaned by 266-nm laser radiation at 35 mJ cm−2 , Backside− Ini – polluted part of the verso side (the untreated base), Backside− Irr and Backside− Irr− 01 – polluted parts of the verso side cleaned by 266-nm laser radiation at 35 and 120 mJ cm−2 , respectively

It should be noted, that the mechanical properties of the backside changed substantially upon treatment. This was obvious due to the fact that the particles scraped oﬀ from the irradiated backside showed a ﬂuﬀy structure. The irradiation of the top side gives a more substantial rise in absorption in the vicinity of 355 nm as compared to the nonirradiated one. These spectral changes may involve photochemical changes just in the varnish layers due to the low penetration depth of the UV light. The inspection of the surface of leather samples before and after laser cleaning with optical microscopy, scanning electron microscopy, secondary ion mass spectroscopy, colorimetry and EDX allowed several conclusions: 1. The density of carbon is substantially higher on the surface than in the bulk. A higher density of Cr and Fe is also observed on the surface. In the bulk, Na, Al, Ca, K are increased. The Ti density is practically the same on the surface and in the bulk. These diﬀerences are connected, most

26 Simultaneous UV–IR Nd:YAG Laser Cleaning of Leather Artifacts

225

Table 26.1. Lab color coeﬃcients

1 2 3

L∗

A a∗

b∗

26.0 25.4 21.9

10.0 9.8 3.0

10.3 11.0 1.3

light source C L∗ a∗ 24.6 24.0 21.7

7.1 7.0 1.4

b∗

L∗

D65 a∗

b∗

8.2 8.9 1.0

24.6 23.9 21.7

7.9 7.7 1.7

8.0 8.7 1.0

Fig. 26.4. The surface of the leather sample after (at the left) and before (at the right) cleaning by nanosecond laser radiation simultaneously with the wavelengths of 266 and 1,064 nm at ﬂuence F266 = 35 mJ cm−2 and F1,064 = 80 mJ cm−2 , respectively

likely, with peculiarities of original leather manufacture, or with airborne contaminants, handling, intrinsic impurities from soil, etc. 2. The analysis of the chemical composition of the contaminated surface before and after laser treatment showed the presence of a large number of chemical elements (C, O, Na, S, Si, K, Ca). There were also some traces of Cl, Al, Mg, Fe. After laser treatment, concentrations of such elements as S, Si, K, Ca, Cl, Al, Mg, Fe decreased, while peaks of O, Al, Na increased. 3. Color coeﬃcients for xyz, XYZ, Lab, and LCH color calculation systems for diﬀerent angles of incidence were measured. Table 26.1 presents Lab

226

S. Batishche et al.

color coeﬃcients at angle of incidence of 10◦ , where color diﬀerences are most distinct, for the original area (1), the polluted area before (3), and after (2) laser treatments at 266 nm. The initial diﬀerence between L∗ a∗ b∗ coeﬃcients of the original area and the polluted area before laser treatment for the light sources A and D65 amounts to 1.5–9 times. The ﬁnal diﬀerence between the color coeﬃcients of the original area and the polluted area after laser treatment at 266 nm becomes small for all of the light sources and angles of incidence. These investigations show that the best laser cleaning results are produced with 266 nm at F = 35–100 mJ cm−2 and with the simultaneous combination of 266 nm and 1,064 nm with F = 35–100 mJ cm−2 and 80–250 mJ cm−2 , respectively. After laser cleaning under optimum conditions the surface gets a dull luster, which can be explained by the partial removal of the varnish layer. Expert evaluation shows that results with 266+1, 064 nm in combination look more attractive (Fig. 26.4). Laser radiation at 532 or 1,064 nm resulted in the damage of the surface in all ﬂuence ranges.

26.4 Conclusions The data obtained suggest that the laser treatment of an ancient polluted leather surface by UV + IR radiation of nanosecond duration at 266 nm and 266 + 1, 064 nm may result in cleaning of the surface, recovering the original color and preserving most of the superﬁcial varnish structure. Expert evaluation shows that the 266 + 1, 064 nm wavelength combination yields the most attractive cleaning result in respect to other wavelengths. Acknowledgments We acknowledge partial ﬁnancial support by the ISTC project B-373-2 “Laser cleaning of art works of metals, paper, parchment, fabric, leather, and painting: research of possibilities, development of technologies and laser equipment”. W.K. thanks P. Noever and M. Trummer of the Austrian Museum of Applied Arts/Contemporary Art (MAK), Vienna, for providing the original samples in the context of the EUREKA project “Laser Cleaning of Paper and Parchment (LACLEPA)” N 1681.

References 1. W. Kautek and E. König (Eds.), Lasers in the Conservation of Artworks I, Restauratorenblätter, Vienna 1997. 2. M. Cooper, Laser Cleaning in Conservation, Butterworth-Heinemann, 1998. 3. W. Kautek, S. Pentzien, P. Rudolph, J. Krüger, and E. König, in Applied. Surface Science Vol. 127–129, 746, 1998.

26 Simultaneous UV–IR Nd:YAG Laser Cleaning of Leather Artifacts

227

4. P. Rudolph, F. J. Ligterink, J. L. Pedersoli Jr., M. van Bommel, J. Bos, H. A. Aziz, J. B. G. A. Havermans, H. Scholten, D. Schipper, and W. Kautek, in Appl. Phys. A, Vol. 9, 181, 2004. 5. J. Kolar, M. Strlic, S. Pentzien and W. Kautek, in Appl. Phys. A 71, 87, 2000. 6. A. Anisimov, S. Batishche, A. Egglezis, C. Fotakis, A. Kouzmouk, P. Pouli, H. Tatur, and V. Zaﬁropulos, in 3-rd International Workshop on New Trends in Laser Cleaning, October 3–4, 2003, Crete, Greece. 7. S. Batishche, A. Englezis, T. Gorovets, A. Kouzmouk, U. Pilipenka, P. Pouli, H. Tatur, G. Totou, and V. Ukhau, in Applied Surface Science, Vol. 248 (1–4), 264, 2005.

27 An Evaluation of Nd:YAG Laser-Cleaned Basketry in Comparison with Commonly Used Methods A. Elliott1 , A. Bezúr2 , and J. Thornton3 1

2

3

Walters Art Museum, Department of Conservation and Technical Research, 600 North Charles Street, Baltimore MD 21201-5185, USA [email protected] Department of Conservation, The Art Institute of Chicago, 111 South Michigan, Chicago IL 60603-6110, USA Art Conservation Department, Buﬀalo State College, 1300 Elmwood Avenue, Rockwell Hall 230, Buﬀalo NY 14222-1095, USA

Summary. While in storage and on exhibition, baskets can accumulate dirt that is aesthetically undesirable and even harmful. The nature of the woven structure, as well as the porosity of organic materials, causes diﬃculty in the removal of accumulated dirt. This chapter presents results from a study of basket-cleaning methods focusing on how Nd:YAG laser-cleaned samples are compare with those cleaned by more commonly used methods. Cleaning tests were performed on stem, bark, and root sample materials in order to examine the eﬀects of cleaning on a variety of plant materials that are commonly encountered with basketry. Photography, optical microscopy, and scanning electron microscopy were used to document and compare the eﬀectiveness and drawbacks of these methods. The results indicated that plant materials with protective cuticle layers can be eﬀectively cleaned using lowtech methods and such ﬁbers would not greatly beneﬁt from laser cleaning. Materials without protective cuticle layers are more sensitive to mechanical cleaning and could possibly be more safely cleaned using lasers.

27.1 Introduction The problem of surface dirt on basketry has been approached from many diﬀerent angles due to the diﬃculty of cleaning. The nature of the woven structure and the ﬁbrous quality of plant materials complicate the cleaning process. Dirt easily becomes imbedded in the rough, uneven surfaces. A study to compare cleaning methods was undertaken to address the scarcity of published information on the cleaning of basketry [1]. This study examined several commonly used cleaning methods including a brush and vacuum, cotton swabs lightly dampened with deionized water, and groomstick [2]. Lasers were included in the study as a possible new cleaning tool because they have been

230

A. Elliott et al.

Fig. 27.1. Epidermis (based on Florian 1990: 8)

used successfully for cleaning other organic materials. In this chapter, we will present the ﬁndings of the laser component of this study in comparison with the most eﬀective commonly used methods. It is important to remember that the irreversible nature of cleaning necessitates a careful approach to treatment, particularly with basketry materials as they often have evidence of ethnographic use. Any residue that could be associated with use should not be disturbed during the removal of post-collection dirt and grime. The ability of cleaning methods to leave residue undisturbed was not investigated. Criteria for evaluating the appropriateness of a cleaning method include (1) the eﬀectiveness of dirt removal, (2) damage to ﬁbers and the weave structure, and (3) the retention/deposition of residues from cleaning materials. As our study conﬁrms, the ﬁrst two criteria are related to the characteristics of the basket, including the type of ﬁber used, the structure of the weave, and condition of the artifact. The inﬂuence of a basket material’s morphology on the success of cleaning cannot be understated. Basketry materials usually fall into one of four categories – roots, stems, leaves, or bark. Stems, roots, and leaves all have an outer layer of epidermal cells which protect their internal structure. These layers have openings called stomata that regulate air and vapor transmission. In addition, the epidermal layers on stems and leaves produce a waxy cuticle layer that covers the structure (Fig. 27.1). The cuticle is not comprised of one homogeneous layer but several layers of diﬀering chemical compositions [3]. The cuticle layers help to reﬂect ultraviolet and infrared radiation, as well as waterproof the plant. If present after processing, these protective layers can help prevent damage to the ﬁber during cleaning and handling. Inner bark on the other hand is found within a woody stem. Its location within the stem eliminates the need for protective layers. When the root and bark ﬁbers are pulled apart during processing, the cells are split longitudinally, leaving a vulnerable structure with no outer protective cells.

This page intentionally blank

27 An Evaluation of Nd:YAG Laser-Cleaned Basketry

231

Fig. 27.2. Stem sample

27.2 Experimental Methods 27.2.1 Sample Materials and Preparation This chapter discusses controlled irradiation with lasers, along with the three most commonly used cleaning methods: brushing and vacuuming, swabbing with cotton lightly dampened with deionized water, and swabbing with groomstick. Three diﬀerent plant materials were sampled for cleaning tests with six diﬀerent methods [4]. An unidentiﬁed stem material, cedar bark, and spruce root were chosen to represent the variety of plant structures encountered on artifacts. The ﬁrst sample set was taken from an Italian basket constructed of an unknown stem material (Fig. 27.2). The ﬁbers appeared to be in excellent condition, with the exception of some areas that appeared more ﬁbrous in texture possibly indicating ﬁber damage. These areas were avoided during sampling. The other samples were taken from artifacts belonging to the Buﬀalo State College Art Conservation Department’s study collection. Both of these artifacts originate from tribes from the northwest coast of North America. One object was a woven mat constructed of the black, orange-red, and tan-colored inner bark from a cedar tree (Fig. 27.3). The other basket consisted of twined spruce root (Fig. 27.4). Both the spruce root and cedar bark artifacts showed some wear and brittleness. The types and degree of soiling varied between the three types of sample materials. The stem material had a heavy layer of gray particulate. The spruce root had a darkened appearance that appeared to be imbedded soiling, along with some light gray particulate soiling on the surface. The cedar bark had very little surface soiling, particularly for the purposes of this study. It was artiﬁcially soiled by mixing dirt collected from artifact storage areas and carbon black pigment. This mixture was dusted onto the lightly dampened surface of the sample set. Three samples measuring approximately 3.18 × 3.18 cm were used for each cleaning method for each material to allow for the observation of variations. In

232

A. Elliott et al.

Fig. 27.3. Cedar bark sample

Fig. 27.4. Spruce root sample

addition, a control sample was used for each sample set. An eﬀort was made to choose samples with similar degrees of soiling. An additional control was used for the stem material. The two-layer construction of the basket maintained an unsoiled woven surface on the inner sides of each layer. A sample was taken from this area for comparison with the uncleaned sample. 27.2.2 Cleaning Procedures The brush and vacuum technique involved lightly brushing the samples with a fan brush while directing particulate into a vacuum nozzle. This technique was performed for 10 s on each of the three sample types. Swabs were dampened with deionized water and blotted to remove the excess. The swabs were rolled across the surface or individual elements using the swab tip to reach into crevices when needed. This technique was performed for 35 s on the cedar bark and stem samples and 25 s on the spruce root samples. A 2.54×0.48 cm section

27 An Evaluation of Nd:YAG Laser-Cleaned Basketry

233

Fig. 27.5. Stem control sample

Fig. 27.6. Water-cleaned stem

of groomstick was applied to the end of a bamboo skewer. The groomstick was rolled across the surface for 25 s on each of the three materials. The laser cleaning was performed using a Lynton Lasers Phoenix Q-switched Nd:YAG laser at the infrared wavelength of 1,064 nm. Fluence was determined by dividing pulse energy by the area of the beam (estimated using carbon paper). A pulse rate of 2 Hz was maintained for all cleaning tests. All samples were cleaned using ﬂuences of 0.39, 0.45, and 0.56 J cm−2 . Additional testing was performed at the ﬂuences of 0.20, 0.29, and 0.35 J cm−2 for the stem material and 0.17, 0.35, and 0.42 J cm−2 for the spruce root material. No further testing was done on the cedar bark material due to a limited supply of sample material. 27.2.3 Documentation and Analysis Samples were documented before and after cleaning using a Nikon D-100 digital camera. Optical microscopy and scanning electron microscopy were used on the control samples and on one of the three samples from each cleaning method and material.

234

A. Elliott et al.

Fig. 27.7. Laser-cleaned stem (0.56 J cm−2 )

Fig. 27.8. Laser-cleaned stem (0.35 J cm−2 )

27.3 Cleaning Results 27.3.1 Stem All four cleaning methods were visually eﬀective in reducing soiling (Fig. 27.5). Brushing and vacuuming produced the most even cleaning because the brush was able to reach the deep interstices of the weave structure (Fig. 27.6). Waterdampened swabs were unable to achieve the same eﬀect but revealed more of the original sheen than other methods. Groomstick removed a moderate amount of dirt. The laser-cleaned samples signiﬁcantly reduced soiling over the entire surface, including the deep interstices, but left an overall dull appearance on some of the samples. Photomacrographs taken before and after treatment show only minor ﬁber disturbances with the commonly used cleaning methods and no change with the laser-cleaned samples. Scanning electron microscopy showed the brush and vacuum and the water-dampened swabs to be slightly abrasive to the cuticle surface, while there was little damage with the groomstick-cleaned sample. Damage was very apparent with laser cleaning at the higher ﬂuences of 0.39, 0.45, and 0.56 J cm−2 . At the highest ﬂuence, the cuticle was largely ablated leaving the underlying cell walls exposed (Fig. 27.7). This damage was also apparent with the 0.45 J cm−2 sample. The cuticle layer of the 0.39 J cm−2 sample appeared to have been partially

27 An Evaluation of Nd:YAG Laser-Cleaned Basketry

235

Fig. 27.9. Cedar bark control

Fig. 27.10. Brush and vacuum cleaned bark

reduced. The samples cleaned with a ﬂuence below 0.35 J cm−2 had larger amounts of dirt remaining but also have intact cuticle layers (Fig. 27.8). 27.3.2 Cedar Bark The three low-tech methods for dirt removal were only marginally eﬀective. Photomacrographs showed damage to the sample surfaces in the form of removed or damaged ﬁbers (Fig. 27.9). Groomstick was clearly damaging and should not be used with this type of material. As with the stem material, brushing and vacuuming removed more of the dirt trapped between the woven elements, while the water-dampened swabs revealed more ﬁber sheen. Lasers were clearly seen as less damaging on this type of material. Loose and damaged ﬁbers were left undisturbed and there was an overall reduction in surface dirt. SEM analysis of the cleaned samples revealed the brush and vacuum method to be the least damaging of the more common cleaning methods (Fig. 27.10), while water-dampened swabs were clearly the most damaging due to the removal of the delicate cell walls (Fig. 27.11). The laser-cleaned samples at the highest ﬂuence of 0.56 J cm−2 showed some disturbance and removal of the cell walls (Fig. 27.12). Damage, if any, at the lower ﬂuences of 0.39 and 0.45 J cm−2 was diﬃcult to distinguish from the original condition of the

236

A. Elliott et al.

Fig. 27.11. Water-cleaned bark

Fig. 27.12. Laser-cleaned bark (0.56 J cm−2 )

ﬁbers. It is likely that no laser-induced damage occurred at the 0.39 J cm−2 cleaned sample. The lack of extra sample material prevented further testing. 27.3.3 Spruce Root Of the low-tech methods, brushing and vacuuming was the only one to produce visually acceptable results. Although swabbing with water-dampened cotton swabs or groomstick removed some light surface dirt, the pressure produced by these methods was too great. The laser-cleaned samples were the only ones to show any signiﬁcant reduction in dirt, likely due to the imbedded nature of the soiling. SEM analysis was diﬃcult to interpret with this material due to the complex structure and apparent damage on the control sample. The results were inconclusive.

27.4 Conclusions This study suggests that the appropriate cleaning method for a basket depends on the type of ﬁber used for construction. Materials that have protective

27 An Evaluation of Nd:YAG Laser-Cleaned Basketry

237

cuticle layers can be cleaned with a wider variety of techniques than materials without cuticle layers. Commonly used methods such as a brushing and vacuuming and swabbing with water-dampened cotton, or a combination of both are suﬃciently eﬀective on materials with protective cuticle layers. This study established a damage threshold for lasers on materials with cuticle layers and those without. Lasers could be beneﬁcial on materials without protective cuticle layers such as spruce root and cedar bark materials. These more fragile materials are easily damaged during cleaning with more traditional materials. The loosely bound ﬁbers of cedar bark are easily lifted and disturbed using traditional methods. Dirt also becomes easily embedded in ﬁbrous materials such as cedar bark and spruce root making them diﬃcult to clean. In addition, the weave structure and condition of many baskets constructed of these types of fragile materials make the pressure of swab cleaning impractical, while lasers do not involve physical contact or pressure. In conclusion, while lasers can provide visibly eﬀective cleaning, cellular damage that is not visible to the unaided eye could occur. Assuming that appropriate ﬂuences are chosen, laser cleaning may be useful for cleaning problematic basket materials and structures, especially since low-tech methods appear to cause comparatively more damage. Acknowledgments We would like to thank the following people for their generous time and support of this project: Dr. Peter Bush, Tony Sigel, Pamela Hatchﬁeld, Ruth Norton, and Claire Munzenrider. Angela Elliott would like to acknowledge funding of her graduate studies and projects by the following entities: the Leo and Karen Gutmann Foundation, Buﬀalo State College, the Getty Grant Program, the Andrew W. Mellon Foundation, the Samuel H. Kress Foundation, the National Endowment for the Arts, the Kenzie Art Conservation Fellowship, and the Museum of New Mexico Foundation.

References 1. Ruth Norton’s section on cleaning in The Conservation of Artifacts from Plant Materials is an extremely useful guide on the topic. M.E. Florian, D.P. Kronkright, and R.E. Norton, J. Paul Getty Trust, Los Angeles, (1990). 2. Groomstick is vulcanized cis-1,4-polyisoprene with titanium dioxide used as a ﬁller. It is described as a molecular trap by the manufacturer and has a very tacky quality. Available from Talas, New York; http://www.talasonline.com. 3. P.J. Holloway, in The Plant Cuticle, Ed. by D.F. Cutler, K.L. Alvin, and C.E. Price, Linnean Society Symposium Series, No. 10, (1980). 4. The two less eﬀective methods not discussed in this paper were the use of a Magic Rub polyvinyl chloride eraser in combination with vacuuming and Smoke Oﬀ sponges, a polyisoprene sponge marketed for the removal of soot. Available from Talas, New York; http://www.talasonline.com.

28 Novel Applications of the Er:YAG Laser Cleaning of Old Paintings A. Andreotti1 , P. Bracco2 , M.P. Colombini1 , A. deCruz2 , G. Lanterna2 , K. Nakahara2 , and F. Penaglia1 1

2

3

Dipartimento di Chimica e Chimica Industriale, Università di Pisa, via Risorgimento 35, 56126 Pisa, Italy Opiﬁcio delle Pietre Dure, Fortezza da Basso, Viale Filippo Strozzi 1, Firenze, Italy Duke University, Department of Chemistry, Durham, NC, USA

Summary. This chapter focuses on the use of Er:YAG laser cleaning technique for the removal of unwanted and/or degraded materials both from a large series of reference standards (overpainting, varnishes, patinas, and restoration materials) which simulate the layering of old paintings, and also examples from old paintings. A series of diagnostic controls (optical microscopy, SEM, FT-IR, GC–MS, and topographic techniques) were designed to study the eﬀects of the laser radiation on the surface components, including morphological, optical, and chemical examination. The most signiﬁcant results show that an eﬀective thin-layer-removal of about 90% is obtained by submitting the painted surfaces to the laser exposure, while the rest of cleaning is rapidly accomplished in safety by applying mild solvents or aqueous methods. Consequently, possible interference with the original substrate can be noticeably minimized. No degradation compound induced by laser energy was formed. The laser cleaning procedure applied on an oil painting canvas “Morte di Adone” (seventeenth century), and on a panel tempera painting “San Nicola e San Giusto” of Domenico di Michelino (ﬁfteenth century) shows that the surfaces cleaned by this system exhibit a morphology quite similar to that obtained by traditional cleaning methods.

28.1 Introduction The cleaning of painted surface is one of the most critical operations in conservation. Contemporary criteria for the cleaning procedures require a selective and progressive removal of materials, which can enable an expert conservator to preserve any external thin original glaze or even old varnishes or patinas. Moreover, for environmental needs and safety reasons it is required to abandon toxic solvents cleaning and to use alternative methods, such as aqueous methods or less-toxic solvent cleaning. Er:YAG laser ablation, in fact, meets these requirements. Actually, laser techniques have demonstrated very promising

240

A. Andreotti et al.

applications for diagnostic and conservation purposes in art conservation [1]. As reported in the literature [2], Er:YAG laser exposure at 2.94 µm on a surface dampened with a liquid containing –OH groups eﬀectively removes old varnish and other encrustations without inducing unwanted chemical or physical changes on the original painted surfaces on canvas and wood panels. On the basis of the previous results [3,4], this chapter presents an exhaustive protocol for the use of Er:YAG laser technique in the removal of superﬁcial layers from paintings to obtain a comprehensive knowledge on laser eﬃciency, ﬂuence, threshold limits, and its practical cleaning approach combined with other ancillary methods. These parameters are tested ﬁrst on the laboratory models and afterwards on old paintings. The study and control of the polychrome surfaces before and after removal of the coatings by Er:YAG laser were performed by a series of diagnostic analyses which provide fundamental information on the morphology and chemical composition of the treated surfaces.

28.2 Experimental Methods Experiments of Er:YAG laser cleaning were conducted ﬁrstly on the laboratory tempera or oil paint models with top layers varying from natural resin varnishes, oil-resin varnishes, synthetic varnishes/ﬁxatives/adhesives, artiﬁcial patinas to overpaintings, which were previously prepared in 1999 [3]. The top layers examined by laser ablation were shellac (natural resin varnish); ketone and vinylic resin (synthetic varnish); mastic and walnut oil mixture varnish, boiled linseed oil varnish (oil based varnish); burnt umbercasein overpainting, Naples yellow-linseed oil overpainting, Naples yellowcasein overpainting, burnt umber-linseed oil overpainting on gypsum/rabbit glue (dark-colour/light-colour overpainting and overpainting on new ground); and EVA-based resin and acrylic resin (BMA and EA/MMA). In order to study the behaviour of the paint layer (tempera/oil) exposed directly to laser, uncoated surfaces were also examined. On the paint model simulating multiple layers (burnt umber-linseed oil overpainting, stucco with gypsum/rabbit glue, mastic varnish, yellow ochre/lead white in egg tempera, ground of gypsum/rabbit glue), laser ablation was tested to prove its gradual thinning action. All the laser tests have been executed under the control of stereo microscope by painting conservators in order to study the action of laser ablation and to determine threshold for each material examined. The laser ablation testing has been proceeded with application of progressively increasing energy to the surface, ﬁrstly starting with dry methods and secondly with wet methods. To study the overall eﬀects of laser, the energy levels beyond the threshold limit were also examined. The auxiliary wetting liquids, applied with small cotton swab, were O–H containing substances (distilled water, ethanol, water– ethanol mixture (1:1, v/v), distilled water with 1% surfactant [Tween20 ]);

This page intentionally blank

28 Novel Applications of the Er:YAG Laser Cleaning of Old Paintings

241

non-O–H containing liquid (light aliphatic hydrocarbons [Ligroine]); non-O– H containing liquid added with O–H containing substances [White Spirit with 15% diethylene glycol]). The range of energies used was between 3 and 200 mJ, with a laser beam diameter of 1 mm at 10 and 15 Hz macropulse frequencies. c of Mona The Er:YAG lasers used were “CrystaLase 2940” and Light Scalpel Laser Inc., USA. A previously published procedure [3, 4] was employed for the operative laser conditions and the collection of ablated materials. These materials were analysed by PY–GC–MS (Hewlett Packard, Palo Alto, CA, USA) and GC–MS (Thermo Electron Corporation, USA) methods for the analysis of synthetic polymers, amino acids, fatty acids, and terpenoids [5, 6]. OM with VIS and UV sources (Zeiss Axioplan) were used to observe the painted specimen surfaces and their modiﬁcation after the laser pulses. SEM (Leica Cambridge) and EDS (Link-Oxford) were used to study the complex morphology changes of the surfaces both in ﬂat samples and in cross sections. FTIR (Thermo-Quest) was used to compare the materials left on the surfaces and the ablated materials collect on the cover-slip glass. µ-Proﬁlometry INOA prototype (Florence, Italy) was used to measure micrometre diﬀerences in depth after the diﬀerent laser pulses.

28.3 Results and Discussion In general, the optimal energy thresholds for thin top layers (thickness ≈ 15 µm), of natural or synthetic resin, is between 8 and 13 mJ with auxiliary liquid containing O–H bond. In particular, synthetic polymers such as Plexisol (n-butyl methacrylate), or Plextol (methyl methacrylate/ethyl acrylate) were successfully removed, while BEVA (ethylene vinyl acetate) was not ablated: however, a deep surface modiﬁcation was provoked by the laser exposure which allowed the cleaning by a simple swabbing technique. Oil-based varnish models resulted quite resistant to laser ablation and showed a morphology (Fig. 28.1) with an increased roughness of the surface after laser ablation. This increase, together with the introduction of chemical agents, permits to achieve a cleaned surface in a shorter time using a low concentration of chemicals. Thick oil-based overpaintings (about 30–50 µm) can be ablated at an energy level between 30 and 100 mJ with auxiliary hydroxyl liquids. Repeated exposure may be performed according to the thickness and nature of the superﬁcial layers, without exceeding the established energy threshold. This allows a gradual removal of the overpainting without causing damage or discoloration of the substrate. Experiments on the multilayer model demonstrated a gradual and very thin laser ablation as shown in Fig. 28.2, where the energy of each passage and the wetting agent is reported. The thick overpainting (burnt umber in boiled linseed oil) was gradually thinned by ﬁve laser applications (100 mJ, 80 mJ, and 50 mJ) using distilled water/ethanol mixture (1:1) except for the last treatment, where white spirit

242

A. Andreotti et al.

Fig. 28.1. Microscope images of the sample YOET 1-1-3 (shellac varnish on lead white/yellow ochre dispersed in egg yolk tempera; gypsum/glue ground). All the images are referred to the same sample observed under optical microscopy and SEM (gold sputtering was used). The fragment was then embedded in resin and observed again in OM and SEM. All the images show the unexposed surface on the left-hand side

Fig. 28.2. The gradual action of Er:YAG laser is shown on a laboratory multilayers sample: burnt umber-linseed oil overpainting on stucco (gypsum/rabbit glue) which covers mastic varnish on yellow ochre/lead white egg tempera

added with 15% diethylene glycol was used. Then the thick stucco with gypsum and rabbit glue was thinned by ﬁve laser applications (150, 80, 50, and 20) using distilled water/ethanol mixture (1:1). The thin layer of mastic varnish as well as the egg tempera with yellow ochre/lead white underneath was well preserved. The laser cleaning was then applied on a ﬁfteenth-century tempera painting on panel, “S. Nicola e S. Giusto”, attributed to Domenico di Michelino, from S. Giusto in Piazzanese church in Prato (Italy), under restoration at the

28 Novel Applications of the Er:YAG Laser Cleaning of Old Paintings

243

4–5 mJ, white spirit + solvent

laser

1. fatty emulsion added with artificial saliva and coccocollagene 2. fatty emulsion pH7 3. ligroin 5–6 mJ, white spirit + solvent 1. fatty emulsion added of resin Soap made with DCA-TEA 2. fatty emulsion pH7 3. ligroin 4. fatty emulsion added of citric acid 5. fatty emulsion ph7 6. ligroin

Old tested area with traditional solvent method

Fig. 28.3. Fifteenth-century tempera painting on panel, “S. Nicola e S. Giusto”, attributed to Domenico di Michelino. Cleaning tests with combined Er:YAG laser and emulsions (top and middle) compared with an old test area (bottom) cleaned with mechanical tools and traditional solvents

Opiﬁcio delle Pietre Dure (Laboratories of restoration in Florence, Italy). The laser ablation was executed on the degraded greyish thin and compact ﬁlm tightly bound to the original layer (a very fragile white tempera). This layer locally contains discontinuous old brown varnish and some residues of greycoloured overpainting in the area of the mitre of Saint Nicolas (Fig. 28.3). The chemical identiﬁcation of the grey patina was performed by analysing the organic material with the GC–MS analysis: the amino acids’ percentage data showed that the patina was mainly constituted by egg with small traces of animal glue. Moreover, the FTIR analysis demonstrated the presence of calcium oxalate on sample surfaces. This area was considered to be suitable for laser test, as it was extremely diﬃcult to execute a selective and safe cleaning even by using sophisticated aqueous methods based on emulsions containing surfactants, resin soaps, or artiﬁcial saliva [7, 8]. After preliminary tests using the energy between 3 and 9 mJ at the frequency of 15 Hz, and wetting agents, white spirit or distilled water/ethanol mixture (1:1, v/v), a combined method with laser ablation and aqueous and mechanical method was considered to be appropriate. Initially the energy level of 8 mJ with white spirit as wetting agent and a clearing agent as water/ethanol mixture (1:1, v/v) after laser ablation seemed to oﬀer a good result as far as removal is concerned. The combined method with laser at 5–6 mJ with white spirit as wetting agent and aqueous methods using an emulsion added with resin soap, followed by a

244

A. Andreotti et al.

milder pH 7 emulsion and by Ligroin one, seemed to be also rather eﬃcient. However, the ﬁrst method did not seem to guarantee a selective and gradual cleaning, judging from observation under stereoscopic microscope especially due to extreme sensibility of the original colour to aqueous solution; the second combined method either was considered inappropriate, as it removed also the antique inner varnish which was considered to be preserved. Finally, the appropriate method consists of: – A laser ablation at 5–6 mJ with wetting agent (white spirit), which permits homogeneous superﬁcial desegregation of the grey-coloured layer – A following cleaning with a fatty emulsion [7] (Brij 35 2 g, artiﬁcial saliva (mucin 0.25 g, tribasic ammonium citrate 0.25 g, deionized water 100 ml) 20 ml, coccocollagen 2.5 ml, ligroin 80 ml) – A treatment with a clearing agent, a pH 7 emulsion (Brij 35 2 g, deionized water 10 ml, Tween 20 (nonionic surfactant) 2 ml, ligroin 90 ml) [8] – A ﬁnal treatment with ligroin alone By this way, it was possible to achieve a real selective removal of the degraded materials without removing the old varnish underneath. This approach is in agreement with the latest criteria of contemporary conservation, oriented towards the minimum intervention and the reduction of toxic chemicals. The laser cleaning approach was also used for removing the insoluble and hard overpaintings on the canvas painting “Morte di Adone” (seventeenth century). Using various cleaning approaches based on polar or basic solvents, acid solution, enzymes, enzyme soaps, chelating agents, and surfactants, any eﬃcient and safe removal of the unwanted materials was observed; nor the scalpel was successful due to the hardness of the overpaintings and their strong adhesion on the original substrate. The investigation on the grey overpainting in the area of the Venere’s shoulder by the GC–MS analysis, highlighted the presence of egg as proteinaceous material and a small amount of beeswax. The analysis of the organic material, laser sampled in diﬀerent areas from a tick overpainting on the Venere’s face, showed that it was mainly constituted by animal glue (Fig. 28.4) with a small trace of a pinacae resin. Laser pulse was repeated several times on the surface, in decreasing energy levels (18–20, 15, and 10 mJ) as shown in Fig. 28.5. This procedure permits the removal of the most of the overpainting and allows a ﬁnal cleaning with application of a mild solvent mixture and a soft employment of the scalpel without removing the old varnish. A noninvasive analysis, µ-proﬁlometry, executed before, during, and after laser cleaning allows to quantify the depth of laser ablation as 3–5 µm (Fig. 28.6). This avoids the need to take samples for cross sections. The operative conditions adopted for the removal of the overpainting were 15–18 mJ energy, wetting liquid isopropanol/ligroine 50%, and isopropanol 27%, followed by a clearance with the same liquid, repeated several times. Results show a very homogeneous and superﬁcial removal of

28 Novel Applications of the Er:YAG Laser Cleaning of Old Paintings

245

3 Database 1 Pos. 3, 2° passage, 15 mJ

2 egg

2 Pos. 3,1° passage, 19 mJ

1 3 Pos. 2,1° passage, 18 mJ 4 Pos. 1,1° passage, 19 mJ

0 −8

−6

−4 animal glue

−2

0

2

−1

−2

−3

4

5 Pos. 1,1° passage, 15 mJ 6 Pos. 2,2° passage, 17 mJ

casein

7 Pos. 3–4,° passage, 10 mJ

isopropylic alcohol + ligroine (1:1) as wetting agent

Fig. 28.4. Principal component analysis of samples from seventeenth century oil painting on canvas “Morte di Adone”: the relative amino acid percentage contents of the ablate material from diﬀerent area of Venere’s face reveal the presence of animal glue

Fig. 28.5. Seventeenth-century oil painting on canvas “Morte di Adone”: four overpainting removal tests realised by combined Er:YAG laser and emulsions actions. Left: tested area; right: close details

Fig. 28.6. Laser micro-proﬁlometry results of ﬁrst and second exposure of Er:YAG laser and emulsion on area No. 1. Left: level diﬀerence between unexposed surface (orange-red) and ﬁrst passage ablation level (yellow-green); centre: diﬀerence between surface and second passage level (cyan); right: relative level diﬀerence between ﬁrst and second passage (5 µm on average)

246

A. Andreotti et al.

irregular top layers after each laser application. However, small changes in laser exposure (about 10 mJ) and in the composition of the solvent mixture (isopropanol/ligroine 1:1), may also give successful results as shown for the diﬀerent areas in Fig. 28.5.

28.4 Conclusions The Er:YAG laser ablation oﬀers a gradual, homogeneous, and OH-selective cleaning with its thinning action of a few micron’s depth. The general mechanism of ablation consists in modifying the surface by breaking up of a few micron’s depth, which enables the conservator to complete the cleaning using less invasive methods which are not eﬃcient if used by themselves. The preliminary tests under microscope are indispensable to obtain the optimal threshold for each speciﬁc material, and to avoid possible surface modiﬁcation or chromatic alteration of the original substrate which may not be directly observed. The macropulse frequency of 15 Hz was conﬁrmed as the optimum to obtain a more homogeneous ablation. No chromatic alteration was observed below the energy threshold, except for certain colours such as yellow ochre, some mineral iron based blue, or Naples yellow in casein. The use of hydroxylwetting agents is appropriate, as it increases the eﬃciency of laser ablation as well as safety of the original substrate. It is important to note that chemical analyses show that ablation under the threshold limits provokes no signiﬁcant variation in chemical composition of both ablated materials and substrate. Analyses also show that the energy levels used below the safety threshold, combined with the conservators’ skills, permit gradual and progressive cleaning. Particularly, repeated laser applications at lower ﬂuencies in respect to the optimal energy thresholds allow the cleaning of degraded. To conclude, we can assert that the combination of Er:YAG laser, used within the energy thresholds of each material, together with the ultimate chemical and biochemical systems, allows the cleaning of a broad variety of unwanted layers. With the suggested cleaning procedure, conservators/restorers have a further and alternative chance to properly solve diﬃcult restorations.

Acknowledgements Ente Cassa di Risparmio of Florence is gratefully acknowledged for the large contribute given for these researches. The authors are very gratefully to Raﬀaella Fontana, Enrico Pampaloni, and Maria Chiara Gambino (INOA, UNIFI) for the µ-proﬁlometry measurements and interpretations, to Fabrizio Cinotti for the macroimages on old paintings, and to Annette Keller for the graphic elaborations.

28 Novel Applications of the Er:YAG Laser Cleaning of Old Paintings

247

References 1. A. deCruz, S. Hauger, and M. L. Wolbarsht, in Opt. Photon. News, Vol. 10, 36, 1999. 2. A. deCruz, S. Hauger, and M. L. Wolbarsht, in Journal of Cultural Heritage, Vol. 1, 173, 2000. 3. E. Adamkiewitz, P. Bracco, M. P. Colombini, A. De Cruz, G. Lanterna, M. Matteini, K. Nakahara, O. Sartiani, and M. L. Wolbarsht, in Journal of Cultural Heritage, Vol. 4, 202, 2003. 4. A. Andreotti. M. P. Colombini, G. Lanterna, and M. Rizzi, in Journal of Cultural Heritage, Vol. 4, 355, 2003. 5. I. Bonaduce and M. P. Colombini, in Journal of Chromatography, Vol. 1028, 297, 2004. 6. I. Bonaduce and M. P. Colombini, in Rapid Communication in Mass Spectrometry, Vol. 17, 2523, 2003. 7. P. Cremonesi, L’uso degli enzimi nella pulitura di opera policrome, Padova 1999. 8. P. Cremonesi, L’uso di tensioattivi e chelante nella pulitura di opere policrome, Padova 2001.

29 A Final Report on the Oxidation and Composition Gradients of Aged Painting Varnishes Studied with Pulsed UV Laser Ablation C. Theodorakopoulos1 , V. Zaﬁropulos2 , and J.J. Boon3 1

2

3

University of London, School of Biological and Chemical Sciences, Birkbeck College, Malet Street, London WC1E 7HX, United Kingdom [email protected] Laboratory of Applied Physics, Department of Human Nutrition & Dietetics, Technological Educational Institute of Crete, I. Kondylaki 46, 723 00 Sitia, Crete, Greece FOM Institute for Atomic and Molecular Physics, (AMOLF), Kruislaan 407, Amsterdam, The Netherlands

Summary. This paper discusses ﬁndings that establish the ageing-induced compositional and crosslinking gradients across the depth-proﬁles of two accelerated aged natural resin varnishes: dammar and mastic, which are commonly applied to paintings. Proﬁle measurements of laser-processed ﬁlms using a KrF excimer laser, as well as online measurements of the C2 emission by laser-induced breakdown spectroscopy (LIBS), showed a signiﬁcant reduction of the ablation step and ablation yield with depth, respectively. Direct temperature mass spectrometry (DTMS) showed that the oxidation products formed upon ageing were gradually eliminated across depth, which aﬀected the depth-wise optical properties of the ﬁlms studied by UV/VIS spectrophotometry. The total ion currents of the DTMS in the electron ionisation mode (EI, 16 eV) demonstrated also a gradual reduction of the pyrolysis yield which corresponded to a gradual depth-wise elimination of the high MW fractions that was conﬁrmed by size exclusion chromatography (SEC). The gradients were established also by surface analyses, such as matrix-assisted laser desorption/ionisation-time-ofﬂight-MS (MALDI-TOF-MS) and attenuated total reﬂection-Fourier transformed infrared spectroscopy (ATR-FTIR), which indicated that the action of the KrF excimer laser is non-destructive to the varnish when optimal ﬂuences are used for the laser cleaning process. The extracted data enabled the quantiﬁcation of the compositional gradients and unraveled a signiﬁcant feature of natural resins, which would not have been possible without the use of KrF excimer lasers.

29.1 Introduction The laser cleaning of old master paintings and other works of art coated with aged and deteriorated varnishes has been proposed since the early 1990s [1–3].

250

C. Theodorakopoulos et al.

The proposed method is based on excimer laser ablation [4–6] and provides superior selectivity owing to a stepwise varnish removal on the micron scale. Since these advances came to light, the main objective has always been the ‘safe’ removal of aged varnish coatings from photosensitive paint substrates [7]. Technically this is diﬃcult, as UV laser irradiation results in paint discoloration once the laser photons reach the paint [8, 9]. Recently, it was shown that whilst the paint medium (oil or egg binding) is not visually aﬀected, discoloration occurs in the presence of inorganic pigments only [10]. Regardless of the minor scale that discoloration occurs, that is a few nm deep in the surface of the exposed pigment particles [11], the damage caused on the surface is irreversible. So far, two ways have been proposed to prevent paint discoloration (a) the use of online control of the laser ablation process by employing techniques that either monitor the plume emission or the treated surface [12–15] and (b) the retention of a thin ﬁlm of varnish on the surface to prevent irradiation of the underlying, photosensitive paint layers [7, 13]. Online control based on analysis has had remarkable results that assisted the improvement of the laser cleaning methods, but in practice it is not always possible to detect damage before its occurrence. We found that several types of analysis can be used at several stages of the cleaning procedure to assess the progress of the treatment and to validate the integrity of the treated surface. Regardless of the use of any complementary analytical assistance, the retention of a thin ﬁlm of the varnish has been proven to be essential and very eﬀective [16], provided that the thickness of the remaining varnish is larger than the optical absorption length of the material at the laser wavelength. Therefore, since the excimer laser cleaning of a coated surface must be terminated before the total ablation of the varnish, the most signiﬁcant factor for the safety of the underlying surface is the chemistry of the varnish. An exceptionally controlled removal of aged varnish with the minimum possible transmission towards the paint is guaranteed once the ‘optimum’ ﬂuence that maximizes the ablation yield per laser photon is being used throughout the cleaning procedure [7]. However, the depth-wise chemical change of aged natural varnishes results (a) in the change of the laser ablation rate [17, p. 164] and (b) in the transition of the ‘optimal’ ﬂuences to higher energy densities as the cleaning proceeds deeper within the varnish [17, p. 174]. This explains the change of the interaction of excimer laser pulses with successive depths of the same varnish ﬁlms using the same ﬂuence [18]. Thus, the optical, oxidation, crosslinking and compositional gradients, which have recently been established across the depth proﬁles of aged natural resin ﬁlms [17], play a dominant role in the laser cleaning of natural varnishes. Herein, we summarize the highlights of a comprehensive investigation of the compositional gradients of accelerated aged dammar and mastic resin ﬁlms, which would not have been possible without the aid of KrF excimer laser ablation.

This page intentionally blank

29 A Final Report on the Oxidation and Composition Gradients

251

29.2 Experimental Methodology Mastic and dammar in xylene 40% (w/v) were spin coated to produce 55-µm thick ﬁlms. After drying, ageing was accelerated in a Sunset CPS, Heraeus R xenon-arc fadeometer. Wavelengths longer than 295 nm were Instruments employed to imitate sunlight exposure [19]. The ﬁlms received a light dosage of 160 Mlux h, then exposed to open air for 45 days, followed by 30 days storage in the dark. R , COMPex series, KrF excimer laser (λ = 248 nm) was A Lambda Physik used with a pulse duration of 25 ns and source energy of 380 mJ. Standard laser ablation rate studies [7] were performed on both ﬁlms. The etching depths R S5P. Laser cleaning was based upon were determined with a Perthometer scanning adjacent areas of the ﬁlms with a pre-calculated number of laser pulses across the Gaussian proﬁle of the laser beam, as described in [7, 16] and [17, p. 159]. Zones of cumulative ∼3 µm steps were carried out as in [18]. Laser-induced breakdown spectroscopy (LIBS) [20] was carried out to determine the carbon dimer emission, which is associated with the degree of crosslinking [7] and condensation [17, p. 87] as a function of depth. Online transmission studies [7] were carried out to quantify the amount of laser light transmitted in the rear of the ﬁlms during the ablation process. Mass spectrometric, chromatographic, and spectroscopic analyses were separated into two groups to determine the depth-wise properties of the aged varnishes (a) those which determined the chemical state of the ablated surfaces, such as attenuated total reﬂection-Fourier transformed infrared spectroscopy (ATR-FTIR) and a speciﬁcally planned methodology for matrixassisted laser desorption/ionisation-time-of-ﬂight-MS (MALDI-TOF-MS) [17, p. 280] and (b) those which determined the mean chemical properties of the remaining ﬁlms across depth, such as UV/VIS spectrophotometry, direct temperature-resolved mass spectrometry (DTMS) [21] – both the summation DTMS mass spectra and the total ion currents (TICs) in EI (16 eV) and NH3 /CI (250 eV) ionisation modes were studied – the multivariant factor discriminant analysis (MFDA) [22, 23] of the DTMS data, as well as high performance-size exclusion chromatography (HP-SEC).

29.3 Results and Discussion A ﬁrst indication of the existence of the gradients across the ﬁlms was the change of the ablation rate against ﬂuence with depth [17, p. 164]. Using constantly ‘optimal’ ﬂuences, which were determined on the surface of the ﬁlms, it was observed that the ablation step was minimized particularly after the removal of the 15 µm surface layers in both ﬁlms [18]. Oxidation gradients were illustrated by the DTMS both in the summation MS and the TICs [17, p. 268, 18]. A good representation of the oxidation gradients detected by DTMS was demonstrated by MFDA. Figure 29.1 shows the molecular modiﬁcations across the MFDA-discriminant function (DF1) vs. the depth of the accelerated aged

C. Theodorakopoulos et al. DTMS discrimination function 1

252

1.2

Depth steps

1.0 0.8 Thresholds of highly aged and less deteriorated material

0.6 0.4 0.2 0.0

fresh dammar: independent signal from film thickness

−0.2 −0.4 0

5

10 15 20 25 30 35 Depth from surface (mm)

40

45

Fig. 29.1. MS projections of MFDA-DF1 in the 15-µm thick accelerated aged dammar ﬁlm. The corresponding DF1 vs. depth plot of mastic has a similar trend [17, p. 279]

Intensity (a.u.)

Dammar

0

10 20 30 40 Mean depth from surface (mm)

50

Fig. 29.2. Intensity of the C2 emission in the laser plume at ∼ 515 nm (ﬁlled circle) and ∼ 470 nm (circle) vs. the mean depth for an accelerated aged dammar varnish observed by LIBS (KrF excimer laser at 0.9 J cm−2 ). A few pulses were required to optimise the laser-coating interaction as described in [6]. The maximum intensity at the 10–15 µm uppermost layers corresponds to a high degree of crosslinking (and condensation) at the surface [7], which eventually decreases with depth. Mastic showed a similar trend [17, p. 180]

dammar ﬁlm. The corresponding MFDA-DF1 plot of mastic had a similar trend. The oxidation gradients have also been detected by micro-FTIR [7] and ATR-FTIR [17, p. 228] both showing a gradual reduction of the carbonyl absorption (∼ 1, 710 cm−1 ) with depth in accelerated aged natural resin ﬁlms. MALDI-TOF-MS indicated that the oxidative change with depth is not only quantitative but also qualitative, since the uppermost 10–15 µm layers had

29 A Final Report on the Oxidation and Composition Gradients

% DTMS-TIC Pyrolysis

no of pulses crosslinking

90

50 40

80 30 70 20 60 10 50

0 0

5

10 15 20 Depth from surface (mm)

No of KrF Excimer Laser Pulses

60

100

253

25

Fig. 29.3. The leap of the ablation step at ∼ 15 µm from the surface of the aged dammar ﬁlm using successive 248-nm laser pulses [18] corresponds to the evident jump of the intensity of pyrolysis during the DTMS investigation of the successive depth-steps of dammar. The latter is a marker for the reduction of the MW [25] with depth. The same observation has been made for the accelerated mastic ﬁlm [17, p. 246]

oxidation secondary products produced by exposure to the ambient UV wavelengths, whereas below the highly oxidised surface there was a zone in the bulk down to 25–30 µm oxidized by the absorption of visible light [17, p. 280]. Both ﬁlms were almost unaﬀected from ageing at depths greater than ∼ 25–30 µm from surface [17, p. 273, 18]. Analysis of the ablation plume with LIBS showed that the carbon dimer emission observed by the C2 Swan band system, the intensity of which has been associated with the degree of crosslinking [7], decreases abruptly after the removal of the 15 µm surface layers (Fig. 29.2) [17, p. 177]. Crosslinking gradients of aged dammar and mastic ﬁlms have been veriﬁed by (a) the depth-wise reduction of the methylene-to-methyl ratios monitored by microFTIR [24] and ATR-FTIR [17, p. 228], (b) the reduction of the intensity of pyrolysis during the DTMS runs of the remaining ﬁlms [17, p. 264, 18] and (c) the gradual decrease of the MW determined by HP-SEC [17, p. 284]. These ﬁndings show that the highly oxidative fraction of the aged varnishes, both in terms of quality and quantity, as well as the high MW crosslinks are generated at the uppermost zone of the ﬁlms. During the laser cleaning, the ablation step per laser pulse had a linear response to the laser-induced etching depth (for a certain ﬂuence), as long as the process was limited at the highly degraded surface layer [18]. Once this layer was removed the ablation step changed, as shown in Fig. 29.3. All data reﬂect theories on the oxygen ‘starvation’ with depth [26, 27], verify Beer’s Law and indicate a faster reduction of the UV compared with the visible wavelengths as the ambient light is transmitted across the depthproﬁle of the ﬁlms. Indeed, UV/VIS spectrophotometric measurements on

C. Theodorakopoulos et al. Optical Absorption Length (mm)

254

50 40 30

Mastic Dammar Threshold of highly aged, UV oxidised material

20 10

non UV oxidized less degraded material

0 250 275 300 325 350 375 400 425 450 Wavelength (nm)

Fig. 29.4. Optical absorption lengths against wavelength of ambient light (250– 450 nm) for dammar and mastic varnish ﬁlms. The corresponding optical absorption lengths of the 248-nm laser are one order of magnitude shorter [7]. Wavelengths shorter than 325–350 nm were completely absorbed by the 10–15 µm surface layers of the ﬁlms [17, p. 227]

the successive laser-ablated zones of the ﬁlms showed that the ambient radiation up to 350 nm is completely absorbed by the 15 µm dammar and mastic varnishes [17, p. 227] (Fig. 29.4). At the beginning of these studies it was shown that yellowing, that is the most readily monitored side eﬀect of resin ageing, was gradually reducing across depth of aged varnishes uncovered by means of nanosecond KrF excimer laser ablation [16]. It could have been suggested that this observation was due to laser-induced bleaching of the yellow chromophores, which are unsaturated quinones that absorb in the blue (400–490 nm) [28]. The presented ﬁndings, however, demonstrated that these chromophores are not generated in the bulk of the aged resin ﬁlms showing that, under controlled circumstances, the possibility of photochemical damage in the photoablated ﬁlms is negligible. According to the UV/VIS data and considering a ‘fresh’ natural varnish suddenly exposed to sunlight (λ > 250 nm), it could be suggested that autooxidative processes, i.e. free radical formation, oxygen uptake and crosslinking, which contribute to the eventual chemical and physical degradation of such a ﬁlm are eliminating with depth. This was eﬀectively reﬂected in the 248-nm ablation of the ﬁlms. During the laser irradiation, the 248-nm photons are ﬁrst absorbed causing (a) electronic excitations across the optical absorption length of the ﬁlm, which in turn causes (b) shock waves, that is photochemical and photomechanical action in the irradiated area, leading to (c) bondbreakage and desorption of excited species into the gas phase [6, 29]. Because of the gradient in absorption and the distribution of diﬀerent amounts and types of chromophores across depth [17, p. 258], the mechanisms that initiate the ablation of natural resin ﬁlms are gradually altered as the thickness of the ﬁlms is eventually reduced. Shock waves are dependent on the number of cova-

29 A Final Report on the Oxidation and Composition Gradients

255

lent bonds that are involved in the preliminary stages of the bond-breakage process [30]. Since both the degrees of absorption, which is indicative of the degree of oxidation, and crosslinking decrease with depth, the intensity of the photochemical and photomechanical actions following the release of the laser pulse also decreases with depth, as is indicated by the change of the ablation step vs. depth (Fig. 29.3).

29.4 Conclusions This paper establishes the ageing-induced compositional and crosslinking gradients formed across the depth-proﬁles of two natural resin varnishes: dammar and mastic, which was possible with the aid of nanosecond KrF excimer laser ablation. The signiﬁcant depth-wise reduction of the degree of oxidation corresponded to a decreasing absorption, which in addition to the reduction of crosslinking with depth, alters the ablation rate across depth of such ﬁlms. The determination of the gradients, especially with surface analysis, showed that laser cleaning with excimer lasers and optimal ﬂuences does not deteriorate further aged coatings made of natural resins. Upon ageing, the 10–15 µm surface layers of the dammar and mastic varnishes were utterly deteriorated absorbing the UV wavelengths of ambient light. The free transmission of the blue light (λ > 400 nm) after having removed the top 25–30 µm layers of the aged ﬁlms indicates that, upon ageing, yellow chromophores are not produced in the bulk of such ﬁlms. Thus, chemical deterioration leading to visual degradation of natural resin ﬁlms is a surface phenomenon. Hence, cleaning of aged (natural) varnish-paint systems can be accomplished by removing surface layers only. Acknowledgements This work would not have been possible without the EU Large Installations Plan DGXII (HCM) program ERBCHGECT920007 at the Ultraviolet Laser Facility at IESL/FORTH, the FOM Program 49 funded by FOM and NWO and the Foundation of the State Scholarships, Greece (IKY), which funded the PhD research of C. Theodorakopoulos. Professor Dr. Wolfgang Kautek, University of Vienna, Austria, and Dr. Klaas Jan van den Berg, ICN, The Netherlands, are acknowledged for their participation in the PhD examination board.

References 1. E.I. Hontzopoulos, C. Fotakis, and M. Doulgeridis, in SPIE: 9th International Symposium on Gas Flow and Chemical Lasers Vol. 1810, Bellingham, Washington, 748, 1993. 2. N. Morgan, Opto and Laser Europe 7, 36 (1993). 3. C. Fotakis, Optics and Photonics News 6, 30 (1995).

256

C. Theodorakopoulos et al.

4. B. Luk’yanchuk, N. Bityurin, S. Anisimov, and D. Bäuerle, in Excimer Lasers, Ed. by L.D. Laude. Kluwer Academic Publishers, The Netherlands. 59, 1994. 5. R. Srinivasan, in Laser Ablation, Ed. by J.C. Miller, Springer Series of Material Science Vol. 28, Springer, Berlin Heidelberg, 107, 1994. 6. D. Bäuerle, in Laser Processing and Chemistry, Third, rev. and enl. edn., Springer, Berlin Heidelberg New York, 2000. 7. V. Zaﬁropulos, in Laser Cleaning, Ed. by B. Luk’yanchuk, World Scientiﬁc, Singapore, 343, 2002. 8. A. Athanassiou, A.E. Hill, T. Fourrier, L. Burgio, and R.J.H. Clark, J. Cult. Herit 1. s209 (2000). 9. V. Zaﬁropulos, T. Stratoudaki, A. Manousaki, K. Melesanaki, and G. Orial, Surf. Eng. 17, 249 (2001). 10. M. Castillejo, M. Martin, M. Oujja, D. Silva, R. Torres, A. Manousaki, V. Zaﬁropulos, O.F. Van den Brink, R.M.A. Heeren, R. Teule, A. Silva, and H. Gouveia, Anal. Chem 74, 4662 (2002). 11. B. Luk’yanchuk, and V. Zaﬁropulos, in Laser Cleaning, Ed. by B. Luk’yanchuk, World Scientiﬁc, Singapore, 393, 2002. 12. I. Zergioti, A. Petrakis, V. Zaﬁropulos, and C. Fotakis, in Proceedings of LACONA I; Restauratotenblätter Sonderband, Vienna, 57, 1997. 13. V. Zaﬁropulos, and C. Fotakis, in Laser in Conservation: an Introduction, Ed. by M. Cooper, Butterworth Heineman, Oxford, 79, 1998. 14. J.H. Scholten, J.M. Teule, V. Zaﬁropulos and R.M.A. Heeren, J. Cult. Herit. 1, 215 (2000). 15. R. Teule, H. Scholten, O.F. van den Brink, R.M.A. Heeren, V. Zaﬁropulos, R. Hesterman, M. Castillejo, M. Martin, U. Ullenius, I. Larsson, F.A. GuerraLibrero, H. Gouveia, and M.B. Albuquerque, J. Cult. Herit. 4, 209 (2003). 16. C. Theodorakopoulos and V. Zaﬁropulos, J. Cult. Herit. 4, 216 (2003). 17. C. Theodorakopoulos, PhD Thesis, RCA, London, with FORTH-IESL, Heraklion, and FOM-AMOLF, Amsterdam, 2005; URL: http://www.amolf.nl/publications/theses/theodorakopoulos/theodorakopoulos.html; 18. C. Theodorakopoulos, V. Zaﬁropulos, C. Fotakis, J.J. Boon, J. van der Horst, K. Dickmann, and D. Knapp, in Springer Proceedings in Physics, Vol. 100, Edited by K. Dickmann, C. Fotakis, and J.F. Asmus. Springer, Berlin Heidelberg, 255, 2005. 19. R.L. Feller, in Accelerated Aging: Photochemical and Thermal Aspects, Ed. by D. Berland. The J. Paul Getty Trust, L.A. 45, 1994. 20. D. Anglos, Appl. Spectrosc. 55, 186A (2001). 21. J.J. Boon, Int. J. Mass Spectrom., Ion Process 118/119, 755 (1992). 22. R. Hoogerbrugge, S.J. Willig, and P.G. Kistemaker, Anal. Chem. 55, 1710 (1983). 23. W. Windig, J. Haverkamp, and P.G. Kistemaker, Anal. Chem. 55, 81 (1983). 24. V. Zaﬁropulos, A. Manousaki, A. Kaminari, and S. Boyatzis, in ROMOPTO: Sixth Conference on Optics, Edited V.I. Vlad, SPIE Vol. 4430, 181, 2001. 25. G.A. van der Doelen, PhD Thesis University of Amsterdam. 28, 1999. 26. A.V. Cunliﬀe and A. Davis, Polym. Degrad. Stabil. 4, 17 (1982). 27. T. Fukushima, Durability Build. Mat. 1, 327 (1983). 28. M.W. Formo, in Baile’s Industrial Oil and Fat Products Vol. 1, Ed. by D. Swern, John Wiley & Sons, New York, 722, 1979. 29. B. Luk’yanchuk, N. Bityurin, S. Anisimov, and D. Bäuelre, Appl. Phys. A 57, 367 (1993). 30. R. Srinivasan and B. Braren, Chem. Rev. 89, 1303 (1989).

30 A New Solution for the Painting Artwork Rear Cleaning and Restoration: The Laser Cleaning ∗

S.E. Andriani1 , I.M. Catalano1 , A. Brunetto3 , G. Daurelio2 , and F. Vona4 1

∗ 2

3

4

InterAteneo Physics Department “M. Merlin” of the University of Bari – InterDepartment Center “Search Laboratory for the Cultural Heritage Diagnostic” – Via Amendola 173, 70126, Bari, Italy [email protected] C.N.R.-I. N. F. M. – National Institute for the Physics of Matter – Regional Lab. L.I.T. Laser Innovation Technology Transfer and Training – c/o Physics Department “M. Merlin” of the University and Polytechnic of Bari, Via Amendola 173, 70126 Bari, Italy Private Restorer & Researcher in Art Restoration of “Restauri Brunetto di Brunetto Anna” ﬁrm, via Tormeno, 63, 36100 Vicenza, Italy Soprintendenza per il patrimonio Storico, Artistico e Etnoantropologici delle provincie di Bari e Foggia, Italy

Summary. Before restoring a painting, in order to assure a good level of adhesion between the canvas and the preparation layer or to reline the painting, it is often necessary to consolidate the canvas by intervening on the painting rear. Traditional cleaning techniques, chemical combined with mechanical ones, show an important drawback: The cleaning process and technique enfeeble permanently the canvas. The present work performs a comparative study for evaluating both the cleaning process eﬃciency and the canvas integrity preservation by using various cleaning methods, including Nd:YAG laser systems and traditional techniques. The eﬀects of a short free running mode (λ = 1,064 nm, pulse duration of 40–110 µs), a long Q-switched mode (λ = 1,064 nm, pulse duration of 200 ns) and a Q-switched mode (λ = 1,064 and 532 nm, pulse duration of 6 ns) of Nd:YAG laser irradiation on the hemp canvas of a seventeenth century painting are investigated. The analyses using FTIR spectroscopy and degradation mapping by optical microscope, with photographs taken before, during and after the cleaning process, were carried out. The work is still in progress.

30.1 Introduction Several physical techniques such as microanalytic methods, optical techniques and laser applications can be employed successfully for diagnosis and maintenance of cultural heritage. During the last few years, many laboratories have tested lasers to solve some cleaning problems. The excellent results of laser cleaning to remove unpleasant and harmful pollution layers from

258

S.E. Andriani et al.

many substrates, mainly limestone and marble, stimulated the researchers to try this method on various kinds of supports with diﬀerent natural or synthetic pigmented surfaces. Many experimental papers have demonstrated the mechanisms operating during laser cleaning and they give a reasonable understanding of the interaction of laser radiation with the complex structure of stone artworks [1, 2]. However, with a polychrome masterpiece, a new variable is introduced that should not be of underestimated and that is the presence of pigments and their related binders [3, 4]. The laser application on painted artworks is more demanding due to the high sensitivity of paint layers (diﬀerent pigments and binding media) to light. Amongst the many interesting challenges to be tackled is the discoloration of pigments upon laser irradiation [4,5]. Some laser application diﬃculties are met with restoration of the back surface of a painting support, where often there is an inhomogeneous polychrome layer. In fact, not only the painting needs restoring but also the support (the painting rear). Alteration layers, past consolidation treatments, adhesives applied on the rear surface, spillage of pigments and binders on the back all damage the rear surface of a painting as well as soil the front painting. In order to assure a good level of adhesion between the canvas and the preparation layer or to reline the painting, it is often necessary to consolidate the canvas by intervening on the back of a painting. Traditional cleaning techniques, chemical combined with mechanical ones, show an important drawback: the cleaning process and technique enfeeble irremediably the canvas. Here various Nd:YAG lasers with diﬀerent pulse duration: Q-switch (τ = 6 ns), long Q-switch (τ = 200 ns), short free running (τ = 40 µs), and diﬀerent wavelengths, fundamental and second harmonic (λ = 1,064 nm and λ = 532 nm) have been tested. Optical microscopy (OM) and stereomicroscopy with an image storage system (Nikon LUCIA) and Fouriertransform infra-red spectroscopy (FTIR) have been used to characterize and evaluate the composition of the untreated and the laser-irradiated canvas samples, as well as to assess the eﬀects induced by the laser radiation. The aims of this innovative research, still in progress, are: – The evaluation of the true eﬃciency of the diﬀerent laser sources employed in cleaning operations compared with traditional ones – The understanding of the principles, the threshold parameters and eﬀectiveness of the laser cleaning method – The research of the optimum working conditions for the materials to be removed, through the establishment of laser parameters (short free running, long Q-switch or Q-switch, wavelength, repetition rate and energy density) without causing damage to the canvas or preparation layer underneath the painting

30 A New Solution for the Painting Artwork

259

30.2 Experimental Methods The rear surface canvas specimens used in the present work derive from the “centre” (150 × 50 cm2 ) of the “S. Sebastiano curato dalle Pie donne” painting. This oil painting on canvas comes from the S. Chiara church, Bari (Puglia-South of Italy) and it dates back to the sixteenth century. The centre was probably an addition of the seventeenth century during the church restoration, in order to increase the painting size. The centre edges, creased and lacerated, showed holes brought by nails that had probably served for ﬁxing the painting on a loom. The back of the canvas sample (Fig. 30.1) shows some cuts, holes and tears. It is covered by a “mash” substance (“beverone”), brown coloured, composed of a mixture of desiccative oils and resins that was brushed on the back of the painting. When this mixture has been applied, it restored the paintings to its original brightness but, subsequently after some time, the same beverone has blackened and defaced the painting. Pigment traces (black and red coloured) and preparation layer (red and brown coloured) that have penetrated crossed the canvas weave of the painting are evident. The preparation and the pigments leakage “spillage” were so thick that they had been leveled, closing the “reading” of the weave in the past (Fig. 30.2).

Fig. 30.1. Centina sample

Fig. 30.2. Beverone and traces of the preparation layer and pigment leakage

260

S.E. Andriani et al.

Paint layer Preparation layer Canvas Mixture: Beverone+Pigment and preparation leakage Fig. 30.3. Typical painting stratigraphy

A schematic cross-section representation of a typical painting is shown in Fig. 30.3, in which a brown oily, resinous layer (beverone) is reported. The usual cleaning process of the rear surface of the canvas is mechanical/chemical and manual, made with the use of emery paper and/or scalpel and, where the alterations are more tenacious, at ﬁrst dampened with H2 O and immediately after “scraped oﬀ” with a scalpel or then dampened with chemical solvents, then scraped with scalpel and buﬀered with alcohol. 30.2.1 Laser Systems and Experimental Procedures The irradiation experiments were carried out with diﬀerent laser systems. Three portable Nd:YAG laser sources called SMART CLEAN, VARIO (produced by EL.EN. Group, Florence, Italy) and PALLADIO (produced by Quanta System-EL.EN. Group, Florence, Italy) were employed for the sample laser cleaning operations. The ﬁrst produces up to 1 J per pulse in a 40-µs pulse duration (SFR), at single shot or 5–10 or 20 Hz repetition rate. The second laser (VARIO) emits up to 350 mJ per pulse in 200 ns (LQS) pulse duration at single shot or 5–10 Hz repetition rate. Both lasers, using only the fundamental wavelength (λ = 1,064 nm), are equipped with an optical ﬁbre delivery system of diﬀerent lengths and use a 1-mm core guiding ﬁbreglass. The ﬁbres terminate with an optical manipulator which permits adjustment of the spot size and ﬂuence (J cm−2 ) onto the irradiated surface. Due to the beam delivery via an optical ﬁbre, the spatial energy proﬁle has a top-hat shape and poses a lower risk for hot spots, thermal eﬀects and discoloration on the surface. The PALLADIO laser system is a Q-switched source that produces up to 450 mJ per pulse in 6 ns pulse duration (QS), at single shot or 5–10 or 20 Hz repetition rate. It oﬀers two choices for the wavelength (1,064 and 532 nm). The laser system delivers the beam through a seven mirror, multi-articulated arm with a handpiece equipped with a focusing lens in order to adjust the spot size and homogeneity. The applied energy densities were calculated from the laser energy measured with a calorimeter (power meter) and the size of the circular laser spot monitored with a thermal sensitive paper. Optical microscopy and stereomicroscopy (OM) with a storage image system (LUCIA) and Fourier-transform infra-red spectroscopy (FTIR) have

30 A New Solution for the Painting Artwork

261

been used in order to characterize the surface to evaluate the composition of the untreated and laser-irradiated canvas samples, as well as to assess the eﬀects induced by the laser radiation. OM (optical microscopy, Nikon ECLIPSE E80i/RT and stereoscopic NIKON Model SMZ-800) was performed on the canvas sample before and after the laser irradiation. FTIR (Perkin Elmer 1600) was employed to determine the inorganic and organic compounds present in the beverone and to evaluate the eﬀects of laser–beverone interaction. For the FTIR analysis, 1 mg of sample was homogenized with 0.2 mg of KBr, and the disks formed were examined in an FTIR spectrophotometer in a transmittance range between 4,000 and 400 cm−1 . 30.2.2 Samples Twelve diﬀerent adjacent zones (4 × 10 cm2 , Fig. 30.4) were selected on rear painting in order to compare chemical, mechanical and laser cleaning eﬃciency as well as drawbacks. FTIR analyses have revealed probably the composition of the beverone: linseed oil and/or animal glue (ν(CH2 ) 2,926 cm−1 , 2,853 cm−1 ), pigment in media and soiling. FTIR analysis was not able to identify the composition of the layer; perhaps the preparation of powder sampling is not the suitable analytical technique in this case. However, the diﬃculty of sampling a homogenous section of sample within even an extremely heterogeneous mash of pigment, media, glue and oil must not be ignored. The basic condition for laser cleaning to be eﬀective without underlying substrate destruction is that the soiling ablation threshold is signiﬁcantly

Fig. 30.4. Painting rear details with diﬀerent test sections: Zones (1, 2, 3, 4, 5, 6, D, C) are laser cleaning areas; Zones (A, B, E and H) employed mechanical–chemical cleaning methods

262

S.E. Andriani et al.

lower in comparison with that of substrate. However, the complexity of the canvas rear alteration layers and the sequence of diﬀerent alteration layers as shown, step-by-step, with OM observations, represent a serious and diﬃcult problem to the application of the laser cleaning process. In order to prevent the direct interaction of the laser beam with light-sensitive paint materials (including pigments and binders) and the risk of excessive material removal (superﬁcial, or even structural, damage), diﬀerent laser sources and exposure parameters (wavelength, energy, pulse duration, spot size, dry/wet cleaning) were tested. Samples: Sections A, B2, H, E The sections A, B2 , H, E (Fig. 30.5) of the hemp canvas were cleaned with traditional cleaning techniques (chemical and mechanical ones) such as emery paper, scalpel and water, scalpel and solvent (1:2 Amile:metilformalamide) and scalpel and emery paper. The use of a scalpel or an emery paper gives a complete superﬁcial cleaning such that all alteration layers were removed even if they generate some canvas mechanical damage. In fact, as shown in Fig. 30.5 (Zone E), the scalpel/emery paper erosion mechanism has not discriminated beverone from canvas so it has completely removed the layers but, at the same time, it irremediably damaged the weave and has not cleaned the weave interstices. The scalpel has levelled the surface, making it unsuitable for the painting relining. Even if scalpel and water are applied, in order to take advantage of the water solvent action, the result is almost the same (Fig. 30.5, zone B2 ). Chemical treatments rely on chemical reactions with the alteration layers; the solvent action gives a more cleaning easiness. In fact, after the chemical reaction, the use of a “soft” abrasive mechanism (scalpel) preserves the canvas weave better as shown by the comparison in Fig. 30.5, zone A–H. Anyway, after a chemical treatment, it is recommended to buﬀer with alcohol in order to interrupt the chemical solvent penetration and relative chemical reaction, preserving the painting layer on the front of the painting. So apply-

B2

A-H

E

Fig. 30.5. Identiﬁcation of zones. B2: scalpel + water; A: emery paper; H: scalpel + chemical solvent; E: (Left) emery paper; (Right) scalpel

30 A New Solution for the Painting Artwork

263

ing this methodology, it is impossible to get a real control of the chemical action into the painting. Samples: After Laser Treatments, Sections 1–6 Some diﬀerent tests, executed by laser, are described next. No examples using the SFR laser are in the following descriptions because this laser showed a high thermal eﬀect that blackened the canvas irremediably. In Table 30.1, the diﬀerent zones with the laser systems employed are correlated. The absorption of the laser radiation and the induced eﬀects depend on the complex stratigraphic composition and inhomogeneity of chromatic characteristics of the beverone. As a result of this complexity, for each individual case studied, some results for the surfaces as well as the laser operating parameters and various experimental conditions are reported on the following. Sections 1 and 2 The working laser parameters are: Vario (τL = 200 ns, φ = 3 mm, ν = 10 Hz, E = 300 mJ, F = 4.2 J cm−2 ) in zones 1a and 2a. The cleaning of the beverone is not complete, in fact, Fig. 30.6a and, in detail, Fig. 30.6b, shows the weave canvas still obstructs so this laser was ineﬃcient on this thick beverone layer. In zone 1b, after the ﬁrst unsatisfactory Vario laser cleaning result with E = 250 mJ, the remaining area was test cleaned with another laser system: Palladio (λ2 = 532 nm, τL = 6 ns, φ = 3 mm, ν = 10 Hz, E = 200 mJ, F = 2.8 J cm−2 ). The results show a complete surface and interstices cleaning. The following OM observation, Fig. 30.6c, reveals the eﬀects of QS λ2 laser cleaning: a canvas ﬁbre abrasion. Section 3 The working laser parameters are Palladio (λ1 = 1,064 nm, τL = 6 ns, φ = 4 mm, ν = 10 Hz, E = 110 mJ, F = 0.8 J cm−2 ). Figure 30.7 shows the complete laser cleaning of canvas and the absence of interstices residues. This Table 30.1. Diﬀerent canvas sections corresponding to laser system used: Short free running (SFR), long Q-switch (LQS), Q-switch with λ = 1,064 nm (QS λ1), Q-switch with λ = 532 nm (QS λ2) and xn , where n indicates the number of laser treatments laser/section

D a

SFR LQS QS λ1 QS λ2

1 b

a

b

2 a

x

x1

x

2

x

x x1

3 a

4 b

5 wet

x 2

x

x

x

x

6

a

b

x

x

a

b

x

x1 x2

264

S.E. Andriani et al.

a

b

c

Fig. 30.6. Sections 1 and 2: (a) zones 1 and 2; (b) zone 1a, after (on the left) and before (on the right) laser treatments; (c) zone 1b, particulars of some canvas ﬁbres. Magniﬁcation, ×10

a

b

Fig. 30.7. Section 3: (a) canvas during the cleaning; (b) microimage (×10) on some canvas weaves, completely cleaned

section represents the best laser exposure parameters for the painting rear cleaning. Section 4 The three zones of Section 4 are irradiated with the same laser system: Palladio (λ2 = 532 nm, τL = 6 ns, φ = 4 mm, ν = 10 Hz) but with diﬀerent laser energy (ﬂuence) values in order to resolve some problems that appeared after cleaning. In fact, in zone 4a, after laser cleaning with E = 120 mJ(F = 1 J cm−2 ) the canvas was completely clean (Fig. 30.8) but a discoloration appears. It seems that the laser, by the ablation process, is responsible of the greenish colour eﬀect (Fig. 30.8I). Other laser tests showed that the greenish colour eﬀect occurs only after encrustation/soiling removal and not after the tests laser irradiation of canvas without beverone. A possible explanation is the physicochemical or thermal alteration of the remaining surface due to a high energy laser beam. By other tests using energy E < 120 mJ, no ablation eﬀects are evident; so it is possible to assume E = 120 mJ is the energy threshold for λ2 = 532 nm. FTIR analysis did not support the physicochemical hypothesis since, as is shown in Fig. 30.9, no modiﬁcation of the layers

30 A New Solution for the Painting Artwork

265

4w 4a I

II

4b

III

Fig. 30.8. (Left) Zone 4 using the Palladio laser, λ2 = 532 nm. (I) Zone a, ×20; (II) Zone b, ×10; (III) Zone w, ×10 magniﬁcation

Fig. 30.9. FTIR spectrum of black and red zones before and after the laser cleaning with λ2 = 532 nm and λ1 = 1,064 nm. No chemical modiﬁcation of the compounds is evident

266

S.E. Andriani et al.

composition are evident. Other hypotheses for the greenish eﬀect are the deposition/redeposition of residues or laser ablation products or residues due to incomplete cleaning. In Zone 4b, if the colour canvas change was due to the incomplete laser cleaning, E = 200 mJ, φ = 3 mm(F = 2.8 J cm−2 ) was used for a few round trips on the zone with the purpose to prevent the greenish eﬀect. This solution did not resolve the problem but a more greenish and abrasive eﬀects on canvas (probably due to a high ﬂuence), were observed (Fig. 30.8II). Another possible solution was tested in Zone 4w: E = 120 mJ, φ = 4 mm(F = 1 J cm−2 ) with some distilled water on the encrustations before irradiating (wet laser cleaning). The water enhanced the cleaning eﬀect, so it is important to wet the surface in those cases where the energy density and number of pulses must be kept to a minimum value to avoid possible substrate damage. At the ﬁrst time, it seems that the water prevents the greenish eﬀect and the canvas abrasion (Fig. 30.8III), perhaps thanks to the lower energy density (F) combined with the water solvent action. However these exposure parameters were ineﬀective with thick beverone island layers. It was also diﬃcult to prevent water penetration/adsorption into the painting and its harmful solvent action. Microimages (Fig. 30.8) were not suﬃcient to ﬁnd a possible explanation of the greenish eﬀect and the work is still in progress. Section 5 The working laser parameters are Vario (τL = 200 ns, φ = 3 mm, ν = 10 Hz) in Zone 1a E = 200 mJ(F = 2.8 J cm−2 ), E = 250 mJ(F = 3.5 J cm−2 ) in Zone 2a. The cleaning results are the same for Section 1a and 2a; no diﬀerences are evident. Section 6 The working laser parameters are Palladio (τL = 6 ns, φ = 3 mm, ν = 10 Hz, λ1 = 1, 064 nm) E = 250 mJ in Zone 6a, and the same in Zone 6b (E = 250 mJ, F = 3.5 J cm−2 ). The cleaning of the beverone is not complete perhaps because of a rapid movement of the optical manipulator due to high ﬂuence employed. Figure 30.10I shows the canvas weave still obstructed and thick beverone red islands remaining. In Zone 6b, after the ﬁrst unsatisfactory laser cleaning (executed by Palladio laser) with the same parameters of Zone 6a, we tried to remove the remaining islands with another laser type: Palladio (λ2 = 532 nm, τL = 6 ns, φ = 3 mm, ν = 10 Hz, E = 150 mJ, F = 2.1 J cm−2 ). The results show a complete surface and interstices cleaning. However, the OM observations reveal the eﬀects of QS λ2 laser cleaning: clearer areas and darker areas, probably due to microabrasion or chromatic alteration of canvas ﬁbres.

30 A New Solution for the Painting Artwork

I

267

II

Fig. 30.10. (I) Zone 6; (II) zone D

Section D The working parameters in Zones Da and Db, were the Palladio laser source (λ2 = 532 nm, τL = 6 ns, φ = 3 mm, ν = 10 Hz, E = 180 mJ, F = 2.5 J cm−2 ) but only with one or two round trips. After cleaning, the greenish eﬀect appears evident (Fig. 30.10II). In Zone Db, another laser irradiation by Palladio (τL = 6 ns, φ = 4 mm, ν = 10 Hz, λ2 = 1, 064 nm, E = 200 mJ, F = 1.6 J cm−2 ) was performed. It is important to underline that the Zone Db, after the second laser treatment, did not show the green eﬀect and the possible explanations are still under investigation.

30.3 Conclusions The behaviour of the cleaning method based on diﬀerent Nd:YAG laser sources gives diﬀerent results depending on the type of laser system and exposure parameters. These preliminary tests show the eﬀectiveness of laser cleaning method comparing with traditional ones and they individualize the optimum working conditions for the substances to be removed. The correct laser parameters by using Palladio source (Q-switch, λ1 = 1, 064 nm, τL = 6 ns, φ = 4 mm, ν = 10 Hz, E = 110 mJ) without causing damage to the canvas and to the preparation layer underneath the paint (front of canvas) are obtained in Section 3. On laboratory tests, some drawbacks were observed due to the incomplete cleaning or to the unknown greenish eﬀect. Further investigations

268

S.E. Andriani et al.

should be done on the cause of greenish eﬀect and on its dependence of wavelength since in Section 6 (Q-switch, λ1 = 1,064 nm) no greenish eﬀect is observed with respect to Section 4 (Q-switch, λ2 = 532 nm). It is important to underline that in all examined experimental laser setups, no alteration or loss of painting fragments are observed on the front of the painting. At least in any case where a laser cleaning process was applied, it is extremely important to ﬁnd “safety” limits for the application of these cleaning methods to prevent irreversible damage on the artworks. Acknowledgements The authors wish to thank Restorers of the Soprintendenza per il Patrimonio Storico, Artistico ed Antropologico delle Province di Bari e Foggia Laboratory for their precious collaboration and help in carrying out the traditional cleaning techniques.

References 1. M. Cooper, Laser Cleaning in Conservation – An Introduction. 2. M. C. Gaetani and U. Santamaria, in Journal of Cultural Heritage, Vol. 1, 199, 2000. 3. A. Sansonetti and M. Realini, in Journal of Cultural Heritage, Vol. 1, 189, 2000. 4. C. Theodorakopoulos, V. Zaﬁropulos, et al. in Lacona V Book of abstracts, 72, 2003. 5. A. Brunetto “L’utilizzo della strumentazione laser per la pulitura delle superﬁci nei manufatti artistici” Edizioni il Prato.

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints by Laser Irradiation at IR, VIS and UV Wavelengths M. Westergaard1,2 , P. Pouli3 , C. Theodorakopoulos3,4 , V. Zaﬁropulos3,5 , J. Bredal-Jørgensen1 , and U. Staal Dinesen1 1 2

3

4

5

School of Conservation, Royal Academy of Fine Arts, Copenhagen, Denmark Art Conservation Centre, Kronborg Castle, Elsinore, Denmark [email protected] Foundation for Research and Technology-Hellas (FO.R.T.H.), Institute of Electronic Structure and Laser, Heraklion, Crete, Greece RCA/ V&A Conservation, School of Humanities, Royal College of Art, Kensington Gore London SW7 2EU, UK Laboratory of Applied Physics, Human Nutrition & Dietetics, Technological Educational Institute of Crete, Ioannou Kondylaki 46, 723 00 Sitia, Crete, Greece

Summary. This study aims to investigate whether laser cleaning may be a valuable method for the removal of soiling from water-based acrylic emulsion paints in comparison to traditional cleaning methods. Acrylic-grounded canvas was painted with three diﬀerent paints (yellow ochre, titanium white and red alizarin) in a polybutyl-acrylate and methyl methacrylate binder. An acrylic binder was used as a reference. The samples were covered with carbon, SiO2 and soot. Cleaning process ablation rate studies were carried out with a Q-switched Nd:YAG laser at 1,064, 532 and 355 nm and a KrF Excimer laser at 248 nm. The energy densities varied from 0.03 to 0.69 J cm−2 . The irradiated tests at 248 nm were monitored by LIBS analysis. On the samples irradiated at 1,064 nm, various analytical methods were carried out. A determined alteration of the titanium white paint resulted in a marked decrease in the glass transition temperature (Tg ). Furthermore, discoloration (yellowing) occurred on the binder and the titanium white paint. The ochre darkened slightly but the alizarin was unchanged. When compared with the samples cleaned with water-based solvents, the samples cleaned with laser appeared cleaner. However, SEM/EDX and ATR showed that SiO2 was still present on the surface after laser cleaning at the tested conditions.

31.1 Introduction The aim of these experiments was to evaluate whether laser cleaning of soiled water-based acrylic emulsion paints could turn out to be a better alternative to a traditional cleaning with water-based solvents, known among conservators to be quite complicated.

270

M. Westergaard et al.

Water-based acrylic emulsion paints are fast-drying paints with a low glass transition temperature (Tg ) [1]. The softness of these paints may lead to the surface attracting dust particles that can become embedded, and thus diﬃcult to remove. Furthermore, the surface of the paint may come oﬀ using traditional water-based solvents, depending on the content of delicate pigments and the amount of dissolving detergents and pigment dispersing agents [2]. In addition, it may be altered mechanically by the movements of the cotton swab when cleaning or just by the grease from ﬁngerprints. A non-contact cleaning method, like laser cleaning, would be preferable. The ablation of varnish on oil paintings at UV wavelengths is a welldocumented and recognized cleaning method [3, 4]. On the contrary, to the authors’ knowledge, there is no documented comparison between surface cleaning of acrylic emulsion paints executed with laser vs. with traditional solvents. Fourrier et al. [5] reported on some good results in relation to UV laser ablation of particles (SiO2 ) from polymers contrary to Real et al. [6] who did some experiments with UV laser ablation of soot, among other materials, from diﬀerent acrylic surfaces. At too high energy densities, the surface went oﬀ while at lower energy densities, the cleaning lacked in suﬃciency. Furthermore, a grey discoloration was observed on the white acrylic primers. A considerable number of previous experiments with laser irradiation, mainly of inorganic pigments, have been executed since the mid 1990s and, in general, these pigments discolour. Some of the pigments, like white lead, red lead and zinc white, will darken only temporarily, that is, from a couple of hours up to a whole week after the laser irradiation. Other pigments like cinnabar decompose and permanently darken [7,8]. According to publications, as by Chappé et al. [9], it is notable that 1,064 nm seems like the wavelength at which an ablation threshold can be found in relation to laser irradiation of inorganic pigments, like zinc white, red lead, ochre, sienna, azurite, malachite, cobalt blue, ultramarine blue, bone black and also the organic alizarin. In this study we intended to get preliminary results on laser ablation of artiﬁcial dust on acrylic emulsion paint. The test materials were chosen with colours of a well-known brand on a primed canvas. The artiﬁcial soiling consisted of known components in a dust layer executed according to the contents of ordinary museum dust.

31.2 Experimental Methods To determine the ablation threshold and evaluate the self-limiting character of the laser cleaning process, ablation rate studies were carried out with a Q-switched Nd:YAG laser at 1,064, 532 and 355 nm and a KrF Excimer laser at 248 nm. The irradiated tests at 248 nm were monitored by Laser induced breakdown spectroscopy (LIBS) analysis. The number of pulses and application of wet or dry cleaning was evaluated and the energy densities varied from 0.03 to 0.69 J cm−2 . Pilot tests were executed at all four wavelengths

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints

271

Table 31.1. Composition of the binder, acrylic paints and soilings 1. titanium white paint (TiO2 ) 2. yellow ochre paint (FeOOH) 3. alizarin paint (1,2-dihydroxyanthraquinone) 4. binder presumably: polybutyl-acrylate/methylmethacrylate P(nBA/MMA) [11]

1. 3.33 g carbon (lampblack) + 6.66 g SiO2 + 0.01 g soot (from beeswax) 2. carbon (lampblack) 3. SiO2 4. 3.33 g carbon (lampblack) + 6.66 g SiO2

in order to test the cleaning eﬀects within the UV, VIS and IR regions. The most promising results of these pilot/preliminary experiments led to the ﬁnal and more detailed experiments at the most promising wavelength. Analyses were executed on the irradiated samples after the ﬁnal experiments: Colour measurements (CIE–L∗ a∗ b∗ ), diﬀerential scanning calorimetry (DSC), scanning electron microscopy (SEM), energy dispersive X-ray analysis (EDX) and infrared spectroscopy/attenuated total reﬂectance (ATR). An evaluation of laser cleaning compared to traditional cleaning with selected solvents on the same samples was also executed. Four primed canvases were each painted with one of the three diﬀerent acrylic paints and with the acrylic binder only (all of the brand “FINITY” from Winsor & Newton). The primer consisted of washed chalk, titanium white pigment and an acrylic binder. The samples were then soiled with artiﬁcial dust (see Table 31.1). The dust was composed according to previous studies [10] on the composition of museum dust concluding that the components are mainly earth (silicates) and carbon/soot and may have a greasy consistency. Laser cleaning tests were carried out on all four reference groups (without dust) and on all three types of paints and binder, each covered by all four types of dust. On the same samples, cleaning with water-based solvents was also carried out. The lasers used in this study were (1) A BMI Q-switched Nd:YAG (series 5022 DNS 10) emitting at 1,064, 532 and 355 or 266 nm. Its pulse duration was in the range of 5–7 ns and the pulse repetition rate could vary from 1 to 10 Hz. (2) A Lambda Physics KrF Excimer (COMPex 110), emitting at 248 nm with a pulse duration of 30 ns and repetition rate up to 100 Hz. The choice of traditional solvents was based on an enquiry among conservators in DK and on an investigation carried out among 29 conservators in Canada in 2002 [12]. It was decided to use two water-based solvents: (1) water (1 dl) added a detergent (“Agepon”) (10 drops) and (2) water added 6% tri-ammonium citrate. Six pilot irradiations were executed on each type of sample at each wavelength and the size of the ablations was 1 × 1.5 cm. The energy densities used with the Nd:YAG at the fundamental wavelength were ranging from 0.1 up

272

M. Westergaard et al.

Table 31.2. Parameters at the further experiments with laser irradiation executed at 1,064 nm Experiment A B a

Energy (mJ)

Spot sizea (cm2 )

81.1 81.1

0.47 0.47

Energy density (J cm−2 ) 0.15 0.15 + H2 O

No. of pulses per square

Hz

40 10–15

2 2

(0.8 cm × 0.75 cm × π/4)

to 0.64 J cm−2 and 1–40 pulses were applied. At 532 nm, the energy densities were in the range of 0.03–0.69 J cm−2 and 1–5 pulses were applied, while energy densities from 0.15 up to 0.26 J cm−2 and 1–10 pulses were used at 355 nm. At 248 nm, 1–7 tests were carried out on each type of paint (two on acrylic binder, four on titanium white paint, three on ochre and seven on alizarin) and the energy densities varied from 0.1 to 0.38 J cm−2 with a beam overlap of 80–90%. Based on the results of these pilot tests, the ﬁnal irradiations were carried out using the 1,064 nm wavelength. For the ﬁnal laser cleaning test, the parameters (“A” in Table 31.2) of most satisfying preliminary tests (evaluated under the microscope) were repeated in a larger area. However, the cleaning eﬀect was insuﬃcient and a thin water ﬁlm was applied prior to the irradiation (“B” in Table 31.2) leading to a much better eﬀect. Apparatus applied for the analyses (1) Optical evaluations. An Axiotech optical light stereomicroscope (Zeiss), ﬁtted with diﬀerential interference contrast (DIC) capability. (2) Colour measurements (CIE–L∗ a∗ b∗ ). A Konica Minolta CM 2600d Spectrophotometer, double rays with Xenon ﬂash as a light source. The measuring geometry was D/8, the specular component was excluded and the lighting was D65 . (3) Diﬀerential scanning calorimetry (DSC). A thermal analysis (TA) instruments DSC Q 1000, coupled with a liquid nitrogen cooling system (LNCS). (4) and (5) Scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDX). A JEOL JSM 5310, LV vacuum scanning electron microscope. Data processing was executed with the analysis programme LINK ISIS from Oxford. (6) Infrared spectroscopy/attenuated total reﬂectance (ATR). A spectrum one, FT-IR spectrometer coupled with a Perkin–Elmer universal ATR sampling accessory with the use of a diamond crystal. Spectra collections and calculations were executed with PE Spectrum software.

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints

273

31.3 Results and Discussion 31.3.1 Visual Observations Even at energy densities as low as 0.1 J cm−2 at 248 nm and 0.15 J cm−2 at 355 nm, discoloration occurred on the binder sample (turned greyish) as well as on titanium white (turned blue-greyish) and ochre (turned brownish). The organic alizarin paint was cleaned without any discoloration at 248 nm (with an energy density of 0.22 J cm−2 applying 1 pulse and 80% overlap for the scanning), and at 532 nm (with an energy density of 0.17 J cm−2 , adding water and 2–3 pulses). Generally, no discoloration occurred on alizarin paint apart from a barely visible bleaching of the reference (no dust) at 248 nm. The tests at 532 nm on the binder, the ochre and the titanium white paint either led to an incomplete cleaning at 0.03 J cm−2 or a more complete cleaning (at an energy density of 0.17 J cm−2 ) but leading to the abovementioned discolorations in the binder, titanium white paint and ochre paint (Fig. 31.1). At 1,064 nm, 40 pulses and an energy density of 0.15 J cm−2 resulted in a much more satisfactory cleaning eﬀect. No discoloration of the binder and the alizarin paint was observed but a slight bluish discoloration of the titanium white paint and a very vague greenish tone on the ochre could be detected. When adding a thin water ﬁlm to the surface before irradiation at 1,064 nm and using an energy density of 0.15 J cm−2 , the cleaning eﬀect seemed complete with only 10–15 pulses and no discoloration could be observed with the naked eye. However, after about a week in the dark, the surface of the titanium white and the binder became somewhat yellow, mostly on samples from which carbon-containing dirt had been removed. The yellowing was more

Fig. 31.1. Microscope image of titanium white paint after laser ablation of carbon and SiO2 at 532 nm with an energy density of 0.17 J cm−2 using 2–3 pulses per irradiated square on a wetted surface. Discoloration of the paint and remains of dirt in the surface can be observed

274

M. Westergaard et al.

pronounced on the titanium white paint. This yellowing bleached to some degree when exposed to daylight for 24 hs. Visual observations indicated that the best results were obtained by cleaning at 1,064 nm, with an energy density of 0.15 J cm−2 , 10–15 pulses and application of a water ﬁlm. By means of the above-mentioned equipment, these results from the test run were analysed. 31.3.2 Colour Measurements Colour measurements showed that the ochre and alizarin paint became somewhat darker, diminishing their L-value by 3.36 and 3.28, respectively, when ablating carbon and SiO2 . This may be due to residues of carbon on the surface. The titanium white reference sample (no dust) also became a little darker after the irradiation, reducing its L-value by 0.63. Laser removal of dust however led to higher b-values indicating a yellowing. b-Values were higher on the titanium white paint than on the binder. Furthermore, these b-values were increasingly higher proportional to the carbonic content of the ablated dust layer, e.g. the ablation of pure carbon on the titanium white paint led to the highest increase of the b-value (+4.79) compared to the increase of the b-value (+0.48) measured after ablation of SiO2 . The discoloured (into yellow) binder and titanium white paint did not bleach/recover completely after exposure to daylight. The discoloured binder after ablation of carbon, SiO2 and soot was bleached/recovered showing a shift in b-value from 3.80 to 2.34. On the titanium white paint after ablation of carbon, SiO2 and soot, the bleaching reduced the yellowing from 4.89 to 2.57 in b-value. As the b-values of the references of the binder and the titanium white paint before laser irradiation were 0.34 and 1.09, respectively, the surfaces were still somewhat yellow. 31.3.3 DSC Measurements It show that the Tg of the three acrylic paints and the binder before laser cleaning ranged from 10.58 to 13.64 ◦ C. This was expected as the Tg of related polymers range from 5 to 17 ◦ C [11,13,14]. The Tg of the binder reference was found to be 11.84 ◦ C; while for the titanium white paint reference it was measured to be 10.58 ◦ C. On the ochre and alizarin acrylic paint reference samples, the Tg was found to be higher, 13.64 and 13.22 ◦ C, respectively. After the laser cleaning at 1,064 nm, only the samples with organic materials, that is, the binder and the alizarin paint were found to have slightly higher Tg : 1.08 and 2.06 ◦ C, respectively. However, the Tg of the paints containing inorganic pigments, namely ochre and titanium white, decreased: ochre paint by 3.09 ◦ C and the titanium white remarkably by 18.60 ◦ C, see Fig. 31.2.

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints Sample: Tio2 -1064nm Size: 4.0000 mg Method: Heat/Cool/Heat Comment: Tio2, 1064 nm 1.0

275

File: C:\TA\Data\DSC\kcs\Tio2-1064nm.001 Operator: kcs Run Date: 06-Mar-03 12:14 Instrument: DSC Q1000 V6.19 Build 227

DSC

Heat Flow (mW)

0.5

0.0

−0.5

−9.60C

−8.02C(1) 0.1043J/g/C(1)

−6.98C −1.0

−1.5 −50

−40

−30

−20

Exo Up

−10

0

10

20

Temperature (C)

30

40

50

60

70

80

Universal V3.5B TA Instruments

Fig. 31.2. DSC measurements on titanium white acrylic emulsion paint after laser ablation of simulated dust at 1,064 nm, with an energy density of 0.15 J cm−2 , 10–15 pulses and application of water. The curve is indicating a decrease in the Tg to −8.02◦ C from 10.58◦ C

31.3.4 SEM SEM was executed on a thin cross section from the yellowed titanium white paint after laser cleaning carbon, SiO2 and soot (at 1,064 nm with an energy density of 0.15 J cm−2 and application of water), but neither in the pigment grains nor in the binder were any morphological changes found. Likewise, no morphological changes were found on the surface neither in the pigment grains of the samples ablated at 1,064 nm. 31.3.5 EDX Mapping EDX mapping on samples after laser ablation of SiO2 alone or of the mixture of carbon, SiO2 and soot showed that SiO2 residues were found on both. When wet laser ablation was performed at 1,064 nm, with an energy density of 0.15 J cm−2 and 10–15 pulses, less SiO2 was found. After the traditional cleaning with water-based solutions, no morphological changes were observed on the surfaces of the samples. 31.3.6 ATR (FT-IR) ATR was executed after the laser cleaning at 1,064 nm but neither on reference samples nor on samples originally coated with carbon, SiO2 and soot any

276

M. Westergaard et al.

change in the binder was observed. No changes could be detected in the pigments either. As regards the traditional cleaning, with a detergent and with triammonium citrate, no changes in the chemical combination of the binder occurred. After wet laser cleaning, residues of SiO2 could still be detected at 1,114 cm−1 , particularly on the binder and the ochre paint samples. No residues of carbon could be observed. After traditional cleaning of the binder and paints, both SiO2 and carbon were detected at 1,114 cm−1 . However, the SiO2 content was higher after laser cleaning than after the traditional cleaning leading to the observation that, under the applied conditions, the laser was less eﬃcient than the traditional cleaning when ablating SiO2 . The results of the analyses indicate that the binder absorbed the irradiation and turned greyish in all the pilot tests except for dry cleaning at 1,064 nm with an energy density of 0.15 J cm−2 and 40 pulses. This could be due to a discoloration of the underlying primer containing titanium white paint turning blue-greyish above the threshold at all four wavelengths in the pilot tests, and to the transparency of the acrylic binder. Ochre paint turned brownish at 248, 355 and 532 nm, indicating transformation into hematite and black magnetite [15]. However, at 1,064 nm (0.15 J cm−2 and 40 pulses), ochre turned greenish, presumably caused by black magnetite mixed with the yellowish of the ochre. When water was applied before irradiation at 1,064 nm, ochre turned slightly darker as revealed from the colour measurements. The permanent grey-bluish discoloration of the titanium white paint at the pilot test is hardly due to a pollution of the pigment with zinc white as Chappé et al. [9] mention that the discoloration of the zinc white is only temporary. The yellowing of the titanium white and the binder appeared on samples ablated at 1,064 nm after cleaning of dirt containing carbon and when applying water. After irradiation also reference samples were kept in the darkness and exposed to temperatures around 30◦ C and they did not show any yellowing. Consequently, the high temperature and the darkness can be excluded as factors causing the yellow discoloration of these polymer paints. According to Strlic et al. [16], the degree of yellow discoloration of various organic surfaces rose in accordance with the thickness of the layer of carbon. But to the author’s knowledge, the application of water (in combination with ablation of carbon) has not yet been reported to induce discoloration. One theory, which is applied in relation to the remaining yellowing patina after laser cleaning of stonework, is relying on changes in the refractive index due to the creation of voids in the surface [17]. However, the indications up to now show that this theory is not applicable on acrylic paints. Firstly, the yellow discoloration recovered to some extent when exposed to daylight and secondly, no morphological changes were noted with SEM, nor did ATR show any alterations in the composition of the acrylic binder after the laser treatment. A possible explanation to the yellowing is that the titanium white pigment particles (in the paint and primer) as well as the carbon particles act as heat sources to the surrounding organic medium. The temperature in the paint can rise signiﬁcantly and trigger a quick decomposition of the organic binder. This

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints

277

decomposition might very well lead to a creation of chromophores and thus a yellowing of the material. Water helps the heat transfer from the particles to the organic binder. Chromophores bleach when exposed to daylight, and so did the examined discoloured samples of binder and titanium white paint. The degree and speed of the polymer recovery is dependent on the Tg . The lower the Tg , the quicker the recovery process. An indication of the decomposition triggered by the titanium white was the remarkable decrease of Tg by 18.60◦ C, leading to a pronounced softness of the paint, due to broken chains in the molecular structure. The binder, on the other hand, turned somewhat yellow too, but increased in Tg by 1.08◦ C. This could indicate that the decomposition and the yellow discoloration were not related. Further tests including more paints containing other pigments should be executed at 1,064 and 2,940 nm with an Er:YAG laser too, as the best results were obtained in the IR area. deCruz et al. [18] reported on good results concerning the cleaning of soot from oil paintings with short pulses and application of water. Also, Bracco et al. [19] ablated varnish from oil paintings at this wavelength. There is a possibility that the lower energy at 2,940 nm combined with the maximum absorption of water at this wavelength might lead to “steaming oﬀ” of acrylic emulsion paints without further heating and therefore without any discoloration. When comparing the cleaning eﬀect of laser cleaning at 1,064 nm (with an energy density of 0.15 J cm−2 , 10–15 pulses and adding water) with the cleaning eﬀect of the traditional cleaning method (triammonium citrate or a detergent in an aqueous solution) by means of optical observations, it was obvious that the samples cleaned with laser seemed much cleaner than those cleaned traditionally. The detergent (Agepon) seemed more eﬃcient than triammonium citrate. Samples traditionally cleaned from carbon were poorly cleaned, while those cleaned from only SiO2 seemed perfectly cleaned. SEM and FT-IR eventually proved the latter to some extent. The same analysis showed that more SiO2 residues could be detected on the laser-cleaned samples.

31.4 Conclusions With the given parameters, 248 and 355 nm, seemed unsuitable for the laser cleaning of acrylic binder, titanium white paint and ochre paint, due to discoloration and to an insuﬃcient cleaning eﬀect. The alizarin paint seemed laser safe, however, and was cleaned without any discoloration at 248 nm, with an energy density of 0.22 J cm−2 , 1 pulse and 80% overlap. Apparently, 532 nm was not suitable for the laser cleaning of acrylic binder, titanium white paint and ochre paint, due to discoloration and insuﬃcient cleaning eﬀect. Optical evaluations indicated that this wavelength and the applied parameters (at an energy density of 0.17 J cm−2 , 2–3 pulses and application of water) seemed adaptable for the cleaning of alizarin paint. At 1,064 nm and an energy density

278

M. Westergaard et al.

of 0.15 J cm−2 , 10–15 pulses and application of water, the cleaning eﬀect was satisfactory but colour measurements showed a slight darkening of the ochre and, after a week kept in the dark, the binder and titanium white had turned yellowish. The yellowing was most pronounced on the titanium white. The yellowing was partly recovered when exposed to daylight and is presumably due to the creation of chromophores. On the titanium white paint samples, the catalysing TiO2 pigment probably intensiﬁed the oxidation process while the transparent binder could be optically inﬂuenced by the TiO2 content in the underlying primer. The catalysing eﬀect of the TiO2 probably contributes to the remarkable decrease of Tg by 18.60◦ C, leading to a pronounced softness of the titanium white paint. This is ascribed to broken chains in the molecular structure of the binder. No morphological changes in the surface of any of the samples were noted with SEM after the laser cleaning at 1,064 nm nor did ATR reveal any changes in the chemical composition of the polymer. When comparing with samples cleaned with water-based solvents, samples cleaned with laser gave a much cleaner result. However, SEM/EDX and ATR showed that SiO2 was still present on the surface after laser cleaning and the yellowing of titanium white, as well as changes in the Tg , questions the suitability of laser cleaning on acrylic emulsion paint, at least at the given parameters. Acknowledgements The authors would like to thank Marie Vest from The Royal Library, Denmark, Jan Jørn Hansen and Jetti van Lanschot from the School of Conservation at The Royal Academy of Fine Arts, in Denmark, Yvonne Shashoua at The National Museum, Denmark, Tom Learner at Tate Gallery in London, England and Costas Fotakis from FO.R.T.H., Institute of Electronic Structure and Laser, Heraklion, Crete, Greece.

References 1. J. Crook, and T. Learner, The Impact of Modern Paints. London: Tate Gallery Publishing Ltd. 200, 2000. 2. C. Lamb, The Conservation of Modern Paintings: Introductory Notes on Papers to be Presented. London 1982. London, United Kingdom Institute for Conservation & Tate Gallery. 6, 1982. 3. V. Zaﬁropulos and C. Fotakis, in Laser Cleaning in Conservation: an Introduction, Edited by M. Cooper, Oxford, 79, 1998. 4. V. Zaﬁropulos, in: Laser Cleaning, Edited by B. Luk’yanchuk. World Scientiﬁc. Singapore, New Jersey, London, Hong Kong, 343–392, 2002. 5. T. Fourrier, G. Schrems, T. Mühlberger, J. Heitz, N. Arnold, D. Bäuerle, M. Mosbacher, J. Boneberg, and P. Leiderer, in Applied Physics A 72, Materials, Science & Processing. 1–6, 2001.

31 Removal of Simulated Dust from Water-Based Acrylic Emulsion Paints

279

6. W. A. Real, I. Zergioti, Y. Spetsidou, and D. Anglos, in The 11th Triennial Meeting: Proceedings. Edinburgh; ICOM Committee for Conservation, 303–308, 1996. 7. P. Pouli, D. C. Emmony, C. E. Madden, and I. Sutherland, in Applied Surface Science Vol. 173, 252, 2001. 8. B. Luk’yanchuk and V. Zaﬁropulos, Chapter 8.3 in: Laser Cleaning, Edited by B. Luk’yanchuk. World Scientiﬁc, Singapore, New Jersey, London, Hong Kong, 393–414, 2002. 9. M. Chappé, J. Hildehagen, K. Dickmann, and K. Bredol, in Journal of Cultural Heritage Vol. 4. suppl. 1., 264, 2003. 10. Y. H. Yoon and P. Brimblecombe, Studies in Conservation, Vol. 45, 127, 2000. 11. T. Learner, O. Chiantore, and D. Scalarone, in The 13th Triennial Meeting Rio de Janeiro: Preprints. Rio de Janeiro 2002.; ICOM Committee for Conservation. 911, 2002. 12. A. Murray, C. Contreras de Berenfeld, P. Y. Chang Sue, E. Jablonski, T. Klein, C. M. Riggs, C. E. Robertson, and A. W. M. Tse, in Materials Research Society Fall Meeting, Materials Issues in Art and Archaeology VI: Proceedings Vol. 712. Boston 2001. Boston: Materials Research Society. 83–90, 2002. 13. M. F. Meckelburg, C. P. Tumosa, and J. D. Erlebacher, Polymer: Preprints. u.p./u.å., Boston: American Chemical Society. Division of Polymer Chemistry, William E. Casp. 297–298, 1994. 14. P. M. Whitmore and V. G. Colaluca, in Studies in Conservation Vol. 40, 51, 1995. 15. A. Athanassiou, A. E. Hill, T. Fourrier, L. Burgio, and R. J. H. Clark, in Journal of Cultural Heritage, Vol. 1, 210, 2000. 16. M. Strlic, J. Kolar, V. P. Selih, and M. Marincek, in Applied Surface Science, Vol. 236, 2003. 17. V. Zaﬁropulos, C. Balas, A. Manousaki, G. Marakis, P. Maravelaki-Kalaitzaki, K. Melesanaki, P. Pouli, T. Stratoudaki, S. Klein, J. Hildenhagen, K. Dickmann, B. S. Luk’yanchuk, C. Mujat, and A. Dogariu, in Journal of Cultural Heritage Vol. 4, 249, 2003. 18. A. deCruz, M. L. Wolbarsht, and P. A. Hauger, Journal of Cultural Heritage, Vol. 1, 173, 2000. 19. P. Bracco, G. Lanterna, M. Matteini, K. Nakahara, O. Sartiani, A. deCruz, M. L. Wolbarsht, E. Adamkiewicz, and M. S. Colombini, in Journal of Cultural Heritage, Vol. 4. suppl.1., 202, 2003.

32 Traditional and Laser Cleaning Methods of Historic Picture Post Cards ◦

M. Mäder1 , H. Holle2 , M. Schreiner1 , S. Pentzien3 , J. Krüger3 , and W. Kautek3,4 1

◦

2

3

4

Institute of Science and Technology in Art, Academy of Fine Arts, Schillerplatz 3, 1010 Vienna, Austria present address: Freiberger Compound Materials, Am Junger Löwe Schacht 5, 09599 Freiberg, Germany Institute of Conservation and Restoration, Academy of Fine Arts, Schillerplatz 3, 1010 Vienna, Austria, [email protected] Division Surface Technologies, Federal Institute for Materials Research and Testing (BAM), Unter den Eichen 87, 12205 Berlin, Germany Department of Physical Chemistry, University of Vienna, Waehringer Str. 42, 1090 Vienna, Austria

Summary. Traditional paper cleaning techniques are not always suﬃcient for cleaning artefacts. The pulsed laser cleaning can oﬀer a valuable tool for solving problematic cases in paper conservation. Comparative studies of traditional and laser cleaning were performed on historic picture post cards. The results demonstrate the possibilities of partial laser cleaning using nanosecond laser pulses at a wavelength of 532 nm.

32.1 Introduction Historic post cards from the beginning of the twentieth century belong to the ﬁeld of applied arts. Today they not only are treasured objects for collectors but also oﬀer an insight in the development of art and culture as well as the printing technology. A couple of such objects show noticeable contaminations on their surfaces, i.e. they are covered with a thin layer of dust and dirt. Therefore, the images and the inscriptions can be hardly recognized. Laser cleaning [1–11] should be applied to remove the impurities without an alteration of the card boards, pigments and dyes as well as the writing media. Before laser cleaning, parts of the objects were cleaned with traditional dry as well as aqueous cleaning methods [12, 13]. Conservation techniques for paper include mechanical scratching and the use of a brush, eraser or draft clean powder (a granulated eraser). Paper may also be cleaned with water, organic solvents, enzymes, etc. Satisfactory results are often

282

M. Mäder et al.

obtained by a treatment using cellulose ethers like carboxymethyl cellulose (CMC), methyl cellulose (MC) or hydroxypropyl cellulose (HPC). Nevertheless, the current aqueous cleaning methods are not always suﬃcient, in particular if the paper is coated with sensitive printing as well as writing media. Additionally, microscopic investigations and X-ray ﬂuorescence analysis (XRF) were carried out in order to characterize the materials and their possible alterations provoked by cleaning.

32.2 Experimental Two historical picture post cards were used in this case study exemplarily: “Moorlake” and “Gruss aus Bad Ems” (Fig. 32.1). Both cards were produced at the beginning of the twentieth century using high-quality chromolithography. Besides the colourful images, they also show handwritten inscriptions of the senders. Additionally, the surfaces of the cards are covered by extensive contaminations accumulated over the years. 32.2.1 Traditional Paper Cleaning First of all, the loose parts of the dirt were removed using a soft brush. Subsequently, draft clean powder was applied to reduce the thickness of the adherent deposits. In the case of the post card “Moorlake”, the left half of the surface area was treated by the granular eraser, whereas the right half of the post card “Bad Ems” was cleaned in this way. In order to test their cleaning behaviour, three diﬀerent cellulose ethers were used in a second cleaning step on the post card “Bad Ems”. For these purposes the surface area was subdivided into three horizontal parts and each cleaning solution was applied to one of these stripes (Fig. 32.2). In the third step small vertical stripes on the left and the right side of the card were treated a second time using the most eﬀective CMC solution.

Fig. 32.1. General view of the picture post cards: “Moorlake” (left) and “Gruss aus Bad Ems” (right)

This page intentionally blank

32 Traditional and Laser Cleaning Methods

283

Fig. 32.2. Wet cleaning treatment using diﬀerent cellulose ethers: after a ﬁrst cleaning step (left), the CMC treatment was repeated on small vertical stripes (right)

Fig. 32.3. Laser cleaning station (Federal Institute for Materials Research and Testing, Berlin)

32.2.2 Laser Cleaning Selected areas of the objects were treated with the radiation of a Nd:YAG laser running at 532 nm wavelength providing a pulse duration of 8 ns. Figure 32.3 shows the computer-controlled laser cleaning station. The laser system with the scanning device is mounted above the object to be cleaned. The sample can be aligned on a movable holder. Volatile ablation products in the form of gas or dust can be exhausted. The whole laser-cleaning workstation fulﬁls the safety requirements of the Laser Class 1 condition. In addition to the laser light, the object may be illuminated by UV or visible light. A camera system allows the observation of the object also during the laser treatment.

284

M. Mäder et al.

The focused beam with a (1/e2 ) Gaussian diameter of 230 µm was scanned over the working area at a ﬁxed laser ﬂuence. The ﬂuences were varied between 53 and 154 mJ cm−2 . The number of laser pulses on every point ranged from 25 to 400. The scan pattern, i.e. the lateral distance between the spot centres, was selected in such a manner that the overlap of spots with a Gaussian distribution of the ﬂuence yielded a nearly homogeneous impact of the laser beam on the working area. 32.2.3 Material Analysis Energy dispersive X-ray ﬂuorescence analysis could be applied to characterize the chemical composition of the cardboards as well as the printing and writing materials in a non-destructive manner. The portable µ-XRF instrument [14] consists of a Mo-tube with a polycapillary lens ensuring a beam spot of about 100 µm. The tube voltage and current were set to 35 kV and 0.6 mA, respectively. X-ray spectra were taken by an electronically cooled drift chamber detector at an acquisition time of 200 or 600 s.

32.3 Results and Discussion The traditional cleaning methods proved suitable for a careful removal of dirt layers from the paper surface. Particularly, the dry techniques allowed a gradual proceeding, but the visible improvement was restricted in the case of extensive contaminations. More pronounced eﬀects could be achieved by treatments with the cellulose ether solutions. The best cleaning result was obtained for the post card “Bad Ems” treated with the 2.5% CMC. A repeated processing with the same solution resulted in a further removal of material from the card surface. However, the result was an “over-cleaning” in this special case, i.e. the original surface of the post card and especially the printing materials were also aﬀected by the second treatment. In preliminary experiments, the laser parameters were optimized in order to get visible cleaning eﬀects. In particular, the laser ﬂuence and the number of pulses per spot were varied. Best results were obtained in a relatively low range of laser ﬂuences between 53 and 100 mJ cm−2 utilizing a low pulse number of 25 per spot. An increase in these values resulted in additional destruction, i.e. the printing material and the covering layer of the card boards were already removed. The general views of both post cards (Fig. 32.1) show clearly the areas irradiated with the laser light. The selected areas of laser treatment are distinguished by an explicit cleaning eﬀect, i.e. the contaminations could be reduced signiﬁcantly. Figure 32.4 depicts a detail of the post card “Moorlake”. The rectangular area was treated with a laser ﬂuence of 53 mJ cm−2 and 25 pulses per spot. The ﬁgure of the man, the printed red letters and the hand-writing are more visible than before. Also microscopic investigations demonstrated

32 Traditional and Laser Cleaning Methods

285

Fig. 32.4. Detail of the post card “Moorlake” showing an area after laser treatment (25 pulses per spot with 53 mJ cm−2 )

that no alterations of the cardboard and the colour of the printed images and letters were observable due to the laser treatment in this case. The other case study on the post card “Bad Ems” indicated that an adherent dirt layer, which partially resisted the cellulose ethers, could be removed by a proper laser treatment. The subsequent visual evaluation of the treated areas revealed that the selected laser ﬂuence of 100 mJ cm−2 already reached the range of destruction in which the printed materials were partially aﬀected. Material analysis using XRF allows the determination of the chemical elements which are characteristic for the materials used for covering, printing as well as writing. The colour palette of the identiﬁed printing materials contains mainly inorganic pigments like cinnabar, red lead, lead white, Naples yellow, various ferric oxides and chromium green. Organic dyestuﬀs were used only in a few cases. Additionally, diﬀerent coatings on the cardboards were detected. Whereas the paper of the post card “Bad Ems” was coated with Permanent White, a chalk containing ground was used for the preparation of the “Moorlake”. An iron gall ink could be identiﬁed as writing material used for the hand-written inscriptions.

32.4 Conclusions Traditional cleaning of paper using dry and/or wet methods oﬀers the possibility of large area as well as partial treatments in order to remove contaminations layer-by-layer from the surface. The dry (mechanical) cleaning with brushes and eraser agents often proves to be insuﬃcient in cases of intense dirt. The application of water or aqueous solvents holds the risk of paper degradation as well as damage (loss) of paints or inks. Partial laser cleaning of picture post cards is demonstrated successfully. The results show that the dirt on the surface, which is particularly resistant

286

M. Mäder et al.

to dry cleaning, can be removed and the obscured pictures and writings could be visualized again. For this purpose, laser ﬂuence levels must stay below the ablation and destruction threshold of the paper substrate as well as the coating materials including paints and inks. Further studies are required to expand the understanding of laser interaction with coated papers in order to avoid any risks for the objects. Acknowledgement We acknowledge partial ﬁnancial support by the European Co-operation in the ﬁeld of Scientiﬁc and Technical Research, “Artwork Conservation by Laser” (COST G7, Short Term Scientiﬁc Mission).

References 1. W. Kautek, S. Pentzien, J. Krüger, and E. König, in Lasers in the Conservation of Artworks I, Restauratorenblätter (Special Issue) Edited by W. Kautek and E. König, Mayer & Comp., Wien, 69, 1997. 2. W. Kautek, S. Pentzien, P. Rudolph, J. Krüger, and E. König, in Appl. Surf. Sci., Vol. 127–129, 746, 1998. 3. P. Rudolph, S. Pentzien, J. Krüger, W. Kautek, and E. König, in Restauro, Vol. 104 (6), 396, 1998. 4. J. Kolar, M. Strlic, S. Pentzien, and W. Kautek, in Appl. Phys. A, Vol. 71, 87, 2000 5. J. Kolar, M. Strlic, D. Müller-Hess, K. Troschke, S. Pentzien, W. Kautek, in J. Cultural Heritage, Vol. 1, 221, 2000. 6. W. Kautek, S. Pentzien, D. Müller-Hess, K. Troschke, and R. Teule, in SPIE, Vol. 4402, 130, 2001. 7. D. Müller-Hess, K. K. Troschke, J. Kolar, M. Strlic, S. Pentzien, and W. Kautek, in Restauro, Vol. 8, 604, 2001. 8. P. Rudolph, F. J. Ligterink, J. L. Pedersoli Jr, M. van Bommel, J. Bos, H. A. Aziz, J. B. G. A. Havermans, H. Scholten, D. Schipper, and W. Kautek, in Appl. Phys. A, Vol. 79, 181, 2004. 9. P. Rudolph, F. J. Ligterink, J. L. Pedersoli Jr., H. Scholten, D. Schipper, J. B. G. A. Havermans, H. A. Aziz, V. Quillet, M. Kraan, B. van Beek, S. Corr, H.-Y. Hua-Ströfer, J. Stokmans, P. van Dalen, and W. Kautek, in Appl. Phys. A, Vol. 79, 941, 2004. 10. H. Scholten, D. Schipper, F. J. Ligterink, J. L. Pedersoli Jr., P. Rudolph, W. Kautek, J. B. G. A. Havermans, H. A. Aziz, B. van Beek, M. Kraan, P. van Dalen, V. Quillet, S. Corr, and H.-Y. Hua-Ströfer, in Lasers in the Conservation of Artworks, Springer Proceedings in Physics, Vol. 100, 11, 2005. 11. E. Pilch, S. Pentzien, H. Mädebach, and W. Kautek, in Lasers in the Conservation of Artworks, Springer Proceedings in Physics, Vol. 100, 19, 2005. 12. O. Wächter, in Restaurierung und Erhaltung von Büchern, Archivalien und Graphiken, Hermann Böhlaus Nachf., Wien Köln Graz, 1982. 13. D. van der Reyden, in JAIC, Vol. 31, 117, 1992. 14. µ-XRF “COPRA”: Prototype for X-ray ﬂuorescence analysis of artifacts, developed and built within an EU-Project (SMT4-CT98-2237).

33 Femtosecond Laser Cleaning of Painted Artefacts; Is this the Way Forward? ∗

P. Pouli1 , G. Bounos1,2 , S. Georgiou1 , and C. Fotakis1,2 1

∗ 2

Institute of Electronic Structure and Lasers (IESL), Foundation for Research and Technology-Hellas (FORTH), P.O. Box 1527, Heraklion, Crete, 71110, Greece [email protected] Department of Physics, University of Crete, Heraklion, Greece

Summary. The laser cleaning of painted artefacts relies on the synergy of thermal, photochemical and photomechanical eﬀects, which are involved in laser ablation. A crucial issue, however, for a successful cleaning intervention is the spatial conﬁnement and control of these eﬀects for safeguarding the original surface from potential damage. Extensive studies have shown that in many cases there is an optimum interplay of laser and material parameters, which resulted in successful laser cleaning applications. The laser pulse duration is an important parameter in this context. The scope of this work has been the exploration of any advantages, which may be oﬀered by using ultrafast UV (248 nm) lasers for the cleaning application of sensitive painted artworks. To achieve this goal, comparative study on the ablation rate and threshold of femto- and nanosecond laser pulses of typical varnishes (dammar, mastic, etc.) have been performed. Femtosecond pulses appear to be superior in terms of the spatial resolution and etching resolution and this fact has been demonstrated for both technical samples and original objects. Additionally, possible induced photochemical modiﬁcations have been investigated by monitoring the photoproduct laser-induced ﬂuorescence of varnish-systems doped with low concentrations of welldeﬁned photosensitive dopants (e.g. PhenI). It is established that irradiation with f s UV laser pulses results in minimal photochemical modiﬁcations. Importantly, the amount of photochemical products is largely independent from the optical properties (i.e. absorptivity) of the varnish. Considering the recent advances in ultrafast laser technology, the use of such lasers appears to provide a viable approach.

33.1 Introduction Femtosecond (fs) laser technology has been demonstrated to oﬀer particular capabilities for material processing overcoming several of the limitations of processing with nanosecond (ns) laser pulses. In particular three major advantages of fs laser processing have been emphasized in previous studies [1]. First, multiphoton processes enable the processing of even nominally transparent substrates [2, 3]. Second, the thermal diﬀusion is minimal thereby enabling

288

P. Pouli et al.

processing with minimal/negligible thermal degradation [1]. Third, there is no plasma shielding so that there is maximum coupling of the laser energy into the substrate [4]. As a result of these factors, the morphological aspects of the processed substrates are far superior to those attained with ns laser pulses [4]. Although the previous advantages are very important in the cleaning of painted artworks [5, 6], there are a number of additional aspects that become important/vital and may be decisive for the success of the femtosecond laser processing. A most crucial issue concerns the extent of photochemical modiﬁcations eﬀected to the substrate. In fact, this question issue is central to all femtosecond laser processing schemes of molecular/organic systems, including biological and medical applications. In relationship with these issues there is almost no work reported in the literature. We report here ﬁrst/preliminary studies in the use of femtosecond laser technology for the processing/restoration of painted artworks. As demonstrated by extensive previous work with nanosecond laser technology [5–7], this is a highly demanding application in which several issues have to be carefully optimised. In particular, given the high photolability of the substrates, photochemical eﬀects upon fs ablation/irradiation become a crucial factor. Although femtosecond laser ablation may considerably improve the morphology, it is unclear to what extent it aﬀects photochemical eﬀects. To explore any advantages which may be oﬀered by using ultrafast pulse lasers for the cleaning application of painted artworks, comparative studies on typical varnishes (dammar and mastic) have been performed. The cleaning result of fs (500 fs) and ns (30 ns) laser pulses at 248 nm was evaluated on the basis of 1. The spatial resolution, i.e. the etching eﬃciency and the ablation threshold (FT H ) of the studied materials in both regimes (fs and ns) 2. The extent of the induced photochemical modiﬁcations to the remaining material, by monitoring the photoproduct laser-induced ﬂuorescence of varnishes doped with very low concentrations of the photosensitive dopants, as a function of the laser ﬂuence and the number of laser pulses 3. The morphology of the ablation spots in order to investigate the extent of “inﬂuence” to the substrate

33.2 Experimental Methods Two types of commercially available varnishes, dammar and mastic of Chios, have been studied. Neat and doped ﬁlms are cast on quartz plates from a dichloromethane solution and are subsequently dried for 48 h. Typical ﬁlm thickness is in the 30 ± 10 µm range. Femtosecond laser pulses (500 fs) at 248 nm were generated using a XeCl excimer pumped dye laser system based on the principle of a distributed feedback dye laser (DFDL). The energy output is 10–30 mJ per pulse while the

33 Femtosecond Laser Cleaning of Painted Artefacts

289

average pulse-to-pulse ﬂuctuation is 15%. The beam was focused perpendicularly on the sample by means of a quartz lens (f = +300 mm). Typical ﬂuence values were between 0.05 and 0.75 mJ cm−2 . Nanosecond laser pulses (10 ns) were emitted from a compact excimer laser (Braggstar 200, TUI Laser) operating at 248 nm (KrF). The maximum energy was 16 mJ while the laser beam was focused on the sample by means of a fused silica plano-convex lens (f = +100 mm). Typical ﬂuences were between 4 and 500 mJ cm−2 . All irradiation experiments were performed in air. A standard laboratory arrangement was used for the pump/probe LIF measurements. For consistency reasons, irradiation of the samples with both fs and ns laser systems was performed under the same recording conditions. Photoproduct ﬂuorescence was induced with the ns laser at 248 nm at ﬂuences of 5–10 mJ cm−2 , in surface area slightly smaller than the area irradiated with the “pump” beam to avoid additional photolysis products. An optical ﬁbre placed nearly perpendicularly to the sample and in close proximity to its surface (∼ 1 cm) was employed to collect the emitted signal while cut-oﬀ ﬁlters were used to minimize detection of laser-scattered light. The emission signal is spectrally analysed in a 0.20 m grating spectrograph equipped with a 300 grooves/mm grating and the spectrum is recorded on an optical multichannel analyzer (OMA III system, EG&G PARC Model 1406) interfaced to a computer. Etch depth measurements, following laser ablation, were performed by a mechanical stylus proﬁlometer (Perthometer S5P, Mahr) and the irradiated areas were examined under an optical microscope (ME600, Nikon).

33.3 Results and Discussion 33.3.1 Spatial Resolution In laser cleaning applications on painted artworks, the major attention is focused on the spatial conﬁnement of the intervention in order to ensure minimal damage risk to the underlying original surfaces [5,6]. This is particularly important in cases where the unwanted surface layers are very thin and thus it is necessary to ensure that the mean removed depth per laser pulse, as well as the thermal load to the substrate, are the minimum possible. It is by now well established [3, 4, 7] that multiphoton processes in the femtosecond laser irradiation result in reduction in the eﬀective optical penetration depth and, as a consequence, in a signiﬁcant decrease to the actual etching rate. Given that the examined varnishes are a mixture of oligomers of some chemical similarity to polymers, it is expectable that their etching eﬃciency presents similar dependence. Figure 33.1 illustrates the etching depths of dammar and mastic by 248 nm excimer laser pulses of 500 fs and 30 ns duration. In the ns case, the etch rate rises proportionally to the applied ﬂuences up to about 150 mJ cm−2 , where the etch rate saturates reaching a plateau. On the other hand, the fs etch

290

P. Pouli et al.

Etch Rate (mm/pulse)

2,0 1,6 1,2 0,8 dammar (fs) dammar (ns) mastic (fs) mastic (ns)

0,4

0

200

400

600

Fluence (mJ/cm2)

Fig. 33.1. Etch rate (in µm/pulse) vs. ﬂuence (in mJ cm−2 ) for dammar (black dots) and mastic (red squares) in the femtosecond (full dots/squares) and the nanosecond region (open dots/squares)

curves appear almost linear, with lower slope in comparison to the ns one. The mean removed depth per pulse is much lower with the shorter laser pulse width and ablation initiates at signiﬁcantly lower ﬂuence values than in the ns regime, suggesting that fs laser ablation may ensure a higher degree of resolution and thus minimal risk to the original artwork surface. Another important observation is that, in the fs case, the etch rate lines of the two studied varnishes are nearly equal. Mastic shows a higher small signal absorption coeﬃcient than dammar at the irradiation wavelength (248 nm) and as a result a smaller optical penetration depth is expected, which justiﬁes a lower etch depth rate in the ns case (Fig. 33.1). The fact that in the fs irradiation these lines appear very close implies that the mean removed material per laser pulse is nearly independent from the optical properties of the material, which may be of great advantage in the speciﬁc cleaning applications. 33.3.2 Photochemical Modiﬁcations Another very important issue in the laser cleaning of painted artworks lays on the extent of the photochemical modiﬁcations that may be eﬀected to the substrate (i.e. the remaining varnish layer) [8]. Based on extensive previous work of our group in polymers [9,10], the induced photochemical modiﬁcations was investigated by monitoring the photoproducts formed in doped varnish systems by means of laser induced ﬂuorescence (LIF) in a “pump-probe” scheme. The approach is based on the following key-points: 1. It was found [7] that the dissociation and subsequent radical activity of the dopant is particularly sensitive to the alterations of the polymer environment and thus it is expected that the study of photoproduct activity of the dopant will lead to a representative picture of the actual photochemistry of the varnish layers.

33 Femtosecond Laser Cleaning of Painted Artefacts

291

Fluorescence Intensity (a.u.)

2000 1800 1600 Dammar F(ns) = 270 mJ/cm2

1400 1200 1000 800 600 400

Dammar F(fs) = 65 mJ/cm2

200 0 300

400 wavelength (nm)

500

Fig. 33.2. Comparison of the product LIF spectra from 9-Iodophenanthrene (1%wt) doped dammar irradiated above the ablation threshold (FT H ) with a single fs (black line) and ns (red line) laser pulses (λ = 248 nm). For consistency reasons in both cases the LIF spectra was obtained with the same probing beam (λ = 248 nm, 10 ns) under the same conditions

2. The employed dopants, usually halo-aromatics, are characterised by a simple photodissociation reactivity pattern resulting exclusively into aryl and halogen radicals. The laser-induced modiﬁcations of these products can be easily and uniquely characterised and quantiﬁed via LIF. For the eﬀective comparison of the extent of photochemical changes induced from the diﬀerent laser systems, it was chosen to use the same probing beam in both experiments. Furthermore in order to ensure one-photon excitation of product, ns pulses were selected, at very low ﬂuences (in the range of 4– 6 mJ cm−2 ). In Fig. 33.2, the product LIF spectra recorded from 9-Iodophenanthrene (1% wt) doped dammar irradiated above ablation threshold (FT H ) with a single fs and ns laser pulse at 248 nm are comparatively illustrated. The emission band at ∼374 nm, which does not appear either before irradiation with the pump laser or upon irradiation of the neat varnish ﬁlms, is attributed to emitting photoproducts. This band is ascribed to the 1 B3u →1A1g transition of PhenH [11, 12]. Given that the illustrated LIF spectra were recorded under the same conditions, it is clear that product formation is much reduced in the fs irradiation as compared with ns irradiation for a single pump pulse with ﬂuence well above the FT H . The superiority of the fs pulses as regards the limited photochemical modiﬁcations was also supported by quantitative characterisation and comparison of the products of the doped varnishes as a function of the ﬂuence of the “pump” beam and the applied number of pulses. Indeed when comparing the product ﬂuorescence intensity upon fs and ns irradiation as a function of the ﬂuence of the “pump” beam [13], it was clearly shown that photochemical

292

P. Pouli et al.

Fig. 33.3. Optical microscope photographs of craters created on thick (∼50 µm) mastic varnish ﬁlm by ns (a) and fs (b) laser pulses at 248 nm. Irradiation parameters: 200 pulses at 350 mJ cm−2 in the ns case and 300 pulses and 265 mJ cm−2 in the fs case

modiﬁcations are remarkably lower when shorter pulses of the same wavelength are applied. Furthermore we observe that the fs irradiation of the two systems results in nearly equal product yields despite their considerable difference in the absorptivity at 248 nm. In other words the amount of product formation is largely independent of the absorptivity of the varnish. This feature can be particularly important for cases where multiple conservation treatments have been applied, as it may enable highly precise treatment, nearly independently of the optical properties of the substrate. Furthermore studies aiming to compare the accumulation of the PhenH photoproducts (at λ = 374 nm) as a function of successive “pump” laser pulses of ns and fs duration [13] have shown that at low ﬂuences (below the ablation threshold) both irradiation regimes show the same behaviour; increasing product formation with increasing number of pulses and ﬂuences. On the other hand a distinctively diﬀerent picture was observed when higher laser ﬂuences (above the ablation threshold) were applied, where product formation in the fs case tends to decrease with increasing laser ﬂuences. The above observations indicate that the thermal eﬀected zone is limited in the fs case even at extensive irradiation conditions (ﬂuence and number of pulses). Thus the importance of using shorter laser pulse width of higher ﬂuence values (signiﬁcantly above the ablation threshold) for an eﬀective and photochemically limited material removal is highlighted. 33.3.3 Morphological Aspects Finally, in laser cleaning of artworks, the morphology of the laser-irradiated surfaces is of major importance as melting phenomena or irregular structures may result into potential damage features on the underlying original surface. In Fig. 33.3, the morphology of the ablation spots created by ns (a) and fs (b) laser pulses on relatively thick varnish layers are comparatively examined under the optical microscope. It is clear that compared to the ns ablation spots, the fs ones result into a sharply deﬁned etch craters with clear edges and no indication of thermal modiﬁcations (melting). This is in agreement with the observations reported by Küper et al. [2] in PMMA and Teﬂon.

33 Femtosecond Laser Cleaning of Painted Artefacts

293

33.4 Conclusions This study aimed to explore any advantages which may be oﬀered by using ultrafast pulsed lasers for the cleaning of sensitive painted artworks. To this end fs and ns laser irradiation of technical varnish samples (dammar and mastic) has been compared on the basis of (a) the spatial resolution, (b) the extent of the induced photochemical modiﬁcations and (c) the morphology of the irradiated surfaces. It was shown that fs pulses ensure higher etching resolution and edge precision without any melting phenomena. Product formation was proven to be much reduced in the fs regime, while it was shown that the amount of the product formation is largely independent of the absorptivity of the material to be removed (varnish). Thus ultrashort lasers can enable a highly precise treatment nearly independent of the optical properties of the substrate, opening new perspectives in the cleaning interventions of sensitive and particularly demanding painted artworks, such as modern paintings. Acknowledgement The authors would like to thank Dr. M. Doulgeridis, director of conservation in the National Gallery of Athens, for valuable discussions and suggestions.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

D. Bäuerle, in Laser Processing and Chemistry; Springer, Berlin, 2000. S. Küper and M. Stuke, Appl. Phys. Lett. 54, 4 (1988). J.K. Frisoli, Y. Hefetz, and T.F. Deutsch, Appl. Phys. B 52, 168 (1991). S. Küper and M. Stuke, Appl. Phys. B 44, 199 (1987). V. Zaﬁropulos and C. Fotakis, Ch. 6 in Laser Cleaning in Conservation: an Introduction, Ed. by M. Cooper, Butterworth Heinemann, Oxford, 79, 1998. V. Zaﬁropulos, Ch. 8 in Laser Cleaning, Ed. B. Luk’yanchuk, World Scientiﬁc, Singapore, 337, 2002. S. Georgiou, Adv. Polymer Sci. 168, 1 (2004). S. Georgiou, V. Zaﬁropulos, D. Anglos, C. Balas, V. Tornari, and C. Fotakis, Appl. Surf. Sci. 127, 738 (1998). M. Lassithiotaki, A. Athanassiou, D. Anglos, S. Georgiou, and C. Fotakis, Appl. Phys. A 69, 363 (1999). G. Bounos, A. Athanassiou, D. Anglos, S. Georgiou, and C. Fotakis, J. Phys. Chem. B 108, 7052 (2004). M. Dzvonik, S. Yang, and R. Bersohn, J. Chem. Phys. 61, 4408 (1974). J.B. Birks, in Photophysics of Aromatic Molecules; John Wiley & Sons: London, 232, 1970. P. Pouli, G. Bounos, S. Georgiou, and C. Fotakis, Appl. Phys. A (submitted )

34 Laser Cleaning of Polyurethane Foam: An Investigation Using Three Variants of Commercial PU Products ∗

U. Staal Dinesen1,2 and M. Westergaard2,3 1

2 ∗ 3

National Workshops for Arts and Crafts, Copenhagen, Denmark/Louisiana Museum of Modern Art, Humlebæk, Denmark School of Conservation, Royal Academy of Fine Arts, Copenhagen, Denmark [email protected]/[email protected] Art Conservation Centre, Kronborg Castle, Elsinore, Denmark

Summary. In this study, tests were undertaken to ascertain whether the laser could achieve a better level of cleaning on polyurethane foam than vacuum cleaning. Optimum laser parameters were found using statistics on data from color measurements. The laser proved to be very eﬀective regarding the removal of dust, but also caused damage on some PU-variants. The laser cleaning has been carried out at National Workshops for Arts and Crafts, Copenhagen, Denmark.

34.1 Introduction With the introduction of polyurethane foam (PU-foam) as an artistic medium, a new challenge is posed in the ﬁeld of conservation of modern art. The cleaning of PU-foam using traditional methods often causes damage or achieves an insuﬃcient level of cleaning. Consequently, tests were undertaken to ascertain whether PU-foam in three variants can (1) be cleaned with an Nd:YAG laser (Phoenix 1200 from Lynton Lasers), and, if so, (2) whether the laser can achieve a better level of cleaning than vacuum cleaning. PU-foam is commercially produced in numerous variants – roughly speaking, in variants of stiﬀ and ﬂexible PU-foam. Generally, the physical structure of stiﬀ PU-foam is closed cells, while the cells of ﬂexible PU-foam are usually open. PU is generally formed by a reaction of the main components isocyanate and polyole. The polyoles are usually based on polyether or polyester (90% are based on polyether) [1]. The isocyanate, TDI (toluen-2, 4-diisocyanat), is usually used in ﬂexible PU-foam. MDI (diphenylmethan-4, 4-diisocyanat) is the most commonly used isocyanate in stiﬀ PU-foam [2]. In this study, three commercial PU-products were used: PU-F: Flexible PU-foam with open cells – based on TDI and polyether polyol

296

U.S. Dinesen and M. Westergaard

PU-S: Stiﬀ PU-foam with closed cells – based on MDI and polyether polyol PU-SB: Stiﬀ PU-foam (same as PU-S, but with the smooth surface cut oﬀ) Most PU-products are degraded at temperatures above 200◦ C. Yellowing and brown/black discoloration can be observed at higher temperatures (>240◦ C) [3]. Yellowing can be a result of the formation of quinon-imides or “photo-Fries rearrangement.” In addition, tests have shown that radiation (<340 nm) can cause the formation of the above-mentioned chromophore groups [4]. Consequently, UV-lasers are considered unsuitable for the cleaning of PU-foam.

34.2 Experimental Methods The aim of the ﬁrst set of tests was to ﬁnd optimum laser parameters for irradiation/cleaning of new, artiﬁcially aged, clean, and soiled variants of the three PU-variants. Optimum laser parameters on each PU-variant were found using statistics (using color measurements, CIE-L∗ a∗ b∗ (Minolta CM 2600D) as response); an analysis of variance, using “JMP Start Statistics” (part of the program “SAS”), gave valuable information about main interactions and two-factor interactions; i.e., clear indications at which wavelength (1,064 or 532 nm), energy density, and number of pulses the cleaning potential is best on each PU-variant. The design of experiments is shown in Table 34.1. The aim of the second set of tests was to compare the eﬀect of laser cleaning using the above-mentioned optimum laser parameters with vacuum cleaning (a traditional cleaning method on PU-foam). The design of experiments is shown in Table 34.2. In addition, examinations were made to ascertain whether laser cleaning caused changes in the chemical or physical structure of the PU-variants by using ATR-FTIR and SEM, respectively. Table 34.1. Design of experiments for the ﬁrst set of tests. Factor Wavelength (nm) aged/unaged Number of pulses (np) Clean/soiled Material Energy density (Jcm−2 )

Level 532 Aged, a 3 Clean, c PU-F 0.25 0.5

PU-S

1,064 Unaged, u 30 soiled, s PU-SB 1 2

For soiling, the artiﬁcial dust “ASHRAE 52/76” (carbon, quartz, cotton, Fe-, Al-, Cr-, Na- and K-oxides) from “Particle Technology” was used. “Atlas Weatherometer CI 3000+” was used for aging (aging corresponding to 36 months with 10 h exposure of 200 lx per day)

This page intentionally blank

34 Laser Cleaning of Polyurethane Foam

297

Table 34.2. Design of experiments for the second set of tests Factor Aged/unaged Cleaning method

Level Aged, a Vacuum cleaning, v

Unaged, u Laser cleaning, l

34.3 Results and Discussion The optimum laser parameters were found through the analysis of several diagrams from the analysis of variance. One surprising ﬁnding was that the cleaning potential on PU-S and PU-SB is best using 1,064 nm, while it is better to use 532 nm on PU-F (Fig. 34.1). Another interesting ﬁnding was that the cleaning potential is better using three pulses per area on both PU-S and PU-SB. The number of pulses is less important regarding the cleaning potential of PU-F (Fig. 34.2). As illustrated in Fig. 34.3 it also became evident that the best cleaning potential, roughly speaking, was achieved using either low-energy density and few pulses or high-energy density and few pulses. In combination with a visual evaluation, the outcome of the analysis of variance resulted in the optimum laser parameters listed in Table 34.3. These parameters were used in the second set of tests. Comparing laser cleaning with vacuum cleaning in the second set of tests (Fig. 34.4), it became clear that the laser cleaned area appears much cleaner than the vacuum cleaned area on all PU-variants; in fact, there is only a minor diﬀerence between the vacuum cleaned area and the soiled area. Similarly, the analysis of variance conﬁrms that laser cleaning achieves a signiﬁcantly better level of cleaning than vacuum cleaning on all PU-variants (Figs. 34.5 and 34.6). As illustrated in Fig. 34.6, vacuum cleaning has to a very little degree caused removal of dust on PU-F. In comparison, the eﬀect of vacuum cleaning is slightly better on PU-S and PU-SB. Laser cleaning, however, did also to some extent cause yellow/brown discoloration – especially on PU-S and PU-SB, discoloration could be observed; most likely as a result of a photothermal reaction (T > 240◦ C). A reason that PU-F does not discolor to the same extent as PU-S and PU-SB could be that there are diﬀerences in the material absorptivity (the thermal conductivity of the PU-products are low – and roughly the same). This observation is supported by the fact that the stability of PU-products (concerning radiation) is related to the isocyanate; the stability of PU-products based on TDI (PU-F) is considered better than the stability of PU-products based on MDI (PU-S and PU-SB) [5]. With the ATR-FTIR analysis it was possible to detect that laser cleaning caused removal of dust (changes at 1,073 and 796 cm−1 – corresponding to Si–O compounds). The analysis, however, showed only few indications of

298

U.S. Dinesen and M. Westergaard Wavelength*Material - ∆E*ab, clean ref 20 15 10 5 0 PU-F

PU-S

PU-SB

532

1064

Fig. 34.1. The two-factor interaction, Wavelength∗ Material, with ∆E ∗ ab as response and 0 representing the clean reference, illustrate that the cleaning potential on PU-F is better using 532 nm. In contrast, 1,064 nm is better on PU-S and PU-SB. Comparing PU-S and PU-SB, it becomes apparent that it is more important to use 1,064 nm than 532 nm on PU-S np*Material - ∆E*ab, clean ref 20 15 10 5 0 PU-F

PU-S

PU-SB

np 3

np 30

Fig. 34.2. The two-factor interaction, np∗ Material, with ∆E ∗ ab as response and 0 representing the clean reference, illustrate that it is important to clean PU-SB and especially PU-S with few pulses per area. On PU-F, the number of pulses is less critical np*Energy density - ∆E*ab, clean ref 20 15 10 5 0 0,25

0,5

1 np 3 ∗

2 np 30

Fig. 34.3. The two-factor interaction, np Energy density with ∆E ∗ ab as response and 0 representing the clean reference, illustrate that the best cleaning potential was achieved at 1 Jcm−2 and three pulses per area

34 Laser Cleaning of Polyurethane Foam

299

Table 34.3. Optimum laser parameters for each PU-variant

PU-F PU-S PU-SB

wavelength (nm)

energy density (J cm−2 )

pulses per area (np )

532 1064 1064

2 1 1

12 3 5

Fig. 34.4. On all PU-variants, the laser cleaned area appears cleaner than the vacuum cleaned area. However, the laser cleaned area does not appear as “clean” as the clean reference

changes in the chemical structure of the PU-matrix, i.e. changes that could relate to the formation of the above-mentioned chromophore groups. With the SEM analysis, however, it became evident, that laser cleaning at given parameters, caused damage to the cell walls in two of the three PU-variants (Fig. 34.7). The damage could only be observed on the laser cleaned areas of the stiﬀ PU-variants (closed cells); on PU-F, the open cell structure appears unaﬀected.

300

U.S. Dinesen and M. Westergaard Main interaction, Material, ∆E*ab, clean ref 20 15 10 5 0 Laser cleaning PU-F

Vacuum cleaning PU-S

PU-SB

Fig. 34.5. The main interaction, Material, with ∆E ∗ ab as response and 0 representing the clean reference, illustrate that laser cleaning achieves a better level of cleaning than vacuum cleaning on all PU-variants – especially on PU-F Main interaction, Material, ∆E*ab, soiled ref 20 15 10 5 0 Laser cleaning PU-F

Vacuum cleaning PU-S

PU-SB

Fig. 34.6. The main interaction, Material, with ∆E ∗ ab as response and 0 representing the soiled reference, illustrate that laser cleaning achieves a better level of cleaning than vacuum cleaning on all PU-variants – especially on PU-F

Fig. 34.7. With SEM analysis it became clear that laser cleaning of PU-variants with closed PU-cells (PU-S (left) and PU-SB (right) resulted in damage to the cells (hole in cell walls)

34 Laser Cleaning of Polyurethane Foam

301

34.4 Conclusions Consequently, laser cleaning has proven to be a very eﬀective method for the removal of dust from PU-foam. However, laser cleaning at given parameters cannot be recommended on two of the three PU-variants as a result of the above-mentioned damage to the foam structure. Contrarily, laser cleaning at given parameters has proven to be very promising on one PU-variant (PU-F); on this PU-variant, laser cleaning did not cause damage to the foam structure and the yellowing was less signiﬁcant than on the other two PU-variants. In addition, the experiments have shown that laser cleaning is much more eﬀective than vacuum cleaning with regards to dust removal. Acknowledgments National Workshops for Arts and Crafts, Copenhagen and Director Ulf Horak for use of laser. Conservator Bent Eshøj, Lic.techn. Jan Jørn Hansen, Geologist Jørn Bredal-Jørgensen, Ph.D. Judith Jacobsen, Designer Lene Harbo Pedersen, Scott Green, Conservator Børge Igor Brandt.

References 1. J. Bützer and K. Kessler, in Kunststoﬀ als Werkstoﬀ: Celluloid und PolyurethanWeichschaum. 105, 2001. 2. G. Oertel, in Polyurethane Handbook, 2. ed. 17, 1993. 3. Y.Shieh, H. Chen, K. Liu, and Y. Twu, in Thermal Degradation of MDI-Based Segmented Polyurethanes. In: Journal of Polymer Science: Part A: Polymer Chemistry, Vol. 37, 4126, 1999. 4. J. F. Rabek, in Photodegradation of Polymers – Physical Characteristics and Applications, 82, 1996. 5. B. Rånby and J. F. Rabek in Photodegradation. Photo-oxidation and Photostabilization of Polymers, 241, 1975.

35 Excimer Laser Ablation of Egg Tempera Paints and Varnishes ∗

P.J. Morais , R. Bordalo, L. dos Santos, S.F. Marques, E. Salgueiredo, and H. Gouveia Instituto de Soldadura e Qualidade, Taguspark, Apartado 012, 2741-901 Porto Salvo, Portugal ∗ [email protected] Summary. In this work a series of egg tempera paint and varnish systems have been prepared, artiﬁcially aged and irradiated with KrF excimer laser at a wavelength of 248 nm. The samples were prepared with pure pigments and selected mixtures. It was found that, for some pigments, the colour changed upon laser irradiation even at low energy densities, below the ablation threshold while for other inorganic pigmented egg temperas the degree of discoloration is very small at moderate ﬂuence of ∼ 0.30 J cm−2 . The varnish systems did not present signs of discoloration. The thickness, superﬁcial roughness and magnitude of the colour changes of the samples were measured. X-ray diﬀraction, Raman spectroscopy and UV/visible spectroscopy were used in order to investigate the changes induced by the KrF excimer laser radiation.

35.1 Introduction Laser radiation has been used to remove unwanted layers from surfaces, not only in technological applications but also in items of cultural value. The ﬁrst examples of lasers into art conservation were explored during the 1970s [1–3], although it was from the early 1990s that the interest and progress in laserbased techniques for the restoration of artworks increased [4–7]. There is a broad range of art materials that have been already investigated, being the stone-based materials extensively studied with the most successful results [8– 10]. The interaction of laser radiation with painting materials has been also a ﬁeld of research due to its potential advantages over traditional methods [6,11– 14]. It has been proven that UV excimer lasers working at λ = 248 nm with proper selection of process parameters are suitable to remove polymerised varnishes or even over-paintings [4, 6, 11, 13–15]. Nevertheless, paint materials can undergo physical and chemical changes during laser irradiation such as discoloration phenomena in pigments [6, 16]. Thus, due to its nature each paint material reacts diﬀerently to the UV excimer laser radiation, which makes it necessary to evaluate the changes induced by the laser light in the

304

P.J. Morais et al.

broad variety of artist’s materials (varnishes, pigments and binding media), in order to provide a safe and eﬃcient process that can improve and complement traditional conservation and restoration practices. The objective of this work was to investigate the eﬀect of KrF excimer laser irradiation (λ = 248 nm) on a series of pigmented egg tempera paints (isolated and mixture), not all previously studied, to the best of the author’s knowledge. In addition, some varnishes were also under investigation. The prepared samples were partially characterised before and after laser ablation, in order to analyse the physical and chemical changes induced by the laser radiation.

35.2 Experimental Methods The presented experiments were carried out on two types of samples: inorganic pigments on egg tempera (as binding medium) and natural varnishes. Isolated and mixed pigmented egg tempera systems were prepared according to the procedure described by Van den Brink [17, 18], which is based on Cennini recipe [19]. In general terms, a homogenous mixture of powder pigment and egg is achieved and mastic resin was added to enhance the adherence to the support. Both types of systems were deposited on sheets of Melinex using a Neurtek ﬁlm applicator capable of depositing 200 µm of wet ﬁlm thickness. The varnishes were also deposited on microscope glass slides. The prepared samples were cured in a dark chamber during one month and artiﬁcially aged with ﬂuorescent daylight lamps for another month, at a constant temperature and humidity. Inorganic pigments and varnishes of historical interest were selected: barium sulphate (BaSO4 ), yellow ochre (Fe2 O3 · H2 O), lapiz-lazuli (complex natrium-aluminium-silicate, containing sulfur), malachite (Cu(CO2 ) · Cu(OH)2 ), white lead (2Pb(CO3 )2 · Pb(OH)2 ), red lead (Pb3 O4 ), Prussian blue (Fe4 [Fe(CN)6 ]3 · 14H2 O) and cobalt green (CoO · ZnO). Representative varnishes were selected: shellac, dammar and mastic. The laser treatment consisted in the irradiation of the samples with a KrF excimer laser (Lambda Physic LPX 240i) emitting 248 nm wavelength with pulses of 14 ns of duration and 160 mJ of energy. The samples were mounted onto a computer-controlled motorized X–Y translation stage. The ﬁnal energy density–ﬂuence (in J cm−2 ) was determined by dividing the incident energy by the irradiated area. The distance from the lens to the sample determined the irradiated area. In order to evaluate the morphological changes, the sample’s roughness and thickness were measured by a mechanical proﬁlometer (Tencor Alpha-Step 200). Scanning electron microscopy (Hitachi S4100) was used to obtain photographs of the egg tempera sample’s surface. Their colour value was measured using the CIELAB system procedure [20]. The crystal phases of the pigmented samples were evaluated by means of X-ray diﬀraction (XRD) measurements (Rigaku Geigerﬂex D/max) with CuKα radiation. Some

This page intentionally blank

35 Excimer Laser Ablation of Egg Tempera Paints and Varnishes

305

of the samples were also analysed by Fourier transform Raman spectroscopy (Bruker RFS 100/S, Nd:YAG 1,064 nm excitation). UV/visible spectroscopy (Shimadzu 3101) was used in the varnished samples in order to evaluate their transparency in the visible region. The laser treatments consisted in the irradiation of the samples with several diﬀerent energy densities. Beginning with a low-energy density, below the ablation threshold, it was stepwise increased to a highest value, well above the ablation threshold. In this way, even when no information was available about the ablation threshold of a speciﬁc material, it was assured that irradiated areas below and above the threshold were present. The tests were stopped each time a colour change was detectable or when a colour change was detected and did not suﬀer any further alteration after successive pulses or even when no physical alteration was observed in the samples.

35.3 Results and Discussion Table 35.1 summarises the laser treatment conditions used on the test systems and the calculated ﬂuences based on the distance between the lens and the samples. After laser treatment it was observed that lead white, yellow ochre, malachite and red lead changed their colour even at low energy densities. The chromatic analysis by colorimetry indicates ∆E (magnitude of the colour change) values of 8.0 and 7.8 for yellow ochre and malachite at low energy densities (<0.08 J cm−2 ), respectively, and 17.6 for red lead at high ﬂuences (<0.55 J cm−2 ), which represents high levels of discoloration. In a recent study with pure pigments (without medium binders), Chappé et al. [21] observed also discoloration in those same pigments when irradiated with a Nd:YAG laser at λ = 266 nm. Possible explanations are related with chemical reactions due to laser-induced heat [21] or to reduction phenomena [22]. Therefore, paint layers based on those pigments may not be suitable for direct laser cleaning. Nevertheless, in some materials, such as the ones based on lead pigments, the discoloration is not permanent and in oxygen-saturated conditions it is possible a recovery towards the initial colour [22]. Cobalt green samples appear to be more stable but with middle ﬂuences the original colour changed while lapis-lazuli showed to be the most stable even at high ﬂuences. The investigated varnishes did not exhibit any colour change upon laser treatment as reported previously by Castillejo et al. [15] and ﬂuences of ∼0.35–0.45 J cm−2 appear the best values for their ablation in a controllable manner. The thickness of the samples was measured before and after laser ablation. Table 35.2 shows the obtained results and the correspondent surface roughness Ra. XRD measurements on the samples that suﬀered colour changes do not indicate any alteration on the crystalline phases, except for the case of red lead paints where a PbO phase (identiﬁed as massicot) can be found even at

varnish

egg tempera paint

System

yellow ochre barium sulphate malachite red lead prussian blue cobalt green lapis-lazuli white lead white lead+malachite (20%–80%) white lead+malachite (50%–50%) white lead+malachite (80%–20%) lapis-lazuli+malachite (20%–80%) lapis-lazuli+malachite (50%–50%) lapis-lazuli+malachite (80%–20%) lapis-lazuli+white lead (20%–80%) lapis-lazuli+white lead (50%–50%) lapis-lazuli+white lead (80%–20%) shellac dammar mastic

Pigment/varnish name

8

Frequency (Hz)

3

10, 25, 50 1, 5 1, 5, 10 1, 5 1, 5 1, 5 1, 5 10, 25, 50 1, 5 1, 5 1, 5

20

Number of pulses

<0.20

0.12

0.08 0.17 0.08 0.09 0.13 0.08

0.46 0.30 0.40 0.23 0.28 0.32 0.3 0.17 0.12 0.06 0.07 0.17 0.17 0.4 0.06 0.07 0.06 0.48 0.41 0.36

Fluence (J cm−2 ) Low Middle

0.86 0.75 0.73

0.3

0.70 0.69 0.58 0.55 0.69 0.62

High

Table 35.1. Excimer laser treatment at 248 nm of pigmented egg tempera and varnish systems. Three data points were selected: below the ablation threshold (low), an intermediate one (middle) and a last point above the ablation threshold (high)

306 P.J. Morais et al.

varnish

egg tempera paint

System

Initial thickness (µm) (±1 µm) 68 58 88 63 56 90 18 21 39

Pigment/varnish name

yellow ochre barium sulphate malachite red lead prussian blue cobalt green shellac dammar mastic 10 9 − 5 8 6 3.0 2.8 4.4

17 15 30 9 16 18 4.4 4.1 5.2

Thickness removed (µm) at middle at high ﬂuence ﬂuence 2.59 0.92 3.2–5.3 1.28 4.20 3.10 0.42 0.33 0.28

Initial roughness Ra (µm)

Roughness Ra (µm) (at middle ﬂuence) 3.43 3.27 7.50 2.56 8.02 4.08 0.32 0.49 0.43

Table 35.2. Thickness and average surface roughness Ra of pigmented egg tempera and varnish systems

35 Excimer Laser Ablation of Egg Tempera Paints and Varnishes 307

308

P.J. Morais et al.

Intensity (a. u.)

(A)

PbO

30

35

(B)

40

45

50

2θ (degrees) Fig. 35.1. X-ray diﬀraction patterns of (a) red lead (Pb3 O4 ) tempera paint and (b) same sample, after laser treatment with 0.23 J cm−2

Intensity (a. u.)

(A)

(B)

500

1000

1500

2000

2500

3000

3500

Raman shift (∆cm−1)

Fig. 35.2. FT-Raman spectra of (a) Prussian blue (Fe4 [Fe(CN)6 ]3 ·14H2 O) tempera paint and (b) same sample, after laser treatment with 0.69 J cm−2

low ﬂuences (see Fig. 35.1). As suggested by Zaﬁropulos et al. [16], the phase changes only occur a few nanometers from the surface (conﬁrmed by optical microscopy), which make them hard to detect by XRD in the conventional geometry conﬁguration. Preliminary results of the FT-Raman spectra of the analysed inorganic pigmented tempera paints indicate a similar behaviour compared with other previously studied pigmented egg tempera systems [14, 15], such as the decrease in the intensity of the pigments bands and the formation of extra bands after laser treatment. The Raman spectra of Prussian blue is shown in Fig. 35.2.

35 Excimer Laser Ablation of Egg Tempera Paints and Varnishes

309

100

Transmittance (%)

80

60

40 without laser treatment 0.39 J/cm2

20

0.86 J/cm2 0 300

400

500 600 Wavelength (nm)

700

800

Fig. 35.3. UV/visible spectra of light-aged shellac varnish before and after laser treatment

The spectrum of the samples without laser treatment was ﬁrst compared with the spectrum of the pure pigments. According to the literature [23], bands at approximately 280, 538, 2,100 and 2, 150 cm−1 indicate the presence of Prussian blue. These bands are the most intense in the spectra of Fig. 35.2 and the ones of higher wavenumbers can be assigned to C–N stretching vibrational modes of the hexacyanoferrate ions [24]. At that same region and at higher frequencies, the egg-binding medium also contributes to the spectra. Figure 35.3 shows the UV/visible spectra of aged shellac varnish before and after laser treatment. The band around 435 nm corresponds to the oxidation and degradation reactions promoted by the ageing treatment, which leads to the yellowing of the varnish. As it can be observed, the transmittance in the visible region decreases for middle-to-high ﬂuences of laser treatment, probably due to alterations induced by the laser radiation to the surface of the varnish, which are not visible with naked eye.

35.4 Conclusions The interaction of KrF excimer laser (λ = 248 nm) with some inorganic pigmented egg tempera paints and varnishes was studied and the samples were analytically characterised. Some of the tested paints changed colour after laser irradiation even at low energy densities, whereas others only suﬀered discoloration at middle energy densities. The tested varnish systems could be removed in a controllable manner at middle ﬂuences. From the practical point of view, laser cleaning of pigmented tempera paints is critical. Thus, a complete knowledge of the interaction between the laser radiation and pigments,

310

P.J. Morais et al.

binding medium and varnish involved is necessary before the actual cleaning procedure. Excimer laser cleaning is a valid cleaning process for the removal of polymerised varnishes and overpaint layers if the direct interaction of the laser with the material is known, as each pigment reacts diﬀerently to the laser radiation. Further analysis on the tested samples is necessary in order to fully understand the interaction of laser–material. Mixtures of pigments should be used in future research so in a later stage real samples could be tested in order to achieve a controllable laser irradiation of a painting with cleaning purposes. In addition, post-ageing procedures should be undertaken in order to investigate possible long-term eﬀects. Acknowledgements The authors would like to thank the following institutions for the access they gave in using their facilities: Universidade Nova de Lisboa, Universidade de Aveiro, Universidade de Coimbra and Centro Tecnológico da Cerâmica e do Vidro.

References 1. J. F. Asmus, G. Guattari, L. Lazzarini, G. Musumeci, and R. F. Wuerker, in Studies in Conservation Vol. 18, 49, 1973. 2. J. F. Asmus, S. G. Murphy, and W. H. Munk: SPIE 41, 19, 1973. 3. J. F. Asmus, IEEE Circuits and Device Magazine 3, 6, 1986. 4. M. Cooper, Laser cleaning in conservation: an introduction (Butterworth Heineman, Oxford, 1998) 5. A. C. Tam, W. P. Leung, W. Zapka, and W. Ziemlich, in J. Appl. Phys., Vol. 71, 3515, 1992. 6. S. Georgiou, V. Zaﬁropulos, D. Anglos, C. Balas, V. Tornari, and C. Fotakis, in Appl. Surf. Sci. Vol. 127–129, 738, 1998. 7. R. Oltra, O. Yavas, F. Cruz, J. P. Boquillon, and C. Sartori, in Appl. Surf. Sci. Vol. 96–98, 484, 1996. 8. M. Cooper and J. Larson, in The Conservator Vol. 20, 28, 1996. 9. S. Siano, F. Margheri, R. Pini, P. Mazzinghi, and R. Salimbeni, in Appl. Optics Vol. 36 (27), 7073, 1997. 10. I. Gobernado-Mitre, J. Medina, B. Calvo, A. C. Prietro, L. A. Leal, B. Pérez, F. Marcos, and A. M. de Frutos, in Appl. Surf. Sci. Vol. 96–98, 474, 1996. 11. V. Zaﬁropulos, A. Galyﬁanali, S. Boyatzis, A. Fostiridou, and E. Ioakimoglou in Optics and Lasers in Biomedicine and Culture, Series of the International Society on Optics within Life Sciences (Springer-Verlag, Berlin, 115–122, 2000. 12. M. Abraham, S. Scheerer, O. Madden, and F. Adar: SPIE 4402, 68, 2001. 13. R. Teule, H. Scholten, O. F. Van den Brink, R. M. A. Heeren, V. Zaﬁropulos, R. Hesterman, M. Castillejo, M. Martín, U. Ullenius, I. Larsson, F. GuerraLibrero, A. Silva, H. Gouveia, and M. B. Albuquerque, in J. Cult. Heritage 4, 209, 2003.

35 Excimer Laser Ablation of Egg Tempera Paints and Varnishes

311

14. M. Castillejo, M. Martín, M. Oujja, J. Santamaría, D. Silva, R. Torres, A. Manousaki, V. Zaﬁropulos, O. F. Van den Brink, R. M. A. Heeren, R. Teule, and A. Silva, in J. Cult. Heritage Vol. 4, 257, 2003. 15. M. Castillejo, M. Martín, M. Oujja, D. Silva, R. Torres, A. Manousaki, V. Zaﬁropulos, O. F. Van den Brink, R. M. A. Heeren, R. Teule, A. Silva, and H. Gouveia, in Anal. Chem. 74, 4662, 2002. 16. V. Zaﬁropulos, C. Balas, A. Manousaki, Y. Marakis, P. Maravelaki-Kalaitzaki, K. Melesanaki, P. Pouli, T. Stratoudaki, S. Klein, J. Hildenhagen, K. Dickmann, B. S. Luk’Yanchuk, C. Mujat, and A. Dogariu, in J. Cult. Heritage 4, 249, 2003. 17. O. F. Van den Brink, G. B. Eijkel, and J. J. Boon, in Thermochim. Acta, Vol. 365, 1, 2000. 18. O. F. Van den Brink, Ph. D. Thesis, University of Amsterdam, 2001. (http://www.amolf.nl/publications/theses/brink/index.html). 19. C. d. A. Cennini: Il Libro dell’Arte (Dover, New York, 1960). 20. R. W. G. Hunt: Measuring Colour (Ellis Horwood, Chichester, England, 1991). 21. M. Chappé, J. Hildenhagen, K. Dickmann, and M. Bredol, in J. Cult. Heritage, Vol. 4, 264, 2003. 22. P. Pouli, D. C. Emmony, C. E. Madden, and I. Sutherland, in J. Cult. Heritage, Vol. 4, 271, 2003. 23. I. M. Bell, R. J. H. Clark, and P. J. Gibbs, in Spectrochim. Acta A, Vol. 53A, 2159, 1997. 24. K. Castro, M. D. Rodríguez-Laso, L. A. Fernández, and J. M. Madariaga, in J. Raman Spectrosc., Vol. 33, 17, 2001.

36 Laser Cleaning of Undyed Silk: Indications of Chemical Change K. von Lerber1 , M. Strlic2 , J. Kolar3 , J. Krüger4 , S. Pentzien4 , C. Kennedy5 , T. Wess5 , M. Sokhan6 , and W. Kautek7 1

2

3

4

5

6 7

Prevart GmbH – Textile Conservation, 8405 Winterthur, Switzerland [email protected] University of Ljubljana, Faculty of Chemistry and Chemical Technology, 1000 Ljubljana, Slovenia National University Library, National Centre for Preservation of Library Materials, 1000 Ljubljana, Slovenia Federal Institute for Materials Research and Testing, Division VIII.2 Surface Technologies, 12205 Berlin, Germany Cardiﬀ University, Structural Biophysics Group, School of Optometry and Vision Science, Cardiﬀ, Wales, UK Imperial College, Materials Department, London SW7 2BP, Great Britain University of Vienna, Institute for Physical Chemistry, 1090 Vienna, Austria

Summary. Three diﬀerent undyed, unweighed silk fabrics (new clean, new soiled, and naturally aged) were cleaned with a computer-controlled Q-switched Nd:YAG laser at 532 nm in 30 combinations of ﬂuence and pulse numbers. They were studied for chemical change by viscometry, X-ray diﬀraction, and FIB-SIMS in combination with temperature calculations. While physical changes only occurred above the tested parameters, chemical changes could be detected as low as 0.2 J cm−2 with four pulses. Yellowing was observed at lower and bleaching at higher ﬂuence/pulse number combinations. Melting was observed in naturally aged silk cleaned with 64 pulses at 4.2 J cm−2 . The temperature reached at 0.1 J cm−2 is suﬃcient to evaporate carbon. Excess energy is transferred into the silk substrate causing thermal degradation. Diﬀerent chemical processes leading to chain scission and to crosslinking seem to occur simultaneously, even at low ﬂuence and pulse number. An increase in pulse numbers also leads to increasing damage.

36.1 Introduction In textile conservation, the removal of strongly attached surface soiling remains a diﬃcult task. As the visual result of conventional cleaning techniques such as vacuum cleaning, various erasing materials and solvents often remains unsatisfying, laser cleaning of silk has been attempted increasingly during the last several years. While for other materials the body of scientiﬁc

314

K. von Lerber et al.

research has been steadily growing; there have been very few papers dealing with the eﬀect of laser cleaning on silk at a level exceeding microscopic or SEM examination [1–4]. Working at a wavelength of 1,064 nm, many authors have reported yellowing [5] and crosslinking in cellulose [6], while working in the UV range on organic materials has been reported to cause a decrease in the degree of polymerization [7]. Working at 532 nm, a diﬀerence in behavior of soiled and unsoiled substrates has been reported [8]. The objective of this study is to investigate the inﬂuence of ﬂuence and pulse number of a laser cleaning treatment at 532 nm on chemical changes induced in undyed silk and to compare theoretical evaporation temperature calculations to heat-induced damage threshold values found in literature and to microscopic observations of the silk samples.

36.2 Experimental Methods 36.2.1 Sample Preparation Three diﬀerent silk fabrics were chosen for testing (1) a new, undyed, unweighted silk (tabby crepe, 60 × 40 threads per cm), (2) the same silk but mechanically soiled with carbon dust, and (3) a naturally aged silk (tabby, 55 × 42 threads per cm) of unknown provenance, unweighted but slightly yellowed. All three silks were treated with a Nd:YAG laser at a wavelength of 532 nm and with 30 combinations of ﬂuence and pulse rates: 0.025, 0.05, 0.1, 0.2, 0.3, 0.5, 1.0, 1.5, and 4.2 J cm−2 with 1, 4, 16, and 64 pulses. The laser treatment was performed on a computerized prototype laser cleaning system based on a diode pumped Q-switched Nd:YAG laser operating with a pulse duration of 8 ns, a repetition rate of 500 Hz, and a maximum energy of 2.5 mJ. The setup consisted of a scanning optical system (410 mm focal length) which delivered a spot diameter of 240.5 µm and maximum ﬂuence in the range of up to F (532 nm) = 4.5 J cm−2 . An integrated exhaust system served to remove volatile debris. The computerized system allowed for maximal repeatability; the overlap of the Gaussian beam distribution was chosen to produce an absolutely uniform distribution of ﬂuence over the entire sample surface. 36.2.2 Viscometry Possible changes in length of polymer chains were studied by viscometry. From 50–85 mg of silk (depending on DP) was dissolved in 8 ml of 10 mol l−1 aqueous solution of LiSCN [9]. The suspension was stirred for 1 h, then left overnight. To the clear solution, 4 ml of water was added, thus obtaining 12 ml of silk solution, 6.67 mol l−1 LiSCN. The solution was centrifuged for 10 min at 3,000 rpm and the clear solution was then used for analysis (“stock solution”). The viscosity of the stock solution was determined, after which 6 ml of it

This page intentionally blank

36 Laser Cleaning of Undyed Silk: Indications of Chemical Change

315

was diluted with 3 ml of 6.67 mol l−1 LiSCN, and a further 3 ml of the stock solution was diluted with 6 ml of 6.67 mol l−1 LiSCN. Viscosities of all these solutions as well as of LiSCN solution were determined with a Unitex-type (Eﬀe-ci, Italy) viscometer at 25◦ C. The degree of polymerisation (DP) was calculated and expressed relative to the same fabric not treated with laser. 36.2.3 X-ray Diﬀraction The samples were mounted in the sample chamber of the NanoSTAR (Bruker AXS, Germany) facility and placed under vacuum. The data collection procedure used followed that described by [10]. Diﬀraction proﬁles were taken over 3 h exposures using a sample-to-detector distance of 4.5 cm. Collected data was corrected for camera distortions, a background image was subtracted, and images were analyzed using in-house software. The two-dimensional detector output was converted to one-dimensional linear proﬁles for analysis. As silk does not give isotropic diﬀraction patterns, only the equatorial regions encompassing the main 210 reﬂection were taken. Each sample has two diffraction images taken at diﬀerent locations on the sample to ensure reproducibility. Linear proﬁles were analyzed using the program PeakFit4 (AISL software). The crystallinity of the peak was then calculated using the methodology of [11]. 36.2.4 Focused Ion Beam, Secondary Ion Mass Spectrometry (FIB-SIMS) A FIB 200 TEM system operating at 30 keV with a gallium primary ion beam from a liquid metal ions source, equipped with SIMS detector and spectrometer has been used. The positive primary ion beam is produced by the application of a strong electrical ﬁeld. By changing of apertures it is possible to change the primary beam current from 10 to 20,000 pA depending on the analysis required, and to adjust the beam focus. The high current is usually set for milling procedure and small current for imaging or mass-spectrometry. The sample is mounted on the circular disc with conducting carbon tape to secure the position and to provide a conducting path with the manipulating stage inside the analysis chamber. The vacuum of the analysis chamber should not be lower than 10−6 Torr. The images of the scanned surfaces are produced by the sputtered secondary electrons or ions but, because the yield of sputtered electrons is much higher than sputtered ions, the resolution of these images is much better. The SIMS detector and spectrometer of the FIB equipment render the unique possibility to combine secondary electron imaging with depth proﬁling, mass spectrometry, and elemental imaging techniques. The SIMS tool on the FIB 200 is operating by the software package: SIMSmap IIIxP Analysis System.

316

K. von Lerber et al.

36.2.5 Theoretical Calculation of Temperature Increase T As the focus of the laser beam (ca. 240 µm) is much larger than the heat aﬀected zone (6 µm), the heat conduction equation can be applied in a onedimensional form [12]. D [2 (1 − R) F0 ] (36.1) T = K (τ π) where R is the reﬂectivity, F0 the laser ﬂuence (J cm−2 ), K the heat conduction coeﬃcient (W cm−1 K−1 ; J s−1 cm−1 K−1 ), D the thermal diﬀusion coeﬃcient [cm2 s−1 ], and τ is the pulse duration (s). The values for D and K can be taken from literature [13]. For graphite, the following temperature increase can be calculated: at 0.5 J cm−2 , T = 1, 119 K (overall temperature including a room temperature of 20◦ C is 1, 139◦ C); at 0.2 J cm−2 , T = 447 K (overall temperature, 467◦ C) and, at 0.1 J cm−2 , T = 223 K (overall temperature, 243◦ C). A ﬂuence on the order of 0.1 J cm−2 has been calculated to be the minimal energy necessary for removal of a carbon dust layer of ca. 0.5 µm thickness. For more details concerning the calculation, see [14].

36.3 Results and Discussion Previous analytical results of the very same samples by means of colorimetry, microscopy, and polarized FTIR [15] suggested that chain scissioning and crosslinking occur simultaneously. At lower ﬂuences and pulse numbers, chain scissioning, leading to polar groups, seems to predominate whereas, at higher ﬂuences and pulse numbers, crosslinking indicated by an increase in conjugated systems seems to predominate. Aged silk is crosslinking at much lower ﬂuences and pulse number than new silk, due to its already deteriorated condition. The carbon soiling seems to enhance the process leading to chain scissioning. Viscometry was only performed on the new silk and the new silk soiled with carbon, as the old silk was too degraded to provide readings above error margin. Viscosity of the carbon soiled silk generally decreased, while new silk increased in viscosity. In order to clarify the overlapping of chain scissioning and crosslinking, Fig. 36.1 compares the results of viscometry (lower two plots) and X-ray diﬀraction (upper two plots) performed on the same samples. The plots compare diﬀerent pulse numbers at the consistent ﬂuence of 1.5 J cm−2 . At four pulses, viscosity of the soiled silk drastically decreased, whereas the clean silk increased in viscosity. Comparison with the X-ray diﬀraction suggests a constant loss of overall crystallinity. This suggests that the supramolecular structure of the silk might be disrupted, and that crosslinking would not lead to a more crystalline, but

36 Laser Cleaning of Undyed Silk: Indications of Chemical Change

60

silk new Sn_2, viscometry silk new soiled Snv_5, viscometry silk new Sn_2, x-ray diffraction silk new soiled Snv_5, x-ray diffraction

40

40

20

20

0

0

x-ray diffraction(%)

change in viscometry (%)

60

−20

−20

−40

317

0

10

20

30

40

50

60

−40

number of pulses at 1.5 Jcm-2

Fig. 36.1. Results of viscometry (below ) and X-ray diﬀraction (above) for new silk and new carbon soiled silk treated at 1.5 J cm−2 with varying number of laser pulses

to a more interlinked molecular system, as has been suggested for tyrosine side groups interlinking and forming melanine- and chinone-like chromophore structures [16, 17], which could possibly account for the observed yellowing and browning. The loss of overall structure is also supported by our previous research, where a disruption within the crystalline structure has been observed by polarized FTIR: the crystallites which in new silk are nicely aligned parallel to the ﬁber axis have tilted out of this axis as of 0.5 J cm−2 and four pulses, suggesting disruption in amorphous regions of the ﬁber. The comparison of viscometry and X-ray diﬀraction also suggests that an increasing number of laser pulses will cause an increase in disruption of the crystalline structure of silk. This suspected loss of supramolecular structure seems to be supported by the temperature calculations: the graphite layer modeling dirt on the silk is very thin (estimated to about 0.5 µm), therefore the linear decrease of temperature upon penetration of the dirt layer will be comparatively small, causing excess energy to dissipate into the silk. The following indications were found in literature [18, 19] (Table 36.1). As these critical temperatures are by far surpassed above 0.2 J cm−2 , the observed phenomena (melting, disruption of crystallinity) seem but a logic consequence. As not all energy introduced by the laser beam can be transformed into movement (ejection of dirt), there will be heat and ultrasonic

318

K. von Lerber et al. Table 36.1. Temperature related changes in silk ﬁbroin

Temperature Observed eﬀect below 100◦ C dehydration above 120◦ C shrinkage due to rearrangement of ﬁbroin molecules in amorphous regions above 210◦ C breaking of side chains, disruption of the β-pleated sheet structure at 215–250◦ C loss of side chains above 250◦ C melting of crystallites within the β-pleated sheet structure, loss of intermolecular H-bonds.

Fig. 36.2. FIB-SIMS image: Melting of the surface of a naturally aged silk ﬁber cleaned with two laser pulses at 4.2 J cm−2 (left: magn. ×25, 000, right: ×50, 000). Silk ﬁbers measure 9–11 µm in diameter. In the right image, the line of holes probably shows melting along a ﬁbroin ﬁbril (1 µm diameter), suggesting disruption of ﬁbril bundle structure

waves propagating into the substrate even at temperatures below the calculated evaporation point for a speciﬁc layer, thus causing, e.g., browning or even melting in the silk substrate. FIB-SIMS provided high resolution images at very high magniﬁcation and made the suspected melting of the ﬁbroin ﬁbers visible, supporting the assumption of supramolecular change being caused by the laser treatment (Fig. 36.2). The data collected, due to ﬁrst uncertainties using the technique on silk, will have to be backed by future research. However, results gained during this project suggest that FIB-SIMS can be a valid tool for measuring change in certain end and side groups of ﬁbroin, indicating change in the condition of the silk. The results support the ﬁndings of the other analytical techniques used in this project not only increasing ﬂuence, but also additional pulses have an impact on the degree of change.

36 Laser Cleaning of Undyed Silk: Indications of Chemical Change

319

36.4 Conclusions A minimal ﬂuence of 0.1 J cm−2 has been calculated to be necessary for the removal of carbon dust, corresponding exactly to visual examination: for four pulses at 0.16 J cm−2 , the cleaning was considered to be just barely suﬃcient but was already causing yellowing. Any excess energy being applied (higher ﬂuence, additional pulses) will cause further thermal degradation of silk, including disruption and melting of the crystalline structure, thus causing loss of tensile strength and discoloration. Such melting can be made visible under very high magniﬁcation. Thermal degradation in silk is reported to include chain scissioning of the peptide bond, leading to a radical chain reaction [20], thus negatively inﬂuencing the long-term preservation of the silk. Therefore laser cleaning for silk substrates seems to pose rather high risks, which will have to be carefully evaluated against risks of other possible cleaning methods. Acknowledgments This study was performed for a diploma thesis entitled Untersuchung zur Reinigung ungefärbter Seide mit Laser, submitted in June 2004 in partial fulﬁllment of the requirements of the degree in textile conservation at AbeggStiftung and Berne University of Applied Sciences, Switzerland.

References 1. M. Strlic, J. Kolar et al., in Applied Surface Sciences, Vol. 207, 236, 2003. 2. P. Garside and P. Wyeth, in Preprints 3rd North American Textile Conservation Conference, Edited by V. J. Whelan, 55, 2002. 3. G. P. Kelly, M. Mollah, and F. Wilkinson, in Journal of the Textile Institute, Vol. 81 (1), 91, 1990. 4. K. von Lerber, S. Pentzien, M. Strlic, and W. Kautek, in ICOM Committee for Conservation 14th triennial meeting The Hague preprints, Editors A. Boccia Paterakis, et al., Vol. 2, 978, 2005. 5. C. Dignard, American Institute for Conservation, Objects specialty group website: http://aic.stanford.edu/sg/osg/dignard_biblio.htm (last accessed 8.9.2005). 6. J. Kolar, M. Strlic et al., in Journal of Cultural Heritage, Special Issue Lacona IV, 185, 2003. 7. J. Kolar, M. Strlic et al., in Applied Physics A – Materials Science & Processing, Vol. 71, 87, 2000. 8. J. Kolar, M. Strlic et al., in Journal of Cultural Heritage, Special Issue Lacona IV, 185, 2003. 9. S. Tse and A. L. Dupont, in Historic Textiles, papers and polymers in museums, ACS Series 779, Editor J. M. Cardamomo and M. T. Baker, 98, 2001. 10. T. J. Wess et al., in Archaeometry, Vol. 43:1, 117, 2001.

320

K. von Lerber et al.

11. M. Burghammer, M. Muller, and C. Riekel, in Recent Research Developments in Macromolecules 7, 103–125, 2003. 12. E. G. Gamaly, A. V. Rode, and B. Luther-Davies, in Journal of Applied Physics, Vol. 85, 4213, 1999. 13. D. Bäuerle, Laser Processing and Chemistry, 3rd ed. 2000. 14. J. Krüger, S. Pentzien, and K. von Lerber (to be published in the postprints of Lacona VI). 15. K. von Lerber, S. Pentzien, M. Strlic, and W. Kautek, in ICOM Committee for Conservation 14th Triennial meeting, The Hague preprints, Editors A. Boccia Paterakis et al., Vol. 2, 978, 2005. 16. R. M. Robson, in Fibre Chemistry-Handbook of Fibre Science and Technology, Vol. IV, Editor M. Lewin and E. Pearce, 647, 1985. 17. A. Timar-Balaszy and D. Eastop, Chemical Principles of Textile Conservation, 1998. 18. P. Garside and P. Wyeth, in Preprints 3rd North American Textile Conservation Conference, Editor V. J. Whelan, 55–60, 2002. 19. M. Nagura and H. Ishikawa, in Polymer, Vol. 24, 820, 1983. 20. S. P. Hersh, P. A.Tucker, and M. A. Becker, in Archaeological Chemistry IV, Advances in Chemistry Series 220, Editor R. O. Allen, 429, 1989.

37 Determination of a Working Range for the Laser Cleaning of Soiled Silk ∗

J. Krüger1 , S. Pentzien1 , and K. von Lerber2 1

∗ 2

Division VIII.2 Surface Technologies, Federal Institute for Materials Research and Testing (BAM), Unter den Eichen 87, 12205 Berlin, Germany [email protected] Prevart GmbH, Oberseenerstr. 93, 8405 Winterthur, Switzerland

Summary. Nanosecond laser (532 nm) cleaning of soiled silk (naturally aged and new fabrics) is discussed in terms of an adequate choice of the processing parameters laser ﬂuence and pulse number to achieve a satisfying cleaning eﬀect and to avoid damage of the sensitive textiles. Experimental limits will be presented and compared to theoretical considerations utilizing graphite as a model for the soil.

37.1 Introduction The removal of soil from fragile silk is a challenging task in textile conservation. Conventional mechanical or chemical treatment often results in an unsatisfying cleaning quality due to a strong adherence of the soil to the silk surface. A laser processing might be an alternative. Recently, a few papers were published concerning that topic (e.g., [1, 2]). A brief review of relevant literature and the main ﬁndings described there is given in [3] with the focus on possible chemical changes of the silk as an unwanted result of laser exposure. In this paper, the determination of a laser working range with respect to laser ﬂuence and number of pulses per spot required for the realization of a successful cleaning procedure is presented. The upper (mechanical and/or thermal damage of the silk) and lower limits (removal of the soil) for a cleaning process employing nanosecond laser pulses at 532 nm wavelength are investigated from a physical point of view.

37.2 Experimental Methods Three diﬀerent unweighted silk samples (two model systems and one naturally aged fabric) were studied. For the model systems, undyed silk (tabby crepe, 60 × 40 threads cm−1 ) was used as received (new clean silk) or mechanically

322

J. Krüger et al.

Fig. 37.1. Prototype laser cleaning system at the Federal Institute for Materials Research and Testing (BAM)

soiled with carbon dust (new soiled silk). The naturally aged silk (tabby 55×42 threads cm−1 ) of unknown provenance showed a marginally yellow color. The experimental setup is depicted in Fig. 37.1. The investigations were performed with a computer-controlled Nd:YAG laser cleaning system (532 nm wavelength) delivering pulses with 8 ns duration at 500 Hz repetition rate. A focus diameter (1/e2 ) of ca. 240 µm was chosen. It was measured with single pulses of diﬀerent energies impinging on photographic paper. The size of the blackened areas was determined using an optical microscope (Eclipse L200, Nikon). The correlation between pulse energy and blackened area was ﬁtted according to a spatial Gaussian beam distribution [4]. The laser spot was scanned over the silk samples resulting in laser-treated areas of 6 × 6 mm2 . For a ﬁxed laser ﬂuence, the number of pulses per spot was changed between 1 and 2,500. The operator can follow the cleaning process on a computer screen (not shown here) by means of a camera. For safety reasons, the whole laserprocessing compartment fulﬁlls Laser Class I conditions. Additionally, an exhaust system is integrated to remove volatile ablation products.

37 Determination of a Working Range for the Laser Cleaning

323

37.3 Results and Discussion For the diﬀerent silk samples, series of experiments with varying laser ﬂuence and number of pulses per spot were conducted. The cleaning eﬀect depending on both parameters was studied. Figure 37.2 shows results for the laser processing of new soiled silk with low numbers of pulses per spot as an example. The photograph indicates that a proper choice of laser ﬂuence and number of pulses per spot is necessary to reach a satisfying cleaning eﬀect. Figure 37.3 displays a comparison of an untreated surface to a laser-cleaned area of new

Fig. 37.2. Laser cleaning of new soiled silk. Photograph of a matrix of laser-treated areas produced with diﬀerent pulse numbers and laser ﬂuences applied to the same spot

Fig. 37.3. Laser cleaning of new soiled silk. Laser ﬂuence 0.5 J cm−2 , 2,500 pulses per spot. Optical microscope image of the border between original (soiled ) and laser-cleaned surface

324

J. Krüger et al.

Fig. 37.4. Laser treatment of new soiled silk. Determination of destruction thresholds in dependence on the number of pulses per spot. A physical damage is present if the color of the laser-treated area changes from ocher to white

soiled silk. The microscopic view demonstrates that a successful cleaning can be achieved from a physical point of view. The application of high laser ﬂuences for a rapid cleaning of the soiled silk (compare Fig. 37.2) is limited by the mechanical and/or thermal destruction thresholds of the textiles. As an example, the results of a laser illumination of new soiled silk with 1.8 and 3.6 J cm−2 for several pulse numbers are depicted in Fig. 37.4. One thousand and ﬁve hundred pulses of 1.8 J cm−2 and 500 pulses of 3.6 J cm−2 are suﬃcient for an irreversible physical damage of the silk. In this case, interleafs were generated (circles in Fig. 37.4). Experiments according to Fig. 37.4 were performed for three diﬀerent textiles. A diagram of destruction thresholds in dependence on the number of pulses applied to the same spot was obtained (Fig. 37.5). First of all, a decreasing destruction threshold can be found for an increasing number of pulses per spot for new soiled silk and naturally aged silk, respectively. This phenomenon is well known as an incubation behavior [5]. Successive laser pulses modify the material and enhance the number of absorbing sites in the sample. The increasing energy absorption in the laser-illuminated volume lowers the threshold values. Only the ﬁrst laser pulse interacts with an unaltered specimen showing the original optical properties. Second of all, a relation for the physical damage thresholds Fth according to Fth (naturally aged silk) < Fth (new soiled silk) < Fth (new clean silk) can be found. The reasons for this inequality might be a various density of absorbing inclusions and defects present in each sample even at the beginning of the treatment (compare new clean silk and new soiled silk) and possible chemical alterations between new and naturally aged silk [3]. In the next part of the paper, the minimum laser ﬂuence needed to remove the soil will be calculated. Graphite serves as a model for the soil.

37 Determination of a Working Range for the Laser Cleaning

325

Fig. 37.5. Physical damage thresholds for 532-nm nanosecond laser processing of textiles

It is assumed that a single laser pulse with a rectangular shape in time (pulse duration τ ) and a laser ﬂuence F0 is incident on a plane solid target. The sample absorbs the pulse energy and will be heated. Losses due to radiation and plume expansion are neglected. If a relation: light penetration depth heat aﬀected zone laser focus diameter is fulﬁlled (graphite: 73.5 nm 6.3 µm 240 µm), the onedimensional heat conduction equation can be applied to determine the maximum temperature rise ∆T at the surface of a bulk graphite target [6] 2(1 − R)F0 D (37.1) ∆T = κ πτ with reﬂectivity R, thermal conductivity κ, and thermal diﬀusivity D. If the temperature at the surface equals the boiling temperature of the graphite Tb , i.e. Tb − T0 = ∆T with room temperature T0 , the ﬂuence threshold for the evaporation of bulk graphite Fth (bulk graphite) is reached and can be written as κ(Tb − T0 ) πτ . (37.2) Fth (bulk graphite) = 2(1 − R) D With the parameters κ = 20 W cm−1 K−1 , Tb = 4,623 K, D = 12.58 cm2 s−1 (all from [5]), T0 = 293 K, τ = 8 ns, R → 0, an evaporation threshold for bulk graphite of 1.9 J cm−2 can be derived. It is evident (Figs. 37.2–37.4), that a laser ﬂuence of 1.9 J cm−2 is sufﬁcient to remove soil from silk. But, the calculated bulk graphite evaporation threshold is by an order of magnitude higher than the minimum ﬂuence of 0.1–0.2 J cm−2 needed for cleaning according to the experimental results shown in Fig. 37.2. The reason for this apparent discrepancy is a transition from bulk conditions (calculation) to thin layers of soil (graphite). If the layer

326

J. Krüger et al.

Fig. 37.6. Optical microscope image of a single ﬁber of new soiled silk. Reference condition without laser illumination

to be removed is thinner than the heat aﬀected zone HAZ = 2(Dτ )∧ 0.5 = 6.3 µm, the evaporation threshold of the layer Fth (layer) scales linearly with the thickness of the layer d according to [7] Fth (layer) = Fth (bulk)

d . HAZ

(37.3)

Utilizing (37.3), the determination of a theoretical evaporation threshold of the soil Fth (layer) is possible if an average thickness d of the dust layer would be known. The average thickness d is estimated using an optical microscopic image of a single ﬁber of new soiled silk (Fig. 37.6). The arrows indicate the soil ﬁlm, and an average thickness below 1 µm(d ≈ 0.5 µm) can be estimated. With the knowledge of the theoretical values Fth (bulk) = 1.9 J cm−2 (37.2), HAZ = 6.3 µm and the thickness d ≈ 0.5 µm, Fth (layer) = 0.15 J cm−2 can be determined (37.3). The theoretical value is in excellent agreement with the experimental ﬁnding that a laser ﬂuence of 0.16 J cm−2 is suﬃcient to achieve a weak cleaning eﬀect (Fig. 37.2). It should be noted that even close to cleaning threshold conditions with ﬂuences of the order of 0.1 J cm−2 , chemical modiﬁcations of the thermally sensitive silk due to a comparatively high heat load cannot be excluded [3].

37.4 Conclusions A laser with 532 nm wavelength and 8 ns pulse duration was employed for cleaning experiments of diﬀerent silk samples. Removal of soil from the fabrics without macroscopic destruction was possible in a laser ﬂuence range

37 Determination of a Working Range for the Laser Cleaning

327

between 0.2 and ca. 2 J cm−2 utilizing pulse numbers < 100. New soiled silk showed higher destruction thresholds than naturally aged silk. An incubation behavior, i.e. a decrease of the physical damage threshold with the number of pulses applied to the same spot, was observed for all textiles. The evaporation threshold of the soil was calculated as the lower limit for a cleaning procedure using graphite as a model substance. The experimental ﬁnding of a soil evaporation threshold of about 0.2 J cm−2 was conﬁrmed theoretically.

References 1. M. Strlic, J. Kolar, V.-S. Selih, and M. Marincek, in Applied Surface Science, Vol. 207, 236, 2003. 2. D. S. McPhail, M. Sokhan, E. E. Rees, B. Cliﬀ, A. J. Eccles, and R. J. Chater, in Applied Surface Science, Vol. 231–232, 967, 2004. 3. K. von Lerber, M. Strlic, J. Kolar, J. Krüger, S. Pentzien, C. Kennedy, T. Wess, M. Sokhan, and W. Kautek, in Conference Proceedings of LACONA VI, this issue. 4. J. M. Liu, in Optics Letters, Vol. 7, 196, 1982. 5. D. Bäuerle, Laser Processing and Chemistry, 3rd Ed., Springer, Berlin, 2000. 6. E. G. Gamaly, A. V. Rode, and B. Luther-Davies, in Journal of Applied Physics, Vol. 85, 4213, 1999. 7. E. Matthias, M. Reichling, J. Siegel, O. W. Käding, S. Petzoldt, H. Skurk, P. Bizenberger, and E. Neske, in Applied Physics A, Vol. 58, 129, 1994.

38 Laser Versus Conventional Cleaning Methods: Do the Costs Outweigh the Beneﬁts? ∗

P. van Dalen1 , R. Broere1 , and H.A. Aziz2 1 ∗ 2

Art Conservation b.v., Vlaardingen, The Netherlands [email protected] Hadeel Abdul Aziz, TNO Bouw, Van Mourik Broekmanweg 6, 2628 XE, Postbus 49, 2600 AA Delft, The Netherlands

Summary. Art Conservation B.V. participates in the PaReLa project, a European Commission funded, cooperative research project entitled ‘Paper Restoration using Laser Technology’ (EVK4-CT-2000-30002). The purpose of this project is to develop a laser system suitable for the accurate, eﬃcient and safe cleaning of paper objects. Other participants include BYB, Guillet, Hai Yen, Susan Corr, TNO, BAM, ICN and FORTH. Our paper discusses the research outcomes and the feasibility of this method in practice. For research purposes we aged a great number of samples, plastered with diﬀerent types of tape, in special ovens. Next we carried out experiments using various types and doses of laser light to determine how the adhesive residues can best be removed. It appears that the laser system can yield excellent results. However, we feel that, in practice, requirements may well diﬀer from those set in the PaReLa project. In order to appreciate the added value of the laser cleaning, we also tested conventional methods. Thus, our paper also provides an assessment of established methods to remove marks, inks, and adhesive discolorations. It is well known that these methods have several disadvantages. For instance, the use of solvents may be harmful to both the restorer and the object, and there is a risk of tidemarking in the long run. Yet the conventional methods are often preferred because they are more familiar to restorers, and, not unimportantly, because they are cheaper. Finally, in a detailed case involving removal of tape and remains of glue from a ﬁlm poster, we will compare laser treatment and a conventional method. In conclusion, laser cleaning has beneﬁts over other, conventional methods. However, as it proves to be more expensive, it is little used in practice. We at Art Conservation indeed wonder whether the researchers’ views on today’s conservation practices are realistic enough.

38.1 Introduction Cleaning of valuable documents, drawings and objects of graphical art is often technically diﬃcult, time consuming, and introduces a risk of causing undesired immediate or long-term side eﬀects. Especially diﬃcult cleaning problems can arise in the context of local cleaning of paper near sensitive media.

330

P. van Dalen et al.

Nevertheless paper conservators often do decide to carry out speciﬁc cleaning treatments on objects made of paper. The decision to carry out a speciﬁc cleaning treatment is based on several considerations. Not all stains, dirt and other types of ‘foreign materials’, actually require removal. In many situations, stains represent important information on the object’s cultural or historical context. In these cases stains are considered to be an integral part of the object and actually add to the object’s value. In other cases however the visual disturbance associated with the presence of a stain can prevail over historic considerations. Stains can be very distractive, are known to be unstable and are expected to develop into states of increasing levels of visual disturbance. In such cases a cleaning treatment of the object might be considered. The decision to carry out a speciﬁc cleaning treatment should always be the result of balancing the anticipated improved aesthetics against possible risks associated with the cleaning action. A speciﬁc cleaning procedure is considered suitable for a speciﬁc application only when suﬃcient removal is expected under the restriction that no signiﬁcant risk of immediate or long term undesired side eﬀects is introduced as a result of the treatment. In this project, laser cleaning would be considered superior with respect to a conventional cleaning technique (solvents/mechanical) either by yielding a better immediate and long term aesthetical result or, in the case of comparable aesthetical results, by some other indirect (i.e. non-aesthetic) improved performance. Examples of indirect improved performance might be a reduction of treatment time, or better long-term mechanical behaviour of the paper substrate. Even the health of the conservator (for example: laser versus toxic organic solvents), can weigh as an indirectly improved performance. To establish speciﬁc applications for which laser cleaning would be suitable and superior over existing conventional cleaning methods, and to aid the design of the PaReLa Laser station, the aesthetic aspects of several conventional and laser cleaning results have been evaluated and compared to the results of conventional cleaning methods. The methodology and results of the aesthetical evaluation of these cleaning trials will be presented here [1–6].

38.2 Assessment of Conventional Cleaning Methods If hardened adhesive has remained on the surface, it is sometimes possible to prise it oﬀ with a scalpel, risking loss of paper ﬁbre. Usually the adhesive has penetrated into the structure of the paper to some degree and a solvent is required. Solvents are an important tool in the conservator’s repertoire and, used carefully, can achieve great results. But there are accompanying risks. All solvents are toxic to the conservator to some degree, some solvents can leave an oily residue in the paper, solvents can soften the binding agent in ink and paint media and solvents can reduce the natural moisture content of the paper ﬁbre at a molecular level. It is not possible to rehydrate paper at

This page intentionally blank

38 Laser Versus Conventional Cleaning Methods

331

Fig. 38.1. Tidemarking

this level. Therefore, the method of solvent application requires consideration sensitive to the media and the paper. Local treatment methods include poultice, exposure to solvent fumes and treatment on the vacuum table. Such treatments reduce the overall amount of solvent required, the solvent can be introduced to the paper in a controlled manner, sensitive media will not be exposed and the dissolved adhesive can be clearly seen as it wicks out onto the absorbing ﬁlter papers/blotters. The disadvantages of localized treatment include the risk that the solvent may carry dissolved material further into the paper ﬁbre, leaving a residue at the interface between the treated and untreated areas. This can result in tidemarking, Fig. 38.1, that appears in later years. Treatment can be slow and the application of the solvent may alter the nature of the adhesive as evaporation occurs. The repeated application of a solvent, particularly on the vacuum table, often causes the adhesive to become resistant to the solvent in use. Newly dissolved adhesive appears to move and ‘harden’ in a translucent band along the innermost part of the tape. This new translucent line requires a stronger solvent for its removal. Total immersion of the paper in a suitable solvent allows for even dispersion of the adhesive into the solvent solution. This reduces the possibility of future tidemarking and ensures that all soluble adhesive can be rinsed out. It can be a quick process and may allow for the treatment of several pieces at once. However, the adhesive will most likely migrate throughout the complete object. In addition, the paper ﬁbre can be saturated with solvent, which can

332

P. van Dalen et al.

remain for long periods of time before complete evaporation, thus increasing the risk of inducing chemical changes in the inks. Finally, total immersion requires a large quantity of solvent for washing and rinsing. It also requires a ventilation system over and above that which may be necessary for local treatment and procedures must be in place for the safe disposal of large quantities of solvent.

38.3 Laser Cleaning As described above, conventional cleaning methods (mechanical, wet) are not always suﬃcient for the restoration of brittle papers, ﬁssures and sensitive inscriptions. Partial cleaning of paper in the vicinity of sensitive media, such as water-colour, is particularly diﬃcult because of the lack of precision using conventional techniques. In these cases, where high spatial accuracy and localized treatments is necessary, laser cleaning might promise to be an additional tool for conservators.

38.4 Results of Laser Applications 38.4.1 Removal of Stamps There are diﬀerent types of stamps and felt-tip pens but, in general, they have the same drawback: in the course of years they can migrate to the front side of the object. Another frequent problem is that the ink has already migrated to the front. In both cases application of the green laser light might be a solution. In the former case, laser reduction of the ink might prevent the ink from migrating. In the second case the mark on the verso can be reduced to such an extent that it becomes less visible on the recto. An alternative is treating the front of the object. The migrated mark can be reduced on the recto without aﬀecting the text or the image. 38.4.2 Removal of Graphite Graphite may be present in the shape of pencil notes. These are usually removed by means of gum in solid or powder form. The harmful eﬀects for the object to be treated consist of leaving a gum residue and surface damage to the paper substrate. Provided the settings are appropriate, laser treatment will not damage the paper substrate. Because nothing is added during treatment, harmful residues will not remain.

38 Laser Versus Conventional Cleaning Methods

333

38.4.3 Removal of Tapes According to Art Conservation, the criteria for assessment of the laser-treated samples were also too high. Comparing these samples with the samples treated by conventional methods, in a number of cases the criteria can be adjusted by considering the following: – Is tape present verso or recto? – Does the object contain a medium that is sensitive to solvents? In a number of cases this would be in favour of the laser treatment. Obviously some conventional techniques were quite successful, such as the removal of scotch tape. However, here we were faced with the problem that the applied ageing method had caused the substrate of several test papers to shrink to such an extent that the substrate had come apart from the adhesive layer. The surface of the paper had consequently been damaged as well. We conclude that this test was of no value, seeing that substrate and adhesive layer had already come apart as a result of the ageing method, and therefore it was no longer necessary to remove the substrate or the adhesive layer. In the case of woody paper, the substrate had not come apart from the adhesive layer. The substrate could be easily removed using a solvent, an isoamylacetate. As we were dealing with synthetic/acrylate glue here, the adhesive layer had to be scraped oﬀ using an isoamylacetate. This treatment carries the risk that the adhesive will further penetrate into the paper ﬁbres. Another problem is that we used blank paper in these experiments. In practice, however, paper usually holds a medium such as printing ink, which may be sensitive to solvents such as acetone or turpentine. By the way, this is obviously the case in all local treatments. In a number of cases, the surface of the laser-treated objects will evidently be damaged mechanically as well. But compared with the samples treated by the conventional method, the results obtained can certainly be called encouraging. Upon ageing of the samples treated by the conventional method, we noticed how harmful this method is. Yet it was always considered as one hundred percent safe to the object, on the premise that treating just the stained position would cause the least possible damage to the rest of the object. Most alarmingly, these methods are widely applied and are seen as reliable and safe. Another method frequently used for the removal of hardened adhesive layers and ink stains on the recto is the so-called ‘mechanical method’. The mechanical method is typically applied using a scalpel knife or a spatula. A layer of the staining is scratched oﬀ. This evidently results in scraping oﬀ a layer of paper as well. A frequent problem associated with this method is damage to the paper as a result of the scalpel knife cutting through the paper.

334

P. van Dalen et al.

38.4.4 Disadvantage of Wet Treatments In wet treatment of objects containing water-sensitive media in the shape of stamp ink or felt-tip pen marks, these media should be removed. Otherwise they could migrate to recto or wick up. The conventional solution for this problem is removal of the marks on the vacuum table using water or a solvent. Laser treatment for removal of such marks is a safe method. 38.4.5 Exposure Levels for the Restorer Environmental eﬀects of laser cleaning of paper samples were compared to the environmental eﬀects of manually treated paper samples. For this purpose, the cleaning of self adhesive tape was demonstrated by means of a UV laser and by means of organic solvents. During both treatments, the emissions of Volatile Organic Compounds (VOCs) were measured. VOCs are emitted when using organic based solvents, for example, to remove unwanted tapes. Examples of VOCs are formaldehyde, heptane and benzene. Based on the obtained results, we conclude that laser cleaning of posters containing pressure sensitive tapes has a signiﬁcant lower impact (approximately 40%) on the exposure to employees of emitted VOCs. None of the compounds found exceeded the maximum acceptable value (MAC level).

38.5 Results of Laser Application in Practice: A Case Study 38.5.1 Method The required laser energy is determined using a matrix on blank paper. The matrix consists of 25 lasered squares with diﬀerent beam spot overlap and energy, ranging from 10% to 80% overlap and from 10% to 80% of maximum laser energy at a ﬁxed pulse rate of 1,000 Hz. In preparation of the actual lasering of the object, tests are performed at the least visible functional sites, in areas measuring a maximum of 5 × 5 mm. However, before testing on the real object, tests are ﬁrst performed on a reference sheet of blank paper containing graphite in order to assess diﬀerences in eﬀect of laser treatment. Not until then is the actual object treated. After microscopic evaluation of the result we will decide on possible further treatment. 38.5.2 Samples We present an object from the collections of the Theater Instituut Nederland, housed in Amsterdam, The Netherlands. The object (Fig. 38.2) shows cellotape along the edges, both on recto and verso. The adhesive has caused

38 Laser Versus Conventional Cleaning Methods

335

Fig. 38.2. Object with Cellotape

browning of the paper and has made it transparent. As a result of this transparency, the laser beam is theoretically not absorbed and ablation will therefore not take place. However, incidentally the adhesive stains would absorb the laser light to some extent, in which case a cleansing eﬀect (ablation) might be reached. For testing, we used a 532 nm Nd:YAG. The liner of the tape is removed using a hot air tool. The adhesive remains attached to the paper. Laser treatment is determined by the two following factors: a. Is material to be removed superimposed? b. Degree of absorption For actual treatment the object is digitally imaged using a Multi-Spectral Camera (MSI). The digital image is processed in such a way that the laser will treat just the parts to be cleaned. 38.5.3 Results In the end we decided not to use laser treatment. On the grounds of test results we expected that lasering would not be eﬀective owing to the paper’s transparency (Fig. 38.3). Surprisingly, ablation occurred on recto as well as on verso. This presented us with a control problem: during treatment we might accidentally remove text on the verso side. As a matter of course, a large part of the staining can be removed, but the time needed for treatment is longer than expected. Thus, treatment would be possible but, on account of its long duration, it is much more expensive than application of a conventional method would be.

336

P. van Dalen et al.

Fig. 38.3. Test result

Fig. 38.4. After laser cleaning

38.5.4 Conventional Method The liner of the tape is removed using a hot air tool. The adhesive remains attached to the paper. Treatment is now determined by the three following factors: a. Type of liner (tape) b. Applied media (e.g. printing ink, water colour paint) c. Solubility of the adhesive layer (rubber-based or synthetic)

38 Laser Versus Conventional Cleaning Methods

337

38.5.5 Method Microscopic evaluation will identify the type of tape applied: rubber-based or synthetic. On the basis of this analytical research and the Solvent Triangle, we can make a well-considered choice from a wide range of solvents. In this case we opt for thinner, which will typically reduce the discoloration caused by the cross-linked adhesive layer. The solvent is then carefully tested on the medium used, in this case, printing ink. After extensive testing we decided to immerse the object totally in a thinner bath. 38.5.6 Result The solvent actually dissolved the adhesive residue, Fig. 38.4. Laser treatment would have carried the risk of removing not only the staining on the recto side, but also text. This is why we ﬁnally opted for the conventional method, i.e. using a solvent with the inherent risks of migrating ink and solvent saturation of the ﬁbre. Laser treatment nevertheless would have been more accurate and would have avoided the latter risks. The choice for the conventional method ﬁnally rested on its lower costs compared with laser treatment.

38.6 Conclusions The practical studies have yielded amazing results which restorers can be all too pleased with. Application of the laser does away with the risk of tidemarking or solvents that may dissolve the media used and may be harmful to the restorer’s health. Yet in practice, we often prefer conventional methods, as we are more familiar with them and they are cheaper. These methods are generally accepted and are considered safe. The long-term eﬀects, however, are easily overlooked. We make a plea for continuing research, therefore, also into the conventional methods. Such research would be unimaginable without the practising restorer.

References 1. R.L. Feller and D.B. Encke, Stages of Deterioration: The Examples of Rubber Cement and Transparent Mending Tape, Science and Technology in the Service of Conservation: Preprints, IIC, Washington D.C., 1982. 2. M. Smith, N.M.M. Jones, S.L. Page, M.L. Dirda, in Journal of the AIC, Vol. 23, No.2, 1984. 3. E. O’Loughlin and L.S. Stiber, A Closer Look at Pressure-Sensitive Adhesive Tapes: Update on Conservation Strategies. Manchester Papers, 1992, 280. 4. L. Spencer, T. Fourrier, J. Anderson, A.E. Hill, Canvas glue removal using a 248 nm Excimer Laser, ICOM, Vol.1, 1999, 336.

338

P. van Dalen et al.

5. J.H. Scholten, P. v. Dalen, Laser cleaning of pressure sensitive tapes on paper. CR, Interdisciplinair tijdschrijft voor Conservering en Restauratie, 2003, 4: 41. 6. H. Abdul Aziz, et al., Environmental impact of laser induced papers compared to manually treated paper samples. Poster presentation at 6th Indoor Air Quality 2004 Meeting, Padova, Italy, 2004.

Part V

Analytical Techniques

39 Raman Spectroscopy: New Light on Ancient Artefacts P. Vandenabeele∗ and L. Moens Ghent University, Laboratory of Analytical Chemistry, Proeftuinstraat 86, 9000 Ghent, Belgium ∗ [email protected] Summary. Raman spectroscopy, being a laser spectroscopic method, is gaining increasingly more interest for applications in the ﬁeld of art and archaeology. The technique is especially appreciated for its non-destructive character, the speed of analysis and the ability to obtain molecular information on a whole range of materials, organic as well as inorganic. Although the Raman eﬀect was observed for the ﬁrst time in 1928, it was not until the end of the 1980s before instrumental improvements enabled the analysis of micro-samples, and thus allowing the application of this method in archaeometry. Next to the identiﬁcation of inorganic materials, organic matter, such as resins and binders, has often been examined by using Raman spectroscopy. Together with the comparison to reference spectra, spectral interpretation is often involved to attribute the Raman bands to speciﬁc molecular vibrations. This is mainly of importance when studying archaeological materials that have suﬀered degradation over time. Recently ﬁbre optics instrumentation became available for the direct and non-destructive analysis of artefacts. Although the approach seems simple and easy to apply, there are several drawbacks that need our attention. For instance, due to the diﬀerent nature of a mediaeval manuscript, mediaeval wall paintings on the vault of a chapel, panel paintings and polychrome sculptures, diﬀerent experimental set-ups are needed to deal with the diversity of artefacts. Moreover, these set-ups need to guarantee suﬃcient stability to allow focusing of the laser beam on the artefact.

39.1 Introduction Raman spectroscopy, being a laser spectroscopic method, enables us to record spectra which reﬂect the molecular vibrations of the material under study. Although this technique was demonstrated experimentally for the ﬁrst time by Sir C.V. Raman in 1928, it is only since the late 1980s that this technique became increasingly available for the investigation of objects of art. During the history of the technique, it is remarkable to see that many evolutions in the technique and its applications have been instrument-driven: the introduction of new light sources in the 1960s (Lasers), the invention of chargecoupled-device (CCD) detectors, while the advent of Fourier-Transform (FT-)

This page intentionally blank

342

P. Vandenabeele and L. Moens

Raman spectroscopy allowed us to record Raman spectra in the infra-red region. Especially the introduction of (confocal) micro-Raman spectroscopy, i.e. coupling the Raman spectrometer with a microscope, enabled the application of the technique in archaeometry. The growing interest in this ﬁeld of applications is nicely illustrated by the attendance of speciﬁc conferences and the rising number of research papers that are annually published on this subject [1]. Recently, the introduction of ﬁbre optics probe heads and the development of mobile Raman instrumentation open new possibilities in this ﬁeld. Indeed, Raman spectroscopy is a micro-analytical technique with certain features which are extremely advantageous for applications in the ﬁeld of art and archaeology. The good spatial resolution (down to ca. 1 µm diameter, down to ca. 5 µm depth resolution) allows us to identify materials in microsamples. By using the confocal properties of the instrumentation, it is even possible to analyse materials on a painting covered by a varnish layer. Moreover, the character of the technique enables us to analyse the samples in a non-destructive way (which keeps them for further investigation with other techniques) and allows us to perform investigations directly on the artwork. Raman spectroscopy is a molecular spectroscopic technique, allowing identifying inorganic as well as organic materials. This implies that someone may study, using this technique, antique textiles as well as twentieth century synthetic pigments and can contribute towards a broad ﬁeld of application of the technique, including pigments, corrosion products, textiles, resins and binding media, gemstones and jewellery, paper and parchment, etc. Moreover, with the advent of ﬁbre optics probe heads the broad range of application for (mobile) Raman spectroscopy becomes within reach. However, despite all these advantages, there are some drawbacks that may hamper the investigations with Raman spectroscopy. One of these drawbacks is the occurrence of ﬂuorescence, overwhelming the Raman spectrum. Indeed, ﬂuorescence radiation is much more intense than the Raman eﬀect, and thus, minute amount of molecules (e.g. from some dirt) in the sampling area may overwhelm the Raman spectrum. When the ﬂuorescence is not caused by the analyte, but it is rather originating from a neighbouring zone, the use of confocal optics may help in avoiding the interference from ﬂuorescence: it is possible to exclude the ﬂuorescent zone from the volume of analysis. Another option to avoid ﬂuorescence is using a diﬀerent excitation laser wavelength. Another drawback often encountered when starting to use Raman spectroscopy, especially when comparing it to some other techniques, is that it may be necessary to cooperate with an experienced Raman spectroscopist. Indeed, the recording of spectra of suﬃcient quality may not be so straightforward as expected for certain samples. Moreover, spectral interpretation can sometimes be complicated, especially when mixtures are involved. In general, several approaches towards the interpretation can be used. One of these approaches is comparing the recorded spectrum with the spectra in an extensive reference database. These databases should preferably be recorded on similar instrumentations

39 Raman Spectroscopy: New Light on Ancient Artefacts

343

and especially the laser wavelength used can be important. Other approaches towards Raman spectral interpretation relay on the assignment of Raman bands towards speciﬁc functional groups. Finally, in some cases, a chemometrics approach can be of interest. After discussing the theoretical background of Raman spectroscopy in this paper, we will give two examples of Raman applications in archaeometry, which will illustrate the approach towards a problem. In the ﬁrst example, it is illustrated how Raman spectroscopy can be used for the investigation of organic materials, such as textile ﬁbres. The second application discusses some possibilities for in situ analysis of mediaeval wall paintings.

39.2 Theoretical Background The Raman eﬀect can be considered as the inelastic collision between photons and molecules. An elaborate description of the Raman eﬀect can be found in several handbooks and is out of the scope of this text. When monochromatic electromagnetic radiation is directed onto a molecular material, diﬀerent eﬀects can occur. Besides absorption eﬀects, the electromagnetic radiation may be scattered. This scattering can be considered as an absorption–emission phenomenon (Fig. 39.1). Consider a molecule in its ground state, when absorbing a laser photon, the molecule is excited towards a virtual energy level. This level is called ‘virtual’ since, according to the laws of quantum mechanics, this is a forbidden level. Therefore, the new situation is an unstable state and the molecule quickly relaxes towards the ground state, emitting a photon of the same energy as the incident photon (Rayleigh radiation). Only about 1 over 106 photons will give rise to a relaxation towards the ﬁrst excited vibrational state instead of the ground state. In this case, a photon is emitted with a lower energy (i.e. a longer wavelength, Stokes Raman scattering). In a Raman spectrum, the energy diﬀerence between the exciting and the scattered photon is present

E = (1 + ½) · hvv

V=1

E = ½ · hvv

V=0 Rayleigh

Stokes

Anti-Stokes

Fig. 39.1. Energy diagram describing the Raman eﬀect

344

P. Vandenabeele and L. Moens

in the abscissa, representing the position of the diﬀerent Raman bands. The Raman band position is typical for the observed molecule. On the other hand, in some cases, before ‘excitation’ the molecule may already be in the ﬁrst vibrational excited state and, when emitting a photon, the molecule may relax towards the ground state. In this case, the scattered (anti-Stokes) radiation is blue-shifted compared to the exciting laser, by the same amount of energy, as Stokes radiation is red-shifted. In general, because of the Boltzmann distribution, anti-Stokes radiation is less intense than Stokes radiation, and therefore, for most analytical applications, only the latter is used.

39.3 Applications of Raman Spectroscopy in Art and Archaeology 39.3.1 Raman Spectroscopic Analysis of Textile Samples

Intensity (Arbitrary Units)

Raman spectroscopy, being a molecular technique, can be applied to study textile materials. From the Raman spectrum, the diﬀerent types of ﬁbres can easily be distinguished. On the one hand, the diﬀerentiation between plant materials and ﬁbres of animal sources (such as silk and wool) can be made on grounds of the presence or absence of features which can be related to proteinaceous materials (Fig. 39.2). Indeed, the presence of the amide I band (ν(CONH) stretching vibration, located at ca. 1, 660 cm−1 ) and the amide III band (δ(NH) bending vibration, located at ca. 1, 300 cm−1 ) are highly indicative for proteins [2]. Many plant ﬁbres, if related to cellulose, do not show features in the Raman spectral region between 1,800 and 1, 500 cm−1 .

a.

b.

c.

1800

1600

1400

1200

1000

800

600

400

Raman Wavenumbers (cm−1)

Fig. 39.2. Raman spectra of textile ﬁbres (300–1, 800 cm−1 ), recorded with a dispersive Raman spectrometer (excitation wavelength: 785 nm, accumulation times ranging between 20 and 120 minutes). a. Silk; b. Sheep’s wool; c. Cotton

39 Raman Spectroscopy: New Light on Ancient Artefacts

345

Indeed, cellulose, being a polysaccharide, does not contain any C=O or C=C functional groups, which usually give rise to features in this spectral range. Next to the distinction between the plant and animal ﬁbres, when carefully examining the spectra of the materials, it is possible to make a more precise assignment of the nature of the material. For instance, when considering the animal ﬁbres of wool and silk, the exact Raman band positions of the amide I and III bands are highly indicative for the tertiary structure of the proteins. Since wool is a keratotic material, consisting mainly of α-helices and random coils, its Raman band positions are shifted in comparison with the spectrum of silk, being a material containing mainly random coils and β-sheets. Besides the identiﬁcation of the materials, Raman spectroscopy is even so well able to provide information on the degradation of the textiles [3–5]. Indeed, since a Raman spectrum is a reﬂection of the molecular composition, the change of molecular composition, originating from degradation, is revealed. Usually degradation, for instance during burial, causes cleavage of the biopolymer-chains, resulting in a mixture of partially degraded chains, monomers and oligomers. As a consequence of this, the Raman spectrum of degraded materials contains several broad-band features and thus Raman bands are less well resolved in comparison with the undegraded textile ﬁbres. Thus, Raman spectroscopy could be considered as a technique to study and monitor the degradation of textile ﬁbres. 39.3.2 In Situ Raman Analysis of Mediaeval Wall Paintings When studying objects of art, the aim always has to be to extract as much as possible information out of the artwork, while minimising the damage (and risk of damage). There are diﬀerent ways to try to achieve this, such as performing several complementary non-destructive investigations on the same micro-sample (maximising the information) [6] or performing direct, non-destructive analysis (minimising the damage) [7]. In this context, using mobile equipment for non-destructive and in situ investigations is of great use [8]. When considering the purchase of mobile Raman equipment, several aspects have to be considered. Next to the spectral response, it is important to select the laser wavelength according to someone’s needs. The use of suitable positioning equipment, which is adapted to the diﬀerent types of artefacts that have to be examined, has to be considered. Besides the macropositioning of the probe head, it is important to consider diﬀerent options for micropositioning and focussing tools. Visualisation (and the use of a proper white light source, avoiding heating of the artefact) of the sample under study is indispensable. It is favourable when the set-up can be adapted to the speciﬁc needs encountered on the location, while on the other hand, since time is often a limiting factor, the speed of setting-up the instrumentation and performing the calibration has to be optimised. Medieaval wall paintings on the vault of the chapel in Ponthoz (Belgium) have been examined by means of mobile Raman spectroscopy during the

346

P. Vandenabeele and L. Moens

winter of 2003-04 [9]. Working conditions strongly interfered with the Raman spectroscopic study, such as the need of scaﬀolding to reach the artefacts, hampering ﬁne positioning and focussing the probehead. Other interferences occurred from cold (freezing of a reference standard), dust and sunlight. Although unfavourable conditions hampered the investigations, several pigments could be identiﬁed with in situ Raman spectroscopy. Calcite was omnipresent. Pigments such as haematite, malachite, vermillion and azurite were detected. Moreover, it was observed that in some areas black smudges occurred from a discoloration of vermillion (transformation of cinnabar towards metacinnabar). Another degradation eﬀect observed was the transformation of azurite towards green substances. These materials could be identiﬁed as green copper hydroxychlorides, such as atacamite and paratacamite. The results from this survey illustrate that, despite practical problems, mobile Raman spectrometry may provide us with interesting information on the composition and degradation of objects of art.

39.4 Conclusions Concluding, it can be stated that Raman spectroscopy is a powerful analytical technique, which due to instrumental improvements came available for research in archaeometry. This molecular spectroscopic technique has several advantageous features such as its speed of analysis, its non-destructive character and its ability to study inorganic as well as organic materials. The technique can be used for the study of modern as well as antique artefacts. Raman spectra can be identiﬁed through comparison with previously recorded spectra from a reference database, but Raman band assignments can even so well be of help, especially when the materials have suﬀered severe degradation. One of the examples in this text illustrates the need of mobile Raman instrumentation adapted for archaeometrical research. Using a ﬁbre optics probe head which, depending on the needs, can be mounted on diﬀerent macro-and micropositioning equipment seems straightforward, but several hostile experimental conditions may interfere. Mobile Raman equipment comes nowadays increasingly available for users in archaeometry. Acknowledgements The authors wish to thank the Research Foundation – Flanders (F.W.O.Vlaanderen), the Research Council of Ghent University (B.O.F. UGent) and the Flemish Government (Department of Culture – Museumdecreet) for their ﬁnancial support of this research. P.V. is especially grateful to the Research Foundation – Flanders (F.W.O.-Vlaanderen) for his postdoctoral grant.

39 Raman Spectroscopy: New Light on Ancient Artefacts

347

References 1. P. Vandenabeele, in J. Raman Spectrosc., Vol. 35, 607, 2004. 2. P. Vandenabeele, B. Wehling, L. Moens, H. Edwards, M. De Reu, and G. Van Hooydonk, in Anal. Chim. Acta, Vol. 407, 261, 2000. 3. H. G. M. Edwards and T. Munshi, in Anal. Bioanal. Chem., Vol. 382 (6), 1398, 2005. 4. H. G. M. Edwards, in Analyst, Vol. 129 (10), 870, 2004. 5. H. G. M. Edwards and P. Wyeth, Case Study: Ancient textile ﬁbres, in: Raman spectroscopy in archaeology and art history (H.G.M. Edwards, J.M. Chalmers, Eds.), The Royal Society of Chemistry, Cambridge, 2005. 6. P. Vandenabeele, A. von Bohlen, L. Moens, R. Klockenkämper, F. Joukes, and G. Dewispelaere, in Analytical Letters, Vol. 33/15, 3315, 2000. 7. P. Vandenabeele, F. Verpoort, and L. Moens, in J. Raman Spectrosc. Vol. 32/4, 263, 2001. 8. P. Vandenabeele, T. L. Weis, E. R. Grant, and L. J. Moens, in Analytical and Bioanalytical Chemistry, Vol. 379/1, 137, 2004. 9. P. Vandenabeele, K. Lambert, S. Matthys, W. Schudel, A. Bergmans, and L. Moens, in Analytical and Bioanalytical Chemistry, Vol. 383/4, 707, 2005.

40 Pigment Identiﬁcation on “The Ecstasy of St. Theresa” Painting by Raman Microscopy ∗

D. Marano1,4,5 , M. Marmontelli2 , G. E. De Benedetto3 , I.M. Catalano1,4,5 , L. Sabbatini2-4 , and F. Vona6 1

2

3

4

5 ∗ 6

Dipartimento di Fisica “M. Merlin”, University of Bari, Via Amendola 173, 70126 Bari, Italy Dipartimento di Chimica, University of Bari, Via Amendola 173, 70126 Bari, Italy Dipartimento di Beni delle Arti e della Storia, University of Lecce, Ex Monastero degli Olivetani, Viale San Nicola, 73100 Lecce, Italy Laboratorio di Ricerca per la Diagnostica dei Beni Culturali (LAUB), University of Bari, www.laub.it CNR – INFM [email protected] Soprintendenza per il Patrimonio Storico, Artistico ed Etnoantropologico dele Province di Bari e Foggia, Via Pier L’eremita 25/A, Bari, Italy

Summary. A study of the pigments of “The Ecstasy of St. Theresa,” a seventeenth century oil painting on canvas, was performed by Raman microscopy. Lazurite was identiﬁed in both Jesus Christ’s and St. Theresa’s mantles as the pigment responsible for the blue coloration. Litharge was identiﬁed inside the black bitumen layer. Usually the bitumen needed a lot of time to dry in the air when mixed with drying oil. Litharge was used by the artist to decrease the oil drying time. A complementary study, using micro-Raman and SEM, allowed us to identify red ochre as the pigment responsible for the red coloration in the altar on the left side of the painting.

40.1 Introduction Raman microscopy is one the more eﬃcient diagnostic technique in the ﬁeld of Cultural Heritage to characterize artistic manufactured articles. This spectroscopic technique is particularly suitable for the identiﬁcation of pigments in complex matrices. Raman microscopy has been used to identify pigments on Egyptian papyri [1], manuscripts [2], polychrome sculptures [3], pottery [4,5], icons [6], mosaics [7], and stamps [8]. As established elsewhere [9], the technique combines high sensitivity and spatial resolution with nondestructiveness, and it is not inﬂuenced by interference from surrounding materials. Moreover, samples do not require any kind of preparation for their Raman analysis.

350

D. Marano et al.

Fig. 40.1. “The Ecstasy of St. Theresa,” oil painting on canvas

In this work we have analyzed some samples from “The Ecstasy of St. Theresa” (shown in Fig. 40.1), an oil painting on canvas, 210 × 250 cm2 , kept in the Santa Teresa dei Maschi Church in Bari (Puglie, South of Italy), starting from the need to characterize the inorganic pigments used in this painting. No previous scientiﬁc work was carried out to study the chemical composition of the layers forming this painting. The author is unknown but, on the basis of a stylistic analysis, it may be thought that this oil painting on canvas was made between the end of seventeenth and the beginning of eighteenth century by a local artist along the lines of a print of Karel de Mallery now at the Bibliothèque Royale of Bruxelles [10]. Therefore this painting is also an evidence of print circulation and cultural exchanges among the artists of northern and southern Europe in that period. The most interesting aspect of this painting is that it never underwent a restoration work in the past because the conservation environment, the church wall on which the painting is at present hung, is good and seems to have been the same for a long time. It may be thought that all the materials found are original except for visible repainting that cover the most damaged parts (probably due to an ancient ﬁre or a traumatic event not yet supplied with documental evidence). The readability of the artwork is quite damaged by

This page intentionally blank

40 Pigment Identiﬁcation on “The Ecstasy of St. Theresa”

351

the deep yellowing of the varnish layer that could be removed in a future conservation work. The aim of this scientiﬁc work was therefore to determine the chemical composition of inorganic pigments by micro-Raman analysis in order to prevent any damage due to an uncritical employment of solvents and chemical products and to enlarge our knowledge about the materials used by the unknown artist. The research conﬁned itself to giving deﬁnition of the chemical composition of inorganic compounds, while varnishes, binding medium, and organic pigments will be analyzed in further studies.

40.2 Experimental Methods We collected some micro-samples of painted materials (each the size of a few mm2 ) in ﬁve zones of the painting, without any visible injury to the artwork. The list of samples is shown in Table 40.1. The measurements were made using a Jobin Yvon portable Raman spectrometer and a Renishaw inVia Raman microscope. First Raman investigation was obtained by focusing the 632.8 nm line of an He–Ne laser onto the samples to a spot of ∼1 µm diameter with a ×50 ultra long working distance microscope objective, 1 mW on the sample. The second one was obtained by using a 1 mW laser diode line at 785 nm, via ×50 objective on the sample. The position of the laser spot on the sample was varied through a mechanical movement system, with micrometric sensibility. A point by point analysis with a micrometric resolution was required so as to study the nature of the pigments which appear as micrometric grains in a complex matrix. The micrometric grains (Fig. 40.2 shows two examples on the blue pigment) of the blue and red pigment on which the analysis concentrated have been individualized through video cameras. Table 40.1. Samples of painted material Sample Description 1 2 3 4 5

Black fragment from a dark tile in the chessboard-like ﬂoor. One black layer superimposed on a priming layer. Black fragment from the borderline between a red ﬂoor tile and a black one. One black layer superimposed on a priming layer. Red fragment from the altar on the left side. One red layer superimposed on a priming layer. Blue fragment from Saint Theresa’s mantle. One blue layer superimposed on a white layer and a priming layer. Blue fragment from Jesus Christ’s mantle. One blue layer superimposed on a white layer.

352

D. Marano et al.

Fig. 40.2. Microphotographs of the blue pigment layer

40.3 Results and Discussion We analyzed, by micro-Raman spectroscopy, the blue pigment in Samples 4 and 5 (see Table 40.1), the red pigment found inside the black bitumen layer of Samples 1 and 2, and the red pigment in Sample 3. Figure 40.3 shows Raman spectra on the blue pigment (A) and of the red pigments (B) in Sample 3. All the pigments were identiﬁed by comparison of their Raman spectra with those in standard pigment database [11]. Blue Pigment The ﬁrst Raman spectrum shows the peak at 548 cm−1 , attributable to −1 , attributthe symmetric stretch of the S− 3 ion, and the shoulder at 583 cm − able to an asymmetric mode of the S2 ion. The Raman spectrum also shows the multiple band at 1, 096 cm−1 (548 × 2) [12]. This Raman ﬁngerprint is typical of lazurite, the blue mineral Na8 [Al6 Si6 O24 ]Sn , a sodium aluminosilicate extracted from lapis lazuli, that was identiﬁed as responsible for the blue coloration. Maybe from practice and his master’s advice, the artist learned that the lack of hiding power from lapis lazuli used in oil could be replaced by spreading out a layer of white lead before laying it on the painting surface. This is the most suitable interpretation to justify the superimposing of lapis lazuli on white lead emerging from the blue fragments cross-sections. The presence of lazurite indicates the importance of the artwork. In fact, the use of lapis lazuli as pigment is old, but quite rare, since precious rock was diﬃcult to ﬁnd, coming from far-oﬀ mines in the mountains of Iran and Afghanistan. Red Pigment Comparison of the Raman spectrum of the red pigment found in the black bitumen layer with the standard pigment database allowed us to identify litharge. Interesting information emerged from this large orange crystal of

40 Pigment Identiﬁcation on “The Ecstasy of St. Theresa”

353

2500

A

Intensity (counts/s)

2000 1500

450

1000

400

500

B

500

350

0

300

−500 0

500

1000

1500

2000

2500

Raman shift(cm−1)

3000

3500

200

300

400

500

600

700

Raman shift

Fig. 40.3. Raman spectra on blue (A) and red (B) pigments

litharge found inside the black bitumen layer. Even though no analysis was carried out on the binding medium, the presence of litharge is meaningful because drying oil such as linseed oil needs the addition of some catalysts to speed up the drying process. This goal was usually reached in the past by adding metallic driers to the oil such as salts of lead, manganese, and cobalt or litharge (lead oxide). This addition was absolutely necessary when the drying oil was employed to dilute a pigment lacking in drying power such as bitumen. It was a black pigment that needed a lot of time to dry in the air when mixed with drying oil. Litharge successfully was able to decrease the oil drying time. In conclusion, the ﬁnding of litharge in a painting layer is an indirect evidence of the employment of drying oil (perhaps linseed oil) as binding medium. To support this hypothesis and conﬁrm the diﬀusion of litharge throughout the bitumen layer, further analyses by means of other diagnostic techniques (SEM/EDS) will be carried out as soon as possible. The second spectrum of Fig. 40.3 led to doubt between Mars red and red ochre, having two similar Raman ﬁngerprints. Further observation carried out by means of SEM revealed the sediment feature of that pigment, including calcite crystals, feldspars, but above all, Fe2 O3 crystals (perhaps hematite), which gave the red colour to the sediment. The surface look combined with SEM observation suggested that red ochre was the original material used by the artist.

40.4 Conclusions We have analyzed the painting materials of “The Ecstasy of St. Theresa” painting in order to identify its inorganic pigments. We have identiﬁed lazurite as responsible for the blue coloration in Christ’s and St. Theresa’s mantles. The presence of lazurite indicates the importance of the artwork as an example of print circulation and cultural exchanges among the artists of northern and southern Europe between the end of seventeenth and the beginning of

354

D. Marano et al.

eighteenth century. Litharge was identiﬁed inside the black bitumen layer: the artist used this pigment to decrease the oil drying time, usually very long for bitumen mixed with drying oil. An SEM analysis, associated with an optical observation, led us to identify red ochre as the pigment responsible for the red coloration. This study shows micro-Raman and SEM as complementary techniques in the ﬁeld of diagnostics of Cultural Heritage.

References 1. L. Burgio and R. J. H. Clark, in J. Raman Spectrosc., Vol. 31, 395, 2000. 2. L. Burgio, D. A. Ciomartan, and R. J. H. Clark, in J. Raman Spectrosc., Vol. 28, 79, 1997. 3. H. G. M. Edwards, D. W. Farwell, E. M. Newton, F. Rull Perez, and S. Jorge Villar, in J. Raman Spectrosc. Vol. 31, 407, 2000. 4. R. J. H. Clark, L. Curri, G. S. Henshaw, and C. Laganara, in J. Raman Spectrosc., Vol. 28, 105, 1997. 5. P. Colomban, in J. Raman Spectrosc., Vol. 34, 420, 2003. 6. L. Burgio, R. J. H. Clark, and K. Theodoraki, in Spectrochimica Acta Part A, Vol. 59, 2371, 2003. 7. P. Colomban, G. March, L. Mazerolles, T. Karmous, N. Ayed, A. Ennabli, and H. Slim, in J. Raman Spectrosc., Vol. 34, 205, 2003. 8. T. D. Chaplin, R. J. H. Clark, and D. R. Beech, in J. Raman Spectrosc., Vol. 33, 424, 2002. 9. R. J. H. Clark, in Chem. Soc. Rev., Vol. 24, 187, 1995. 10. M. Marmontelli, Un dipinto inedito nella chiesa di Santa Teresa dei Maschi(Bari): analisi micro-Raman e S.E.M.-E.D.S. applicate allo studio dei materiali pittorici de “L’Estasi di Santa Teresa”, degree thesis, 2005. 11. L. Burgio and R. J. H. Clark, in Spectrochim. Acta, Part A, 57, 1491, 2001. 12. P. Colomban, in J. Raman Spectrosc., Vol. 34, 420, 2003.

41 Colorimetry, LIBS and Raman Experiments on Renaissance Green Sandstone Decoration During Laser Cleaning of King Sigismund’s Chapel in Wawel Castle, Cracow, Poland ∗

A. Sarzynski , W. Skrzeczanowski, and J. Marczak Military University of Technology, Institute of Optoelectronics, 2 Kaliskiego str., 00-908 Warsaw, Poland ∗ [email protected] Summary. Measurements aimed at optical characteristics of the basic building material of the Renaissance dome of King Sigismund’s Chapel in Wawel Castle are described in the paper. Colorimetric measurements, LIBS investigations and Raman eﬀect measurements were carried out. LIBS spectra of encrusted and clean (fresh fractured) grey-greenish Myslenicki sandstone are presented. Raman spectra and reﬂection coeﬃcients as well as L, a, b coordinates for several samples are also shown.

41.1 Introduction Laser cleaning technology, developed in 2003 in the framework of EUREKA E!2542 RENOVA LASER project, was included into the full program of conservation and restoration of Sigismund’s Chapel at Wawel Castle in Cracow, Poland, covering more than 800 m2 of decorative, sixteenth century sculptured surfaces. The conserved object – the Renaissance chapel – was made of a greygreenish sandstone excavated in Myslenice near Cracow. Damage and cavities in the original sandstone were repaired with green painted gypsum or a grey concrete during former restoration works in eighteenth and nineteenth centuries. During current renovation, several samples of the grey-greenish sandstone, green painted gypsum and concrete were investigated. Measurement results aimed at optical characterization of these samples are shown below. For comparison, the measurement results of a ﬂuorite (or amethyst) sample are also included.

41.2 Experiment LIBS measurements were carried out in the experimental setup shown in Fig. 41.1. Similar arrangements were used in Raman and colorimetric

356

A. Sarzynski et al.

Fig. 41.1. Experimental setup for LIBS measurements

measurements. Colorimetric measurements were carried out using a Minolta CM-2600d spectrophotometer, LIBS investigations using a Mechelle 900 – SensiCamFS arrangement (resolution λ/∆λ = 900; spectral range, 300–1,000 nm; Nd:YAG laser, 1,064 nm, 30 mJ, 10 ns), while Raman eﬀect measurements used the Ocean Optics R-3000 system (diode laser, 250 mW; 785 nm; resolution, 8 cm−1 ). Experimental details can be found elsewhere [1, 2].

41.3 Results and Discussion A picture of an original encrusted sandstone is shown in Fig. 41.2. Brighter areas have been cleaned using RENOVA laser system. LIBS spectra of encrusted and clean (fresh fractured) sandstone are shown in Figs. 41.3 and 41.4. It was very easy to collect LIBS spectra with the use of the Mechelle 900 spectrometer, but it was very diﬃcult to interpret them unambiguously. Because of a low spectral resolution of the instrument used, at least 10 (but sometimes even more than 100) transitions seemed to be related to the one spectral line. We were not able to decide whether investigated crystal was ﬂuorite or rather amethyst. In spite of this, presence of lead on the surface of a dirty sandstone (Fig. 41.3) and green-painted gypsum was discovered. The lead spectral lines disappeared for clean sandstone (Fig. 41.4) as well as white gypsum spectrum. The lead could appear in these samples as an impurity. It could also be added in a form of lead white into the green paint (painted gypsum). The green color of sandstone is caused by the presence of copper compounds (Figs. 41.3 and 41.4). Our LIBS experiments showed complex spectra of sandstone sample. They showed a few tens of spectral lines belonging to over 20 elements among which

41 Colorimetry, LIBS and Raman Experiments

357

Cu ll 766.5, K l 766.5, 769.9 O l 777.2-777.5, Cu ll 779.0

N l 742.3, 744.1, 746.9

Ca l 714.8

Ca l 643.9, 646.3, 649.4, Cu ll 644.3, 644.9 H l 656.3

Ca l 610.3, 612.2, 616.1-617.0, Cu ll 610.0, 617.2

Ca l 559-560

2000

Ca l 487.8 Ca ll 500.1, Pb l 500.5, Ca l 504.2 Cu ll 518.3 Ca l 518.9 Ca l 526-527, Cu ll 527.0 Ca l 534.9

4000 Pb l 357.3, 364.0, 369.3

Intensity (counts)

6000

Ca ll 369.4, 370.6, 373.7, Pb l 374.0 Mg l 382.9, 383.2, 383.8 Ca ll 396. 8 Pb l 405.8, 406.2, Cu l 406.3 Ca l 422.7, Cu ll 423.0 Ca l 429.9, 430.3 Ca l 442.5, 443.5, 445.5

Cu ll 393.3, Ca ll 393.4

Ca l 585.7 Na l 589.0, 589.6, Pb l 589.6, Cu ll 589.8, 600.0

Fig. 41.2. Encrusted Myslenicki sandstone

0 300

350

400

450

500

550

600

650

700

750

Wavelength (nm)

Fig. 41.3. LIBS spectrum of encrusted Myslenicki sandstone. Apart from Ca, Mg, O, N, H, K, Na and Li, the spectral lines of Pb and Cu can be seen

Al, Ba, Br, Ca, Cd, Cl, Cu, Fe, Gd, H, I, K, Mg, N, Na, O, Pb, S, Si, Ti, Y, Zr were identiﬁed (spectra not shown in this paper). As the typical chemical composition of this greenish Myslenicki sandstone includes elements such as Al, C, Cu, Na, K, Si, Mg, Ca, O, H, Fe, rest of the identiﬁed constituents are probably impurities originating from the two large industrial facilities: steel and aluminum plants, situated quite close to Wawel Castle.

Cu ll 766.5, K l 766.5, 769.9 O I 777.2-777.5, Cu ll 779.0

N l 742.3, 744.1, 746.9

Ca l 559-560

1000

Ca l 487.8 Ca ll 500.1, Ca l 504.2 Cu ll 518.3, Ca l 518.9 Ca l 526-527, Cu ll 527.0 Ca l 534.9

2000

Ca l 714.8

Ca ll 396.8

3000

O l 407.B Cu l 406.3 Ca l 422.7, Cu ll 423.0 Ca l 429.9, 430.3 Ca l 442.5, 443.5, 445.5

4000

Ca ll 369.4, 370.6, 373.7

Intensity (counts)

5000

Ca l 643.9, 646.3, 649.4, Cu ll 644.3, 644.9 H l 656.3

Cu ll 393.3, Ca ll 393.4

6000

Ca l 610.3, 612.2, 616.1-617.0, Cu ll 610.0 617.2

A. Sarzynski et al.

Ca l 585.7, Na l 589.0, 589.6, Cu ll 589.8, 600.0

358

0

300

350

400

450

500

550

600

650

700

750

Wavelength (nm)

411.98

Fig. 41.4. LIBS spectrum of a clean (fresh fractured) Myslenicki sandstone. Lead spectral lines disappeared (for example, at 357.3, 364.0 and 369.3 nm)

463.00

4000

3000

1109.76

754.11

356.32

1000

217.44

99.85

2000

0 0

200

400

600

800 1000 1200 Counts/Raman Shift (cm−1)

1400

1600

1800

Fig. 41.5. Raman spectrum of a ﬂuorite sample. The only spectrum containing peaks typical for molecular species

The typical Raman spectrum was obtained only for a ﬂuorite sample (Fig. 41.5). In all other cases, strong ﬂuorescence dominated over a weak Raman signal (Fig. 41.6). Figure 41.7 shows reﬂection coeﬃcients and L-, aand b- values for several samples.

41 Colorimetry, LIBS and Raman Experiments

359

Fig. 41.6. Raman spectrum of encrusted sandstone. Because of strong ﬂuorescence, the spectrum does not contain peaks typical for molecular species

Fig. 41.7. Reﬂection coeﬃcients of several samples. (1) ﬂuorite (parameters L, a and b were 14.31, 14.06 and −5.0, respectively), (2) dirty sandstone (L, a, b – 49.04, 0.22, 10.66), (3) concrete (L, a, b – 55.65, 0.14, 7.83), (4) clean sandstone (L, a, b – 60.24, −0.98, 6.95) and (5) green painted gypsum (L, a, b – 58.16, 0.55, 12.09)

41.4 Conclusions Optical characteristics for decoration elements of the dome of King Sigismund’s Chapel were found. However, spectral resolution of the Mechelle 900 system is a bit too low for unambiguous interpretation of LIBS spectra. Instrument with one order higher spectral resolution is needed. In the investigated range (300–1,000 nm), acquired spectra also contained many spectral

360

A. Sarzynski et al.

lines belonging to atmospheric gases (oxygen and nitrogen). The future measurements should be moved to the UV range (190–400 nm) to avoid presence of these lines. In the future experiments, we intend to improve laser stability and homogeneity of energy density distribution. Acknowledgements This work was supported by the Ministry of Science and Information Society Technologies – project 120/E-410/SPB/COST/T-11/ DWM726/2003–2005.

References 1. A. Sarzyński and W. Skrzeczanowski, Mechelle 900 Spectrometer with Fast SensiCamFS Camera for Pulsed Radiation Sources Studies, 5th Int. Conference MECHATRONICS 2004, September, 23–25, 2004, 294, Warsaw, Poland 2. A. Sarzynski, W. Skrzeczanowski and J. Marczak, LIBS experiments on artworks at the Institute of Optoelectronics, The 7th Symposium of Optoelectronics, SIOEL 2004, October 28–29, 2004, Bucharest, Romania, 12 (2004), in print.

42 Non-Destructive Observation of the Laser Treatment Eﬀect on Historical Paper via the Laser-Induced Fluorescence Spectra K. Komar and G. Śliwiński Polish Academy of Sciences, Photophysics and Laser Lab. Institute of Fluid-Flow Machinery, Fiszera 14, 80–952 Gdansk, Poland [email protected] Summary. The ﬂuorescence spectra of historical and model paper samples, previously irradiated with the laser beam at wavelengths of 1,064, 532, 355, and 266 nm, are recorded under excitation at 266 nm, and the nonirradiated samples are used for reference. The spectral proﬁles obtained for the laser-treated model papers made of cotton and/or linen only reveal diﬀerences compared to the reference ones. After irradiation at 532 and 1,064 nm, a decrease of the band intensities of the entire spectral proﬁle is observed. In contrary, the UV irradiation at 355 nm of the same samples results in the increase of bands centered at 341 and 370 nm compared to the visible region only. Prolonged treatment at 266 nm results in the marked increase of band intensities in the visible region and corresponds to the independently observed yellowing.

42.1 Introduction In conservation of the historical and delicate paper objects, the sensitive and reliable techniques allowing for non-destructive analysis and diagnostics are wishful and a search on this ﬁeld is reﬂected in numerous works performed recently. Among various optical techniques such as the Raman, SERS, and DRIFT spectroscopy, which are recognized tools in the analysis of organic objects, the laser induced ﬂuorescence (LIF) can deliver information derived from spectral proﬁles in the UV–vis region by means of relatively simple instrumentation. For excitation of the LIF spectra, a relatively low laser ﬂuence of the order of 0.1 J cm−2 is applied. Examples of application of the LIF technique for pigment analysis [1] as well as observation of the laser cleaning eﬀect of historical paper can be found in the literature [2, 3]. For paper excited by the UV laser at 266 nm, typically a structure of a few broad spectral bands centered at 410, 440, 480, 530, and 612 nm is observed [3].

362

K. Komar and G. Śliwiński

In this work the application of the LIF technique is considered for detection of possible damage and changes in the surface structure caused on historical paper documents by laser irradiation.

42.2 Preparation of Paper Samples For the sample extraction, ﬁve types of historical and model paper substrates were selected. The microscopic inspection of substrates performed prior to laser irradiation conﬁrmed the rough appearance of the surface, the locally non-homogeneous structure and, in the case of historical paper, also the presence of surface contamination. Therefore a number of samples suﬃcient to provide averaging and reliable data of the experiment were extracted. The samples of historical papers made of wood pulp were taken from the archive document (end of nineteenth century), and magazine printed paper (early twentieth century). The model samples were extracted from high purity, hand-made, non-bleached paper substrates of the following: – Cotton – Cotton, gelatine-sized – 90% cotton + 10% linen The sample collection was divided into two parts composed of the same number of representatives of each substrate type. Some of the samples were subjected to laser irradiation and others were used as reference material. For irradiation the pulsed Nd:YAG laser (Quantel) of 6 ns pulsewidth operated at frequency of 20 Hz was applied. The dose applied to a given sample area was dependent on the sample and the laser wavelength of 1,064, 32, 355, or 266 nm. The irradiation parameters are listed in Table 42.1.

Table 42.1. The laser irradiation parameters of the historical and model paper samples Wavelength [nm]

Fluence (J cm−2 )

Cotton/ linen

Cotton with/ without gelatine

Archive document

Magazine printed paper

Total irradiation dose/area (J cm−2 ) 1,064 532 355 266

2 1.6 0.2/1.18 0.16

810 20 25 15

1,215 320 120 100

100 25 10 10

360 55 10 10

42 Non-Destructive Observation of the Laser Treatment Eﬀect

363

42.3 Experiment To observe the laser induced ﬂuorescence, the single pulse excitation of samples at 266 nm provided by the above mentioned laser was used. The required, mild excitation conditions were assured at ﬂuence around 50 mJ cm−2 selected carefully in regard to the damage threshold of the substrate. The detection path consisted of a band pass ﬁlter to suppress the laser excitation wavelength, the focusing lens (f = 50 mm), and the optical ﬁber mounted at the entrance slit of the 0.5 m spectrograph (Acton). The spectrograph was operated with a grating of 150 grooves mm−1 (blaze at 500 nm). For recording the spectra and data processing, a Peltier-cooled CCD camera (CVI) synchronized with the excitation by means of a pulse generator (SRS), and a PC-based data acquisition unit were used. The recording time frame of 100 ms was selected. The LIF spectra were accumulated over three to ﬁve excitation pulses and the spectra of each sample were measured on three diﬀerent sample areas and averaged.

42.4 Results The LIF spectra measured for all investigated paper types prior to laser treatment are shown in Fig. 42.1. Under excitation at 266 nm the ﬂuorescence reveals a characteristic structure composed of a few broad bands located in the ultraviolet and visible spectral region [3]. This structure is similar for all samples. Considerably higher ﬂuorescence intensity in the entire spectral region is observed for the gelatine-sized cotton paper as well as for samples of

442

1,0 403

intensity [a.u.]

0,8

479

cotton paper cotton gelatine sized cotton/linen archive document (end of XIX c.) magazine printed paper (early XX c.)

laser 532nm 609

0,6 370 0,4 341 0,2

300

400

500 600 wavelength [nm]

700

Fig. 42.1. LIF spectra of various paper samples before treatment; excitation at 266 nm

364

K. Komar and G. Śliwiński

the historical paper. The cotton/linen paper reveals ﬂuorescence higher in UV (bands centered at 341 and 370 nm) and lower in the visible region compared to historical paper made of the wood pulp. The laser irradiation of historical samples at 266, 355, 532, and 1,064 nm does not result in observable changes of the spectra in comparison to reference samples. This is probably due to surface contamination decreasing the signalto-noise ratio and also low irradiation intensities applied well below the damage threshold of these papers. Only in the case of model substrates made of cotton and of cotton/linen were diﬀerences in the UV region of the LIF spectra, measured before and after irradiation, observed. A signiﬁcant increase of the band intensities at 341 and 370 nm is observed accompanied by an intensity decrease of bands located in the visible and near-infrared part of the spectrum. The increase of the UV ﬂuorescence bands is most pronounced in the case of nonsized cotton paper, see Figs. 42.2 a–c. The dependence of this eﬀect on the irradiation dose leads to the conclusion that the spectral diﬀerences are closely related to chemical changes in the paper structure caused by the irradiation. A decrease of bands centered at 403, 442, 479, and 526 nm due to laser irradiation at 355 nm is observed particularly for the nonsized cotton paper (Figs. 42.2a and b) and does not depend on irradiation dose. After prolonged irradiation at 266 nm, yellowing is observed by naked eye and is accompanied by a marked increase in intensity of the bands located in

intensity [a.u.]

1,0

442 479

0,8

laser 532 nm 403

0,6

341 370

b)

609

0,4

442 479

1,0 intensity [a.u.]

untreated 355 nm treated (10 s) 355 nm treated (30 s)

a)

403

0,8

laser 532 nm

0,6 0,4

untreated 355 nm treated (30 s)

609

370 341

0,2

0,2 300

400 500 600 wavelength [nm]

700

300

700

untreated 355 nm treated (15 s)

c) 1,0 intensity [a.u.]

400 500 600 wavelength [nm]

laser 532 nm 442 479

0,8 0,6 0,4

609 370 341

0,2 300

400 500 600 wavelength [nm]

700

Fig. 42.2. Changes in the LIF spectra of model papers due to laser irradiation at 355 nm: (a) cotton paper, (b) cotton, gelatine-sized, (c) cotton/linen; excitation at 266 nm

42 Non-Destructive Observation of the Laser Treatment Eﬀect

1,0 intensity [a.u.]

b)

untreated 266 nm treated (30 s)

479

0,8

442

0,6

609

370

0,4

untreated 266 nm treated (30 s)

1,0 laser 532 nm

341

intensity [a.u.]

a)

479 0,8

400 500 600 wavelength [nm]

laser 532 nm

442 0,6

609

403 0,4

370 341

0,2

0,2 300

365

700

300

400 500 600 wavelength [nm]

700

Fig. 42.3. Changes in the LIF spectra of model paper substrates due to laser irradiation at 266 nm: (a) cotton paper, (b) cotton gelatine-sized; excitation at 266 nm

a)

b)

band intensity @ 479 nm[ a.u.]

40

36

non-treated sample treated @ 266 nm 355 nm 532 nm 1064 nm

65

266 nm

reference

60 532 nm 55

32 50 45

28

355 nm 1064 nm

40 10

12

14

16

18

20

22

14

16

18

20

22

24

band intensity @ 341nm [a.u.]

Fig. 42.4. Band intensity ratio (visible/UV) of the LIF spectra recorded for (a) cotton, and (b) cotton gelatine-sized paper substrates, after laser irradiation at 266, 355, 532, and 1,064 nm

the visible spectral region (compared to the spectra of nonirradiated samples), see Fig. 42.3. The intensity in the UV part of the spectra is lower compared with the untreated samples. In general, spectral proﬁles are similar to those of historical samples which are originally yellowish due to their age, compare Fig. 42.1. This diﬀerent response to laser irradiation observed through the intensity changes of spectral bands allows to resolve two band sets: the UV bands located at 341 and 370 nm and bands located in the visible region at 403, 442, 479, and 526 nm. The intensity ratios of the most intense bands centered at 341 and 479 nm of both groups are similar for the model cotton papers and are shown in Fig. 42.4. Ratios calculated for the cotton samples

366

K. Komar and G. Śliwiński

irradiated at 355 nm (Fig. 42.4a) and those of cotton gelatine-sized and irradiated at 266 and 355 nm (Fig. 42.4b) conﬁrm the observed diﬀerences. This is in agreement with the conclusion that changes in shape of the LIF spectra correspond to changes in the chemical structure of the paper substrate due to laser irradiation.

42.5 Conclusions The LIF spectra of historical and model paper samples were recorded and analyzed in order to observe changes due to laser treatment. For samples untreated by laser the characteristic bands centered at 341, 370 nm and 403, 442, 479, 526, and 609 nm were observed. However, relative intensities between these bands revealed dependence on the type and age of paper. For the gelatine-sized, historical and model paper samples a considerably higher ﬂuorescence intensity in the entire spectral region was observed. Laser irradiation of historical samples at 266, 355, 532, and 1,064 nm at ﬂuences below damage threshold did not change their LIF spectra. Only for model papers made of cotton and of cotton/linen were diﬀerences in LIF spectra measured before and after UV irradiation observed . The treatment at 355 nm led to a signiﬁcant increase in the ﬂuorescence band intensities at 341 and 370 nm. The intensity ratios of the most intense visible and UV bands revealed similar decrease after irradiation at 532 and 1,064 nm. In contrary, the laser treatment at 355 and 266 nm resulted in generally higher band intensities in the visible spectral range. It indicated the chemical changes due to laser interaction, which are dependent on the irradiation wavelengths. It was shown that the LIF spectra can deliver reliable data for analysis of the paper surface. In case of historical documents, however, a more sensitive detection could be advisable. This is due to surface contamination, stain, and ageing diminishing the signal-to-noise ratio required for non-destructive analysis. Acknowledgements This work is supported by the Ministry of Science and Information Society Technologies (MNiI) via project 0075/H01/03/25.

References 1. D. Anglos, M. Solomidou, I. Zergioti, V. Zaﬁropulos, T. G. Papazouglou, and C. Fotakis, in Applied Spectroscopy, Vol. 50, 1331, 1996. 2. K. Ochocińska, A. Kaminska, and G. Śliwiński, in Journal of Cultural Heritage, Vol. 4, 188, 2003. 3. K. Ochocińska, M. Martin, J. Bredal-Jørgensen, A. Kamińska, and G. Śliwiński, SPIE Proceedings 5229, 296, 2003.

43 Eﬀects of LIBS Measurement Parameters on Wall Paintings Pigments Alteration and Detection R. Bruder1 , D. Menut1 , and V. Detalle2 1 2

DPC/SCP/LRSI CEA, Saclay 91191, Gif-sur-Yvette, France LRMH, 29 rue de Paris, 77420 Champs-sur-Marne, France [email protected]

Summary. LIBS is a very eﬃcient tool for pigment analysis since it is a rapid, noncontact and nearly non-destructive technique. This work focussed on the particular context of wall paintings analysis. Six common pigments were studied: ultramarine blue, red lead, green earth, charcoal, red and yellow ochre. Two complementary approaches were tested: macro- and micro-LIBS. Micro-LIBS enabled us to verify pigment distribution on a small area, thanks to its excellent spatial resolution and analytical capabilities. For macro-LIBS, the inﬂuence of laser energy and focal length on the crater size, induced by laser-material interaction and on plasma emission signal, were studied to evaluate their importance on sample alteration and pigment detection. It appeared that varying the focal length induced modiﬁcation on the crater size without change in signal. Moreover, all pigments showed similar behaviour in terms of analytical signal. Laser energy and focal length also induced variations on crater diameters, suggesting a beam treatment to get a better control on crater dimension. Raman microscopy was used as a diagnosis tool to check the preservation of the pictorial layer after a LIBS analysis.

43.1 Introduction Mural paintings are major artworks in France, since several monuments have been ornamented with this technique, which constitutes a valuable patrimony needing constant attention to ensure its preservation. Fresco or derivative techniques based on the use of lime have been employed in several castles or churches. Pigments are diluted in water and applied on a fresh layer of lime. A carbonation mechanism is then induced by reaction between carbon dioxide and lime to produce calcite and water. Pigments are thus inserted in calcite layer. In derivative techniques, pigments could be diluted in lime, or applied on wetted calcite [1]. For conservation and restoration of these artworks, knowledge of pigments used is essential. Indeed, pigment nature inﬂuences methods of restoration and provides useful information about artwork history.

368

R. Bruder et al.

Many techniques are employed to achieve pigment identiﬁcation on various types of works of art, as XRF, XRD, FT-IR spectroscopy, Raman spectroscopy, laser induced ﬂuorescence (LIF) or laser induced breakdown spectroscopy (LIBS) [2–6]. LIBS is a powerful technique: it could be used in situ, without sample preparation, giving elemental information, thanks to a rapid non-contact process. The ﬁrst LIBS experiment was published by Brech in 1962 [7]. This analytical tool was ﬁrst used for cultural patrimony by Asmus [8] and the technique is now used for an increasing number of studies: analysis of ceramics, archaeological metals [9], pigments [2–4, 10, 11], where LIBS enables users to identify elemental composition of samples. Other works focus particularly on the quantitative potential of the technique [12,13], while another axis of development consists in making the technique portable [14]. The principle of LIBS has been well described in several publications [2, 15]. One of the major features of LIBS is its potential to obtain valuable spectra with a single laser shot. To ensure this, parameters of laser–matter interaction must be controlled and optimized. Two complementary approaches were tested in this study: micro-LIBS mapping and macro-LIBS. Micro-LIBS mapping provides a detailed view of the sample surface, with craters 3 µm in diameter, but remains a laboratory instrumentation that can be applied on painting fragments. The second approach consisted in interaction at the macroscopic level, which enables users to use it on site, with a direct analysis of wall paintings. Damage to the sample with a single laser shot would be greater (few tens of micrometers in size) but remain tolerable compared to ﬂake dimensions. Employing LIBS technique for wall paintings analysis requires speciﬁc development and optimization. The purpose of this work is to study the eﬀects induced by the variation of parameters governing laser–matter interaction on damage caused on the wall painting sample and on plasma signal for pigment detection. In the case of wall paintings analysis, a prior diﬃculty could be the degradation or removal of pigments by LIBS laser shot. To verify this possibility, Raman microscopy was used as a complementary tool, on the unaltered sample surface and at the crater bottom.

43.2 Experimental 43.2.1 Micro-LIBS Instrument The laser used is a Minilite II Q-Switched Nd:YAG laser, of 5 ns pulse duration (Continuum, USA). Fundamental radiation and its harmonic frequencies were available. The 266 nm radiation was used for this work, since it gives the best coupling between laser and matter and less absorption of incident laser beam by the plasma. This feature is essential in the case of small crater and plasma emission. The laser shots were repeated at 20 Hz and 120 µJ energy. An optical microscope focuses the laser beam onto the sample (×15), located on

43 Eﬀects of LIBS Measurement Parameters

369

micrometric stages (X–Y motorized, Z manual), to create the micro-plasma. The laser spot size is set to 10 µm for the considered measurements. Mapping consists of a square array of 170 × 170 shots (8.5 × 8.5 mm2 ). Plasma emission was carried through optical ﬁbres to two monochromator spectrometers (1 m focal length), in the range from 378 to 398 nm (resolution: 26 pm) and from 401 to 409 nm (resolution: 13 pm). The signal was recorded by ICCD cameras (delay: 400 ns, exposure time: 1,100 ns). 43.2.2 Macro-LIBS Instrumentation LIBS apparatus used for the measurements was comparable to the micro-LIBS instrumentation, without microscope. The laser employed was still a Minilite II Q-Switched Nd:YAG laser. The fundamental radiation at 1,064 nm was selected for several reasons: in one hand, the detection system is not sensitive in the infrared domain, so, is not disturbed by the incident laser wavelength. In the other hand, infrared wavelength is less absorbed by material than 266 nm, and thus has a lower sample ablation rate. Indeed, for macro-LIBS, plasma emission signal was strong enough to ensure pigment detection, and allow paying attention to ablation rate. The shot-to-shot energy delivered by the laser is controlled by a photodiode, previously calibrated to ensure the stability of the energy. For this study, laser energy was successively set to 5, 20, 35 and 50 mJ per shot, inducing four craters at diﬀerent locations on the sample surface. The laser beam, focussed by a lens (f : 100 or 200 mm), reached the sample surface at normal incidence. Emission was collected with an angle of 45◦ with respect to the incident beam by a lens (f : 100 mm), and focussed at the entrance of an optical ﬁbres bundle, connected to the entrance slits (5 µm width) of three spectrometers, HR2000 (Ocean Optics, USA), which enabled us to cover a range of wavelengths between 200 and 940 nm. Emission spectra were recorded with an internal CCD detector (integration time: 2.1 ms). 43.2.3 Raman Microscopy Raman microscopy measurements were performed with a LabRam spectrometer from Jobin-Yvon. Raman excitation was obtained using the second harmonic of Nd:YAG laser, at 531.9 nm. For these experiments, conﬁgurations delivering 5 and 0.5 mW power at the sample surface were selected. The beam diameter was set to 150 µm. The incident laser radiation was focussed onto the sample surface by an optical microscope. Radiation backscattered were collected by the microscope and transmitted to an optical spectrometer (grating: 1, 800 mm−1 ; range: 1, 500 cm−1 ). 43.2.4 Samples To simulate conditions encountered on site, samples were specially made with the fresco technique for this work. Table 43.1 presents their Raman bands

370

R. Bruder et al.

Table 43.1. Key LIBS emission lines and Raman bands of the pigments studied Pigments Raman bands

Wavelengths LIBS emission lines

Ultramarine blue (Kremer 548, 583, 806, 45010) 1,357 cm−1 Red lead (Kremer 42500) Brentonico green earth (Kremer TO 501000)

Charcoal (Kremer 47800) Red ochre (Okhra)

Yellow ochre (Okhra)

Calcite

Na 588.99/589.59 nm; Al 308.21/ 309.27/394.40/396.15 nm; Si 288.16/ 390.55 nm Pb 220.35 (II) / 280.20 / 283.30/ 142, 228, 390, 363.96/368.35/405.78 nm 478, 547 cm−1 684, 1,284, 1,389, K 766.49/769.90 nm; Al 308.27/ 394.40/396.15 nm; Fe 238.20 (II)/ 1,507, 248.33/382.04/404.58 nm; Mg 279.55 1,540 cm−1 (II)/ 280.27 (II)/285.21 nm; Si 288.16/ 390.55 nm C 247.85 nm 1,350, 1,600 cm−1 224, 404, 608, Fe 238.20 (II)/248.33/382.04/ 1,330 cm−1 404.58 nm; Al 308.21/309.27/394.40/ 396.15 nm; Si 288.16 / 390.55 nm Fe 238.20 (II)/248.33/382.04/ 304, 402, 563, 695, 1,335 cm−1 404.58 nm; Al 308.21/309.27/394.40/ 396.15 nm; Si 288.16/390.55 nm Ca 315.89 (II)/317.93 (II)/370.60 158, 283, 715, (II)/373.69 (II)/393.37 (II)/396.85 1,086, (II) / 422.67 nm 1,438 cm−1

and atomic emission lines. For micro-LIBS mapping, the sample consisted in a fragment of fresco external layer, i.e. calcite layer containing pigments. The probed zone was composed of three pigments: red lead, ultramarine blue and green earth. In the case of macro-LIBS, samples were concrete plates on which pigments were applied with the fresco technique. Red lead, green earth, charcoal, ultramarine blue (Kremer, Germany), yellow and red ochre (Okhra, France) were considered here.

43.3 Results and Discussion 43.3.1 Micro-LIBS Mapping A sample with three pigments was selected for micro-LIBS mapping, in order to evaluate analytical selective capability of LIBS, at a micrometric scale. Ultramarine blue ((Na − Ca)8–10 Al6 Si6 O24 S2–4 ), green earth (K[(Al,Fe),(Fe,Mg)] (AlSi3 , Si4 )O10 (OH)2 ) and red lead (Pb3 O4 ) were present on the surface. A characteristic emission line of each considered pigment was ﬁrst selected. The mapping was then performed on 170 × 170 points (crater size: 10 µm). Each probed spot gave a spectrum, in which speciﬁc emission

43 Eﬀects of LIBS Measurement Parameters

371

Fig. 43.1. (left) Photography of a zone similar to the probed area (black lines show cracks occurred during sample preparation) and (right) a pseudo-colour map built with lines intensity of Si (390.55 nm), Pb (405.78 nm) and Fe (404.58 nm) (dimensions: 170 × 170 shots, 8.5 × 8.5 mm2 ; dark lines are sample cracks)

lines could be obtained. A map was build by plotting selected line intensity as a function of position on the sample. In the present case, silicon (from ultramarine blue) presents a characteristic line at 390.55 nm, lead (indicating red lead) at 405.78 nm and iron (green earth characteristic) at 404.58 nm, which enables creating a pseudo-colour map based on Si, Pb and Fe. Figure 43.1 shows a photograph of a zone similar in pigment distribution to the probed area by micro-LIBS, and the map built by recording selected lines intensities in the probed zone. Areas presented in black may indicate that the pictorial layer is irregular, with lack of pigment in some regions. So, micro-LIBS is an interesting tool for pigment analysis. However, it could induce painting layer removal, since high ﬂuency is requested to obtain a useful signal. Indeed, valuable spectra were acquired with an excellent spatial resolution. Pigment can be identiﬁed, thanks to their spectrum, and distribution of diﬀerent pigments on the sample was veriﬁed. 43.3.2 Macro-LIBS Experiments The interest of LIBS instrumentation would be for conservation specialists to perform an in situ, rapid, micro-destructive technique for pigment identiﬁcation. However, a number of parameters need to be ﬁtted for an optimal use: emission signal must be analytically exploitable, with minimal eﬀects on the sample. Six pigments were selected to begin this large study, and tested with four laser pulse energies, for two focal lengths, to determine the inﬂuence of the settings on signal and the impact on sample. Figure 43.2a shows signalto-noise ratio (SNR) of diﬀerent emission lines for ultramarine blue plotted vs. the laser pulse energy (f : 100 mm). It clearly demonstrates that the line intensities increase with laser energy, for all characteristic elements present in

372

R. Bruder et al. 10

4

S/N Ultramarine blue S/N Green earth S/N Red ochre

9 S/N Si (288, 16 nm) S/N Al (309, 27 nm)

8

S/N Na (588, 99 nm)

7 S/N ratio

S/N ratio

3

2

S/N Red lead S/N Charcoal S/N Yellow ochre

6 5 4 3

1

2 1 0

0 3

a

8

13

18

23

28

33

Laser energy (mJ)

38

3

43

b

5

7

9

11

13

15

17

19

21

Laser fluency (J.cm−2)

Fig. 43.2. (a) Evolution of SNR of three lines for ultramarine blue vs. laser pulse energy. (b) Evolution of SNR for Ca line (317.933 nm) for all pigments vs. laser ﬂuency

the sample, indicating that the ablation of the sample is more eﬃcient and, at these energies, the laser beam is not masked enough by the plasma to saturate the interaction. Laser pulse energy also inﬂuences the crater diameter. For ultramarine blue, crater diameter increases from 340 µm for a 5 mJ laser pulse to 465 µm for 50 mJ. This phenomenon could be explained by the pseudo-Gaussian shape of the beam: sides of the beam, with increasing energy, could reach ablation threshold to mark the sample. It could also be due to diﬀerent eﬀects such as laser or thermal diﬀusion in the matrix. These hypotheses will be veriﬁed in a future work. Laser ﬂuency was then rather used to compare between diﬀerent shots. A standard aluminium alloy (ref. 970-Péchiney) was used to characterize the beam by measuring the crater surface for the diﬀerent settings chosen. A reference ﬂuency was deﬁned as the ratio between laser energy and the reference area (given by the crater dimensions) in aluminium alloy for a comparable energy. The ﬂuency vs. SNR were plotted for the calcium line at 317.933 nm, on Fig. 43.2b, for all pigments. This ﬁgure shows an enhancement of SNR with increasing laser ﬂuency, which conﬁrms the signal improvement, for all tested pigments. This behaviour is similar for all pigments, indicating that the absorption of calcite matrix is comparable in those cases. Fluency could also be adjusted with the focal length of the lens. Table 43.2 illustrates the inﬂuence of focal length on the crater size. A theoretical model for Gaussian and pseudo-Gaussian beam enables comparison between theory and experiments. It predicts the beam diameter, after a focussing lens, at the focal plane: 4λf (43.1) w = M2 πw0 where w is the laser beam diameter at the beam waist; M 2 , the quality factor; λ, wavelength; f , focal length; w0 , the beam diameter before focussing.

43 Eﬀects of LIBS Measurement Parameters

373

Table 43.2. Inﬂuence of focal length on crater size for diﬀerent energies Ultramarine blue; crater surface (cm2 ) Focal length (mm) 100 200

5 mJ 8,92E−04 1,69E−03

20 mJ 35 mJ 1,24E−03 1,53E−03 2,37E−03 2,35E−03 Yellow ochre; crater surface (cm2 ) 5 mJ 20 mJ 35 mJ 7,77E−04 1,02E−03 1,32E−03 1,62E−03 2,07E−03 2,71E−03

Focal length (mm) 100 200

50 mJ 1,70E−03 2,83E−03 50 mJ 1,51E−03 3,04E−03

Ultramarine (surface) Ultramarine (crater) Calcite

Intensity

Ca

Mg Si Al

0

200

400

600

800

1000

Wavenumber (cm−1)

1200

1400

1600 275

280

285

290

295 300 305 Wavelength (nm)

310

315

320

325

Fig. 43.3. LIBS and Raman spectra of ultramarine blue sample. (left) Raman: 0.5 mW excitation, 4 × 40 s. (right) LIBS: 35 mJ laser shot, crater diameter: 440 µm

The model is not completely respected in such conditions. As in the case of increasing crater diameter with laser energy, this could be explained by the fact that peripheral zones of the pseudo-Gaussian beam could not reach pigment ablation threshold, apparently reducing crater dimensions. Hypothesis of laser or thermal diﬀusion could also be evoked. Regarding signal intensity, SNR obtained for similar laser energy at both focal lengths were comparable. These ﬁrst results seem to indicate that a beam treatment could be necessary to reliably control crater dimensions. For settings optimization, Raman microscopy was used as a diagnosis tool to control the preservation of pigment integrity after LIBS analysis. Raman Microscopy The purpose was to evaluate the feasibility of controlling the pigment removal by LIBS with Raman microscopy. For each energy tested, most pigments were preserved and remain unchanged. Raman spectra of ultramarine blue (acquired with a 0.5 mW laser power, 4 × 40 s exposure time), obtained on the surface and at the bottom of the crater (induced by a 35 mJ laser shot, 440 µm diameter) are presented in Fig. 43.3. Characteristic features of pigment are clearly visible, even at the bottom of the crater, after LIBS analysis, at 548, 583, 806, 1,096 and 1, 357 cm−1 . The calcite spectrum is also visible at

374

R. Bruder et al.

283 and 1, 088 cm−1 . These results indicate that the pigment is not modiﬁed and not completely removed by the LIBS laser shot. Other pigments, such as charcoal, red ochre or red lead, show the same behaviour, demonstrating that Raman spectroscopy could be a very eﬃcient way to control the eﬀects of LIBS on the sample and to check the preservation of the pictorial layer. It could help to optimize LIBS settings, to ensure single layer analysis and enable stratigraphic measurements.

43.4 Conclusions LIBS can be a very eﬃcient tool for wall paintings analysis, but for this purpose, signal collection must be maximal, with minimal eﬀect induced on the sample. In this context of optimization, this work has shown that micro-LIBS is an eﬃcient technique that gives valuable analytical results with a remarkable spatial resolution (crater diameters of 3 µm are possible), but induces damage because of necessary high ﬂuency conditions. For macro-LIBS approach, signal is improved by increasing laser ﬂuency. However, enhancing laser energy also increases crater diameter, due to pseudo-Gaussian shape of the beam, since peripheral zones could reach the sample ablation threshold. Improving beam shape with diaphragm could be an eﬃcient way, which will be tested in future work, to solve this problem and to improve instrumentation spatial resolution. In the case of wall paintings, it is clearly demonstrated that identiﬁcation is possible with a single laser shot. Characteristic emission lines of the pigments are identiﬁed, with the contribution of calcite. Raman microscopy, used as a complementary tool, showed that, even in high ﬂuency conditions, the pictorial layer could be preserved, allowing successive LIBS and Raman analysis on the same location on the sample.

References 1. M. Stefanaggi, Les techniques de la peinture murale, International course on wall paintings conservation, Ravello (Italy), 1997. 2. D. Anglos, S. Couris, and C. Fotakis, in Applied Spectroscopy, Vol. 51, 1025, 1997. 3. D. Anglos, C. Balas, and C. Fotakis, in American Laboratory, Vol. 31, 60, 1999. 4. M. Castillejo. M. Martín, M. Oujja, D.Silva, R. Torres, C. Domingo, J. V. García-Ramos, and S. Sánchez-Cortés, in Applied Spectroscopy, 55, 992, 2001. 5. B. Gilbert, S. Denoël, G. Weber, and D. Allart, in The Analyst, 128, 1213, 2003. 6. B. Hochleitner, V. Desnica, M. Mantler, and M. Schreiner, in Spectrochimica Acta Part B ; 58, 641, 2003. 7. F. Brech and L. Cross, in Applied Spectroscopy, Vol. 16, 59, 1962. 8. J. F. Asmus, G. Guattari, L. Lazzaroni, G. Musumeci, and R. F. Wuerker, in Studies in Conservation, Vol. 18, 49, 1973.

43 Eﬀects of LIBS Measurement Parameters

375

9. F. Colao, R. Fantoni, V. Lazic, and V.Spizzichino, in Spectrochimica Acta B, Vol. 57, 1219, 2002. 10. M. Castillejo, M. Martín, D. Silva, R. Torres, and F. Guerra-Librero, in Journal of Cultural Heritage, Vol. 1 (Suppl.1), 293, 2000. 11. L. Burgio, R. J. H. Clark, T. Stratoudaki, M. Doulgeridis, and D. Anglos, in Applied Spectroscopy, Vol. 54, 463, 2000. 12. D. Bulajic, M. Corsi, G. Cristoforetti, S. Legnaioli, V. Palleschi, A. Solvetti, and E. Rognoni, in Spectrochimica Acta Part B, Vol. 57, 339, 2002. 13. A. Jurado-Lopez and M. D. Luque de Castro, in Applied Spectroscopy, Vol. 57, 349, 2003. 14. B. C. Castle, A. K. Knight, K. Visser, B. W. Smith, and J. D. Winefordner, in J. Anal. At. Spectrom., Vol. 13, 589, 1998. 15. L. J. Radziemski and D. A. Cremer, Laser-Induced Plasmas and Applications, Marcel Dekker, Inc.: New York (US) and Basel (CH), 295, 1989.

44 A Parametric Linear Correlation Method for the Analysis of LIBS Spectral Data E. Tzamali and D. Anglos

∗

Institute of Electronic Structure and Laser, Foundation for Research and Technology-Hellas (IESL-FORTH), P.O. Box 1385, 71110 Heraklion, Crete, Greece ∗ [email protected] Summary. A parametric linear correlation method that enables the identiﬁcation of the diﬀerent elements in multicomponent LIBS spectra based on the comparison against reference spectra of individual elements is proposed. The method is described and preliminary tests are presented with spectral data obtained during the analysis of archaeological bronze.

44.1 Introduction The use of LIBS in the examination of many diﬀerent types of art objects has been extensively investigated and the technique has been shown to be very eﬀective in identifying the elemental content of a wide variety of materials including pigments in paintings, metals, glass, ceramics, etc. [1–4]. The information on the elemental composition relies on the assignment of characteristic emission lines in the LIBS spectrum to the corresponding elements. However, in certain cases, which give rise to rather complex emission spectra, such as those resulting from multicomponent materials, the above straightforward approach may become a tedious task leading to ambiguities regarding the composition of materials. A systematic and reliable method to decompose a multicomponent spectrum and identify all the constituents is, therefore, essential in LIBS analysis and several methods have been used toward this direction. A spectral line-matching approach can be quite successful provided that a proper spectral line database is available (such as the one provided by NIST) and that the LIBS spectra are recorded at high resolution to minimize mismatch of closely lying lines. Alternatively, a pattern recognition approach that relies on visual spectra matching between the analyzed material’s spectrum and standard spectra corresponding to individual elements or certain types of reference materials can be used. In a more systematic approach, linear correlation methods have been successfully used in LIBS analysis of materials, including the sorting of diﬀerent

378

E. Tzamali and D. Anglos

types of steel samples [5], the discrimination among pottery samples of diﬀerent origin [3], or even the control of laser ablation in micromachining [6] or laser cleaning applications [7]. The use of correlation methods for spectral discriminant analysis is beneﬁcial in the sense that the correlation coeﬃcient is invariant to linear transformations of either or both of the two data sets for which it is calculated. This implies that detector sensitivity and baseline level variations, as long as they aﬀect the whole spectrum uniformly, do not aﬀect the coeﬃcient value and the identiﬁcation decision thereupon. A spectral database of well-deﬁned materials can be constructed once and compared with any sample spectrum referred to in the same spectral bandwidth. Nevertheless in the case of multicomponent spectra in which the individual spectral constituents may be present in arbitrary proportions (e.g., pigment mixtures, archaeological metal alloys), a complete reference spectra database would have to include a very large (ideally inﬁnite) number of spectra, which is obviously a problem. If, on the other hand, the multicomponent spectra are correlated against single component ones, the presence of analyte lines of other than the reference components aﬀects the correlation coeﬃcient. The resulting drop of the coeﬃcient, caused by the coexisting emission line patterns, is advantageous in cases where purity certiﬁcation of the reference material is the main objective, indicating a mixing in the material under investigation. However, in spectral decomposition analysis, where the identiﬁcation of all the constituents is the main objective, the resulting drop of the correlation coeﬃcient leads to ambiguities. A reasonable solution to the problem of coexisting spectral patterns is the shifting of “attention” to each one of the corresponding reference spectral patterns. The method proposed herein utilizes a modiﬁed linear (parametric) correlation approach to decompose and identify the constituents of a multicomponent spectrum. This chapter describes thoroughly the parametric correlation method followed by some preliminary results from data analysis on LIBS spectra collected during the examination of archaeological copper alloys.

44.2 Description of the Correlation Method 44.2.1 Linear Correlation This method emphasizes the degree to which a linear model may describe the relationship between two variables X and Y . The Pearson product-moment correlation coeﬃcient is the measure of how well a linear equation describes the relation between X and Y . If a series of n measurements of X and Y written as xi and yi where i = 1, 2, . . . , n, is assumed, then the Pearson product-moment correlation coeﬃcient (r or rxy ) can be used to estimate the correlation of X and Y . The Pearson coeﬃcient is also known as “sample linear correlation coeﬃcient.” It is especially important if X and Y are both

This page intentionally blank

44 A Parametric Linear Correlation Method

379

30000 rCu = 0.974

rSn = 0.007 20000

4000

ISn (a.u)

ICu (a.u)

6000

2000

0

10000

0 0

5000

10000

15000

ISample (a.u)

0

5000

10000

15000

ISample (a.u)

Fig. 44.1. Correlation graphs of a LIBS spectrum obtained from a Cu–Sn sample (1% by weight in Sn) against the corresponding spectra from pure copper (left) and pure tin (right)

normally distributed and is then the best estimate of the correlation of X and Y . The mathematical expression of the Pearson correlation coeﬃcient is given by (xi − x ¯)(yi − y¯) (xi − x ¯)(yi − y¯) = i , (44.1) rxy = r = i (n − 1)Sx Sy (xi − x ¯)2 (yi − y¯)2 i

i

where x ¯ and y¯ are the sample means of xi and yi , respectively, Sx and Sy are the sample standard deviations of xi and yi and the sum is from i = 1 to n. The linear correlation coeﬃcient is a number between −1 and 1, which measures how close to a straight line the set of points (xi , yi ) falls. A correlation coeﬃcient close to one and zero implies that the variables X and Y are correlated and uncorrelated, respectively. In the case of LIBS data, X and Y represent the two spectra to be compared; usually one is the reference and the other is the sample spectrum. The spectral data xi and yi represent emission intensity values accordingly, at wavelength λi (in practice the index i represents the pixel number on the detector). Figure 44.1 shows graphs indicating the correlation of a LIBS spectrum obtained with a copper–tin (Cu–Sn) alloy sample (Sn content of 1%) against two reference spectra, one from pure copper (very good correlation) and the other from pure tin (poor correlation). 44.2.2 Parametric Correlation The proposed method relies on the selective correlation of a raw data spectrum with the reference spectra in the database. The comparison is accomplished by applying the linear correlation method, focusing speciﬁcally on the characteristic lines of the database reference spectrum. This is accomplished by

380

E. Tzamali and D. Anglos

masking both the sample and reference spectra according to the emission lines of the reference spectrum. The mask is an array of zeroes and ones, where ones indicate the positions of the selected emission peaks taking into account the spectral line width. Masking means a multiplication of each of the spectral data points with the mask array. Furthermore, in order to select the characteristic lines from a reference spectrum, a threshold based on the statistical characteristics of each reference spectrum is applied and a mask array with zeroes and ones is set accordingly, thus the term “parametric” linear correlation. Following operation of the mask array on both the sample and reference spectra, the correlation method is applied. The mathematical expression that describes the above idea is similar to (44.1) with the exception of the binary parameter ai : ai (xi − x ¯)(yi − y¯) i , ai ∈ {0, 1}. (44.2) ra = ai (xi − x ¯)2 ai (yi − y¯)2 i

i

The LIBS spectrum of raw data is compared against all1 reference spectra individually on the basis of parametric correlation coeﬃcient and one can thus identify which elements/materials are presented in the spectrum. This approach has been employed by Galbacs et al. [8] to process LIBS spectral data in relation to quantitative analysis of binary alloys producing calibration curves, which rely on the dependence of the masked correlation coeﬃcient (corresponding to several analytical lines from both elements of the alloy) on the composition of the alloy.

44.3 Results and Discussion 44.3.1 Method Testing In order to test the performance of the method and investigate its sensitivity and limitations, tests were carried out on a series of simulated spectra produced by combining real LIBS spectra (rms explain this normalized) from pure copper and tin in diﬀerent proportions. Figure 44.2 shows the parametric correlation coeﬃcients of the composite spectra against the reference spectra of Cu or Sn as a function of the proportion of each spectral component. Similar correlation responses are observed for both components, which are quite high (r > 0.9) when the component in question is found in a proportion over 10%. It is noted that a thorough study of parameters involved in the parametric correlation that has tested the probability for false positive or false negative hits suggests that a coeﬃcient above 0.5 indicates a positive 1

The process may be simpliﬁed by limiting the comparisons to those elements/materials that are sought for.

44 A Parametric Linear Correlation Method

381

1,00

Parametric r

0,75 0,50 0,25 rCu rSn

0,00 −0,25 0

20

40 60 80 spectral proportion

100

Fig. 44.2. Dependence of the parametric correlation coeﬃcient on the spectral composition in simulated LIBS spectra of Cu–Sn alloys Cu

rCu = 0.884 rSn = 0.764

Cu

rPb = 0.618 rAu = −0.100

Sn Sn

Cu

Sn

Sn Pb

Fe

260

280

Pb

Sn Sn

Sn

Ag Cu Ag

Cu

300

320

340

Wavelength (nm)

Fig. 44.3. LIBS spectrum from a Minoan bronze scraper

identiﬁcation. This observation enables reliable identiﬁcation of the minor spectral component down to proportions ∼3%. For realistic purposes, namely analysis of actual bronze objects in which the weight concentration of Sn ranges from 0.5 to 20% and in a few cases up to 35%, the relevant simulated spectra show very good behavior for both Cu and Sn. 44.3.2 Analysis of Archaeological Bronze A Minoan bronze scraper was examined to demonstrate the method’s potential. The sample was chosen since it contains several diﬀerent elements and constitutes a characteristic example of archaeological ﬁndings. The spectrum obtained (Fig. 44.3) was correlated against the reference spectra of Cu, Sn,

382

E. Tzamali and D. Anglos

Pb, and Au. The parametric correlation method veriﬁed the presence of Cu, Sn, and Pb, and properly conﬁrmed the absence of Au in the sample.

44.4 Conclusions The parametric correlation method described in this work is a promising tool for assisting in the analysis and interpretation of LIBS spectral data enabling the reliable identiﬁcation of individual elements in multicomponent LIBS spectra. Acknowledgments The work was supported in part by the General Secretariat for Research and Technology (Project: LASTOR, contract: Π14/2004) and the European Commision (Project PROMET, contract: FP6-2002-INCO-MPC-1).

References 1. D. Anglos, in Appl. Spectrosc. Vol. 55, 186A, 2001. 2. M. Corsi, G. Crisoforetti, V. Palleschi, A. Salvetti, and E. Tognoni, in Eur. Phys. J. D 13, 373, 2001. 3. J. M. Anzano, M. A. Villoria, I. B. Gornushkin, B. W. Smith, and J. D. Winefordner, in Canadian Journal of Analytical Sciences and Spectroscopy, Vol. 47, 134, 2002. 4. F. Colao, R. Fantoni, V. Lazic, and V. Spizzichino, in Spectrochim. Acta Part B, Vol. 57, 1219, 2002. 5. B. Gornushkin, B. W. Smith, H. Nasajpour, and J. D. Winefordner, in Anal. Chem. Vol. 71, 5157, 1999. 6. T. Tong, J. Li, and J. P. Longtin, in Applied Optics. Vol. 43, 2004. 7. M. Lentjies, K. Dickmann, and J. Meijer, “Low resolution LIBS for online monitoring during laser cleaning based on correlation with reference spectra” in Book of Abstracts, 6th International Congress on Lasers in the Conservation of Artworks, Vienna, 21–25 September 2005. 8. G. Galbacs, I. B. Gornushkin, B. W. Smith, and J. D. Winefordner, in Spectrochim. Acta Part B, Vol. 56, 1159, 2001.

45 Investigation on Painting Materials in “Madonna col Bambino e S. Giovannino” by Botticelli ∗

D. Bersani1 , P.P. Lottici1 , A. Casoli2 , M. Ferrari2 , S. Lottini1 , and D. Cauzzi3 1

∗ 2

3

Dipartimento di Fisica, Università di Parma, Parco Area delle Scienze 7/a, 43100 Parma, Italy [email protected] Dipartimento di Chimica Generale e Inorganica, Chimica Analitica, Chimica Fisica, Università di Parma, Parco Area delle Scienze 17/a, 43100 Parma, Italy Soprintendenza per il Patrimonio Storico Artistico e Demoetnoantropologico, via Belle Arti, 56, 40126 Bologna, Italy

Summary. A study on the painting materials (pigments and binders) of the famous painting “Madonna col Bambino e S. Giovannino” by Sandro Botticelli, located in the Museo Civico of Piacenza (Italy), was performed before a recent restoration. The painting materials were investigated by the analysis of ﬁve millimetric samples taken in damaged regions. The pigments were determined using the micro-Raman spectroscopy, with the 632.8 nm line of a He–Ne laser. Despite the strong ﬂuorescence background, the nature of the ground layer (gypsum and anhydrite) and of most pigments (i.e. goethite, lapis lazuli, white lead) was determined. Gas chromatography coupled with mass spectroscopy (GC/MS) was used to determine the organic binder media, and in particular proteinaceous and lipid materials. Egg and animal glue were found, while no siccative oils were detected.

45.1 Introduction The recent (2004) restoration of “Madonna col Bambino e S. Giovannino” by Sandro Botticelli provided a unique opportunity to study by means of microsamplings the materials used by the painter. The painting, dated 1480–1485, is conserved since 1903 in the Musei Civici of Palazzo Farnese in Piacenza, Italy, and is a property of the Piacenza municipality since 1862, when it was donated by the Oratory of the Bardi Castle (Parma, Italy). The artwork represents a Madonna adoring her Child with the little Saint John and was executed on four wooden boards shaped in circular form (“tondo”). The Holy Virgin is knelt down, with the hands joined and the look turned to her son. On the other side is San Giovannino, dressed with goat

384

D. Bersani et al.

skins and a red cape and carrying a cross made with canes. A green ﬁeld with some cut roses, two bushes of blooming red roses and a landscape complete the artwork. The painting is attributed to Botticelli since 1890 by Pollinari while the original wooden frame is attributed to the Giuliano da Maiano studio.

45.2 Experimental Methods A very small amount of painted material (four samples, each a few mm2 in size) was taken from damaged regions near the frame of the painting, without any further visible injury to the artwork. Every sample was taken in order to contain all the paint layers, from the outer surface to the ground layer. One more sample was taken from the original wooden frame in order to identify the binders under the gilding layer. The samplings were performed at the beginning of the restoration work, before cleaning the painting. The sampling points are shown in Fig. 45.1. Part of the samples was then used to obtain cross sections in resin, necessary for a detailed study of the ground and paint layers. After the observation with the optical microscope the raw samples and the cross sections were analysed by means of Raman micro-spectroscopy, using a Jobin-Yvon Horiba Labram apparatus, equipped with a motorized xy-stage and using the 632.8 nm line of a He–Ne laser as excitation. The spectral resolution is about 2 cm−1 , while the spatial resolution is on the order of 2 µm. After the nondestructive micro-Raman analysis, part of the samples was used to identify the organic compounds by means of the gas chromatography/mass spectroscopy

Fig. 45.1. “Madonna col Bambino e S. Giovannino” by Botticelli and the sampling points

45 Investigation on Painting Materials

385

(GC/MS). A HP 5890 Series II gas chromatograph coupled to a HP-5971A mass selective detector was employed [1].

45.3 Results and Discussion 45.3.1 Sample 1 The sample was taken from a green region of a leaf painted near the right bottom rim. The cross section shows two diﬀerent layers: the white ground and a green painted layer. This outer layer contains green and blue grains, with some yellow areas (Fig. 45.2). The micro-Raman analysis performed on the blue grains, both in the raw fragment and in the cross section, shows the characteristic spectra of lazurite (Na, Ca)8 (AlSiO4 )6 (O, S, SO4 )1−2 which is the main component of lapis lazuli [2, 3]. Green grains on the outer surface of the raw sample show Raman spectra of copper sulphate (brochantite or posnjakite – main bands at 386, 427, 972 cm−1 ) [3], probably due to the sulphation of malachite CuCO3 · Cu(OH)2 [4], a very common green pigment. Malachite is indeed clearly detected by the Raman analysis (Fig. 45.3 – peaks at 178, 218, 267, 429, 509, 1,050, 1, 090 cm−1 ) of similar green grains at greater depths in the cross section, where no sulphation products are found. The yellow zones visible in the cross section show the Raman spectra of lead tin yellow (Type I), Pb2 SnO4 (Raman peaks at 130, 197, 274, 292, 456, 525 cm−1 ), a synthetic pigment in use by the twelfth century [3] (Fig. 45.3). The Raman measurements performed on the white ground layer show clearly the anhydrite (CaSO4 ) spectrum (main peak at 1, 017 cm−1 ) [2] (Fig. 45.3). The calcium sulphate, both in anhydrous (anhydrite) and hydrous (gypsum) form, was often used to obtain the ground layers in paintings on wood. As to the organic binders, the GC/MS analysis shows evidence of amino acids. No fatty acids were detected (Fig. 45.4). The proteinaceous material was identiﬁed as egg and animal glue (revealed by hydroxyproline).

Fig. 45.2. Cross section of Sample 1

386

D. Bersani et al.

Abundance (Arb. Units)

Fig. 45.3. Raman spectra of some compounds found in Sample 1

NVal

NLeu

Gly

Hyp Pro

Ala Ser

Glu Asp

Len

Phe

Thr 6.50

7.00

7.50

8.00

8.50

9.00

9.50

Time (min)

Fig. 45.4. GC/MS proﬁle of Sample 1: Ala = alanine, Gly = glycine, Thr = threonine, Ser = serine, NVal = norvaline (internal standard), Leu = leucine, Nleu = norleucine (internal standard), Pro = proline, Hyp = hydroxyproline, Asp = aspartic acid, Glu = glutamic acid, Phe = phenylalanine

45.3.2 Sample 2 Sample 2 was taken from the hip of San Giovannino on the right rim of the tondo: it is of pink colour. The cross section evidences six diﬀerent layers (Fig. 45.5): a white ground with maximum thickness of about 140 µm; four white thin layers (thickness from 5 to 10 µm), intercalated by thinner layers of yellow-red, ﬁnely ground (1–2 µm) pigments; and a red-brown, thick (20–40 µm) external layer seemingly due to a varnish paint.

45 Investigation on Painting Materials

387

Fig. 45.5. Cross section of Sample 2

The Raman measurements performed on the white ground show the two characteristic Raman features of gypsum (CaSO4 · 2H2 O) at 1, 005 cm−1 and of anhydrite CaSO4 at 1, 017 cm−1 . All the intermediate white layers are composed of white lead (biacca, 2PbCO3 · Pb(OH)2 ). The Raman spectra obtained from the yellow pigments in the innermost pigmented layer show the features of goethite α-FeOOH (Raman bands at 205, 242, 297, 394, 477, 546, 1, 006 cm−1 ), the main component of yellow ochre [5] only, while in the other pigmented layers, the presence of massicot (PbO) (main peaks at 145 and 289 cm−1 ) was revealed, together with goethite. The outer red-brown layer gives a strong ﬂuorescence, probably due to the varnish paint or to a lake, which hides most of the Raman signal. Only one measurement in a red area gave the strong Raman bands of cinnabar (vermilion, HgS) (bands at 254, 285, 345 cm−1 ). As in the Sample 1, the GC/MS analysis of the organic materials shows egg and animal glue. 45.3.3 Sample 3 Sample 3 was taken from a deep pink area in the second rose, from the top, belonging to the left bush, probably painted on the sky background. Five layers are identiﬁed by the cross section (Fig. 45.6): an innermost white ground layer; a uniform white priming layer, 20 µm thick; a layer, 35–40 µm thick, in which large blue crystals are embedded in a white matrix; a reddish thin (9–10 µm) layer; and an outermost thin (9–10 µm) red glazing layer. Also in this sample, the ground layer is made of calcium sulphate. The Raman signal obtained from the white priming is typical of white lead. The blue grains give the Raman spectrum of lazurite (Fig. 45.7 – peaks at 266, 547, 584, 814, 1, 093 cm−1 ). Near the blue crystals, in some cases, smaller gold-like grains are present: their Raman spectrum is that of pyrite FeS2 (Fig. 45.7 – sharp Raman peaks at 342 and 376 cm−1 ), a mineral usually present in the natural lapis lazuli, forming golden veins inside the blue lazurite. The large size of the lazurite crystals (7–8 µm) indicates the high quality of the pigment. The white matrix in which the blue grains are embedded is found to be white lead (Fig. 45.7 – main band at 1, 052 cm−1 ). All the Raman measurements performed in the red outer layers were dominated by a

388

D. Bersani et al.

Fig. 45.6. Cross section of Sample 3

Fig. 45.7. Raman spectra of some compounds found in Sample 3. W = white lead

strong ﬂuorescence signal, so it was not possible to identify the red pigment. As concerns the organic binders, the GC/MS analysis shows, as before, egg and animal glue. 45.3.4 Sample 4 Sample 4, of green colour, was taken from a leaf of the rose bush at the left-hand side of the painting. Above the priming, three layers have been identiﬁed (Fig. 45.8): an inner layer belonging to sky background where blue grains are found in a white medium; a dark layer, from the priming of the grass background; and a green painting layer where green grains are observed in a yellow medium. The Raman analysis shows that the inner layer is composed of a white lead matrix in which lapis lazuli grains are embedded. The dark intermediate layer gives the Raman signal of carbon black (bands around 1,320 and 1, 570 cm−1 ) [2]. The green grains in the outer paint layer give the Raman features of malachite, while the yellow part of the same layer shows the spectrum of lead tin yellow (Type I). For this sample separate GC/MS analysis of the paint and ground layers were performed. The results show the presence of egg and animal glue in the paint layer and in the ground layer, respectively.

45 Investigation on Painting Materials

389

Fig. 45.8. Cross section of Sample 4

45.3.5 Sample 5 The sample has been taken from the frame, by removing both the gilding layer and the underlying bole (ﬁne clay, usually red, mixed with organic binders), which is the base for the application of the gold leaf. The cross section reveals three diﬀerent layers: the ground; a thin red-orange layer with regular thickness (9–10 µm) (bole); and the gilding. For this sample, only the organic components were studied by means of GC/MS: the analysis reveals, as in the paint round table, the typical proteins found in animal glue and egg and no trace of lipid material. Separate measurements indicate that the animal glue has been used as a binder in the ground before the application of the bole which had been prepared with egg, probably albumen.

45.4 Conclusions The combined use of the Raman micro-spectroscopy and of the gas chromatography coupled with mass spectroscopy enabled us to identify both inorganic and organic components of the painting materials, resulting in wider knowledge of the pigments and binders used by Botticelli in his “Madonna con Bambino e S. Giovannino” painting. The information is fundamental for the study of the painting technique of the artist. The nature of the ground (gypsum, anhydrite and animal glue) is typical of all Botticelli’s artworks. The pigments used for the priming are characteristics of the painter: white lead for the light areas and pure carbon black in dark areas. In the painter’s palette (Table 45.1) only one earth colour was found (yellow ochre), while red ochre or green earths have not been identiﬁed in the samples. This reveals the choice of Botticelli for bright colours. These results may conﬁrm the debated attribution of the whole artwork to Botticelli. As to the binders for pigments, while siccative oils were found in many artworks of Botticelli, here the GC/MS analysis suggests the presence of the egg proteins only.

390

D. Bersani et al. Table 45.1. Painting materials found in the samples investigated #

micro-Raman

GC/MS

1

ground: anhydrite pigments: brochantite/ posnjakite, malachite, lapis lazuli, lead tin yellow (type I) ground: gypsum, anhydrite priming: white lead pigments: goethite, massicot, cinnabar ground: gypsum priming: white lead pigments: lapis lazuli, pyrite, white lead priming: white lead, carbon black pigments: lapis lazuli, malachite, lead tin yellow (type I) –

animal glue–egg

2

3

4

5

animal glue–egg

animal glue–egg

animal glue (ground) egg (paint layer)

animal glue (ground) egg (bole)

Acknowledgements The authors want to thank the restorer Donatella and Paola Zari and Carlo Giantomassi for their technical assistance.

References 1. A. Casoli, P. Cremonesi, G. Palla, and M. Vizzari, in Annali di Chimica, Vol. 91, 727, 2001. 2. L. Burgio, and R. J. H. Clark, in Spectrochimica Acta. Part A, Vol. 57, 1491, 2001. 3. M. Bouchard, D. C. Smith, in Spectrochimica Acta. Part A, Vol. 59, 2247, 2003. 4. D. Bersani, P. P. Lottici , G. Antonioli, E. Campani, A. Casoli, and C. Violante, in J. Raman Spectrosc. Vol. 35, 694, 2004. 5. S. Scardova, P. P. Lottici, D. Bersani, G. Antonioli, G. Michiara, and C. Pezzani, in Studies in Conservation, Vol. 47, 24, 2002.

46 Laser-Induced Plasma Spectroscopy for the Analysis of Roman Ceramics Terra Sigillata ∗

A.J. López , G. Nicolás, M.P. Mateo, V. Piñón, and A. Ramil Laboratorio de Aplicacións Industriais do Láser. Dpto Enxeñaría Industrial II, Universidade da Coruña, Mendizábal s/n 15403 Ferrol (A Coruña), España ∗ [email protected] Summary. Roman ceramics Terra Sigillata from diﬀerent production areas have been analyzed by means of “laser induced plasma spectroscopy” (LIPS). The overall objective of this study is to show the capability of LIPS to classify these archaeological ceramics as a function of their provenance. The use of linear correlation methods allows us to cluster the samples by quantitative comparison of their LIP spectra, leading to a reliable assignment of Terra Sigillata pieces to their regions of origin and establishing reference groups for the purpose of assigning future pieces.

46.1 Introduction Terra Sigillata is the name of ﬁne tableware produced in Roman times and characterized by a red sintered slip, obtained from a suspension of very ﬁne clay of suitable composition, which was deposited before ﬁring on the surface of the body. The combination of ﬁring conditions with grain size and composition of the clay led to the development of a highly sintered coating. The production of Terra Sigillata began in Central Italy in the ﬁrst century B.C. and from there it spread to many areas of the Roman Empire, being popular until the ﬁfth century AD. Due to extensive trading in the ancient world, ﬁndings of Terra Sigillata from an excavated site may include products from diﬀerent workshops and periods that frequently diﬀer in technological features. Scientiﬁc examination of the pottery sherds can provide valuable information for assigning provenances and rediscovering production technologies [1]. Various analytical methods have been used to study ancient ceramics providing important physical and chemical insight to the structure of objects and materials. The most common methods include molecular, atomic, and electron spectroscopy as well as X-ray and nuclear techniques. Moreover, in recent years, “laser induced plasma spectroscopy” (LIPS) has also been increasingly applied for this purpose [2–7]. Its main advantages over more conventional

392

A.J. López et al.

methods could be the possibility to obtain in-depth compositional proﬁles, the virtual nondestructiveness of this technique, and the fast analysis speed. Commonly LIPS is used to extract elemental intensity peaks out of the plasma radiation; however, a diﬀerent approach is to consider the LIP spectrum as a spectral “ﬁngerprint”. Some authors have been applying this method in combination with linear correlation to identify materials with very similar LIPS spectra [8]. The aim of this chapter is to show the capability of LIPS for the analysis and classiﬁcation of Terra Sigillata sherds as a function of their provenance. With this purpose in mind, 25 Terra Sigillata samples, from diﬀerent production areas, have been analyzed by LIPS and their spectra are compared by using linear correlation analysis. The methodology developed in this work allows us to obtain reliable classiﬁcation of the samples as a function of their provenance and to establish reference groups for the purpose of assigning future pieces.

46.2 Experimental Methods A Q-switched Nd:YAG laser source operating at the third harmonic, 355 nm, was used to irradiate the samples in air at room temperature and pressure. The laser beam was focused onto the sample surface at normal incidence with a quartz plano-convex lens (300 mm focal length) leading to an analyzed area of 0.5 mm2 . The emission of the plasma was collected and guided to the spectrograph (Oriel, model MS257) with a ﬁber optic. Light was dispersed by using the 600 grooves/mm grating of the spectrograph. An intensiﬁed solidstate, two-dimensional charge-coupled device (Andor, model DH5H7-18F-03) was used to detect the plasma image. The CCD consisted of 512×512 elements and presented an active area of 12.3 mm × 12.3 mm. The type and morphology of the material to be analyzed were taken into account to select the optimal working parameters. The energy density, the number of laser shots, and other experimental parameters in relation with the detection device were adjusted to get a high signal to noise ratio of the measured atomic lines and best reproducibility. However, in order to evidence the transition from slip to body in Terra Sigillata samples, low energy density values need to be used. Furthermore, these gentle irradiation conditions preserve the archaeological specimen from signiﬁcant damage. For these reasons, energy density values around 2.0 J cm−2 were chosen as a compromise between spectral signal and depth resolution. R , which facilitates the selecSpeciﬁc software was developed using Matlab tion of the spectral window and performs the analysis of the LIP spectra; in brief, the software reads spectrometer data ﬁles and calculates all mutual correlation coeﬃcients between the spectra. Moreover, in order to improve the calculations, a ﬁne tuning of the wavelength is automatically performed for each pair of spectra.

46 Laser-Induced Plasma Spectroscopy

393

46.3 Results and Discussion On the basis of archaeological information, the 25 Terra Sigillata sherds were classiﬁed as can be seen in Table 46.1: 13 pottery samples (H1–H13) were assigned to Hispanic workshops in Tricio (La Rioja) and Andújar (Andalucía) and seven pieces were attributed to the Gaulish centre of La Graufesenque (G1–G7). All of them were dated in Higher Roman Empire (from ﬁrst to second century AD). In addition, two samples (HL1–HL2) from the Hispanic workshops in Tricio and three samples from the north-African centers (A1–A3) dated from the Lower Roman Empire (from fourth to ﬁfth century AD) were also analyzed. In previous works [9, 10], morphological and compositional diﬀerences between slip and body in Terra Sigillata pieces have been studied by means of SEM-EDX and compositional depth proﬁles have been obtained by LIPS. SEM images revealed important morphological diﬀerences between the thickness and the degree of sinterization of the slip in these pottery samples. Those corresponding to ﬁrst and second century AD present a well sintered, well adhering thick slip in contrast with pieces manufactured in fourth to ﬁfth century AD, suggesting a decline of technological skill. In addition, LIPS depth proﬁles showed the existence of compositional gradient from slip to body, which was clearly assessed in the case of calcium and iron contents. Based on these results and in order to obtain a reliable classiﬁcation of the samples, analyses of the slips and bodies have been performed separately. In a ﬁrst stage, each one of the 25 ceramic sherds was analyzed at ﬁve random points of a fresh fracture by accumulating ten laser shots over the same sample position. The average of the ﬁve measurements in the spectral window from 260 to 340 nm was taken as the “LIPS-ﬁngerprint” of the sample. Classiﬁcation of the ceramic bodies and lack of standards in these archaeological materials force us to establish representative pottery of a production area. The criteria followed was based on the choice of the sample in each provenance group (Hispanic, Gaulish, and African) that showed the lowest dispersion in the ﬁve LIP spectra, that is, the sample with most homogeneous body composition. Table 46.1. Archaeological classiﬁcation of the 25 Terra Sigillata samples analyzed by LIPS Region

Workshops

Samples

Period

Hispanic

Tricio (La Rioja) Andújar (Andalucía) Tricio (La Rioja) La Graufesenque (Aveyron)

H1–H13

1st–2nd century AD

HL1–HL2 G1–G7

4th–5th century AD 1st–2nd century AD

A1–A3

4th–5th century AD

Hispanic Gaulish North-Africa

1.5

Ca(I) 315.887 Ca(I) 317.993

Ti(II) 323.452

Intensity [A.U.]

2

AI(I) 308.215 AI(I) 309.271

Fe(II) 274.948

x106

Mg(I) 285.213 Si(I) 288.158

A.J. López et al. Mg(II) 279.55

394

a)

1

b)

0.5 c) 0 260

270

280

290 300 310 Wavelength [nm]

320

330

340

Fig. 46.1. LIPS spectra of the samples: (a) H5, (b) G4, and (c) A2 selected as references of Hispanic, Gaulish, and African groups, respectively

Figure 46.1 shows LIP spectra taken as references of the diﬀerent production areas: Hispanic, Gallic, and African. In this spectral window, emission lines corresponding to Fe, Mg, Si, Al, Ca, and Ti can be observed. It should be noted that all the samples analyzed in this study present similar elemental composition in terms of major constituents, only diﬀerences between calcium content (in the spectrum interval from 315 to 320 nm) of African reference as compared to the others is noticeable. Consequently, the subtle diﬀerences between LIP spectra consist of variations in the ratios of line intensities that are not obvious in a visual examination. The “LIPS-ﬁngerprints” of all the pottery sherds were compared with the references by means of linear correlation techniques that measure the association between variables by a linear correlation coeﬃcient R, expressed in the following form: (xi − x) (yi − y) i (46.1) R = 2 2 (xi − x) (yi − y) i

where x is the mean of xi values and y is the mean of yi values. The absolute value of R can be used as an indication of the association (correlation) between the x and y data sets, since values around zero indicate uncorrelated data sets while R = 1 and R = −1 correspond to complete positive and negative correlation, respectively, when the data points lie on a perfect straight line with positive (R = 1) or negative (R = −1) slope. In this study, the correlation method has been applied to Terra Sigillata samples to quantify the similarity between their spectra. The spectral intensities measured in one sample (yi ) were plotted versus intensities of LIP

46 Laser-Induced Plasma Spectroscopy H G A

0.2 0.15 1− RA

395

0.1 0.05 0 −0.04 −0.03 −0.02 −0.01

0

0.01

0.02

0.03

0.04

RG − RH

Fig. 46.2. Discriminating plot of variables V1 = 1 − RA vs. V2 = RG −RH . Pottery samples are clustered into three well-separated groups. Ellipses correspond to the 90% conﬁdence level

spectrum (xi ) of the references. Values of RH , RG , and RA (correlation coefﬁcients with respect to Hispanic, Gallic and African references, respectively) were obtained. To provide a method of quantiﬁcation, two variables V1 and V2 that are linear combinations of RH , RG , and RA were deﬁned: V1 = 1 − RA and V2 = RG − RH . In Fig. 46.2, the discriminating pattern result of plotting V1 vs. V2 is shown. As it can be observed, the samples are clustered into three well-separated groups, corresponding to Hispanic, Gallic, and African provenance. Because of the small number of samples analyzed, the Student’s t probability density function was assumed in V1 and V2 values [11]. Ellipses in the plot were obtained and deﬁne the 90% of conﬁdence level for variables V1 and V2 . The ﬁgure reveals that, analyzing the ceramic bodies, all the samples can be ascribed with a 90% of conﬁdence to one of the three production areas. Therefore, the initial archaeological classiﬁcation based on the provenance region of the pottery was largely conﬁrmed by the LIPS studies. Once the bodies of Terra Sigillata sherds were analyzed, the slips have also been considered. By subsequent ablation of the sample surface at the same irradiated spot, the in-depth LIPS proﬁle (from slip to body) was carried out. Series of 40 pulses were performed in six random points of the sample surface. The ﬁrst spectrum of the series was selected as “slip-reference” and an averaged spectrum of the body was taken as “body-reference”. The correlation coeﬃcients corresponding to samples H2 and HL1 are shown in Fig. 46.3a, b, respectively. These coeﬃcients were calculated by comparing each spectrum of the series with references of the slip and the body. The intersection point between both curves in Fig. 46.3 can be taken as the transition point between slip and body. This point occurs for a diﬀerent number of pulses depending on the sample, showing great diﬀerences in slip

396

A.J. López et al. 1

a)

correlation coefficient

correlation coefficient

1 0.95 0.9

0.85 0

10 20 30 number of pulses

40

b) 0.98

0.96

0.94

0

10 20 30 number of pulses

40

Fig. 46.3. Values of correlation coeﬃcient between each spectrum of the series and references of the slip (point) and body (x-mark) of Terra Sigillata samples from the Hispanic workshops in Tricio, (a) H2 and (b) HL1

number of pulses

30 H HL G A

20

10

0 sample

Fig. 46.4. Transition point from slip to body (i.e., slip thickness) for each Terra Sigillata sample analyzed by LIPS: H (Hispanic), HL (Hispanic Late), G (Gallic), and A (Northern-African)

thickness between samples corresponding to diﬀerent periods, in agreement with images obtained by SEM. Figure 46.4 represents the number of pulses corresponding to the transition point (i.e., thickness of the slip) taken as an average of the six analyzed points of each sample. Diﬀerences in this value allow us to distinguish between samples from the same production area but diﬀerent periods as in the case of groups H (from ﬁrst to second century AD) and HL (from fourth to ﬁfth century AD). It should be noted that the error bars are, in some cases, very large showing important variability in the slip thickness of the pottery sherds.

46.4 Conclusions The results presented in this chapter demonstrate the applicability of LIPS technique combined with linear correlation methods to classify Terra Sigillata samples as a function of their provenance. Reference spectra of Hispanic,

46 Laser-Induced Plasma Spectroscopy

397

Gallic, and African groups have been established and can be used in future work for rapid classiﬁcation of Terra Sigillata sherds from excavated sites. LIPS depth proﬁles give information about the position of the transition between slip and body, i.e., slip thickness, which could be used to distinguish between Terra Sigillata samples from the same production area but diﬀerent periods. Moreover, the methodology developed here can be applied to other archaeological studies focused on the assignment of provenances with the advantage of its virtual nondestructiveness and fast analysis speed. Acknowledgments The authors wish to thank Dr. M.C. López Pérez and Museo de Prehistoria e Arqueoloxía de Vilalba for providing pottery samples. This work was partially supported by Palladium Estudio de Arqueología S.L. and Xunta de Galicia through Project PGIDIT03CCP04E.

References 1. M. S. Tite in Proceedings of the International School of Physics Enrico Fermi, Physics Methods in Archaeometry, IOS Press, The Netherlands, 2003. 2. D. Anglos, in Appl. Spectrosc. Vol. 55, 186A, 2001. 3. Y. Yoon, T. Kim, M. Yang, K. Lee, and G. Lee, in Microchem. J., Vol. 68, 251, 2001. 4. K. Melessanaki, M. Mateo, S. C. Ferrence, P. P. Betancourt, and D. Anglos, in Appl. Surf. Sci., Vol. 197–198, 156, 2002. 5. F. Colao, R. Fantoni, V. Lazic, and V. Spizzichino, in Spectrochim. Acta Part B, Vol. 57, 1219, 2002. 6. J. M. Anzano, M. A. Villoria, I. B. Gornushkin, and B. W. Smith, in Can. J. Anal. Sci. Spectrosc., Vol. 47, 134, 2002. 7. V. Lazic, F. Colao, R. Fantoni, A. Palucci, V. Spizzichino, I. Borgia, B. G. Brunetti, and A. Sgamellotti, in J. Cult. Heritage Vol. 4, 303, 2003. 8. I. B. Gornushkin, B. W. Smith, H. Nasajpour, and D. Winefordner, in Anal. Chem., Vol. 71, 5157, 1999. 9. A. J. López, G. Nicolás, M. P. Mateo, V. Piñón, M. J. Tobar, and A. Ramil: “Compositional analysis of Hispanic Terra Sigillata by laser spectroscopy”. In Spectrochimica Acta Part B (In press). 10. A. J. López, G. Nicolás, M. P. Mateo, V. Piñón, A. Ramil, and A. Yáñez: “Análisis de cerámicas romanas Terra Sigillata mediante espectroscopia de plasmas inducidos por láser (LIPS)”. In Bol. Soc. Esp. Ceram. V. (In press). 11. P. G. Hoel in Introduction to Mathematical Statistics, Wiley, New York, 1984.

47 Laser-Induced Fluorescence Analysis of Protein-Based Binding Media ∗

A. Nevin1,2 , S. Cather1 , D. Anglos2 , and C. Fotakis2 1

2

∗

Conservation of Paintings Department, Courtauld Institute of Art, University of London, Somerset House, Strand, WC2R ORN, London, U.K. Institute of Electronic Structure and Lasers, Foundation for Research and Technology Hellas (IESL, FORTH), P. O. Box 1527, Heraklion, 71110 Crete, Greece [email protected]

Summary. Laser-induced ﬂuorescence of intrinsic ﬂuorophores of organic media found in paintings (casein, animal glue and egg proteins) provides a means of characterising general classes of media depending on the amino acid composition and presence of degradation cross-linkages. Wavelength dependence of spectra is investigated for non-destructive and non-invasive analyses of thin ﬁlms of protein-based binding media.

47.1 Introduction The characterisation of painting materials is crucial to their conservation. Paintings consist of multiple layers of pigments and colourants bound in a medium on a support and are often coated with a varnish. Protein-based materials have traditionally been used as binding media and their identiﬁcation is critical for conservation. Analysis of proteins and other binding media typically involves taking samples so that characterisation using non-invasive methods is clearly advantageous. While non-invasive techniques are available, for example portable Fourier transform infrared spectroscopy (FTIR), results can be compromised by a lack of speciﬁcity, resolution and sensitivity. More reliable and speciﬁc information can be obtained from analysis of small samples. Sensitive techniques, such as high performance liquid chromatography (HPLC) and gas chromatography-mass spectroscopy (GC-MS), are required for precise identiﬁcation of binding media, but are destructive; however, sampling is complicated by stratigraphic variation and analysis by sample preparation [1]. Laser-induced ﬂuorescence (LIF) provides an alternative non-destructive means of investigating and analysing proteins found in works of art. Fluorescence spectroscopy is a widely applied technique in the analysis of proteins, which exhibit intrinsic ﬂuorescence due to the presence of aromatic

400

A. Nevin et al.

Table 47.1. Excitation and emission maxima of ﬂuorescence spectra, extinction coeﬃcient (εmax ), lifetime of emission and quantum yield of autoﬂuorescent amino acids [7]

Phenylalanine Tryptophan Tyrosine

λexcitation (nm)

λemission (nm)

εmax (10−3 )

Lifetime (ns)

Quantum yield

257 280 274

282 348 303

0.2 5.6 1.4

6.4 3.6 2.6

0.04 0.20 0.14

amino acids tryptophan, tyrosine and phenylalanine [2]; optical properties of the free amino acids diﬀer signiﬁcantly (Table 47.1). Diﬀerences in extinction coeﬃcient and quantum yield explain the dominating ﬂuorescence observed with tryptophan and tyrosine in the presence of phenylalanine. Especially sensitive to changes in the structure and environment of the proteins, emissions of amino acids are often used to follow dynamic reactions of known proteins in solution rather than for the identiﬁcation of speciﬁc proteins, as can be desirable in the analysis of binding media found in works of art. In addition, a large range of ﬂuorescent degradation products from aged proteins have been identiﬁed, some of which are speciﬁc to particular proteins. These include both photo-oxidation products from the combination and modiﬁcations of amino acids and cross-linkage reaction between amino acids with free glucose and other sugars (the Maillard reaction) [3]. Glycation related cross-linkage products have been documented in systems of proteins that are the sources of binding media and include collagen [4], milk casein [5] and eggs [6], although only collagen presents such cross-linkages prior to ageing. Initial work using spectroﬂuorimetry on artists’ materials has focused on the ﬂuorescence of ﬁlms of oil-based media [8], as well as on pigments bound in various media [9,10]. Photoluminescence of solutions of proteins from selected binding media has been ascribed to intrinsic ﬂuorescence of proteins to amino acids components [11]. Investigation of LIF of protein-based binding media have been limited to few instances [12], and more often LIF is used as a monitoring tool, as in the case of laser cleaning of parchment [13] or egg tempera [14]. Additionally, ﬂuorescence lifetime imaging (FLIM) has exploited emission lifetime for the analysis of binding media mixed with strongly ﬂuorescent inorganic pigments [10, 15]. However, interpretation of ﬂuorescence spectra is a common problem, not only due to characteristically broad signals, but also due to the lack of a systematic investigation of LIF for the analysis of proteins. The aim of this research is the analysis of protein samples of binding media found in works of art using both LIF and complimentary spectroﬂuorimetry. Binding media were selected based on recipes from original treatises and artists’ accounts [16, 17] and included milk, eggs (white and yolk) and animal glues, predominantly composed of casein and lactalbumin, ovalbumin and lysozyme and collagen, respectively.

This page intentionally blank

47 Laser-Induced Fluorescence Analysis of Protein-Based Binding Media

401

47.2 Experimental Methods Rabbit skin glue (Kremer Pigments) was swollen and dissolved in a warm water bath in nanopure water to give a solution of 5% /w. Parchment glue was prepared as described by Cennini [16]: parchment samples were cut into ﬁne strips and these were washed in hot water, and then left to swell in water for 24 h (original concentration 5% by weight); the mixture was then heated for 6 h at 90 ◦ C in a covered beaker, which yielded a transparent colourless solution of partially dissolved collagen (glue). Egg white was beaten to form stiﬀ peaks and left for 24 h; foam was skimmed and removed and a solution of 50%/w clear egg white in nanopure water was prepared. Egg yolk was extracted from egg by piercing the yolk and allowing the liquid yolk to drip from the encasing ﬁlm; the yolk was diluted in nanopure water to give a 50%/w emulsion. Casein was swollen in water for 24 h to give a 1.5%/w solution; dilute ammonia solution was added until the swollen gel dissolved and excess ammonia was left to evaporate. In addition, solutions of 0.1% N -acetyl tryptophan and 0.1% tyrosine ester monohydrate were prepared. Solutions of 0.1%/w of each binding medium were analyzed using a Jobin-Yvon/Horiba Fluoromax-P Spectroﬂuorimeter, integration time 0.2 s, intervals of 0.5 nm and excitation/emission slits between 1 and 5 nm. Films of proteins were cast on polished fused-silica disks to give ﬁlm thickness of ca. 15 µm (Pethometer S5P proﬁlometer). Films were examined 3 months after their preparation. Films of proteins were analyzed at 30◦ from the sample-axis using excitation/emission slits of 1 nm. Fluorescence of ﬁlms of proteins was recorded using 248 nm excitation using a KrF Excimer laser (TUI Laser AG BraggStar 200), 10 ns pulse duration, maximum of 500 pulses and ﬂuency of 5 mJcm−2 per pulse, spot size 1 × 1 mm. Spectra were collected using a ﬁbre-optic placed at 60◦ to the sample axis coupled to an Ocean Optics HD4000 Spectrophotometer with approximately 0.13 nm resolution. A 290 nm cut-oﬀ ﬁlter was placed in front of the ﬁbre optic. Further spectra were recorded using Q-switched, 355 nm, Nd:YAG 3rd harmonic excitation (Spectron Laser Systems), 10 ns pulse duration, a maximum of 300 pulses and ﬂuency of 5 mJcm−2 per pulse.

47.3 Results and Discussion Analysis of solutions of binding media gives a range of ﬂuorescence emission and excitation spectra, dependent on the protein, the excitation wavelength employed and the source of radiation. Spectra can be rationalized on the basis of the amino acid composition of the media (Table 47.2) as well as the presence of degradation and deterioration products present in the prepared glue. Most signiﬁcant for ﬂuorescence are diﬀerences between collagens that is free of tryptophan and the other tryptophan-containing proteins. In addition, various degradation products associated with the preparation and natural ageing

402

A. Nevin et al.

Table 47.2. Autoﬂuorescent amino acid composition (percent of amino acids) of materials used as binding media: egg [18], casein [19] and collagen [20] Egg white

Egg yolk

Casein

Collagen

5.3 1.2 3.9

4.3 1.2 4.5

3.2 1.4 2.7

2.3 n/a 1.0

Phenylalanine Tryptophan Tyrosine

0.1 % casein (aq) 0.1 % n-acetyl tryptohpan (EtOH)

(a)

Intensity (a.u.)

300

350

400

(b)

450

500

0.1 % parchment glue (aq) 0.1 % ethyl tyrosineester (aq)

300

350

400

450

500

(c) 0.1 % parchment glue (aq)

250

300

350

400

450

500

Wavelength (nm)

Fig. 47.1. Emission spectra of binding media and solutions of amino acid derivatives; (a) λex 280 nm, λmax emission casein (354), n-acetyl tryptophan (348); (b) λex 260 nm, λmax emission parchment glue (301), ethyl tyrosine ester; (c) synchronous scan with ﬂuorescence emission set-oﬀ 50 nm with respect to excitation wavelength, emission max occurs at excitations of 260 nm and secondary peaks at 335 and 375 nm

of protein-based binding media are responsible for diﬀerent and signiﬁcant contributions to ﬂuorescence spectra of proteinaceous binding media. A similarity between the emission spectra of solutions of binding media and amino acids tryptophan and tyrosine can be observed (Fig. 47.1). A red-shift in the emission spectra of casein as compared with that of n-acetyl tryptophan is ascribed to the local environment of tryptophan in the protein that is particularly susceptible to local pH and polarity [21]; contributions from tyrosine (Table 47.2) in casein are not clearly apparent due to the reduced emission of tyrosine at excitation of 280 nm as compared to its excitation maximum (260 nm) and the higher absorption coeﬃcient and quantum yield of tryptophan (Table 47.1). The principal emission from solutions of parchment glue (excited at 260 nm) matches with that found in ethyl tyrosine ester; contributions from additional ﬂuorophores are present, especially noticeable

47 Laser-Induced Fluorescence Analysis of Protein-Based Binding Media

403

at longer wavelengths. Through the use of synchronous ﬂuorescence scanning, where both excitation and emission are varied simultaneously, three separate ﬂuorophores can be observed in the solution of parchment glue, with excitation (emission) maxima (nm) at 260 (310), 335 (385) and 375 (415). The ﬁrst emission can be ascribed to tyrosine; other emissions are likely due to age-related cross-linkages in the parchment (including pentosidine) [22] and di-tyrosine [23]. Rabbit skin glue, in contrast to parchment glue, is extracted at higher temperatures and often involves base hydrolysis of collagen from animal skin; loss of tyrosine has been implicated in photo-oxidation [24] and the accumulation of degradation products and cross-linkages as a result of ageing and processing are also autoﬂuorescent [3, 22, 25]. The excitation and emission spectra of rabbit skin glue indicate two diﬀerent ﬂuorophores in the excitation spectrum (Fig. 47.2a). Peaks are found at 300 and 385 nm in the excitation spectrum, similarly reﬂected in the emission spectrum where two peaks are found at ca. 285 and 300 nm. The broad peak centred at 380 nm can be ascribed to the presence of cross-linkages pentosidine and pyridonoline, which have been isolated in gelatine and aged collagen following electrophoresis [22]. Egg white, similar in amino acid composition to casein, has contributions from both tyrosine and tryptophan (Fig. 47.2b), where diﬀerent excitation wavelengths can be used to excite either tryptophan (emission maximum at 335 nm) or derivatives of tyrosine (emission at 415 nm). Signiﬁcant diﬀerences between fresh egg white and egg yolk solutions and ﬁlms were not observed; however, with cross-linking reactions associated with the polymerization and drying of egg yolk, a tenfold increase in intensity of ﬂuorescence with maximum at ca. 440 nm dominates the spectrum, as found in studies of artiﬁcially aged tempera ﬁlms [14]. LIF provides a portable means and rapid analysis of ﬁlms and reveals similar diﬀerences between diﬀerent binding media as found using a traditional excitation source within the ﬂuorimeter. Most importantly, examination with 248 nm can be used to classify binding media into those exhibiting λemission = 390 nm

(a)

λexcitation = 280 nm

(b)

λexcitation = 248 nm

λexcitation = 355 nm

Intensity (a.u.)

30000 20000 10000 0 250

300

350

400

450

300

400

500

Fig. 47.2. (a) Excitation and emission spectra from 0.1%/w aqueous solution of rabbit skin glue, (b) wavelength dependence of ﬂuorescence emission from ﬁlms of egg white

404

A. Nevin et al.

Table 47.3. Fluorescence emission (nm) of media as a function of laser excitation wavelength Excitation (nm)

248

Collagen glues

355 305 340 415

Egg & casein

2000

355 pentosidine tyrosine (parchment) tryptophan di-tyrosine

440 pyrodinoline and 415 di-tyrosine 415 di-tyrosine

600

(a) parchment glue

(b) 400 200

0 300

400

500

600

0 400

500

600

400

500

600

400

500

600

800

Intensity

2000

rabbit skin glue

600 400 200 0

0 300

400

500

5000

600

400

casein egg white

0

200

0 300

400

500

600

Wavelength (nm)

Fig. 47.3. Emission spectra from diﬀerent protein ﬁlms recorded with laser excitation of (a) 248 nm and (b) 355 nm

a strong peak centred at around 340 nm, which is ascribed to tryptophan (casein and egg proteins) and those which contain peaks centred at around 380 nm attributed to pentosidine (collagen-derived proteins), (summarised in Table 47.3). Seen in Fig. 47.3, a weaker contribution from tyrosine (emission at 305 nm) is found in rabbit skin glue, as compared to that from parchment glue, using 248 nm excitation. Emission from tyrosine-related emission at around 415 nm is more pronounced than in solution, due to competition from quenching in the latter. Tryptophan is most apparent in ﬁlms of casein, as compared to egg white, where tyrosine and tryptophan emissions are of similar intensities, in contrast to spectroﬂuorimeter excitation at 280 nm (Fig. 47.2) where only contributions from tryptophan are observed. With 355 nm excitation (Fig. 47.3b), intense ﬂuorescence from rabbit skin glue is assigned to combinations of lysine-adducts (maximum at 400 nm) [4], di-tyrosine and other photo-oxidation products; egg white and casein have a maximum peak at 415 nm, ascribed to di-tyrosine [25].

47 Laser-Induced Fluorescence Analysis of Protein-Based Binding Media

405

47.4 Conclusions Initial results demonstrate that LIF can be used to distinguish between proteins in art works, and that emission have a strong wavelength dependency. Particularly signiﬁcant for laser-based analysis is that LIF allows determination of organic media, not aﬀorded by other laser-based applications (LIBS, XRD, XRF.) Within the context of organic analysis in conservation, LIF is advantageous as it is a non-invasive, non-destructive, simple and rapid technique that can be used to distinguish between diﬀerent proteins on the basis of their general class; although LIF spectra cannot be used to distinguish between casein and egg white, and likely not between various collagen-based glues, results suggest that it may be possible to discriminate among the two groups. Further work should focus on the inﬂuence of the ageing of protein media, pigment-media interactions and investigations of ﬂuorescence lifetime. Acknowledgements Research was supported by: European Union 6th Framework Programme Marie Curie Early Stage Training Fellowship as part of the ATHENA Project (MEST-CT-2004-504067).

References 1. M. P. Colombini and F. Mondugno, in Journal of Separation Science, Vol. 27, 147, 2004. 2. A. Ladhokin, in Encyclopedia of Analytical Chemistry, Edited by R. Meyers, 2000. 3. M. J. Davies, S. Fu, H. Wang, and R. T. Dean, in Free Radical Biology & Medicine, Vol. 27, 11/12, 1999. 4. D. G. Dyer, J. A. Blackledge, S. R. Thorpe, and J. W. Baynes, in Journal of Biological Chemistry Vol. 266, 18, 1991. 5. F. Morales, C. Romero, and S. Jimenéz-Pérez, in Food Chemistry, Vol. 57, 3, 1996. 6. U. Tagami, et al., in Journal of Mass Spectrometry, Vol. 35, 2, 2000. 7. C. Cantor and P. Schimmel,Biophysical Chemistry, Part 1: The Conformation of Biological Macromecules, W.H. Freeman, New York, 1980. 8. E. R. de la Rie, inStudies in Conservation, Vol. 27, 1, 1982. 9. T. Miyoshi, in Japanese Journal of Applied Physics, Vol. 27, 4, 1988. 10. D. Anglos, et al., in Applied Spectroscopy, 50, 1996. 11. M. Tseitlina and N. Kozhukh, in ICOM Committee for Conservation 9th triennial meeting: Dresden, German Democratic Republic, 26–31 August 1990, Edited by K. Grimstad, ICOM Committee for Conservation, London. 12. L. J. Larson, K. S. K. Shin, and J. I. Zink, in Journal of the American Institute for Conservation, Vol. 30, 1, 1991. 13. W. Kautek, et al., in Journal of Cultural Heritage, Vol. 4, 2003. 14. M. Castillejo et al., in Journal of Cultural Heritage, Vol. 4, 2003.

406 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

A. Nevin et al. D. Comelli, et al., in Applied Optics, Vol. 43, 10, 2004. Cennino Cennini, The craftsman’s handbook, Dover, London, 1960. Theophilus,De diversis artibus, Dover, London, 1961. Supplement-Agriculture Handbook No. 8, Human Nutrition Information Service, USDA, 1989. P. Fox, Editor,Advanced Dairy Chemistry, Vol.1. Proteins, Chapman and Hall, New York, 1992. J. S. Mills and R. White, The organic chemistry of museum objects, Butterworth, London, 1987. E. A. Burstein, in Intrinsic Protein Luminescence, VINTI, Moscow, 1977. Z. Deyl, I. Miksìk, and J. Zicha, in Journal of Chromatography A, Vol. 836, 1, 1999. W. Liu, et al. in Polymer, Vol. 41, 7589, 2000. M. P. Colombini, et al., in Studies in Conservation, Vol. 43, 1, 1998. C. Cole and J. Roberts, in Centenary Conference of the International Union of Leather Technologists and Chemists, London, 1997.

48 Applications of a Compact Portable Raman Spectrometer for the Field Analysis of Pigments in Works of Art ∗

S. Bruni and V. Guglielmi Dipartimento di Chimica Inorganica, Metallorganica e Analitica, Università degli Studi di Milano, Via G. Venezian 21, 20133 Milano, Italia ∗ [email protected] Summary. The importance of Raman micro-spectroscopy for the identiﬁcation of pigments in works of art is well established. In recent times, portable Raman spectrometers have been introduced which allows users to perform ﬁeld analysis directly where the artefacts are placed (churches, museums, archaeological sites, etc.). The present work reports results obtained by a remarkably compact instrument, in particular, on frescoes and illuminated parchments.

48.1 Introduction In the last two decades, Raman micro-spectroscopy has gained increasing importance in the study of artistic materials [1], particularly but not only pigments in painted layers, thanks to two main advantages in comparison with other analytical techniques. First of all, it is virtually non-destructive since samples do not require any form of manipulation. Moreover, it is suitable for the characterization of a wide range of materials, especially inorganic media, as the entire vibrational spectrum is accessible through this technique. Until a few years ago, Raman micro-spectroscopy was a very useful means mainly for the investigation of artistic objects that could be easily moved to a laboratory. The development of solid-state lasers, ﬁbre-optics technology and CCD detectors allowed users in recent times to assemble portable Raman spectrometers and therefore to perform in situ measurements directly in the ﬁeld. In the present work, applications of a particularly small-sized system [2] are described for the identiﬁcation of pigments in ﬁfteenth century frescoes inside a northern Italian Romanic church (S. Alessandro, Lasnigo, Como) and in illuminated parchments inside the “Archivio Storico dell’Ospedale Maggiore” in Milano.

408

S. Bruni and V. Guglielmi

48.2 Instrumentation Our equipment allows the use of two diﬀerent laser sources, a frequencydoubled Nd:YAG laser emitting at 532 nm and a diode laser emitting at 785 nm, to obtain the best compromise amongst diﬀerent variables (i) the light-absorption properties of the examined pigments, i.e. their colour; (ii) the intensity of the Raman signal that increases according to the fourth power of the exciting frequency; and (iii) the ﬂuorescence background that can be signiﬁcant in the analysis of painted artefacts due to the pigments themselves or, more frequently, to the binders with which they were mixed, and that can be reduced using a near-IR excitation wavelength. The spectrograph (Lot Oriel MS125TM ) is particularly compact (15 × 19 × 8 cm3 ) and can accommodate alternatively two diﬀerent diﬀraction gratings, namely, a 1,200 or 1,800 lines/mm having, respectively, 750 or 500 nm as blaze wavelength. A CCD detector is used for the signal collection. The laser radiation is brought to the sample and the Raman scattered radiation is collected through one or the other of two small micro-probes (Jasco RMP-100, 15 × 5 × 5 cm3 ). Each microprobe is equipped with a notch ﬁlter and an interference ﬁlter (for one of the two excitation wavelengths), a microscope ×50 objective and a CCD camera for sample observation. Fibre optics carry the laser radiation to the microprobe and the scattered radiation from the microprobe to the spectrograph. A notebook computer is used for the acquisition of the spectrum and for the visualization of the sampled area. The instrumental setup is completed by a halogen lamp with a ﬁbre directing the light onto the region from which the Raman spectrum is to be collected, and by a tripod on which a xyz stage is mounted to support the microprobe allowing its micrometric positioning. Remote controls are also available to adjust ﬁnely the position of the probe and to switch between visualization and measurement modes. All optical ﬁbres and cables for remote control are about 3 m long for the easiest accommodation of the probe. Figure 48.1 shows the micro-probe in place during measurements on a fresco.

48.3 Results and Discussion 48.3.1 Application to Frescoes Raman micro-spectroscopy is particularly suitable for the analysis of pigments in frescoes, as problems related to ﬂuorescence background are not as relevant as in paintings where organic binders were used (e.g. tempera or oil paintings). On the other hand, wall paintings are typical examples of those works of art that require a portable instrument for a truly non-destructive investigation. In a previous work [2], we reported results elucidating the artist’s palette obtained on a fresco in a northern Italian church. The study described below was performed on ﬁfteenth-century frescoes in a Romanic church in Lombardy,

48 Applications of a Compact Portable Raman Spectrometer

409

probe

xyz stage

Fig. 48.1. The Raman micro-probe on the tripod equipped with a xyz stage during ﬁeld measurements

namely S. Alessandro in Lasnigo (Como, Italy) and is a part of a wider Swiss– Italian project for the recovery of the church (the Swiss partner of the project is the University of Applied Sciences of Southern Switzerland, Department Environment, Construction and Design, whilst the Italian partner is the parish of S. Alessandro itself). Our study aimed in particular to the analysis of blue and green pigments in the frescoes since some of the blue painted areas seem to have turned green as an eﬀect of a degradation phenomenon. Figure 48.2 (left side) shows a detail of the fresco representing the Cruciﬁxion situated on the wall of the apse. The tunic worn by the saint is white (point 1 in the ﬁgure) with drapery evidenced in light blue (point 2), whilst the internal side of the cloak is light green (point 3). Figure 48.3 shows the spectra obtained from points 1 and 2 in the fresco of the apse. The white colour has, as expected, the composition of calcite (spectrum a, bands at 1,087 and 282 cm−1 ) whilst the blue colour was obtained using azurite (spectrum b, bands at 1,577, 1,424, 1,095, 940, 836, 766, 545 and 402 cm−1 ) [3, 4]. In spectrum b, bands at about 1,630, 1,490 and 1, 464 cm−1 were also observed and assigned to calcium oxalate monohydrate (whewellite) [5, 6]. Point 3 gave a spectrum where the only detectable bands were located at about 1,084 and 1, 007 cm−1 . Both bands have been reported by Bell et al. for green earth [3], but a third band of comparable intensity should be observed for that pigment around 670 cm−1 whilst it is lacking in our spectrum. As the frequencies of those bands correspond also to the strongest signals, respectively, of calcite and gypsum, we cannot interpret the spectrum unambiguously. It should be noted however that all spectra were characterized by a ﬂuorescence background that was corrected in the ﬁgures

410

S. Bruni and V. Guglielmi

•3 •1 •2

Fig. 48.2. (Left). Detail of the fresco on the wall of the apse in the church of S. Alessandro in Lasnigo. (Right). Fresco on the right side of the triumphal arch in the same church. The box gives an enlarged view of the deterioration of the blue painting in the cloak of the Holy Virgin: light grey areas in the photograph correspond to a light green colour, whilst dark grey areas correspond to a bright blue colour

a

b

1600

1200 Raman Shift

800

400

(cm-1)

Fig. 48.3. Micro-Raman spectra of: (a) point 1 (white tunic of the saint) and (b) point 2 (blue drapery of the tunic) of the fresco in the apse of S. Alessandro in Lasnigo. Legend: ﬁlled-circle azurite, ﬁlled-diamond whewellite. λ0 = 532 nm

48 Applications of a Compact Portable Raman Spectrometer

411

for an easier visualization of the bands, but that could hide the signals of those compounds that are weaker Raman scatters. On the right side of Fig. 48.2, another fresco is shown, situated on the triumphal arch in the same church and representing the Enthroned Virgin with Child and Saints. The peculiarity of this painting, of less artistic value if compared with the Cruciﬁxion of the apse, is associated with the degradation observed for the blue-painted areas that appear to be mixed with large green zones. Either the alteration of the original blue colour or possibly the intentional use of diﬀerent pigments, even in diﬀerent times, was suggested as a possible explanation. The Raman spectra obtained, respectively, on blue and green areas (Fig. 48.4) seem to point to the ﬁrst hypothesis. Bands observed for the blue colour (spectrum a) correspond to those expected for azurite [3,4], but signals due to whewellite and gypsum (1, 007 cm−1 ) are also evident. Spectrum b, resulting from a green area, shows again the bands of calcium oxalate and sulphate, but other signals of medium intensity are observed at about 504, 411 and 320 cm−1 , whilst weaker ones appear at ca. 970, 920 and 884 cm−1 . With the exception of the 320 cm−1 band, the other signals could ﬁt with the spectral pattern reported in the literature for the basic copper chloride clinoatacamite [3, 7]. Our frequencies correspond more closely with those reported in [7] but, in this respect, it should be noted that a certain confusion exists in the literature as regards the Raman spectra of basic copper chlorides, also

a

b

1600

1200

800

400

Raman Shift (cm-1) Fig. 48.4. Micro-Raman spectra of: (a) blue area and (b) green area in the cloak of the Virgin in the fresco on the triumphal arch in S. Alessandro in Lasnigo. Legend: ﬁlled-circle azurite, asterisk clinoatacamite, ﬁlled-diamond whewellite, ﬁlled-square gypsum. λ0 = 532 nm

412

S. Bruni and V. Guglielmi

due to the fact that atacamite and clinoatacamite are polymorphs and can be found in mixture. Moreover, if it is to be supposed that in the painted layer under examination basic copper chlorides derive from degradation of azurite, it is possible that diﬀerent compounds of similar chemical composition are present at the same time, also representing diﬀerent stages of the degradation sequence that should end indeed with clinoatacamite [7]. Finally, it is probable that in our spectrum, the band around 504 cm−1 also contains a component associated with whewellite [5, 6] and the signal at 411 cm−1 a residue of the most intense band of azurite. 48.3.2 Illuminated Parchments in the Archives Illuminated manuscripts are certainly amongst the artefacts most studied by Raman micro-spectroscopy, as demonstrated especially by the numerous papers by Clark and co-workers (see, i.e. [8]). In principle, they can be moved to the laboratory in many cases but, in practice, this could imply risks of damage and insurance expenses. Therefore the portable instrumentation may again be really useful to perform the analyses directly where the manuscripts are usually kept. In this way we could perform measurements on illuminated parchments (Fig. 48.5) in the “Archivio Storico dell’Ospedale Maggiore” of Milano (Italy) simply placing them on an easel and keeping them ﬂat by means of magnetic holders. As exempliﬁed in Fig. 48.6, rather satisfying results were obtained using both excitation wavelengths at our disposal, notwithstanding the limitation associated with the notch ﬁlters mounted in the microprobe for 785 nm that, at present, do not allow us to acquire signals below about 300 cm−1 . In a parchment dating to 1,610 and consisting in a diploma of Carlo Emanuele I, the use of cinnabar (band at 346 cm−1 ) for red and minium (bands at 548

•1 •2

•3

Fig. 48.5. Details of a 1,610 parchment (a diploma of Carlo Emanuele I, Archivio Storico dell’Ospedale Maggiore di Milano). Numbers indicate the points at which Raman spectra were recorded

48 Applications of a Compact Portable Raman Spectrometer

413

a c

b

650

550 450 350 Raman Shift (cm-1)

1700 1400 1300 1100 Raman Shift (cm-1)

Fig. 48.6. Micro-Raman spectra of: (a) a red detail (point 1), (b) an orange detail (point 3) and (c) a black detail (point 2) in the parchment of Fig. 48.5. For spectra a and b, λ0 = 785 nm; for spectrum c, λ0 = 532 nm

and 392 cm−1 ) for orange was recognized by near-IR excitation, whilst carbon black was identiﬁed by visible excitation.

48.4 Future Perspectives The above examples show just some of the potential of the portable microRaman spectrometer. Other artefacts have been analysed for the identiﬁcation of pigments by means of this instrument and amongst them ceramic fragments should be cited [2]. The measurements on sherds or very small grains were performed in laboratory, but, as the portable instrumentation was used in any case, it will be obviously possible to extend them also to intact or large-sized ceramic objects in a museum. Acknowledgements The authors wish to thank: Dr Giovanni Cavallo, SUPSI, as co-ordinator of the diagnostic work in the project for the recovery of S. Alessandro in Lasnigo; Dr Paolo Galimberti, director of the “Archivio Storico dell’Ospedale Maggiore di Milano”, for his hospitality and enthusiast support to our work and for the idea to use magnets to keep parchments perfectly ﬂat; Dr Luca Consolandi, Centro di Riﬂettograﬁa Infrarossa e Diagnostica dei Beni Culturali, Università degli Studi di Milano, who took wonderful photographs of the parchments; Mr. Gianmaria Manvati, who took photos of our instrument whilst measuring in the church of S. Alessandro.

414

S. Bruni and V. Guglielmi

References 1. F. Cariati and S. Bruni, in Modern Analytical Methods in Art and Archaeology, Edited by E. Ciliberto and G. Spoto, Wiley & Sons, New York 2000, 255, and references therein. 2. S. Bruni, F. Cariati, and V. Guglielmi, in Raman Spectroscopy in Archaeology and Art History, Edited by H. G. M. Edwards and J. M. Chalmers, Royal Society of Chemistry, Letchworth, 2005, 142–151. 3. I. M. Bell, R. J. H. Clark, and P. J. Gibbs, in Spectrochimica Acta Part A, Vol. 53, 2159, 1997. 4. M. Bouchard and D. C. Smith, in Spectrochimica Acta Part A, Vol. 59, 2247, 2003. 5. H. G. M. Edwards, D. W. Farwell, R. Jenkins, and M. R. D. Seaward, in Journal of Raman Spectroscopy, Vol. 23, 185, 1992. 6. S. S. Potgieter-Vermaak, R. H. M. Godoi, R. Van Grieken, J. H. Potgieter, M. Oujja, and M. Castillejo, in Spectrochimica Acta Part A, Vol. 61, 2460, 2005. 7. R. L. Frost, in Spectrochimica Acta Part A, Vol. 59, 1195, 2003. 8. K. L. Brown and R. J. H. Clark, in Journal of Raman Spectroscopy, Vol. 35, 217, 2004.

49 Classiﬁcation of Patinas Found on Surfaces of Historical Buildings by Means of Laser-Induced Breakdown Spectroscopy ∗

C. Vázquez-Calvo1 , A. Giakoumaki2 , D. Anglos2 , M. Álvarez de Buergo1 , and R. Fort1 1

∗ 2

Instituto de Geología Económica (CSIC-UCM), RTPHC, Facultad de Ciencias Geológicas, Universidad Complutense de Madrid, C/José Antonio Nováis 2, 28040 Madrid, Spain [email protected] Institute of Electronic Structure and Laser, Foundation for Research and Technology Hellas (IESL-FORTH), P.O. Box 1527, 71110 Heraklion, Crete, Greece

Summary. This study deals with the analysis and characterisation of historic patinas on Spanish buildings from the sixteenth and seventeenth century by means of laser induced breakdown spectroscopy (LIBS). Indicative results from the analysis of these samples are shown. LIBS can be proven as an appropriate technique to characterise patinas and assist in a rapid and facile classiﬁcation. The prospect of developing portable LIBS equipment makes the technique attractive for carrying out analyses on site.

49.1 Introduction The term patina is referred to here as a yellow-to-brown coloured thin layer that covers external stone surfaces (Fig. 49.1). It accounts for previous treatments [1–3], which may have aimed to protect and/or colour the stone. However, contributions due to biological activity cannot be ruled out [1, 4–6]. Some descriptions on the way patinas were prepared can be found in the literature, but still the information is not enough to cover suﬃciently the diﬀerent types of materials and recipes employed. The composition and ingredients of these ﬁlms were supposed to be linked to local practice and fashion at diﬀerent times, and, in many cases, this information has been lost forever. The major components of such patinas that cover limestone and marble in a large number of monuments are usually calcium oxalates, calcite, gypsum, iron oxide pigments and various silicates [7–9]. Calcium phosphate can also

416

C. Vázquez-Calvo et al.

Fig. 49.1. (a) Church of Santa Ana, Peñaranda de Duero, Burgos, Spain (sixteenth century), the image shows a detail of the façade and one of the patina fragments that were sampled; (b) Thin section of a patina (on the right-hand side of the image) is found over a limestone substrate

be found as a constituent of the patinas composition [10, 11]. The ingredients and substances used in the making up of these patinas undergo chemical and mineralogical transformations in the course of time. As a result it is important to use appropriate mineralogical and chemical analysis to infer the original composition of the patinas. Intentional removal of these patinas during cleaning interventions has resulted in greater damage in the period of only a few years than that caused by natural agents over centuries. Moreover, the loss of the patina layers eliminates part of the construction history of the building and makes it harder to get the right and proper “reading” of the monument. Thus, it is necessary to determine the present and original composition of the patinas, to identify its origin and to determine if this covering is protecting the stone surface or, on the contrary, is damaging it. All these aspects should be considered in advance before making the decision of the patina removal, and in order to guarantee its further conservation. Although there are several speciﬁc works about this subject including two international symposia [12, 13], there are not many studies realised with nondestructive techniques. A ﬁrst approach to the characterisation of this kind of material by LIBS has been reported by Maravelaki-Kalaitzaki et al. [14]. In the present study, a series of samples from Spanish historical monuments from the sixteenth and seventeenth century were analyzed by LIBS in an eﬀort to assess the potential of the technique, to identify the patina components and to provide analytical information leading to the discrimination between diﬀerent types of patinas. These samples were analysed in previous studies [15] using other techniques including optical microscopy, scanning electron microscopy (SEM) and portable X-ray ﬂuorescence (XRF).

49 Classiﬁcation of Patinas Found on Historical Buildings

417

49.2 Experimental Methods A standard laboratory arrangement was used for all the LIBS measurements. A Q-switched Nd:YAG laser (Spectron Laser Systems) operating at the fundamental wavelength (1,064 nm) was used. The duration of the laser pulses was 10 ns. The laser beam was focused by means of a plano-convex lens (f = +75 mm) on the sample surface. The light emitted from the plume was collected through a quartz optical ﬁber (diameter: 0.6 mm, length: 1.5 m), located about 15 mm from the sample surface at a 45◦ angle with respect to the normal, and introduced into a 0.32 m imaging spectrograph (TRIAX320, ISA) equipped with three interchangeable diﬀraction gratings (600, 1,800 and 2,400 grooves mm−1 ) oﬀering diﬀerent spectral resolution (0.4, 0.15 and 0.1 nm). The spectrum was recorded on an intensiﬁed charge coupled device (ICCD) detector (DH520, Andor Technology), which was gated by means of a pulse generator (DG-535, Stanford Research) in order to discriminate the atomic emission from the continuum background present at early times. Typical delay and gate time values of 300 and 800 ns, respectively, were used in these measurements. The values of the laser ﬂuence employed in the experiments were in the range of 10–20 J cm−2 with the irradiated spot diameter around 150 µm. For the semi-quantitative analysis of minor elemental constituents, such as phosphorus (P), strontium (Sr) and lead (Pb), present in the patinas, the preparation of samples of known composition was necessary for producing appropriate calibration curves. These samples were prepared in pellet form, having calcium carbonate as the matrix and calcium phosphate, strontium sulphate and lead carbonate as minor components with weight concentrations in the range of 0.5–15 %.

49.3 Results and Discussion The results of LIBS analysis of several patinas show that calcium (Ca) is a major element in all samples while magnesium (Mg), silicon (Si) and aluminium (Al) are present in most cases reﬂecting the major components of the patina (Fig. 49.2a). These data are in agreement with analyses carried out using an XRF instrument and point to the presence of minerals such as calcite, dolomite and feldspars commonly found in building materials. In several cases, emission due to iron was observed in the LIBS spectra (Fig. 49.2a). The presence of iron substances can be ascribed to pigments that were used to give colour to the patinas. In several patinas, low or medium quantities of elements such as strontium or lead were identiﬁed by LIBS analysis (Fig. 49.2b). In Fig. 49.3, two spectra taken from the analysis of the patina and the stone substrate are presented. While both spectra show similarities, it is noted that only the patina spectrum shows distinct emission from phosphorus.

418

C. Vázquez-Calvo et al.

a)

b) Ca

Ca Ca

Mg Ca

Ca Si Fe Mg

Al

Ca

Ca PbSr 260

280

300

320

400

340

Wavelength(nm)

Ca 420

440

460

Wavelength(nm)

Fig. 49.2. LIBS spectra from patina samples from (a) Church of Santa Ana, Peñaranda de Duero, Burgos, Spain; (b) Collegiate church of San Pedro and San Pablo, Lerma, Burgos, Spain

a)

b) Si

Si

Si

Si Si C Si

C

P

Si

P

Fe Fe

Al Al

245

250

Wavelength (nm)

255

Si

Si Fe

240

Si

Fe Fe

240

Al

245

250

Fe

255

Wavelength (nm)

Fig. 49.3. LIBS analysis of patina (a) and stone substrate (b) of a sample from the Monastery of Santa María de La Vid, Burgos (Spain)

The presence or absence of certain elements may be used to discriminate diﬀerent types of patinas and preparations. For instance, phosphorous might indicate the use of dairy products such as milk in the original manufacturing of the coating, while strontium could be correlated to the presence of gypsum. A rough estimate of the concentration of these elements in some of the patinas examined was carried out on the basis of calibration curves obtained by analyzing, under our experimental conditions, several model samples. Such an indicative curve for phosphorus is shown in Fig. 49.4.

49 Classiﬁcation of Patinas Found on Historical Buildings

419

8000 7000

Intensity

6000 5000 4000 3000 2000 1000 0 0

2

4

6

8

10

Calcium phosphate (%)

Fig. 49.4. Calibration curve for phosphorus obtained from calcium phosphate– calcium carbonate model samples. The emission line of P at 253.56 nm was used

49.4 Conclusions According to the preliminary studies presented in this paper, LIBS appears as a quite promising technique to characterise patinas and assist to a rapid and facile classiﬁcation, which is based on marker elements present in low concentrations. Such tool will support historians, architects and conservators with valuable information about the origin and formation of the patinas. The prospect of developing portable LIBS equipment makes the technique even more attractive for carrying out analyses on site. Acknowledgements C. Vázquez-Calvo has been supported through the Marie Curie Host Fellowship project at IESL-FORTH (fellowship No. HPMT-GH-00-00177-21), under the “European Commission Human Potential programme". A. Giakoumaki receives a fellowship from IKY (Hellenic Foundation of Scholarships). This study has been ﬁnanced by project BIA2003-0473 (Spanish Ministry of Education and Science) and by a “Ramon y Cajal” contract (MAdB) (Ministry of Education and Science).

References 1. C. Cipriani and L. Franchi, in Bollettino del Servicio Geologico, Vol. 79, 555, 1958. 2. M. Franzini, C. Gratziu, and E. Wicks, in Rendiconti della società italiana di mineralogia e petrologia, Vol. 39(1), 59, 1984.

420

C. Vázquez-Calvo et al.

3. V. Fassina, in The Science of The Total Environment, Vol. 167(1-3), 185, 1995. 4. V. Alunno Rossetti and M. Laurenzi Tabasso, in Problemi di conservazione, Bologna, 375, 1973. 5. M. Del Monte and C. Sabbioni, in Environmental Science and Technology, Vol. 17(9), 518, 1983. 6. M. Del Monte, C. Sabbioni, and G. Zappia, in The Science of The Total Environment, Vol. 67(1), 17, 1987. 7. G. Alessandrini, R. Bugini, and R. Peruzzi, in La Certosa di Pavia: passato e presente nella facciata della chiesa, Edited by CNR Roma, 291, 1988. 8. F. Guidobaldi, M. Laurenzi Tabasso, and C. Meucci, in Proccedings of the Fourth International Congress on Deterioration and Preservation of Stone Objects, Lousville (KY), USA, Edited by U. Giovanni, 175, 1982. 9. E. Previde Massara and G. Perego, in Proceedings 9th International Congress on Deterioration and Conservation of Stone, Venice, Edited by V. Fassina 425, 2000. 10. K. Polikreti and Y. Maniatis, in The Science of The Total Environment, Vol. 308(1–3), 111, 2003. 11. K. Kouzeli, C. Lazari, A. Economopoulos, and C. Pavelis, in II International Symposium: The oxalate ﬁlms in the conservation of works of art. Milan, Edited by M. Realini and L.Toniolo, 81, 1996. 12. Proceedings of the Symposium: The oxalate ﬁlms: origin and signiﬁcance in the conservation of works of art, Milan. Edited by Centro C.N.R. “Gino Bozza” & Politecnico di Milano, 1989. 13. Proceedings of the II International Symposium: The oxalate ﬁlms in the conservation of works of art, Milan, Edited by M. Realini and L. Toniolo, 1996. 14. P. Maravelaki-Kalaitzaki, D. Anglos, V. Kilikoglou, and V. Zaﬁropulos, in Spectrochimica Acta, Vol. B56, 887, 2001. 15. C. Vázquez-Calvo. Estudio de pátinas históricas en materiales pétreos del patrimonio arquitectónico de Lerma, 2005. Unpublished work.

50 Laser-Induced Breakdown Spectroscopy of Cinematographic Film M. Oujja1 , C. Abrusci2 , S. Gaspard1 , E. Rebollar1 , A. del Amo3 , ∗ F. Catalina4 , and M. Castillejo1 1

∗ 2

3 4

Institute of Physical Chemistry Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain [email protected] Department of Microbiology III, Faculty of Biology, Universidad Complutense de Madrid, José Antonio Novais 2, 28040 Madrid, Spain Filmoteca Española, Magdalena 10, 28012 Madrid, Spain Institute of Polymer Science and Technology, CSIC, Juan de la Cierva 3, 28006 Madrid, Spain

Summary. Laser-induced breakdown spectroscopy (LIBS) was used to characterize the composition of black-and-white, silver-gelatine photographic ﬁlms. LIB spectra of samples and reference gelatine (of various gel strengths, Bloom values 225 and 75 and crosslinking degrees) were acquired in vacuum by excitation at 266 nm. The elemental composition of the gelatine used in the upper protective layer and in the underlying emulsion is revealed by the stratigraphic analysis carried out by delivering successive pulses on the same spot of the sample. Silver (Ag) lines from the light-sensitive silver halide salts are accompanied by iron, lead and chrome lines. Fe and Pb are constituents of ﬁlm developers and Cr is included in the hardening agent. The results demonstrate the analytical capacity of LIBS for study and classiﬁcation of diﬀerent gelatine types and the sensitivity of the technique to minor changes in gelatine composition. In addition LIBS analysis allows extracting important information on the chemicals used as developers and hardeners of archival cinematographic ﬁlms.

50.1 Introduction Since the advent of motion pictures in the nineteenth century until the coming of professional video in the 1970s, the use of emulsion-coated, transparent plastic-base ﬁlm has been the main carrier for production, dissemination and preservation of motion picture contents. During almost 12 decades, an enormous amount of moving images (feature ﬁlms, documentaries, commercials, artworks, etc) have been produced on ﬁlm, and became part of ﬁlm collections stored in archives, representing a unique body of cultural, historical and artistic documents, which are a fundamental part of memory and cultural heritage.

422

M. Oujja et al.

As cinematographic ﬁlms constitute an important part of cultural heritage legacy, its characterization is essential to design adequate conservation conditions [1]. A cinematographic ﬁlm is a complex multicomponent system consisting of a layer of photographic emulsion overcoating a polymeric support and covered by a transparent superﬁcial crosslinked gelatine layer applied for protective purposes. The photographic emulsion is composed basically of gelatine (60–70 %) where the light-sensitive silver halide salts, the ﬁlm developer’s constituents and the hardening agents are in suspension. Gelatine is a natural protein derived from collagen, which consists of amino acids chained together by a peptide bond. Gelatine is a useful biological material with both complex structure and special functions. In general gelatines may be classiﬁed according to purpose as edible, pharmaceutical or photographic. The application of gelatine revolutionized photography [2, 3] due to its favourable physical and chemical properties which have caused gelatine to remain the dominating vehicle for silver halide photographic materials for more than a century. In the present work, study is limited to gelatine used in photographic manufacturing of type-B with high quality grade and Bloom value B225. Although laser techniques have been applied on many types of organic materials (paper, paintings, tissue, leather, etc) of cultural heritage for cleaning and analysis of elemental constituents and its stratigraphy [4–7], the applications of laser techniques to photographic or cinematographic heritage are scarce [8]. The objective of this research was to explore the applicability and potential of Laser-induced breakdown spectroscopy (LIBS) for elemental analysis of composition of diﬀerent gelatine types and of the emulsion of the photographic ﬁlms. Stratigraphic analysis, performed by applying successive laser pulses on the same spot, allowed detection of the ﬁlm layers.

50.2 Experimental Methods This work was carried out on two types of samples to assess the capability of LIBS to characterize the gelatine material. We prepared a ﬁrst set of samples consisting of commercial type-B gelatine ﬁlms with two gel strengths, Bloom values 75 and 225 (e.g. B75 and B225) and two crosslinking degrees (1 and 24 h). The second type of samples were obtained by cutting 2 × 2 cm2 pieces from cinematographic black-and-white, silver halide ﬁlms. For the ﬁrst type of samples, self-standing ﬁlms of 40 µm were prepared using aqueous solutions (6.67 wt.%) of gelatine (Aldrich) followed by solvent evaporation at 37◦ C. Type-B gelatines with two diﬀerent gel strength values were used: B75 and B225. In a previous work [9], the corresponding relative viscosities in water solution (ηrel ) and molecular weight average numbers (Mn ) of both gelatine grades were determined and resulted in ηrel = 3.64 cP and Mn 62.3 kDa for B75 and ηrel = 5.50 cP and Mn 77.3 kDa for B225. For the

50 Laser-Induced Breakdown Spectroscopy

423

preparation of crosslinked gelatine, B225 ﬁlms were immersed in a 4% aqueous solution of formaldehyde during 1 and 24 h to reach two diﬀerent crosslinking degrees and then the ﬁlms were washed repeatedly in distilled water and dried at 37◦ C until constant weight. In order to characterize the crosslinking degree of the ﬁlms, discs of 8 mm diameter were weighted (W0 ) before being immersed in water at 37◦ C under stirring. At predetermined time intervals, the wet ﬁlms were weighted (Wt ) immediately after removal of water from the disc surface by blotting brieﬂy between two pieces of ﬁlter paper. Crosslinked gelatine ﬁlms reached a steady-state of swelling within 1 h. The crosslinked ﬁlm characterization was carried out by measuring their weight swelling ratio q [10], deﬁned as q = Wt /W0 . Crosslinking for 1 and 24 h resulted in a swelling ratio of 7.53 and 4.04, respectively. The longer treatment of gelatine with formaldehyde gives a higher degree of crosslinking and subsequently a lower swelling ratio was reached. The black-and-white cinematographic ﬁlm samples consist of a layer of photographic emulsion (of 15–20 µm) overcoating a polymeric support of triacetate of cellulose of about 100 µm thickness and a transparent superﬁcial crosslinked gelatine layer of about 1 µm thickness that covers the emulsion for protection purposes. Analysis was performed separately in black-and-white zones of the ﬁlms. The LIBS system used for the experiments has been described previously [11] and a short description is given here. Laser irradiation of the gelatine and photographic ﬁlms was carried out with the fourth harmonic of a Q-switched Nd:YAG laser (Quantel Brilliant B, pulses of 6 ns, repetition rate of 10 Hz, 266 nm). The samples, located inside a vacuum chamber evacuated down to 3 Pa, were irradiated using a f = 10 cm lens allowing to achieve ﬂuences up to 17 J cm−2 . The ﬂuence of irradiation was determined by measuring the pulse energy with an ED-200 Gentec joulemeter and the area of the beam print on an unplasticized PVC ﬁlm. The shot-to-shot laser energy ﬂuctuation was less than 10%. The emitting plasma was collected at right angles to the laser beam by means of two quartz lenses (f1 = 4 cm, f2 = 10 cm) and a periscope inserted between the lenses. The purpose of the periscope is to rotate the plasma image by 90◦ so that the expansion direction lies along the orientation of the entrance slit of the spectrograph allowing a more eﬃcient signal collection [12]. The emitting plasma was analysed in the 250–900 nm wavelength range with a 0.30 m spectrograph (TMc300 Bentham, 1,200 grooves mm−1 , 500 nm blaze) coupled to a time-gated intensiﬁed charged-coupled device (2151 Andor Technologies), with temporal and spectral resolutions down to 2 ns and 0.2 nm respectively. The temporal gate was operated at zero time delay and temporal observation window of 3 µs. For the results presented here, a 300 nm longpass ﬁlter was installed in front of the spectrograph to prevent second order diﬀraction peaks appearing in the spectra.

424

M. Oujja et al.

50.3 Results and Discussion

OI NI

OI NI

C2

C2

C2

HI

OH (2nd order)

Na I

Ca II CN

Ca I CH HI

OH

Ca I CN Ca II

CI

OH

400

500

OI

HI

600

OI

Na I

C2

C2

C2

CH

HI

CN Ca II Ca I Ca I CN

OH

CI

300

OH (2nd order)

b) B225, crosslinked 1 h OH

Intensity / a.u.

a) B75

HI

The UV absorption spectra of the gelatine ﬁlms used in this work have been reported previously [13]. Absorption below 240 nm can be attributed to the nonaromatic amino acids of the gelatine structure, whereas at longer wavelengths, in the 250–350 nm range, absorption is due to aromatic amino acids. The ﬁlm absorption spectra of the four gelatine types were measured using a UV/VIS spectrometer (Perkin Elmer, Lambda 35) to determine the eﬀective optical absorption coeﬃcients α at the irradiation wavelength (266 nm) used in this work. The derived optical penetration depths 1/α obtained for the gelatines B75, B225, B225 crosslinked 1 h and B225 crosslinked 24 h are 30, 55, 36 and 66 µm, respectively. Characteristic LIB spectra of the gelatine ﬁlms are shown in Fig. 50.1. The spectra of the four gelatine types show the presence of atomic C I (247.85 nm), C II (387.64, 426.72 nm), Ca I (336.19, 422.67 nm), Ca II (373.69, 393.36, 396.84 nm), Na I (588.99, 589.59 nm), O I (777.19, 777.41, 844.63 nm), N I (742.36, 746.83, 859.40, 862.92 nm) and H I (434.04, 486.13, 656.28 nm) and molecular emissions OH (A2 Σ+ − X2 Π, ∆ν = +1, 288 nm, ∆ν = 0, 306 nm), CN (B2 Σ+ − X2 Σ+ , ∆ν = +1, 359 nm, ∆ν = 0, 388 nm, ∆ν = −1, 421 nm), C2 (d3 Πg − a3 Πu , ∆ν = +1, 473 nm, ∆ν = 0, 516 nm, ∆ν = −1, 563 nm). These species derive from the amino acid structure of the gelatine and from impurities left in the elaboration process or present in the cattle bones used as raw material in the manufacturing process.

700

800

900

Wavelength / nm

Fig. 50.1. LIB spectra obtained in vacuum upon ablation at 266 nm of ﬁlms of gelatine: (a) B75; (b) B225 crosslinked 1 h. Vertical units are common in both spectra

50 Laser-Induced Breakdown Spectroscopy

425

Table 50.1. Intensity of atomic and molecular spectral emissions (relative to OH emission) obtained in LIB spectra of gelatine ﬁlms recorded upon excitation at 266 nm Emitting species (wavelength in nm)

B75

B225

B225 crosslinked 1h

B225 crosslinked 24 h

C I (247.85) C II (387.64) OH (306) Ca II (393.36) CN (388) C2 (473) H I (486.26) Na I (588.99) H I (656.28) O I (777.19) N I (746.83)

0.75 1.15 1.00 4.60 0.60 0.50 1.70 4.06 5.3 1.20 0.05

0.45 0.70 1.00 1.60 0.30 0.45 1.40 1.75 3.10 0.60 0.05

0.75 1.10 1.00 0.35 0.50 0.40 0.90 0.10 5.80 0.70 0.03

0.80 0.90 1.00 0.75 0.45 0.55 1.20 0.50 1.30 1.20 0.10

The ablation plumes of crosslinked gelatines are less luminous than those of the ﬁlms not subjected to crosslinking. Crosslinking results in higher mechanical strength and rigidity of the material, which is revealed in the LIB spectra by giving rise to less amount of material ejected and therefore to lower emission line intensities. Spectra of the diﬀerent gelatine types also diﬀer in the relative intensity of the bands as listed in Table 50.1. Gelatine grades B75 and B225 yield intense Ca and Na lines when compared with the corresponding line intensities in crosslinked gelatines. These diﬀerences are due to the higher content of alkaline residues left in the alkali pretreatment of type-B gelatine. In fact, B225 gelatine, of higher purity than B75, should contain less alkali residues; accordingly the intensities of Ca and Na are lower in the former. In the crosslinked gelatines, Ca and Na lines are notably less intense, this eﬀect being related with the fact that, in the crosslinking process described before, the swelling hydrogel is washed out several times with distilled water and these ionic metals are more eﬃciently extracted from the material. Also it is important to point out that the intensity of the Ca and Na lines in the crosslinked gelatines follows the reciprocal order of their swelling ratio (q). Gelatine B225 with high degree of crosslinking is used in photographic applications due to its high purity and less content of alkali residues. The density of the intermolecular bridges has an inﬂuence on the extraction of impurities from the materials. LIB spectra of black-and-white silver-gelatine photographic ﬁlm are shown in Fig. 50.2. The spectrum of the black emulsion (Fig. 50.2a) corresponding to the ﬁrst pulse of the laser, shows the presence of the same atomic and molecular emissions obtained in the spectra of the crosslinked B225 gelatine (Fig. 50.1b). In addition to these, emission lines corresponding to silver (Ag I: 328.06, 338.28, 520.90, 546.55 nm), Fe I (404.58, 406.35, 438.35, 440.47 nm),

C2

Na I

Ag I

Ag I

C2

HI Fe I

CN Ca II Fe I Pb I Ca I Cr I H I C II Fe I

Ag I

OH Ag I

OH Pb I

CI

a) Black emulsion

HI

M. Oujja et al. OH (2nd order)

426

1st 2nd 4th

Ag I

HI

HI Fe I C2

C2

CN Ca II Pb I Fe I Ca I Cr I

1st

Cr I

Ag I Ag I

OH

Pb I

OH

b) White emulsion

Na I

30th

CI

Intensity / a.u.

6th

2nd 4th 6th 30th

240

300

360

420

480

540

600

660

Wavelength / nm

Fig. 50.2. LIB spectra obtained in vacuum upon ablation at 266 nm of black-andwhite silver halide photographic ﬁlm: (a) black area; (b) white area

Pb I (283.30, 405.78 nm) and Cr I (391.91, 425.43 nm) are observed. Silver is present in the light-sensitive halide salts used for preparing the emulsion, while Fe and Pb are constituents of ﬁlm developers and the Cr is included in the composition of hardening agents used to improve the mechanical properties of the emulsion. The LIB spectra of the white emulsion (Fig. 50.2b) are similar to the ones corresponding to the black emulsion, except that emissions corresponding to silver, which are relatively absent (except in the ﬁrst pulse), indicating the eﬃcient extraction of silver halides during the developing process of the photographic ﬁlm. The weak intensities observed for the emissions corresponding to the Ca and Na lines indicate that the gelatine used in the photographic ﬁlms corresponds to crosslinked B225 type. The stratigraphy of the black emulsion (Fig. 50.2a) was studied by delivering successive pulses on the same spot of the sample to yield the composition of diﬀerent layers of the ﬁlm. Emissions from the ﬁrst to the ﬁfth pulse give rise to lines corresponding to the composition of the crosslinked gelatine (used

50 Laser-Induced Breakdown Spectroscopy

427

in the upper protective layer), silver and constituents of the ﬁlm developers and the hardening agents present in the emulsion. The emissions obtained from the sixth until the 30th pulse, although weak, correspond to the material of the polymeric support, triacetate of cellulose. Thirty-two laser pulses are required at the ﬂuence of 17 J cm−2 to drill a hole in the photographic ﬁlm, yielding an etching rate of 4 µm per pulse. The stratigraphy of the white emulsion (Fig. 50.2b) is similar to that of the black emulsion, except that the intensity of Ag lines is very much reduced in the ﬁrst laser pulse. In fact this result is expected, as in the white emulsion the silver halides have been extracted in the developing process.

50.4 Conclusions The results prove that LIBS is a valuable technique for the classiﬁcation of diﬀerent gelatine types and qualities (B75, B225 and crosslinked B225) being sensitive to minor changes in their elemental composition. The composition of the emulsion of the photographic ﬁlm could also be characterized by LIBS enabling extraction of important information about chemical products used as developers and hardeners agents in the developing process of archival ﬁlms. The stratigraphic analysis performed by means of this low invasive analytical method on the photographic ﬁlm allowed us to distinguish between the diﬀerent constituent layers (protective, emulsion and the polymeric support) while the damage only occurs on a micrometric spot of the surface. These results allow us to obtain relevant information to improve the storage conditions of the cinematographic archival ﬁlms. Acknowledgements Work funded by Project MCYT BQU2003-08531-C02-01, Spain. MO and ER thank CSIC I3P program for a contract and a fellowship respectively and SG the EU sixth FP for a Marie Curie EST Fellowship (MESTCT-2004-513915). CA acknowledges Filmoteca Española and Fotoﬁlm Madrid (CSIC-UCM-FEFOTOFILM) for her fellowship.

References 1. A. T. Ram, in Pol. Degrad. Stab. Vol. 29, 3, 1990. 2. T. H. James (Ed.), The Theory of the Photographic Process, 4th ed., Ed. James, T. H., Macmillan Publishing Co., Inc. (New York), 1977. 3. J. M. Burnhan and C. G. Gray, Physical properties of photographic materials, in: P. Z. Adelstein (Ed.), SPSE: Handbook of Photographic Science and Engineering, Wiley, New York, 1973. 4. J. Kolar, M. Strlic, S. Pentzien, and W. Kautek, in Appl. Phys. A , Vol. 71, 87, 2000.

428

M. Oujja et al.

5. K. Melessanaki, A. Mastrogiannidou, and D. Anglos, in LACONA V Proceedings, Osnabrueck, Germany, Sept. 15–18, 2003, Springer Proceedings in Physics, Band 100, K. Dickmann, C. Fotakis, J.F. Asmus (Eds.). 6. J. Kolar, M. Strlic, D. Müller-Hess, A. Gruber, K. Troschke, S. Pentzien, and W. Kautek, J. Cult. Herit. Vol. 1, 221, 2000. 7. M. Castillejo, M. Martín, M. Oujja, D. Silva, R. Torres, A. Manousaki, V. Zaﬁropulos, O. F. van den Brink, R. M. A. Heeren, R. Teule, A. Silva, and H. Gouveia, in Anal. Chem., Vol. 74, 4662, 2002. 8. V. V. Golovlev, M.J. Gresalﬁ, J. C. Miller, G. Romer, and P. Messier, in J. Cult. Herit. Vol. 1, 139, 2000. 9. C. Abrusci, A. Martín-González, A. Del Amo, T. Corrales, and F. Catalina, in Polym. Deg. Stabil., Vol. 86, 283, 2004. 10. N. A. Peppas, in Biomaterials, Vol. 11, 635, 1990. 11. M. Castillejo, M. Martin, M. Oujja, E. Rebollar, C. Domingo, J. V. García Ramos, and S. Sánchez-Cortés, in J. Cult. Herit. Vol. 4, 243, 2003. 12. J. Siegel, G. Epurescu, A. Perea, F. J. Gordillo-Vázquez, J. Gonzalo, and C. N. Afonso, in Opt. Lett. Vol. 29, 19, 2228, 2004. 13. C. Abrusci, A. Martín-González, A. Del Amo, F. Catalina, P. Bosch, and T. Corrales, in J. of Photochem. and Photobiol. A: Chemistry, Vol. 163, 537, 2004.

51 Online Monitoring of the Laser Cleaning of Marbles by LIBS Sulphur Detection ∗

V. Lazic1 , F. Colao1 , R. Fantoni1 , V. Spizzichino1 , and E. Teppo2 1 ∗ 2

ENEA, FIS-LAS, Frascati (RM), Italy, [email protected] Alphacon, Inc., 956 Summer Ridge Road, Bozeman, MT 59715, USA

Summary. In the present work, the feasibility of sulphur detection was demonstrated during laser ablation of the gypsum layers on the polluted surface of aged marbles. The laser used for cleaning and generation of the LIBS signal was a Q-Switched Nd:YAG emitting dual pulses at 1,064 nm. Diﬀerent sulphur emission lines were identiﬁed through comparative LIBS measurements on gypsum encrustation and K2 SO4 and MgSO4 reference materials, but only few, not weak, sulphur lines resulted suitable for monitoring the cleaning of marble, due to the overlap of most features with emissions from bulk marble constituents. Sulphur vertical distribution in the encrustation was then monitored by LIBS during the cleaning. Results of LIBS sulphur stratigraphy were comparable with those obtained by SEM–EDX measurements. In the present study it was possible to establish by LIBS the boundary between bulk marble and its encrustation containing gypsum and, consequently, to determine the optimal point for the interruption of the laser cleaning at the end of the heavily sulphated crust.

51.1 Introduction LIBS technique is already widely used for analyses of the materials and objects related to the Cultural Heritage [1]. It can be applied in situ, without sample preparation and also allows for subsurface micro-stratigraphy, which is particularly important for characterization of painted surfaces and of outer object layers due to the aging of the artifacts [2–4]. Marble is a widely used material in the European cultural heritage and its preservation and restoration is the subject of numerous studies. Aged marble is covered with an encrustation whose composition depends strongly on the environmental conditions to which the artifact was exposed during its history, i.e. on the environmental pollution [5]. Both in traditional marble cleaning (by abrasion, chemical agents, steam) and in laser cleaning, it is of importance to characterize the composition of the encrustation and its stratigraphy before the intervention, in order to establish the optimal restoration procedure. In the case of laser cleaning, it is preferable to have the possibility to

430

V. Lazic et al.

perform online monitoring in order to preserve the protective marble layers located near the original surface before reaching the bulk, thus avoiding over-cleaning, which could damage and/or induce fast aging of the artifact surface [6]. LIBS has been successfully applied to monitor removal of the marble encrustation containing some characteristic elements (e.g. metals) [7, 8] but the control of the ablation of the gypsum layer from marble has not been reported up to now. The main diﬃculty in gypsum recognition by LIBS is related to the diﬃculties to detect sulphur emission in air due to its high reactivity with oxygen. In the present work a feasibility of sulphur detection by LIBS in open air atmosphere has been demonstrated. LIBS was also applied for measurements of sulphur depth distribution inside marble encrustation, and to establish the point where the clean marble surface was reached by laser cleaning. The double pulse technique was applied which was previously demonstrated to be suitable for trace detection of diﬀerent elements in marble samples [9].

51.2 Experimental LIBS measurements were performed employing a Q-Switched (QS) Nd:YAG laser (Quanta System, model Handy Nd:YAG) operated at 1,064 nm, with a repetition rate of 10 Hz and maximum energy of 320 mJ. The laser QS trigger has been externally controlled in order to extract two laser pulses during the same lamp ﬂashing, each one with duration of about 8 ns [10]. After the LIBS signal optimization, the laser pulse energies were set to E1 = 82 mJ and E2 = 180 mJ, corresponding to the time delay between the lamp trigger and the QS successive triggers: t1 = 145 µs, t2 = 55 µs. The laser beam was focused upon the sample surface by a 150 mm focal length planoconvex lens. The light emitted by the plasma in the UV – visible region (230–1,000 nm) was collected by wide-angle receiver optics, at approximate angle of 30◦ with respect to the laser beam axis. The collected signal is carried by an optical-ﬁbre bundle, 0.1 mm wide at the exit, to the entrance slit of a 550 mm monochromator (Jobin-Yvon, model TRIAX 550), equipped with 1,200 grooves mm−1 grating (2 mm slit). At the exit plane, a gated ICCD (Andor Instaspec IV) was used to record the LIB spectra. The marble sample examined was original from Veneto region (Italy) and covered with thick (up to 2 mm), rough, black encrustation. This sample was ﬁrst characterized by SEM–EDX, which revealed a high sulphur content (about 10%) in the encrustation. This element might be expected to be in the form of gypsum, thus reducing strongly the mass concentration of calcium in the crust. The composition of both marble crust and bulk, as measured by SEM–EDX, is reported in Table 51.1.

51 Online Monitoring of the Laser Cleaning of Marbles

431

Table 51.1. Concentrations (c) of the elements in the marble measured by SEM–EDX with its conﬁdence level Element C Ca O Fe Mg Al Si S K

BULK

CRUST

c (%)

Conf. +/−

c (%)

Conf. +/−

22.74 25.97 50.18 0.07 0.19 0.01 0.16 0.21 0.47

2.14 1.28 4.84 0.05 0.08 0.04 0.05 0.06 0.05

21.59 9.98 55.69 0.37 − 0.59 1.71 9.78 0.56

1.97 0.43 5.00 0.05 – 0.06 0.10 0.45 0.03

In order to establish the capability of sulphur detection by LIBS, ﬁrst we acquired full spectra on sulphate reference samples, namely, MgSO4 and K2 SO4 powders previously pressed into pellets.

51.3 Results By applying double pulse laser excitation, we detected diﬀerent ionic sulphur emission lines in the range 380–405 nm (Fig. 51.1), and a weak atomic emission around 920 nm. In the near UV range dominated by sulphur ionic lines, some diﬀerences in K2 SO4 spectra with respect to MgSO4 and crust might be attributed to the very low ionization energy of potassium, which gives rise to the signiﬁcant diﬀerences in plasma parameters. Other sulphur atomic lines, given in NIST database, could not be detected or assigned, either because they were too weak or too close to emission lines from oxygen. Some diﬀerent ionic lines also coincided with emissions from marble constituents. However, in the near UV region there are two S+ lines that could be used to distinguish clean marble from the gypsum layer (Fig. 51.2), one around 403 nm and another around 399 nm. These lines are close to Ca ionic lines (around 390 nm) and Ca atomic emission (422.7 nm and band around 410 nm), so the ratio S+ /Ca could be used as a control parameter during the laser cleaning. In LIBS spectra acquisitions carried on with diﬀerent gate delays from the second laser pulse, we observed that ionic sulphur emission is stronger at the plasma beginning, but on K2 SO4 and MgSO4 samples it could be still detected after 5 µs. In order to monitor other elements present in the encrustation, we increased the ICCD gate width from the optimum value for sulphur (500 ns) to 1,000 ns. In this case for the chosen monochromator position, corresponding to the spectral range 398–424 nm, sulphur emission from the degraded marble layer is still detectable but also Fe (404.6 nm and 406.3 nm) and Sr (ionic emission at 407.8 nm and 421.6 nm) were detected (Fig. 51.3). Regarding iron,

432

V. Lazic et al.

Intensity (a.u.)

12000

K2SO4 crust

S+

10000

S+

S+

8000

MgSO4

K

6000 4000 2000 0 380

385

390

395

400

405

410

415

420

Wavelength (nm) Fig. 51.1. Identiﬁcation of sulphur emission by LIBS – comparative measurements on K2 SO4 , MgSO4 and marble encrustation. Acquisition parameters: delay from second pulse 300 ns, gate width 500 ns

Intensity (a.u.)

18000

Clean marble-crust

Ca+ crust marble

12000 S+

S+

6000

0 380

385

390

395

400

405

410

415

420

Wavelength (nm)

Fig. 51.2. LIBS spectrum from clean marble and its encrustation acquisition parameters: delay from second pulse 300 ns, gate width 500 ns

Marble Crust

1.0 Normalised intensity

Ca Sr+

0.8 0.6

Ca+

Sr+

0.4 Fe Ca

0.2 S+

S+

0.0 400

405

410 415 Wavelength (nm)

420

Fig. 51.3. LIBS spectrum from clean marble and its encrustation. Acquisition parameters: delay from second pulse 300 ns, gate width 1,000 ns

51 Online Monitoring of the Laser Cleaning of Marbles

433

which is a possible catalyst for sulphation [11], we were particularly interested to check whether this element has the same depth distribution as sulphur in the crust. For the depth proﬁle analysis of sample crusts, the spectra were registered at each laser shot. Sulphur (line at 399 nm and 403 nm) and iron (line at 404.6 nm) distributions were measured after the line peak intensity normalization to Ca emission (410 nm). Although sulphur emission around 403 nm coincides with the manganese emission band, the same depth distributions were obtained as when considering sulphur emission at 399 nm, so we might conclude that the Mn presence in the examined sample might be neglected in LIBS measurements, both on the crust and in the bulk. On the basis of the depth distribution of sulphur obtained, whose plot is given in Fig. 51.4 (single point local measurements), the crust could be considered removed with less than about 25 shots. In fact with the application of a larger number of laser shots the surface composition remained practically constant, thus indicating that the bulk material was already reached. However, the results presented in Fig. 51.4 are relevant only to the chosen point, because the crust thickness changes signiﬁcantly from one point to another. Although iron emission reaches a stable value after about the same number of applied shots as sulphur, correlation between sulphur and iron emission is very weak, namely 0.58 for the reported example. This means that iron might play a role in the surface sulphation, but iron distribution in the crust is not an appropriate indicator for sulphur (i.e. gypsum) distribution.

Normalized Peak ratio

1.0 S+/Ca Fe/Ca

0.8

0.6

0.4

0.2

0.0 0

10

20

30 Shot number

40

50

Fig. 51.4. Depth distribution of sulphur and iron inside the crust

434

V. Lazic et al.

51.4 Conclusions In the present work, it was demonstrated for the ﬁrst time that sulphur emission from gypsum encrustation can be detected by LIBS and in open air atmosphere. Emission lines from sulphur were identiﬁed through comparative LIBS measurements on marble crust containing gypsum, and pressed MgSO4 and K2 SO4 materials. The relative intensities of the identiﬁed lines are very diﬀerent from those reported in NIST database. Until now, LIBS monitoring of crust removal from stones was based on detection of other elements present in the encrustation, and gypsum distribution was deduced indirectly, through the presence of transition metals which may act as catalysts for the transformation from marble to gypsum. However, the vertical distribution of these metals do not follow exactly the sulphur content in the crust, as it has been demonstrated here for iron. Direct sulphur detection by LIBS opens the possibility to monitor in situ removal of gypsum layers and/or to rapidly obtain its distribution in the degraded stone surfaces. The latter is useful as well for restorers to establish correct restoration procedure and also if applying traditional cleaning methods. During laser cleaning of stones it would be also possibly to implement a reliable automatic control for gypsum layers. Further developments of the present LIBS analyses shall regard sulphur concentration measurements by LIBS, after generation of an appropriate calibration graph on calcium carbonate matrices doped with sulphate salts. The quantitative sulphur determination can still be a diﬃcult task, due to the few lines suitable to the purpose and the need to establish the validity of Saha equation when relying on ionic lines. However the eﬀort might be worthwhile whenever part of the original sulphated layers must be preserved after the cleaning. Acknowledgements The present work supported by the Italian MIUR PON-FESR Project TECSIS (Diagnostic Technologies and Intelligent Systems for Developing Archaeological Parks in Southern Italy).

References 1. D. Anglos, in Appl. Spectroscopy, Vol. 55 (6), 186, 2001. 2. F. Colao, R. Fantoni, V. Lazic, L. Caneve, A. Giardini, and V. Spizzichino, in J. Anal. Atom. Spectrom. Vol. 19, 502, 2004. 3. F. Colao, R. Fantoni, V. Lazic, and V. Spizzichino, in Spectrochim. Acta B, Vol. 57, 1219, 2002. 4. P. Maravelaki-Kalaitzaki, D. Anglos, V. Kilikoglou, and V. Zaﬁropulos, in Spectrochim. Acta B, Vol. 56, 887, 2001.

51 Online Monitoring of the Laser Cleaning of Marbles

435

5. G. Amoroso and V. Fascina, “Wet and dry deposition of air pollutants on stone and the formation of black scabs, in Stone Decay and Conservation” Elsevier Science, Amsterdam 1983. 6. P. Gaspar, C. Hubbard, D, McPhail, and A. Cumming, in J. of Cultural Heritage, Vol. 4, 294, 2003. 7. V. Lazic, R. Fantoni, F. Colao, A. Santagata, A. Morone, and V. Spizzichino, in J. Anal. Atom. Spectrom. Vol. 19, 429, 2004. 8. P. Maravelaki-Kalaitzaki, D. Anglos, V. Kilikoglou, and V. Zaﬁropulos, in Spectrochim Acta B, Vol. 56, 887, 2001. 9. P. Maravelaki-Kalaitzaki, V. Zaﬁropulos, and C. Fotakis, in Appl. Surf. Science, Vol. 148, 92, 1999. 10. V. Lazic, F. Colao, R. Fantoni, and V. Spizzichino, in Spectrochimica Acta B, Vol. 60, 1014, 2005. 11. P. Elfving, I. Panas, and O. Lidqvist, in Appl. Surf. Sci,. Vol. 78, 373, 1994.

52 Low Resolution LIBS for Online-Monitoring During Laser Cleaning Based on Correlation with Reference Spectra ∗

M. Lentjes1 , K. Dickmann1 , and J. Meijer2 1

∗ 2

Laser Centre FH Münster (LFM), University of Applied Sciences Münster, Stegerwaldstrasse 39, 48565 Steinfurt, Germany [email protected] Faculty of Engineering Technology, University of Twente, The Netherlands

Summary. Based on a commercial miniature spectrometer, we have built a LIBSsystem for online monitoring and controlling during laser cleaning of artworks. In contrast to common LIBS set-ups with ICCDs, our system oﬀers less sensitivity and λ-resolution. This system is unsuitable for high resolution elemental identiﬁcation but it is applicable for the detection of “spectral ﬁngerprints”. A powerful method of comparison online spectra during laser cleaning process with reference spectra is the analysis of correlation. Based on “Pearson’s correlation”, the linear relationship between entire data sets of both spectra are classiﬁed by the correlation coeﬃcient r. In practise of laser cleaning a permanent comparison of the spectrum from the ablated layer with a reference spectrum of the layer to be preserved is necessary. Thus, online monitoring of the cleaning process is enabled by permanent estimation of r. In case of closed-loop laser cleaning the ablation will continue until r exceeds a predeﬁned value and subsequently the ablation process is automatically stopped.

52.1 Introduction Laser-induced breakdown spectroscopy (LIBS) is nowadays a frequently used method in a lot of diﬀerent areas of application. The best-known method is the qualitative and also quantitative elemental identiﬁcation of samples with unknown composition [1]. The advantage of the LIBS-technique over classical methods is the possibility to analyse all types of materials in every state of aggregation without the need of sample preparation. In common a high intense pulsed laser beam is focussed on the sample and intensely heats a small volume of material, which results in a transient plasma above the irradiated area. The spectrum emitted by this plasma plume depends on the elemental composition of the ablated material. This can be analysed by a spectrograph.

438

M. Lentjes et al.

A totally diﬀerent approach is to consider a LIBS-spectrum as a spectral ﬁngerprint. Winefordner et al. applied this method in combination with linear correlation to identify plastic and solid materials [2]. Here the spectrum of the “unknown” sample is compared with a library of spectra, and identiﬁed by the weight of correlation. In cultural heritage, LIBS is mostly applied as a diagnostic tool to identify the elemental conﬁguration, separately or in combination with laser cleaning [3]. A practical example is the identiﬁcation of pigments by elemental conﬁguration [4]. As well LIBS can be used to control a laser cleaning process to avoid over-cleaning. This was already successfully applied by Zaﬁropulos et al. in diﬀerent applications [5]. Identiﬁcation of the layers is done by elements recognition. In common they used a conventional high resolution spectrometer with intensiﬁed CCD-array. These spectrometers have a high signal to noise ratio and sensitivity. Disadvantages are high investment costs, environmental sensitivity and complexity [6]. In this study the plasma radiation induced during laser cleaning is used to monitor or online control the process. The plasma radiation is recorded with a low resolution miniature ﬁbre optic spectrometer characterised by a large spectral bandwidth, low λ-resolution and sensitivity. In contrast to conventional LIBS-systems, this system is low cost and easy to handle. These properties make this system unsuitable for high resolution elemental identiﬁcation but it is applicable for the detection of spectral ﬁngerprints. It has turned out from former investigations concerning controlled laser cleaning of artworks that elemental identiﬁcation is not urgent necessary [7]. Each layer has its individual spectral ﬁngerprint and therewith it is distinguishable from other layers in a multi-layer arrangement. The application of correlation analyses enables the recognition of layers (ﬁngerprints), measured with a low resolution spectrometer on the basis of reference spectra. Through the use of optimal laser parameters during laser cleaning, the laser energy cannot be set arbitrary high in order to induce a bright plasma emission. Thus, in case of less plasma emission intensity the detection becomes critical or even impossible with a miniature spectrometer.

52.2 Experimental The miniature spectrometer was integrated into a KrF-Excimer laser cleaning system to test its capability as a feedback sensor for monitoring or controlling the process automatically. We applied a user-conﬁgured “Ocean Optics HR2000 UV-VIS” miniature ﬁbre optic spectrometer. For collecting the plasma radiation into the spectrometer, a 2 m long, 600 µm glass ﬁbre in combination with a collimator is applied. The groove density (300 grooves mm−1 ) and the entrance slit of 25 µm led to a spectral range of 200–1,100 nm with a resolution of 2 nm. The exposure time of the electronic “shutter” is ﬁxed on 2 ms.

This page intentionally blank

52 Low Resolution LIBS for Online-Monitoring

439

Fig. 52.1. Schematic visualisation of the set-up for monitoring and controlling laser cleaning via miniature spectrometer

The experimental set-up is built around a KrF-Excimer laser (Lambda Physics LPX 305i, 248 nm, 1–50 Hz, 20–40 ns) with a standard excimer laser based mask illumination and imaging optical set-up. Collimator and glass ﬁbre are placed behind the last dielectric mirror, which gives obvious advantages for the handling in praxis. For positioning and moving the samples, the set-up is equipped with an x–y stage (travelling range 80 cm and 7 µm resolution in both axes). Figure 52.1 schematically visualises the electronic connection of the individual parts. A computer equipped with a DAQ-card (NI PCI-6221) and in LabView written software controls the automated laser cleaning process. Time oﬀset between laser pulse and opening the spectrometer “shutter” is arranged by the delay generator programmed in LabView.

52.3 Recognition of Layers by Linear Correlation LIBS spectra of diﬀerent layers (materials) are unique ﬁngerprints, with diﬀerences in spectrum and intensity. There are various methods to compare these spectra with a reference spectrum and assign it to a layer. In this study we used Pearson’s linear correlation. The linear correlation coeﬃcient is a weight for the linearity between two sets of data and has a dimensionless index r in the range of −1 and 1. A value of r = 1 represents a correlation of 100f %, r = 0 means no linear correlation and r = −1 represents 100% negative correlation. The linear correlation coeﬃcient is given by n

r=

(xi − x ¯) (yi − y¯) , n n 2 2 (xi − x ¯) (yi − y¯) i=1

i=1

i=1

(52.1)

440

M. Lentjes et al.

Fig. 52.2. (left) Intensities of the LIBS spectrum measured at the ﬁrst laser ablation pulse vs. the intensities of the reference spectrum. There is a low linear correlation, r = 0.712. (right) Intensities of the LIBS spectrum measured at the eighth laser ablation pulse vs. the intensities of the reference spectrum. There is a high linear correlation, r = 0.993

where x ¯ is the mean of all xi -data and y¯ the mean of all yi -data. The intensity of the CCD-pixels measured during recording the reference spectrum forms the x data set while the y data set is given by the spectrum of the irradiated layer in process. A total data set consists of 2,048 points representing the number of pixels of the spectrometer CCD-array, so i = 1 · · · 2, 048. Any pixel forms a xi -value (reference spectrum) and a yi -value (actual spectrum) which can be plotted in an x–y graph (Fig. 52.2). The graph in Fig. 52.2 (left) shows less linearity between the single points than the data points of the x–y graph in Fig. 52.2 (right). Here the spectrum measured at the ﬁrst laser ablation pulse is correlated with a reference spectrum from another layer (diﬀerent material). Both spectra have a low linear correlation with each other, which is visible in the chaotic point distribution and results in a small linear correlation coeﬃcient, r = 0.712. Figure 52.2 (right) shows the linear association between a spectrum recorded during the eighth laser ablation pulse and a previous measured reference spectrum. The correlation coeﬃcient approximates the value of 1, which means a high linear correlation between the spectrum of the eighth laser ablation pulse and the reference spectrum. The single data points nearly all lay on the linear ﬁt line, which is calculated by the least square method. This results in the conclusion that these spectra are almost identical and in many cases belong to the same layer (material). An advantage of this correlation method is that intensity ﬂuctuations of the spectra (e.g. caused by pulse-to-pulse variations) do not highly aﬀect the coeﬃcient r.

52.4 Process Monitoring by Correlation Analysis First tests of monitoring laser ablation by the weight of the linear correlation coeﬃcient are applied on a deﬁned artiﬁcial multi-layer arrangement (homogenous parallel laminates, construction depicted in Fig. 52.3). First of

52 Low Resolution LIBS for Online-Monitoring

441

Fig. 52.3. Correlation coeﬃcients r related to the reference spectra “black,” “adhesive” and “paper” with increasing penetration depth. The arrows indicate the maximum of r, appearing at the transitions

all a reference spectrum had to be recorded from each layer. The correlation coeﬃcient is calculated after any laser ablation pulse between the recorded spectrum and the three reference spectra. Figure 52.3 shows the corresponding correlation coeﬃcients r calculated during the ablation vs. ablation pulse number. The bar at the top of the graph visualises in which layer the laser ablates material. When surveying the line that belongs to the correlation with the black reference spectrum, it is apparent that the correlation coeﬃcient approximates the value one while reaching the black layer. The highest value for r is reached after the same number of pulses (8) as the reference spectrum is recorded. Correlation with the other two spectra results also in the highest r at the transitions of the layers correlated with. This demonstrates that the correlation coeﬃcient can be applied to stop a laser cleaning or laser ablation process deﬁned at a given layer (level) in a homogenous layer arrangement. The suitability of this principle during laser cleaning of complex samples is tested on “real objects”: parchment, glass and iron. These experiments exhibit that it is possible to stop a laser ablation process during cleaning of an inhomogeneous polluted sample by correlation analysis. An important parameter for stopping the cleaning process on the original surface is the diﬀerence between the spectra of each layer. Tests with rusty iron have led to a negative result. The spectra of rusty and “clean” iron (recorded with the low resolution spectrometer) are too similar to distinguish them by correlation analysis.

442

M. Lentjes et al.

52.5 Closed-Loop Cleaning of Areas In a next step the correlation process is automated in a closed-loop system which is able to clean a predeﬁned area automatically. As already described all components in the laser cleaning set-up are directly or indirectly connected with one controller-PC. This closed-loop process is controlled by software written in LabView. The ﬁrst step, in the case of automatic cleaning, is recording a new LIBS reference spectrum or using a stored reference spectrum. The area to be cleaned, laser spot size and spatial overlap are programmed on the x–y positioning controller. After entering the relevant data, the controller is waiting idle for an input signal from the PC. Laser parameters for cleaning are directly entered in the controller of the laser. The ultimate control parameter of this closed-loop process is the comparing criterion. The correlation coeﬃcient will be estimated after each laser pulse by correlating the actual spectrum with the reference spectrum. If this value exceeds the value of the compare criterion, the ablation process is paused and the translation table moves the sample to a “new” polluted position. The cleaning process resumes until the correlation coeﬃcient again exceeds this criterion. This process repeats until the predeﬁned area is completely cleaned. The ﬁrst trials of automatic laser cleaning are applied on polluted parchment and colour pigment smalt with encrustation (binder: linseed oil, encrustation: artiﬁcial black gypsum). In both cases controlled cleaning of the predeﬁned area was possible. Figure 52.4 (left) shows the polluted pigment smalt with the controlled laser cleaned section. This rectangular area (4.5 × 6 mm2 ) is cleaned with a 1.5 × 1.5 mm2 square laser spot in 1 min time. The number of laser ablation pulses used per spot and the meander movement of the sample is visualised in Fig. 52.4 (right). The corresponding correlation coeﬃcients r calculated during the ablation of encrusted smalt vs. ablation

Fig. 52.4. (left) Encrusted smalt pigment with automatically laser cleaned area: 4.5×6 mm2 . (encrustation: artiﬁcial black gypsum, linseed oil binder, 248 nm, 42 mJ, 35 ns) (right) Meander movement (dashed line) during cleaning of smalt pigment. The numbers represent the amount of ablation pulses per spot

52 Low Resolution LIBS for Online-Monitoring

443

Fig. 52.5. Variation of correlation coeﬃcient r vs. ablation pulse number per spot calculated during automatic laser cleaning of encrusted smalt. When r exceeds the compare criterion (0.98), the process is paused and continues at the next spot position

pulse number per spot is visualised in Fig. 52.5, the horizontal dashed line indicates the compare criterion (0.98). After exceeding this value, the process is paused and continued at the next spot.

52.6 Conclusion Experiments on artworks and multi-layer arrangement have shown that a laser cleaning or ablation process can be controlled or monitored by linear correlation analysis via low resolution miniature spectrometer. This method has a high potential to stop laser ablation on a predeﬁned level. The process is automated by applying a controller-PC with LabView written software which connects all single modules into a closed-loop system. In combination with LIBS reference spectra, this system is able to automatically clean a predeﬁned area by using linear correlation. The use of linear correlation analysis gives the advantage of less inﬂuence from plasma intensity ﬂuctuations on the signal interpretation caused by laser pulse-to-pulse variations. The correlation coeﬃcient is not estimated by comparing single spectral lines but by the linear association of two complete spectra. In cases of almost identical LIBS spectra of two diﬀerent layers, identiﬁcation via linear correlation is diﬃcult or even impossible.

444

M. Lentjes et al.

References 1. F. Colao, R. Fantonni, V. Lazic, and A. Paolini, in Applied Physics A, Vol. 79, 143, 2004. 2. J. M. Anzona, I. B. Gomushkin, B. W. Smith, and J. D. Winefordner, in Polymer Engineering & Science, Vol. 40, 2423, 2000. 3. S. Acquaviva, M. L. De Giorgi, C. Marini, and R. Poso, in Applied Surface Sciences, Vol. 248, 218, 2005. 4. D. Anglos, S. Couris, and C. Fotakis, in Applied Spectroscopy, Vol. 51, 1025, 1997. 5. J. H. Scholten, J. M. Teule, V. Zaﬁropulos, and R. M. A. Heeren, in Journal of Cultural Heritage, Vol. 1, 215, 2000. 6. J. E. Carranza, E. Gibb, B. W. Smith, D. W. Hahn, and J. D. Winefordner, in Applied Optics, Vol. 42 No. 30, 6016, 2003. 7. M. Lentjes, D. Klomp, and K. Dickmann, in Lasers in the Conservation of Artworks, Edited by K. Dickmann, C. Fotakis, and J. F. Asmus, 427, 2005.

53 Pigment Identiﬁcation on a XIV/XV c. Wooden Cruciﬁx Using Raman and LIBS Techniques ∗

M. Sawczak1 , G. Śliwiński1 , A. Kaminska2 , M. Oujja3 , M. Castillejo3 , C. Domingo4 , and M. Klossowska5 1 ∗ 2

3 4 5

Polish Academy of Sciences – IFFM, Fiszera St. 14, 80–231 Gdansk, Poland [email protected] Agency for Integration of Conservation Activities – AICA, Dubois St. 55, 80–419 Gdansk, Poland Instituto de Química Física Rocasolano, CSIC, Serrano 119, 28006 Madrid, Spain Instituto de Estructura de la Materia, CSIC, Serrano 123, 28006 Madrid, Spain National Museum in Gdansk, Torunska St. 1, 80–822 Gdansk, Poland

Summary. The Raman and laser-induced breakdown spectroscopy (LIBS) techniques were applied for the pigment identiﬁcation in polychrome layers on a fourteenth/ﬁfteenth century wooden cruciﬁx. In the Raman spectra, characteristic bands associated with compounds of the pigment samples taken from diﬀerent areas of the object are observed. Groups of bands corresponding to the original white, red, and green pigments allow the identiﬁcation of chalk, vermilion, red lead, malachite, and azurite. From the presence of bands ascribed to Prussian blue (282, 538 cm−1 ) and chrome yellow (338, 360, 403 cm−1 ) known since eighteenth century, retouching of some statue parts can be concluded. The elemental composition is obtained from LIBS proﬁles recorded under excitation at 248 and 266 nm. The gold-leaf technique is identiﬁed and the presence of Cu, Pb, Cr, Fe, CN, C2 , and Ca agrees with the pigment composition applied for re-touching and observed in the Raman bands.

53.1 Introduction During the last decade, an increasing interest in applications of micro-Raman and laser-induced breakdown spectroscopy (LIBS) for analysis of artworks is observed. These sensitive methods provide comprehensive information on the compounds and elemental composition of the surface layer and, due to negligible sampling, nearly non-destructive analysis is possible. This is conﬁrmed by experimental data on historical pigments published recently [1–3]. In this work the complementary use of the above mentioned techniques is applied for the pigment identiﬁcation in the polychrome layers of a fourteenth/ﬁfteenth century wooden cruciﬁx, (Fig. 53.1) during restoration in the National Museum in Gdansk. The sculpture depicts the cruciﬁed Christ with arms extended in the shape of “V”. The body shaped along with legs and feet from one piece of wood is 165 cm high and is garmented in a red perizonium

446

M. Sawczak et al.

Fig. 53.1. The wooden cruciﬁx (XIV/XV c.) during the consolidation process

draped in representative gothic arrangement of folds. The crown has the form of twisted rope with wooden thorns. The complexion has natural ﬂesh colour with some grey shade probably originating from surface soiling. Wounds on hands, the right side of the body, and feet are rust-coloured. The remaining polychrome shows the characteristic scheme of gothic stratigraphy, i.e. the priming and overlying polychrome.

53.2 Experiment The Raman spectra were recorded by means of the micro-Raman stand (Renishaw 1000) under diode laser excitation at 785 nm of samples (2 × 2 mm2 ) taken from each colour of the polychrome layers: ﬁgure body (A), gold of the perizonium (B), green of the crown of thorns (C), red of the side wound (D) and the perizonium (E), and its inner fold (F). All samples were divided into two parts resulting in two sample sets. One of them was prepared by immersion in resin and subsequent polishing in order to expose the cross sections and the second set was used for measurements on the outermost surface layer of the sample. The elemental analysis was carried out by means of the LIBS technique. Measurements of the laser-induced plasma emission were performed by focusing a KrF laser beam (248 nm, 20 ns pulse) on the sample surface. The laser ﬂuence was in the range from 0.8 to 1.5 J cm−2 . The plume emission was dispersed by a 0.3 m spectrograph (Bentham; grating 1,200 grooves mm−1 ) and recorded by a time gated ICCD detector (Andor Technology). For measurements on selected samples, laser excitation at 266 nm (6 ns) and ﬂuence in the range from 2 to 5 J cm−2 were applied. In that case the emission was collected by the 0.5 m spectrograph (Acton RC, grating 300 grooves mm−1 ) and recorded by a CCD detector (SBIG ST-6). In both cases, the time integration mode was used.

53 Pigment Identiﬁcation on a XIV/XV c. Wooden Cruciﬁx

447

Fig. 53.2. LIBS proﬁle of sample A – polychrome of the hand, and the Raman spectrum (inset)

53.3 Results and Discussion For sample A, taken from the polychrome of the ﬁgure hand, the homogeneous layer observed in the cross section corresponds to the greyish complexion of the ﬁgure body. In Fig. 53.2, the lines of Zn (330.26; 334.5; 468.01; 472.21; 481.05; 518.20) and Ba (413.06; 455.4; 493.41; 585.36) in the LIBS proﬁle indicate the presence of lithopone (ZnS + BaSO4 ). Since this white pigment was ﬁrst manufactured in 1874, the conclusion of retouching is in order. The evidence of the original pigment component is revealed by the Raman bands at 1,087 and 283 cm−1 which correspond to chalk (CaCO3 ) (Fig. 53.2, inset). The chalk, known since Antiquity, was used together with a binder as priming [4,5]. Pigments such as ochre, greens, reds, or blacks, usually added to white in order to reproduce the skin colour, are not observed in the spectra [6]. Apart from numerous Au lines observed for sample B taken from the gold coated perizonium, strong bands of Ca are also present and ascribed to the boliment (Figs. 53.3a,b). This follows from comparison of proﬁles recorded for the ﬁrst, second, and third excitation pulses showing gradual removal of the very thin Au layer (Fig. 53.3a), inset. Spectra obtained for the layer under the gold plating reveal the presence of elements like Si, Mg, Fe, and Al of the earthen oxides in the boliment (Fig. 53.3b). For sample C, extracted from the crown of thorns, the green colour dominates on the surface. However, besides that green forming a homogenous bulk, some blue and yellow particles are also visible in the sample cross section. The Raman spectra relevant for the green and yellow coloured regions

448

M. Sawczak et al.

Fig. 53.3. (a) LIBS proﬁle of the gold sample surface; (inset) decrease in intensity recorded for the ﬁrst, second, and third excitation pulse, and (b) spectrum recorded for the poliment

are shown in Fig. 53.4a, b. For the pure green layer, numerous bands of malachite Cu2 (CO3 )(OH)2 are observed with the strongest ones located at 178, 268, 433, 1,051, 1,085, and 1, 492 cm−1 (Fig. 53.4a). Besides, the presence of band groups corresponding to ivory black

53 Pigment Identiﬁcation on a XIV/XV c. Wooden Cruciﬁx

449

Fig. 53.4. Raman spectrum of the thorn crown of sample C with (a) recorded for the bulk of the green layer and (b) for the yellow grain inclusion

Ca3 (PO4 )2 + C + MgSO4 and azurite Cu3 (CO3 )2 (OH)2 are observed for the base green and blue inclusion (not shown here) as well. The azurite often occurs with malachite due to similar origin and structure. Moreover, the green layer shows bands ascribed to the Prussian blue pigment Fe4 [Fe(CN6 )]3 · 14–16H2 O known since 1704. This observation allows us to conclude that the green polychrome area was retouched after that date [7]. The proﬁle corresponding to the yellow grain inclusion consist of strong, wide bands at 275 and 457 cm−1 and a weaker one at 525 cm−1 . These can be doubtless ascribed to lead tin yellow, Pb2 SnO4 (Fig. 53.4b). The collection of the sharp, strong bands at 309, and 353 cm−1 accompanied by broader ones at 220; 381 cm−1 and the weak one at 230 cm−1 is ascribed to orpi-

450

M. Sawczak et al.

Fig. 53.5. LIBS spectra for green pigment of the crown of thorns of sample C. The elemental composition observed in the visible spectral range by means of the LIBS technique agrees with the pigment compounds found in the Raman spectra

ment, As2 S3 . Both yellow pigments are known to be used in painting during thirteenth to eighteenth century [8]. In addition, the strong bands located at 338 and 403 cm−1 , together with a weaker one at 360 cm−1 , are assigned to chrome yellow, PbCrO4 . This pigment began to be used in painted artworks in 1809 [9]. Its evidence, together with the Prussian blue, conﬁrms the conclusion on retouching of the crown polychrome performed probably more than 100 years ago. For consistency, also the elemental composition of sample C can be considered (see Fig. 53.5). The atomic emission lines observed at 333.78; 570.02, and 578.21 nm are assigned to Cu and ascribed to malachite and azurite. Groups of Pb and Cr lines correspond to compounds of chrome and tin yellow pigments. The group of Fe lines and bands of CN (382 nm) and C2 (470; 516; 555 nm) are ascribed to Prussian blue and ivory black pigments, respectively, [2]. Numerous lines of Ca conﬁrm the presence of chalk-based ground revealed in the Raman spectra of sample B, too. The presence of Na (588.99; 589.59 nm) originates from surface pollutants often revealed by the LIBS technique when applied to historical objects [3]. The above results indicate that the elemental composition of sample C supports the assignment of pigment compounds identiﬁed in the Raman spectra. Spectra of the red pigmented samples D, E, and F are shown together in Fig. 53.6. The intense, red pigment extracted from the perizonium (sample D) reveals the strong, wide Raman band located at 253 cm−1 , and two slightly weaker ones at 284 and 343 cm−1 , all ascribed to vermilion, HgS (Fig. 53.6a).

53 Pigment Identiﬁcation on a XIV/XV c. Wooden Cruciﬁx

451

Fig. 53.6. Raman spectra of the intense red pigment layer of perizonium, and the inside part of haunch festoon: (a) samples D, E and (b) of the side wound, sample F

Another sample E, taken from the inner fold of the perizonium, is characterized by weaker light reﬂection and dull colour. Its microscopic examination shows some darkened, brown, and red areas. In the spectrum the bands of vermilion are present similarly to sample D but the signal/noise ratio is much weaker. The darker appearance of this polychrome area may be due to metacinnabar which originates from the light and humidity induced transformation of vermilion [10]. Bands recorded for sample F are readily assigned to vermilion (HgS), and red lead (Pb3 O4 ) used since Antiquity (Fig. 53.6b). Besides, rich collection of bands located at 181, 202, 220, 230, and 309 cm−1 conﬁrm the presence of the yellow pigment orpiment (As2 S3 ) which was used extensively in Antiquity and during the Middle Ages.

452

M. Sawczak et al.

53.4 Conclusions The LIBS and Raman techniques were applied for analysis of the historical pigments of polychrome paints on a wooden cruciﬁx from fourteenth/ﬁfteenth century. The pigment compounds used originally for the ground and paint layers such as chalk, vermilion, red lead, Mars red, malachite, and azurite were identiﬁed. From the presence of bands corresponding to Prussian blue and chrome yellow, known since the eighteenth century, the partial retouching of some statue areas was concluded. Using two independent LIBS stands, the elemental analysis was performed upon laser excitation at 248 and 266 nm and numerous pigment compounds were identiﬁed in agreement with the relevant Raman bands. Results of this work made possible the mapping of the polychrome pigments and also the layer stratigraphy and indicate that both techniques supply reliable information on the composition of historical paint layers. Acknowledgements Authors acknowledge support of the Polish Ministry of Scientiﬁc Research and Information Technology via grants H01/DWM119 and KN/DWM102, and also the bilateral exchange between CSIC and Pol. Acad. Sci. M. Oujja appreciates the Thematic Network on Cultural Heritage of CSIC (I3P Program) for a contract.

References 1. M. Castillejo, M. Martin, D. Silva, T. Stratoudaki, D. Anglos, L. Burgio, and R. Clark, J. Molec. Struct. 191, 550 (2000). 2. P. Vandenabeele, T.L. Weis, E.R. Grant, and L. Moens, J. Anal. Bioanal. Chem 379, 137 (2004). 3. K. Ochocinska, M. Martin, J. Bredal-Jørgensen, A. Kamińska, and G. Śliwiński, Rad. Phys. Chem. 68/1–2, 227 (2003). 4. J. Hopliński, Farby i spoiwa malarskie, Ossolineum, Warszawa, 139, 1990. 5. Ars Longa Vita Brevis. J. Flik (ed.), N. Copernicus Univ. Press, Toruń, 405, 2003. 6. E. Berger, Quellen und Technik, Muenchen, 47, 1912. 7. M. Bouchard and D.C. Smith, Spectrochim. Acta A 59, 2247 (2003). 8. M. Doerner, Materialy malarskie i ich zastosowanie, Arkady, Warszawa, 1975. 9. D.A. Anfam, M. Beal, E. Bowes, et al. Techniques of great masters of art, Arkady, Warszawa, 2004. 10. K. Keuene, PhD Thesis, Institute of Atomic and Molecular Physics, Univ. of Amsterdam, 93, 2005 (www.amolf.nl/publications/theses/).

54 MOLAB, a Mobile Laboratory for In Situ Non-Invasive Studies in Arts and Archaeology ∗

B.G. Brunetti1 , M. Matteini2 , C. Miliani1 , L. Pezzati3 , and D. Pinna4 1

∗ 2

3

4

Centre SMAArt & INSTM, Department of Chemistry, University of Perugia, Via Elce di Sotto 8, 06123 Perugia, Italy [email protected] CNR – Istituto per la Conservazione e Valorizzazione dei Beni Culturali, Via Madonna del Piano 10 – Ediﬁcio C, 50019 Sesto Fiorentino (FI), Italy CNR – Istituto Nazionale di Ottica Applicata, Largo E. Fermi 6, 50125 Firenze, Italy Opiﬁcio delle Pietre Dure, Ministero per i Beni e le Attività Culturali, Via Alfani 78, 50125 Firenze, Italy

Summary. Mobile laboratory (MOLAB) is a unique joint collection of portable equipment for non-destructive in situ measurements. MOLAB activities are carried out within the frame of the Eu-ARTECH Integrated Infrastructure Initiative of the sixth F.P. In situ measurement is quite useful because it eliminates any risk connected to moving artworks or other precious objects to a laboratory. MOLAB instruments are accessible to European researchers through a peer-review selection of proposals. Starting from July 2004, MOLAB enabled non-destructive in situ studies of many precious artworks, such as paintings by Perugino, Raphael and Leonardo.

54.1 Introduction The application of non-destructive scientiﬁc methodologies to the conservation of cultural heritage is a very active ﬁeld in Europe due to the increasing value attributed by each nation to the preservation of its artistic and historical patrimony. These activities are currently carried out in public and private institutions, through cooperative inter-disciplinary programmes among scientists, conservators, archaeologists and art-historians. Within the framework of European science for conservation, great advancements are expected by cooperation among institutions, with the aim of structuring of a common European Research Area in this ﬁeld. Therefore, a Consortium among 13 distinguished European infrastructures has been established with the name Eu-ARTECH (Access Research and Technology for the conservation of the European Cultural Heritage). The Consortium is coordinated by the Italian Centro di eccellenza SMAArt of the University of Perugia, and includes the Centre de Recherche et de Restauration des Musees de France (C2RMF), the Scientiﬁc Department of

454

B.G. Brunetti et al. Table 54.1. List of Eu-ARTECH Consortium infrastructures

1 2 3 4 5 6 7 8 9 10 11 12 13

University of Perugia, Perugia, Italy (Coordinator) Centre de Recherche et de Restauration des Musées de France, Paris CNR – Ist. per la conservaz. e valorizzazione dei beni culturali, Firenze, Italy The Scientiﬁc Department of the National Gallery of London, UK Opiﬁcio delle Pietre Dure, Firenze, Italy Bayerisches Landesamt für Denkmalpﬂege, Munich, Germany Ormylia Art Diagnosis Centre, Ormylia, Greece Instituut Collectie Netherlands, Amsterdam Laboratorio Nacional de Engenharia Civil, Lisbon, Portugal Institut Royal du Patrimoine Artistique, Bruxelles, Belgium University of Aachen, Germany University of Bologna, Italy CNR – Istituto Nazionale di Ottica Applicata, Firenze, Italy

the National Gallery of London (NGL), two institutes of the Italian National Research Council (CNR-ICVBC and CNR-INOA), the Opiﬁcio delle Pietre Dure (OPD) of Firenze and other important European institutions (for the complete list, see Table 54.1). The activities of the Consortium are articulated as follows: – Networking, with the goal to promote exchange of personnel and sharing of knowledge, and to foster the adoption of common good practices. – Joint research, devoted to improve the performances of the participating infrastructures in artwork studies and conservation. – Trans-national access, devoted to strengthen the high-level research in Europe by promoting the access to advanced instrumental resources (laboratories) and know-how (competences). Within the Eu-ARTECH project, two trans-national accesses have been planned. The ﬁrst is to the Accélérateur Grand Louvre d’Analyse Élémentaire (AGLAE) laboratory of C2RMF, located at the Palais du Louvre in Paris, oﬀering the possibility of non-destructive elemental composition studies by PIXE, PIGE, RBS and other nuclear techniques. The second is to the Mobile Laboratory (MOLAB), consisting of a unique collection of portable equipment, available for in situ applications, belonging to a joint group of four Italian infrastructures (Centre SMAArt of the University of Perugia, CNR – ICVBC, CNR – INOA and Opiﬁcio delle Pietre Dure). Access can be obtained through proposals presented to Eu-ARTECH, where an international peerreview committee evaluates the scientiﬁc relevance of the proposed projects. For accepted proposals, Eu-ARTECH supports the researchers for travel and subsistence. Access to the MOLAB has very speciﬁc characteristics: Its main feature is that the access “route” is reversed. In the traditional approach, researchers (and objects) are moved to the laboratory to use “immovable” equipment. In the MOLAB case, the laboratory itself is moved to the site where the

54 MOLAB in Arts and Archaeology

455

artwork is located, together with a team of researchers. This reversed access, certainly unusual for European trans-national programs, is now recognised as indispensable in the case of non-destructive studies of artworks, monuments or other precious objects, which are rarely moved from their places of keeping.

54.2 The Mobile Laboratory The MOLAB combines several diagnostic instruments and techniques, most of which are optical, using radiation from the X-rays to the infrared (IR). These are complementary, allowing thorough investigation of works of art, from their morphological structure (both visible and hidden) to their elemental and molecular composition. Here follows a brief description of the main instruments currently within the MOLAB portfolio. X-Ray Fluorescence (owned by SMAArt). XRF is a well-established technique for the identiﬁcation of key elements, in applications where the integrity of the sample is a basic requirement. The elemental surface mapping, for elements heavier than silicon, is achieved by a portable X-ray instrument with a spatial resolution of 4 mm. In pigment identiﬁcation, the detection of key elements such as copper, mercury, lead and cobalt supports educated guesses on pigment composition while the identiﬁcation of key impurity elements, such as nickel, zinc and manganese may provide information on provenance or even manufacturers, as it has been recently shown in a study of about 50 Perugino’s paintings [1]. The information on elemental composition can be complemented by information provided by molecular spectroscopies such as FTIR and micro-Raman [2, 3]. Fibre optic mid-FTIR (SMAArt). This portable spectrophotometer (JASCO VIR 9500) is equipped with a mid-infrared ﬁbre optic sampling probe. The system has an excellent signal-to-noise ratio in the range 900–4,000 cm−1 with the exception of the 2,050–2,200 cm−1 region. The non-contact probe is kept perpendicular to the painting surface (0◦/0◦ geometry) at a distance of about 6 mm. Owing to the probe geometry, reﬂectance mid-FTIR spectra can present large distortions, both in band shape and absorption frequency, so that it is diﬃcult to compare these spectra with available database transmission spectra. For this reason, SMAArt carried out a careful laboratory study of the performance of the reﬂectance set-up, according to diﬀerent experimental conditions and substances to be identiﬁed. This approach allowed SMAArt to carry out successful in situ measurements on several artwork types from marble sculptures [4] (Fig. 54.1) to wall paintings [5, 6]. Fibre-optic micro-Raman (SMAArt). This instrument is a prototype developed and assembled in cooperation between SMAArt, University of Milano and JASCO Europe Company, working with a Nd:YAG laser source. Its spectral range is about 2, 000 cm−1 with a maximum spectral resolution of about 8 cm−1 . The measurement is contact-free and the maximum laser power at the

456

B.G. Brunetti et al.

Fig. 54.1. MOLAB analyses on Michelangelo’s David

surface is 5 mW. Fibre-optic micro-Raman has demonstrated its potential for the identiﬁcation of pigments and alteration compounds with eﬃcient Raman scattering. Fibre optics UV–VIS ﬂuorescence (SMAArt). This technique employs another assembled prototype and it is used as a ﬁrst approach to the study of organic substances. It has recently been used experimentally for the study of dyes and lakes on dyed tapestries and on easel paintings. Punctual analyses achieved by the ﬂuorimeter can be complemented by imaging ﬂuorescence surveys performed by the portable ﬂuorescence system owned by OPD. In this case ﬂuorescence images are recorded by a 5-megapixel digital camera, using a Spectralon target for reference. IR/colour scanner for reﬂectography (INOA). IR reﬂectography is one of the most eﬀective optical techniques applied to non-destructive analysis of ancient paintings. The examination of a painted surface by IR light reveals hidden features underlying the pictorial layer or inside it, due to the transparency of ancient colours to radiation in the near-IR region of the electromagnetic spectrum. The MOLAB includes the prototype of the IR/colour scanner developed by INOA (Fig. 54.2). This equipment allows recording both colour and IR images, free of distortions and with the best resolution available in the ﬁeld. From the collected data it is possible to compare directly, on the monitor of a PC, the visible image of the object under study and its IR reﬂectogram. This allows the users to study the execution technique of a painter [7, 8] and, as in a recent study of Perugino’s technique, to clearly

54 MOLAB in Arts and Archaeology

(a)

457

(b)

Fig. 54.2. The IR/colour scanner measuring the Estasi di S. Cecilia by Raﬀaello (a), and the Madonna Benois by Leonardo at the State Hermitage Museum (b)

identify hand-drawings, pentimenti, and cartoon re-use [9]. Besides, it is also possible to identify old restorations or to infer the presence of some pigments. For this purpose, the imaging survey can be usefully complemented by punctual spectral analyses carried out by ﬁbre optic VIS-NIR spectroscopy or other elemental techniques. Laser micro-proﬁlometry (INOA). The laser micro-proﬁlometer is an interferometric device to measure, with high precision, the relief of small objects or almost ﬂat surfaces. Laser micro-proﬁlometry can be applied to measure and monitor the surface of many artworks ranging from panel and canvas paintings, to analyse alterations by restoration interventions or pigment microdetachments, to marble artefacts, to monitor the roughness before and after the restoration intervention [10]. Recently, a 3-D survey has been carried out on Perugino’s painting surfaces revealing features such as parallel grooves impressed during scraping and smoothing of the panel ground, thus providing a starting point for a detailed study of the painting history [11]. R (SMAArt). The portable instrument (Nuclear MagEUREKA-MOUSE netic Resonance Mobile Universal Surface Explorer, registered trademark at the University of Aachen) is a unidirectional NMR relaxometer purposely created for the study of cultural heritage and for NMR characterisation of unmovable objects. It takes advantage of the principles of magnetic resonance and inside-out-NMR. The instrument can be used to measure paper [12], parchment or wood degradation, detachment of painted surfaces from a wall or other support, and presence of water in porous stones [13], frescoes or ceramics.

458

B.G. Brunetti et al.

Drilling resistance measurement system (CNR-ICVBC). DRMS is a new portable system for directly determining stone mechanical features, such as the hardness, by measuring its drilling resistance. The DRMS has no competitors for its application in comparative in situ tests and has been suggested as standard method, to UNI Normal (Italy), for quality assessment of consolidating treatments. The test is essentially non-destructive. By this system, the cohesion proﬁle of a stone material can be determined on the basis of aﬀordable and sensible data on its mechanical and abrasive properties [14].

54.3 National and Trans-National Access: Examples and Results During its ﬁrst year of diagnostic activities, the MOLAB collected many interesting results, acquired through several recent national and transnational study campaigns. A relevant example is the study carried out on Michelangelo’s David prior to its recent restoration. Reﬂectance FT-IR spectroscopy pointed out the diﬀuse presence of gypsum, distributed all over the surface, and calcium oxalate in lower quantities, with a more localised distribution pattern [4]. Laser micro-proﬁlometry allowed for monitoring the marble surface state of conservation, by measuring the roughness over six sample areas [10]. Last, the ﬂuorescence investigation carried out by OPD and CNR-ICVBC revealed the presence of a yellow luminescence, distributed on a large fraction of the surface [15] and probably due to beeswax, as supported by time-resolved ﬂuorescence imaging and ﬁbre optic mid-FTIR carried out on several points [4]. Another example of trans-national access was the study of the painting by Agnolo Bronzino, Lamentation on the Dead Christ (1545), conserved at the Musée des Beaux-Arts et d’Archéologie de Besançon (F). The MOLAB study allowed restorers to gather detailed information on the underdrawing (pentimenti were found), and on regions where retouchings were carried out. Precious indications on varnish composition and on the nature of binder(s) used in the execution of the artwork were also obtained. Of course, all information was obtained without moving the painting and without any sampling, i.e. without any damage or any risk connected with the transportation of the artwork to a laboratory. Probably the most relevant results of the MOLAB’s ﬁrst year of transnational access was the examinations of three Leonardo paintings: the Madonna Benois and Madonna Litta at the State Hermitage Museum in St. Petersburg and the Vergine delle Rocce at the National Gallery of London (Fig. 54.3). In the latter case, an unexpected underdrawing by Leonardo has been revealed in the ﬁnest detail, allowing the researchers of the National Gallery to gather new insights about this controversial painting. The discovery, published and thoroughly discussed in a recent work in the Burlington

54 MOLAB in Arts and Archaeology

(a)

459

(b)

Fig. 54.3. Measuring the Vergine delle Rocce by Leonardo at the National Gallery in London: two MOLAB scanners during a measurement (a) and a detail of the IR image showing the unexpected drawing of a female head (and a left hand) under the painted Virgin (b)

Magazine [16], has been the object of a world-wide press campaign this last summer.

54.4 Conclusions The objective of the MOLAB activity is to sponsor new opportunities for researchers in conservation science, conservators or archaeologists, oﬀering access to a unique pool of advanced portable equipment and competences for in situ non-destructive studies of artworks. Applications are welcome from postgraduate researchers to senior professors, who are working in universities or research centres, or from institutions such as Museums or Conservation Institutes, or even from public and private SME’s working in the ﬁeld of conservation/restoration. A ﬁnal relevant note: the applicant of the proposals selected by the peer-review committee will use the portable equipment completely free of charge and will be the intellectual property owner of the produced scientiﬁc data. Applications can be made using the online form available at http://www. eu-artech.org. Deadlines are arranged with six-month periods from June 2004 until May 2009.

460

B.G. Brunetti et al.

References 1. I. Borgia, B. G. Brunetti, P. Moioli, C. Seccaroni, and A. Sgamellotti, in The Painting Technique of Pietro Vannucci, called the Perugino, LabS TECH Workshop Proceedings, Perugia, April 13th –14th , Nardini Editore, 2003. 2. C. Ricci, I. Borgia, C. Miliani, A. Sgamellotti, B. G. Brunetti, C. Seccaroni, and P. Passalacqua, in J. of Raman Spectroscopy, Vol. 35, 616, 2004. 3. F. Rosi, C. Miliani, I. Borgia, B. Brunetti, and A. Sgamellotti, in J. of Raman Spectroscopy, Vol. 35, 610, 2004. 4. C. Miliani, A. Sgamellotti, B. G. Brunetti, I. Borgia, and C. Ricci, in Exploring David – Diagnostic tests and state of conservation, Giunti Florence-Milan, 165, 2004. 5. C. Miliani, C. Ricci, F. Rosi, I. Borgia, B. Brunetti, and A. Sgamellotti, in Proceedings of IRUG6 Conference, Florence 2004. 6. E. Carretti, F. Rosi, C. Miliani, and L. Dei, Monitoring of Pictorial Surfaces by mid-FTIR Reﬂectance Spectroscopy: Eﬃciency of Innovative Colloidal Cleaning Agents, in Spectroscopic Letters, in press. 7. L. Marras, M. Materazzi, L. Pezzati, and P. Poggi, in Proceedings of SPIE International Symposium on Optical Metrology for Arts and Multimedia, Munich, June 2003. 8. M. Materazzi, L. Pezzati, and P. Poggi, in Venere e Amore / Venus and Love a cura di F. Falletti e J. K. Nelson, ed. Giunti, 244, 2002. 9. L. Pezzati, M. Materazzi, and P. Poggi, in The Painting Technique of Pietro Vannucci, called the Perugino, LabS TECH Workshop Proceedings, Perugia, April 13th – 14th , Nardini Editore, 2003. 10. R. Fontana, M. C. Gambino, M. Greco, L. Marras, M. Materazzi, E. Pampaloni, and L. Pezzati, in Exploring David – Diagnostic tests and state of conservation, Giunti Florence-Milan, 141, 2004. 11. R. Fontana, E. Pampaloni, and C. Seccaroni, in The Painting Technique of Pietro Vannucci, called the Perugino, LabS TECH Workshop Proceedings, Perugia, April 13th –14th , Nardini Editore, 2003. 12. B. Blümich, S. Anferova, S. Sharma, A. L. Segre, and C. Federici, in J. of Magnetic Resonance, Vol. 161 n. 2, 204, April 2003. 13. S. Sharma, F. Casanova, W. Wache, A. Segre, and B. Blümich, in Magnetic Resonance Imaging, Vol. 21, 249, 2003. 14. J. Delgado Rodrigues, A. Ferreira Pinto, and D. Rodrigues da Costa, in J. of Cultural Heritage, Vol. 3, 117, 2002. 15. A. Aldrovandi, M. Massi, and S. Porcinai, in Exploring David – Diagnostic tests and state of conservation, Giunti Florence-Milan, 150, 2004. 16. L. Syson and R. Billinge, in The Burlington Magazine, Vol. 147, n. 1228, 450, 2005.

Part VI

Scanning Techniques

55 From 3D Scanning to Analytical Heritage Documentation M. Schaich ArcTron – 3D Surveying Technology and Software Development Ltd, Ringstrasse 8, 93177 Altenthann, Germany [email protected] Summary. During the last few years, the number of historical and archaeological items recorded using innovative, three dimensional surveying technologies has increased considerably. Comprehensive digital, photo-realistic 3D recording and modelling yields a huge range of new possibilities for documenting, analysing and safeguarding items of cultural importance. ArcTron GmbH has specialised in electronic surveying and the development of CAD and database information systems for heritage and archaeology for over 12 years employing total stations, laser pantographs and photogrammetry, as well as a variety of laser scanning systems, to generate 3D recordings. Innovative complementary combinations of these technologies allow their respective strengths to be fully exploited. The data resulting from recording techniques of these kinds can be used as a basis for producing documentation of outstanding quality. 3D digital models with photo-realistic texturing, orthophotos, coloured point clouds, 3D damage mapping, rapid prototyping models and multi-media presentations are just some of the potential resulting forms of documentation. Historical monuments and items ranging from huge sections of terrain down to the smallest artefact can be reproduced with such accuracy and to such a high level of realism that documentation of this kind is not only extremely useful for archaeologists, conservators and architectural historians etc. but practically indispensable.

55.1 Introduction During the last few years, the employment of innovative three-dimensional surveying technologies has given rise to a diversity of new occupational ﬁelds for institutes that specialise in documenting cultural goods and monuments. The use of high-tech 3D scanning technologies and high-resolution, threedimensional computer tomography is becoming more and more prevalent within the ﬁelds of archaeology and heritage. We are, however, still emerging from the initial stages of this trend and the items being surveyed and modelled are mainly monuments of importance to world heritage, which are especially representative, spectacular or marketable [1–7].

This page intentionally blank

464

M. Schaich

55.2 Do Archaeology and Heritage Need 3D? These days, no manufacturer of laser scanners lets the opportunity slip by to mention the potential uses of their technologies for archaeology and heritage. These innovative technologies provide unprecedented opportunities within these ﬁelds for recording monuments and ﬁnds with photo-realistic precision and in three dimensions. Data such as this can be stored in virtual 3D archives to ensure that the information is maintained for posterity. Photo-realistic 3D documentation is of great beneﬁt not only to investigative archaeologists and architectural historians but also to conservators and restorers who may be required to reconstruct objects destroyed by environmental inﬂuences, natural disasters, vandalism or war. Three-dimensional documentation can also be made available to scientists and members of the public who are unable to visit monument sites in person and who must, therefore, rely on virtual, computer-supported access. Three-dimensional recordings are practically the only source of reliable basic data for structural analysis and for the planning of restorative or renovative measures. Thus, the question, especially when dealing with more complex historical monuments, should no longer be whether 3D technology is necessary for conservation projects, but rather which technologies and software solutions are currently available or must be developed.

55.3 First-Rate Standard of Documentation 3D scanning technologies are still relatively expensive procedures for museums and arts sciences but are increasingly being used for the documentation of every-day ﬁnds and heritage objects. 3D scanning technology was employed during the documentation of outstanding sculptures very early in its short history, for example, Michelangelo’s David in Florence. Europe’s largest monumental bronze, the 19 m tall Bavaria in Munich (Fig. 55.1) erected in 1850, was recorded by us in 2002 using a combination of various 3D scanning technologies. Many other outstanding archaeological and architectural monuments around the world have now been documented using procedures such as laser scanning and 3D photogrammetry. For over a decade, ArcTron Ltd. (http://www.arctron.com) has specialised in electronic surveying techniques and the development of surveying instruments with connected CAD/database information systems for archaeology and heritage. Starting in 1993, we mainly employed total stations, laser pantographs and 3D photogrammetry techniques. Since 2000, we have invested heavily in 3D laser scanning technologies. We now have diﬀerent scanners that enables us to provide a comprehensive three-dimensional surveying service. This ranges from the documentation of small ﬁnds, accurate up to 0.05 mm, all the way up to large-area surveying and topographical recordings of several hectares.

55 From 3D Scanning to Analytical Heritage Documentation

465

Fig. 55.1. The Bavaria. A monumental sculpture in Munich, almost 19 m tall, which was surveyed in its entirety using 3D scanning techniques in 2002. 3D data processing was carried out in connection with a 3D damage mapping system and used c ArcTron Ltd; Polygon to generate a complete, photo-realistic textured model ( Technology; Engineering Consultant, Dr. Koenig)

55.4 Time of Flight Laser Scanners The 3D laser scanners used by our company (Fig. 55.2; items 1, 2) are from the Austrian scanner manufacturer, Riegl-Laser Measurement Systems. These scanners are distinguished by their high accuracy and scanning ranges as well as being weatherproof and robustly built for use in the ﬁeld. One feature of these scanners that deserves particular mention is their interface to an external, high-resolution digital SLR camera mounted on the scanning head. In this conﬁguration, incidentally the ﬁrst of its kind, two complementary technologies are ideally combined. The hybridisation of photogrammetry and active 3D laser scanning in one system presents us with unprecedented opportunities to exploit the respective strengths of the two surveying techniques for use in the evaluation of 3D data. During laser scanning, 3D surveying is carried out “at the speed of light”. Technical details and speciﬁcations are given in the paper by F. Zehetner and N. Studnicka. 3D scanners can, of course, only capture areas that are within their ﬁeld of vision. The backs of walls, for instance, must be recorded from another position. This means that the scanner must be set up multiple times to record various areas from diﬀerent angles and positions to eliminate scan shadows. Archaeological excavations are frequently complex in nature with masonry features often being convoluted, narrow, undercut or deep. It is, therefore, particularly important to record features such as these from a number of positions to guarantee as high a level of survey coverage as possible.

466

M. Schaich

Fig. 55.2. 3D Scanners. Illustration of the systems employed by ArcTron. 1: Riegl LMS Z-420i with Nikon D100, 2: Riegl LPM 25-HA with Canon EOS D300, 3: Stripe light scanner, PTM-1024, from the Fraunhofer Institute

The accuracy of an individual scanned point measurement depends on many factors. For information concerning the comparative accuracies of different laser scanners, test results recently published by Professor Wolfgang Boehler from the Institute for Spatial Information and Surveying Technology at Mainz College of Applied Sciences are relevant. A special obstacle course for laser scanners has been set up at the institute. It contains a range of stations that disclose information concerning sensor noise, distance and angle accuracies, spot size, edge behaviour and the inﬂuence of diﬀerent surfaces on the various scanners. Errors such as these occur in all instruments to a greater or lesser extent and should always be taken into consideration during data evaluation. One fundamental factor for consideration is that laser scanning can only reproduce sharp edges with a certain amount of blurring. This is caused partly by the intrinsic sensor noise from the instruments and partly by the eﬀects of beam divergence on the accuracy of distance measurements. The edge eﬀect, in particular, can lead to incorrect or unclear distance measurements of certain geometric shapes. This is especially the case when the laser spot hits sharp edges, steps or sloped surfaces and can only return correct reﬂection values for some areas.

55.5 Triangulation Scanners: High Resolution 3D Recordings For the reasons cited earlier, the accuracy and resolution achieved by laser scanners is not suﬃcient to generate precise documentation of very complex archaeological objects such as skeletons, small ﬁnds or complicated architectural features such as capitals and so on. In cases such as these, high-precision triangulation scanners are required, which have accuracies in the sub-millimetre range. We employ and market a structured light scanner that was developed at the Fraunhofer Institute for

55 From 3D Scanning to Analytical Heritage Documentation

467

Computer Graphics (Fig. 55.2; item 3). Based on the principle for surveying using structured light, it has a maximum resolution of 0.05 mm and is thus ideally suited to record artefacts that are no bigger than a coin. Surveying with structured light is based on the principle of triangulation. The process is automatic and hands-free. In our system, a projector, two digital video cameras and the object location form a triangle in which the focussing directions of the cameras converge on the object at a ﬁxed angle known as the triangulation angle. The result of the individual recordings is digital images of the surfaces illuminated with structured patterns of light. These are then converted into point coordinates using specialist algorithms.

55.6 Data Post-Processing: A Complex Workﬂow Data post-processing (Fig. 55.3) is a relatively time-consuming and complicated procedure. We aim to produce high-resolution, photo-realistically textured 3D models, which can be used for further detailed analysis. One of the basic stages of processing is to check the point clouds for incorrect measurements and remove any errors. Likewise, any data that does not belong in the scan, such as passers-by, are removed. The point clouds are then transferred into a common co-ordinate system. This can be carried out automatically using various reﬂectors set up around the site that act as tie points

DATA ACQUISITION

BASIC DATA PROCESSING (Point Cloud Processing)

TOF Laser Scanner

SINGLE SCAN SEGMENTATION PURGING

Phase-Based Laser Scanner

3D Trisngulation Scanner

ADVANCED DATA PROCESSING

REGISTRATION (Common Coordinate System)

BASIC 3D DATA (POINT CLOUDS)

PHOTOREALISTIC DATA PROCESSING

DATA FILTERING POINT THINNING

PREPROCESSING IMAGES

TRIANGULATION POLYGONAL MESHING

MESH CHECKS MESH EDITING

3D MODEL / CAD (DXF, VRML, STL, IGES

TEXTURE MAPPING

MESH REDUCTION

TEXTURED 3D MODEL (3Ds, Vrml -> interactive)

Fig. 55.3. Sketch of the workﬂow: data processing with the software solutions that c ArcTron Ltd) are marketed or developed by ArcTron Ltd (

468

M. Schaich

to merge scans recorded from diﬀerent positions. A second way of doing this is to use a registration system of three or more points. This entails marking at least three naturally occurring points that are present in two overlapping scans. The software then aligns the point clouds of the two scans using these markers and transfers them into a common coordinate system. The end result of this stage is a seamless point cloud of the entire scanned area, which can be used for further evaluation and which should be archived as a preliminary result. The next stage in the procedure is advanced data processing, which can incorporate many further methods of evaluation. These include importing the data into CAD software, which, for instance, can display portions of point clouds and add ﬂoor plans or cross sections. If scanning has been carried out in combination with photogrammetry, it is now also possible to import the photographic information along with the point cloud and to evaluate the two sets of data concurrently. Individual 3D lines and bodies can now be manually deﬁned and traced. Procedures such as these, i.e. ﬁtting regular geometries into point clouds, are only of use when dealing with clear and simple structures, such as when surveying industrial or architectural objects. These techniques are only rarely of use when recording archaeological objects as the features encountered often have amorphous, irregular surfaces. A larger number of points is required to adequately describe such surfaces. In cases such as these, the point clouds must be triangulated and used to generate a suitably complex wire mesh. For this process, we employ and market the software, QTSculptor, which was developed at the Fraunhofer Institute for Computer Graphics. It is a specialist program for processing amorphous surfaces and is capable of correctly triangulating complex undercuts and recesses. The software works like a sculptor, carving 3D bodies out of a solid block. QTSculptor evaluates point clouds taking into account the respective position of the scanner. An algorithm, known as the Marching Cube algorithm, is employed to create a very ﬁne 3D mesh, which is subsequently optimised and reduced in various stages. In the resulting RAM-optimised mesh, areas in which a small number of triangles are suﬃcient to describe the surface are modelled with correspondingly large triangles whereas edges and intricate regions are given suitably ﬁne triangle structures. The resulting mesh is subsequently checked and 3D surface reduction is carried out to produce the initial ﬁnished geometric model. At this stage we also usually process the models in multiple Levels of Detail. The number of polygons in the model is reduced in stages from the highest resolution down to a representation with markedly reduced detail. This reduction process relies on the ability of the software to discriminate between essential and non-essential information. The software uses processes such as edge ﬁnding and geometric analysis to make these distinctions and thus ensures that the mesh can be reduced and optimised without loss of vital data.

55 From 3D Scanning to Analytical Heritage Documentation

469

Once this stage is complete, the water-tight 3D models are available in STL or DXF output formats. These data can be used as a basis for generating reproductions using rapid prototyping technologies. Surface reconstruction can also be carried out at this stage, for use in industrial CAD environments, for example. The concluding stage of modelling is the addition of photo-realistic textures to the polygon mesh, a process that is extremely involved and time-consuming. This stage is of utmost importance for the purposes of archaeology and heritage. In these disciplines, the quality of photographic realism and high-resolution texturing of objects surveyed using 3D scanning are far more important than geometric descriptions being accurate to within a millimetre. Photographic textures essential information for the contextual interpretation and classiﬁcation of objects. The resulting textured 3D model (Fig. 55.4) is the end product of processing. It can now be viewed in standard gaming engines, used in animation programs or even in immersive 3D virtual-reality environments. One of the most important questions is what the technicians or archaeologists involved in the project can ﬁnally achieve using this 3D data. If those involved have access to suitably powerful, expensive and complex 3D software, the data can be worked with directly in their own systems. However, this is often not the case. For precisely this reason, we have been developing a 3D information system called aSPECT 3D for several years (Fig. 55.5).

Fig. 55.4. Complete photo-realistic 3D survey of the buildings and excavations at c ArcTron Ltd) Castle Useldingen, Luxembourg (

470

M. Schaich

Fig. 55.5. 3D information management for heritage and archaeology. Screenshot of c ArcTron Ltd) the software aSPECT 3D (

55.7 aSPECT 3D: 3D Information Management for Heritage The software program, aSPECT 3D, enables archaeologists, conservators, excavation technicians and restorers, even those who are inexperienced in working with 3D data processing, to handle their 3D data directly and easily. The program is an interface in which photo-realistic 3D models can be structured, dissected or used to generate mapping. It can also generate orthophotos of diﬀerent views of masonry features, plans, etc. to be used as a basis for printed documentation. An integrated freeware 3D viewer allows other users to access the ﬁnished object and its layer structure. We believe that this program enables even those archaeologists and excavation technicians who do not own highly sophisticated 3D software or do not have the opportunity to carry out the necessary intensive training to access 3D documentation. In this program, the user is in direct contact with the 3D objects. Models can be revolved or rotated and zoomed to view even the smallest details. The program enables the user to draw directly onto 3D surfaces in order, for example, to mark, colour or cut out areas of damage. All objects are numbered and can be given further attribute content (metadata).

55 From 3D Scanning to Analytical Heritage Documentation

471

Individual 3D structures, such as the diﬀerent building phases of a masonry feature, can be accurately separated into individual 3D objects. The program features freely conﬁgurable layer control functions, which allow 3D information to be structured and organised on various internal layers. Filters can be used to selectively display certain types of object; water pipes or particular object numbers, for example. One of our further developments is a database interface for use in long-term monument monitoring projects and for data evaluation in 3D geographic information systems. It enables individual 3D elements to be linked to external databases. It is thereby possible to generate thematic mapping by using appropriate database queries. To conclude, it is re-iterated that this kind of documentation allows users to realise radically new methods of documenting ﬁnds and features for the purposes of heritage conservation. It is often the case in the world of archaeology and heritage that valuable features have to make way for new building measures or other projects. For these reasons, it seems advisable to employ and continue to develop the best available methods of documentation. 3D scanning provides users with an outstanding new foundation to build upon.

References 1. J. Albertz (Ed.), Proceedings of the XVIIIth International Symposium CIPA 2001. Berlin (2002). 2. O. Altan (Ed.), CIPA 2003 International Symposium. The ISPRS International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Volume XXXIV-5/C15 (2003). 3. W. Böhler (Ed.), Proceedings of the CIPA WG6 International Workshop on Scanning for Cultural Heritage. Corfu, Greece (2002) 4. V. Coors and A. Zipf (Eds.), 3D-Geoinformationssysteme. Grundlagen und Anwendungen. Heidelberg (2005). 5. T. Luhmann (Ed.), Optische 3D-Messtechnik – Photogrammetrie – Laserscanning. Beiträge der Oldenburger 3D-Tage 2004. Heidelberg (2004). 6. M. Schaich 3D-Scanning for Archaeology and Cultural Heritage. Preserving History with Geospatial Technology. In: J. Thurston (Ed.), Geoinformatics. Magazine for Surveying, Mapping & GIS Professionals 6, Vol. 7, 18, 2004. 7. M. Schaich, 3D Scanning Technologies and Data Evaluation in an Archaeological Information System. In: F. Niccolucci (Ed.), Proceedings of the CAA 2004 in Prato, Italy 2004.

56 Cleaning of Painted Surfaces and Examination of Cleaning by 3D-Measurement Technology at the August Deusser Museum, Zurzach P.-B. Eipper1 and G. Frankowski2 1

2

Alte Galerie am Landesmuseum Joanneum, Raubergasse 10, 8020 Graz, Austria, [email protected] GFMesstechnik, Warthestrasse 21, 14513 Berlin-Teltow, Germany

Summary. Grime and dirt are hazards to oil paint surfaces. To remove these impurities, paintings are usually cleaned dry, or wet with surfactants in aqueous medium. Historic paint material (oil-wax colors produced by Schoenfeld Lukas, Düsseldorf) used by the Rhenish painter August Deusser (1870–1942) were obtained and studied. To examine the eﬀects of diﬀerent cleaning methods, paint surfaces were treated dry and wet. The surfaces of the treated paints were examined by 3D-measuring technology. This new, transportable technology provides measurements in seconds during the cleaning process and produces measurable images that show changes on the surface and craquelure. Some aqueous cleaning systems can increase craquelure up to ﬁve times as much as dry cleaning methods on oil paint surfaces. However, dry methods are not suﬃcient to completely clean the surfaces. Therefore, modiﬁcation of aqueous cleaning methods are necessary and include using mild nonionic surfactants, thickening of the solutions used, reduction of contact humidity, and increasing temperature and, pH.

56.1 Introduction Most of the hundreds of surfactant solutions in use worldwide by conservators as well as other substances [1–12] have been shown to be harmful for treating art objects [13–16]. Oil- and oil-resin paints should not be eroded, leached out, weakened, or discolored by the use of improper or unsafe materials. To take account of diﬀering binding media, conservators want and need the potential to modify a cleaning solution. The wet cleaning of originally unvarnished paint surfaces is a frequently performed activity in conservation work. By the term “wet cleaning”, only the removal of water soluble or loosely adhering surface impurities is implied here (such as grime and particles, which have accumulated over time) and not the removal of additional layers, such as overpaint and varnish. It has been reported in the literature that the use of certain detergents can be

474

P.-B. Eipper and G. Frankowski

harmful and provoke serious damage to paint ﬁlm [14, 15, 17]. In collaboration with the Medical University of Hanover and the Max Planck Institute, Stuttgart, for example, we found that aqueous cleaning of surfaces always provokes microtears in the surface of oil paints [18, 19]. After ﬁnding reaction diﬀerences between newer and older oil paints and oil-resin colors, a cleaning solution was developed and examined. Criteria for choosing surface active substances include carbon chain length, critical micelle concentration (CMC), maximum foaming temperature (MFT), ethylene oxide number (EO/mol), hydrophobicity index (HI), hydrophilic lipophilic balance (HLB), and Zein number. It must be taken into account that the selected surfactant should not tend to discolor the oil paint or catalyse the decomposition of the painting material. A combination of 2% methyl cellulose and 0.2–0.3% of linear, nonionic, pH-neutral, minimally hygroscopic, nondegradable, nondiscoloring fatty alcoR 1618/25 (C-chain length of the alcohol between 16 hol ethoxylate, Marlipal and 18), produced by Sasol (formerly Hüls, then CONDEA, Marl, Germany), in aqueous solution (100 ml water) was developed and found to be exceptionally good for cleaning oil paint surfaces. In the years following, we have attempted to demonstrate the safety and usefulness of this solution using SEM [18, 19], ESEM [22], laser proﬁlometry [20, 21], and 3D-stripe projection [22] to detect any undesirable changes on the paint surface. It is well known that there are diﬀerences in the formulation of various oil paints used during the span of art history. Nevertheless, in the last years, we have found increasingly more similarities between many oil and oil-resin paints [21, 22].

56.2 Apparatus This new, transportable technology, called MicroCad, was presented to the public for the ﬁrst time in 2001 in Washington DC, USA and Düsseldorf, Germany. The 3D-Stripe projection technique based on micromirrors was developed at the GFMesstechnik, Berlin-Teltow, Germany. The micromirror projection can be diﬀerentiated from previous such stripe projections primarily through the use of digital projection. This measurement system produces and projects on the measured area diﬀerent grey values, which may be regulated by the use of micromirrors. This permits local reﬂections from various areas on the object surface to be balanced, in contrast to the use of traditional laser proﬁlometry, which causes artefacts. The micromirror technology is based on a development by Texas Instruments, Dallas. 1024 × 768 micromirrors of 16 × 16 µm2 size, are arranged in arrays (13×8 mm2 ). Each of these micromirrors can be regulated in brightness and/or color by a computer. In that way, various colors and/or light intensities can be produced. The advantage of this technology is that the computer is calculating a pattern that is then projected and ﬁnally an image is recorded with

This page intentionally blank

56 Cleaning of Painted Surfaces and Examination of Cleaning

475

a CCD-camera. The diﬀerence measurement between the computer-generated and projected pattern and that which is recorded by the camera leads to the so-called measurement eﬀect. An additional important advantage of the micromirrors is that one has a great deal of light available for measuring the paint surface. These types of micromirror projectors were developed for the use in Power Point-type presentation projectors. In 1996/1997 the idea for using the micromirror projectors for measurement technology was developed by GFM. The diﬀerence between the GFM-MicroCad technology compared to usual stripe projection may be summarized as follows: 1. With micromirror projection, one can project digitally, which means that through speciﬁc control of the micromirrors, measurement points of deﬁned grey values between 0 and 255 bits can be produced and projected. Thus diﬀerent local reﬂections from surfaces can be equalized. 2. Through the use of micromirrors for light projection, one observes (when compared to conventional projection systems such as LCD’s)that up to 90% of the light intensity from the lamp is directed onto the object to be measured. Thus one is in the position to carry out optical measurement techniques in normally lighted rooms without darkening. 3. Through a special development of GFM and Texas Instruments, a so-called high speed projection is feasible; that is, by using conventional lighting with structured white light, the video pulse time of the projectors/camera used are able to produce a 3D-proﬁle in 68 ms. In contrast to the 64 h measuring time needed by laser proﬁlometry for the same area, this is of great advantage. The MicroCad projects stripe patterns on the surface using either microscopic or macroscopic digital stripe projections reﬂected from micromirrors (referred to as digital micromirror devices). The light used, produced by a cold halogen source of 270 W (about 20 000 lx ) is conveyed through the projector to the surface to be measured for 300 and 1 000 ms, respectively. Following the measurement of the surface with light projection only, stripe patterns are then projected on the surface of the object. These intersecting patterns are photographed by a CCD-camera of 1300 × 1024 pixels and displayed within seconds on the computer screen. The measured surface is shown as intersecting patterns of the lines projected on the object. The distance between measurement points is 1.5 µm per pixel in the lateral (x−, y−) resolution and 0.3 µm in the vertical (z−) resolution with 1,000 points per mm (∼500 × enlargement). The photo technique delivers color and black/white scaled camera photos (the latter for orientation in the measuring ﬁeld), a photorealistic representation (which is very similar to SEM) as well as scaled colour microtopographic 3D-pictures. The last two, because they are based on 3D-point clouds, produce pictures that are both objectively measurable. Currently, the smallest measurement ﬁeld is 2 × 1.5 mm2 .

476

P.-B. Eipper and G. Frankowski

The outstanding advantage of the MicroCad for conservators is that this device allows immediate in situ examination and measurement within seconds during the cleaning process and it is producing measurable images that can show surface and craquelure changes before, during, and after treatment. This mobile measuring technique allows one to vary the surfactant solutions used during the cleaning process. The exact, scaled color microtopographic images give exact information concerning possible surface changes, which can also be useful in legal court cases. The MicroCad can also be used for measuring whole picture surfaces, for determining the edges of overpainted areas and comparing diﬀerent brush strokes of questionable attributions. In the future, no damaging paint sampling will be needed to be performed and no surface replicas will have to be produced.

56.3 Experimental To evaluate the impact of cleaning methods (wet/dry), as well as various types of wet cleaning agents on the surfaces of paint samples, the MicroCad technology was applied. Measurements at 10–15 cm working distance above the samples were carried out before and after treatment as described later. 56.3.1 Materials Paint Samples Historic painting material (“Lukas”-oil-wax colors from Schoenfeld, Düsseldorf) was taken from a painting by the Rhenish artist August Deusser (1870– 1942) [23]. To examine the eﬀects of diﬀerent cleaning methods, the paint surfaces were ﬁrst treated dry with a latex sponge (“wishab”, “akapad”) [24], and then wet (separately) with: 1. 2. 3. 4. 5. 6.

Tap water Demineralized water Saliva 2 g methyl cellulose (MC) in 100 ml demineralized water 2 g carboxymethyl cellulose (CMC) in 100 ml demineralized water R R 1618/25 (0.2 g Marlipal 1618/25 in The nonionic surfactant, Marlipal 100 ml demineralized water) R 1618/25 in 7. A combination of 2 g methyl cellulose and 0.2 g Marlipal 100 ml demineralized water 8. The anionic surfactant, SDS 9. A combination of methyl cellulose and 0.2 g SDS in 100 ml demineralized water

Before and after treatment, surfaces were examined using the MicroCad.

56 Cleaning of Painted Surfaces and Examination of Cleaning

477

Basic Properties of the Wet Cleaning Agents A wide variety of agents were used and their properties and characteristics are summarized here. Demineralized water It has little or no cleaning eﬀect and cannot dissolve oily or fatty substances. As is already known [4, 25], treatment of surfaces with demineralized water can, however, produce changes in the surfaces of oil paintings. Since the publication of these studies, we have been aware that water and especially demineralized water interacts in an unspeciﬁable way with dirt and oil paints and can thereby harm the surface of paintings. In practical terms, this means that it is not advisable to use demineralized water. It is better to use, depending on the pH-value of the dirt on the painted surface, tap water that contains carbonate, calcium, and possibly magnesium ions. Water with iron, copper, and manganese ions on the other hand cannot be recommended. Reducing the amount of water, for example, by producing pastes is advisable. In our present study, demineralized water has been used in order to compare it with nondemineralized water as well as to compare it with the eﬀectiveness of surfactants. Saliva The use of saliva has demonstrated generally good results when used for the cleaning of paintings. The unspeciﬁable cleaning eﬀect of saliva must be considered. Saliva contains a number of ingredients, e.g., enzymes, digesting lipids, and carbohydrates, which can decompose binding media. Saliva solubilizes polar components from varnishes and binding media [26–29]. Methylcellulose (MC) The usual Na-methylcellulose is made of wood or cotton cellulose and (caustic soda) sodium hydroxide solution. MC soluble in cold water comes in several types, some of which are alkaline. MC is a bacteriocide and fungicide. In the cleaning of surfaces of paintings, MC is used to lower the surface tension of water, or to thicken surfactant solutions. Carboxymethylcellulose (CMC) CMC is, like MC, a derivate of cellulose. CMC can form complexes, that is, CMC can carry dirt in solutions, even in absence of electrolytes; this is the reason why CMC is used in detergents. In the cleaning of paintings, CMC is used to lower the surface tension of water or to thicken surfactant solutions, although it is not as eﬀective in thickening surfactant solutions as MC.

478

P.-B. Eipper and G. Frankowski

R Marlipal 1618/25 powder

It is still problematic ﬁnding an ideal surfactant for conservators. According to the information obtained from chemists at Sasol, Germany’s largest producer of surfactants, only one out of the huge range of nonionic surfactants may be R 1618/25-powder, a sebum-fatty used for cleaning paint surfaces. Marlipal alcohol surfactant (alkyl-polyethyleneglycolether) has a water-like character. Its speciﬁc chemical details are as follows: 1. 2. 3. 4.

Chain length of C 16–C 18 EO/mol number 25 HLB number approximately 16 Reaches critical micelle concentration at a concentration of 0.2% at pH 6.5–7.0 5. 90% biologically degradable 6. Very low adipose eﬀect 7. HI number approximates 1; therefore, it is hygroscopic only to a small extent R 1618/25 powder was stored To examine this HI quality, 1 cm3 of Marlipal at 100% relative humidity. After 14 days, it was swollen and sticky, but only on its surface. In contrast to this, 1 cm3 concentrated granulate of SDS powder was wet after only 6 h. For this examination, solutions containing 0.2% R 1618/25 powder were used, because it reaches the critical micelle Marlipal concentration at this point. If surfactant concentrations exceed the critical micelle concentration, the cleaning eﬀect will not be more eﬀective [22].

Sodium Dodecyl Sulphate (SDS) + SDS (C12 H25 OSO− 3 Na , a monoester of sulphuric acid with dodecanol-1) is an anionic surfactant. Some of its properties are mentioned below:

1. 2. 3. 4.

HLB number is approximately 35 Reaches its critical micelle concentration at 0.2% Its pH ranges between 6.5–7.0 (0.5% in water) Its Zein number is 640

SDS can be considered an aggressive cleaning agent for paintings. Since SDS is a monoester of sulphuric acid, the ester binding may not be stable. It is known that catalytic agents, such as heavy metals found in paints, may hydrolyse SDS over time. The product is highly reactive and very hygroscopic sulphuric acid, provided the paint layer and ground layer of the support do not contain enough cations to buﬀer the sulphuric acid. SDS interacts with the hydrophobic side chains of proteins (molecular weights >5, 000 Dalton) and through its hydrophilic site it will solubilize formerly insoluble proteins. [30–32] This must be taken into account. It is feasible to choose SDS for the surface cleaning of paint ﬁlms and paint layers that are made of linseed oil, walnut oil, poppyseed oil, soya oil, or hempseed oil.

56 Cleaning of Painted Surfaces and Examination of Cleaning

479

For this examination, aqueous solutions of 0.2% SDS content were used, because SDS reaches the critical micelle density at this point.

Experimental: Cleaning Tests In accordance with our experience using computer-assisted laser proﬁlometry [20, 21, 33–38], paint areas were cleaned with the above-mentioned solutions. The greyish dirt layer consisted of loose dirt and grime. Sampling was done according to the treatment given below. Each treatment above was done for 20 s. Dry cleaning using a latex-sponge (“wishab”, “akapad”), Fig. 56.1 Wet cleaning using saliva on cotton swabs, Fig. 56.2 Wet cleaning using tap water on a microporous sponge, Fig. 56.3 Wet cleaning using demineralized water on a microporous sponge, Fig. 56.4 Wet cleaning using MC (2 g/100 ml demineralized water) on a microporous sponge, Fig. 56.5

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

1. 2. 3. 4. 5.

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Hohe (µm)

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Hohe (µm) −20

−15

−10

−5

0

−20

5

−15

−10

−5

0

5

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0 Länge (mm)

Länge (mm)

Fig. 56.1. (Left) Before; (right) after, dry cleaning with “wishab” latex sponge

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0 0.0 Hohe (µm)

0.3 −15

0.6 −10

0.9 1.2 Breite (mm) −5

1.5

1.8

0.0 0.0 Hohe (µm)

0

5

10

15

0.3 −15

0.6 −10

0.9 1.2 Breite (mm) −5

0

5

1.5

10

1.8

15

Fig. 56.2. (Left) Before; (right) after, wet cleaning with saliva on cotton swabs

P.-B. Eipper and G. Frankowski 1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0 Länge (mm)

Länge (mm)

480

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0 0.0 Hohe (µm)

0.3 −15

0.6 −10

0.9 1.2 Breite (mm) −5

1.5

0.0 0.0

1.8

Hohe (µm) 0

5

10

15

0.3 −15

0.6 −10

0.9 1.2 Breite (mm) −5

0

1.5

5

1.8

10

15

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

Fig. 56.3. (Left) Before; (right) after, wet cleaning with tap water on microporous sponge

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

0.0 0.0

1.8

Höhe (µm)

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm) −10 −8 −6 −4 −2

0

2

4

6

−10 −8 −6 −4 −2

8

0

2

4

6

8

Fig. 56.4. (Left) Before; (right) after, wet cleaning with demineralized water

1.6

1.6 1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

1.4

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm)

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm) −10 −8 −6 −4 −2

0

2

4

6

8

−10 −8 −6 −4 −2

0

2

4

6

8

Fig. 56.5. (Left) Before; (right) after, wet cleaning with methyl cellulose

56 Cleaning of Painted Surfaces and Examination of Cleaning

481

6. Wet cleaning using CMC (2 g/100 ml demineralized water) on a microporous sponge, Fig. 56.6 R 1618/25 (0.2 g/100 ml demineralized water) 7. Wet cleaning using Marlipal on a microporous sponge, Fig. 56.7 R 1618/25 (0.2 g/100 ml demineralized water) 8. Wet cleaning using Marlipal and MC (2 g/100 ml demineralized water) on a microporous sponge, Fig. 56.8 9. Wet cleaning using SDS (0.2 g/100 ml demineralized water) on a microporous sponge, Fig. 56.9 10. Wet cleaning using SDS (0.2%) and methylcellulose (2 g/100 ml demineralized water) on a microporous sponge, Fig. 56.10

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

The 3D stripe projection technique, described earlier, generated the ﬁgures in what follows.

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

0.0 0.0

1.8

Höhe (µm)

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm) −8 −6

−4

−2

0

2

4

−8 −6

6

−4

−2

0

2

4

6

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

Fig. 56.6. (Left) Before; (right) after, with carboxymethyl cellulose

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

0.0 0.0

0.3

0.6

0.9

1.2

1.5

1.8

Höhe (µm)

Höhe (µm) −12

−9

−6

-3

0

3

6

−12

−9

−6

-3

0

3

6 R

Fig. 56.7. (Left) Before; (right) after, wet cleaning with Marlipal microporous sponge

1618/25 on

P.-B. Eipper and G. Frankowski 1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

482

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

0.0 0.0

1.8

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm)

Höhe (µm) −9

−6

−3

0

3

6

−9

9

−6

−3

0

3

6

9

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

R Fig. 56.8. (Left) Before; (right) after, wet cleaning with Marlipal 1618/25 and methyl cellulose on microporous sponge

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm)

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

Höhe (µm) −6 −4

−2

0

2

4

−6 −4

6

−2

0

2

4

6

1.6

1.6

1.4

1.4

1.2

1.2

1.0

1.0

Länge (mm)

Länge (mm)

Fig. 56.9. (Left) Before; (right) after, wet cleaning with SDS on microporous sponge

0.8 0.6

0.8 0.6

0.4

0.4

0.2

0.2

0.0 0.0

0.3

0.6

0.9 1.2 Breite (mm)

1.5

1.8

0.0 0.0

0.3

0.6

0.9

1.2

1.5

1.8

Breite (mm) Höhe (µm)

Höhe (µm) −10 −8 −6 −4 −2

0

2

4

6

8

−10 −8 −6 −4 −2

0

2

4

6

8

Fig. 56.10. (Left) Before; (right) after, wet cleaning with SDS and methyl cellulose on microporous sponge

56 Cleaning of Painted Surfaces and Examination of Cleaning

483

56.4 Results Dry cleaning of oil paints with a latex sponge did not provoke new craquelure in the oil paint (Fig. 56.1). However, dry cleaned surfaces did not appear clean enough. Unfortunately, aqueous cleaning harms the surface of oil paints. The wet treated surfaces of oil paints showed more homogeneous proﬁles after cleaning than the dry cleaned ones. We discovered, similar to previous studies [18, 19, 22], that all wet cleaned surfaces demonstrated a huge range in the width of the craquelure. The craquelure, before cleaning, was 10–20 µm wide and, after wet cleaning and time to dry, 20–50 µm wide. We found that demineralized water produced very wide craquelure in the paint surfaces (Fig. 56.4). Nondemineralized water, which contains carbonate, calcium, and magnesium ions, is recommended over plain demineralized water. We found that runny, unthickened solutions provoke a wider craquelure in the surface than thickened cellulose ether pastes. R 1618/25 Surprisingly, we found that a combination of Marlipal (0.2 g/100 ml demineralized water) and methylcellulose (2 g/100 ml demineralized water) used on microporous sponge is the most sensitive cleaning agent for oil paints (Fig. 56.8). This paste reacts more sensitively than saliva. A rinse cleaning with demineralized water does not have to be carried out; tap water with carbonate, calcium, and magnesium ions is suﬃcient. Cellulose ethers (MC or CMC) oﬀer a careful, safe surface cleaning (Figs. 56.5, 56.6). Cellulose ethers tend to accumulate on surfaces of resinous oil paints [22]. Therefore, a second cleaning with water as above is necessary. It has been observed throughout the cleaning process that the sponge should follow the direction of the brushstrokes of the paint. Tissue paper should not be used for drying the surfaces. Cotton swabs provoke scratches on the paint surface. Microporous sponges made of polyvinyl acetate (e.g., “blitzﬁx”, “wondersponge”) should be used. R 1618/25 (0.2 g/100 ml water) combined with methyl cellulose If Marlipal (2 g/100 ml water) used on a microporous sponge does not produce suﬃciently clean surfaces, it is possible to prolong the time of contact, increase the temperature of the solution, or change the pH-value of the solution (to maxima of 6 or 8, respectively), depending on the pH of the surface impurities, to increase eﬃciency. This would be better than using a more aggressive anionic surfactant. It should be pointed out that shiny, rough, and dark surfaces of oilwax paints were problematic for the MicroCad because their high reﬂectivity caused interruptions during the measurements. Smooth and/or nonshiny paint surfaces are well suited for these examinations [39]. Acknowledgments The authors would like to thank PD. Dr. Julia Welzel (Medical University of Lübeck), Dipl.-Ing. Gert Opielka (Max Planck Institute for Metal Research,

484

P.-B. Eipper and G. Frankowski

Stuttgart), Dr. Claus-Dierk Hager (Sasol, Marl), Prof. Dr. Gebhard Reiss (University of Witten-Herdecke), Dipl.-Ing (FH) Silvio Zepke (GFMesstechnik, Berlin-Teltow), Samuel A. Westbrook, B.S., M.M.Sc. for translation and Dipl.-Chem. Matthias Appelt for revising.

References 1. D. Erhardt and J. J. Bischoﬀ, Resin soap components, in V. Todd et al.: Preprints to Restoration ’92 Conservation. Training, Materials and Techniques: Latest Developments, held at the RAI International Exhibition and Congress Centre, Amsterdam (1992), UK Institute of Conservation, London, 77, 1992. 2. B. Ford and A. Byrne, Australian Institute for the Conservation of Cultural Materials Bulletin, Vol. 17, 51, 1991. 3. B. Gilsa, Zeitschrift für Kunsttechnologie und Konservierung Vol. 1, 48, 1991. 4. G. Hedley, M. Odlyha, A. Burnstock, J. Tillinghast, and C. Husband, IIC Summaries of the Posters at the Brussels Congress (1990) Cleaning, Retouching and Coatings, 98, 1990. 5. N. Khandekar, Reviews in Conservation Vol. 1, 10, 2000. 6. K. L. Mittal and P. Bothorel (Eds.), Surfactants in Solution: Proceedings of two international symposia on surfactants in solution, Plenum Publishing Corporation 1–562, 1986. 7. M. A. Urbaneja, J. R. L. Arrondo, A. Alonso, and F. M. Goñi, The Interaction of Triton X-100 with Multilamellar Phosphatidylcholine Liposomes, Mittal & Bothorel 759, 1986. 8. M. M. Watherston, The cleaning of colourﬁeld paintings. The great decade of American abstract modernist art 1960–1970, The Museum of Fine Arts, Houston, Texas, 1976. 9. R. C. Wolbers, Notes for workshop on new methods in the cleaning of paintings, Getty Conservation Institute, 1–158, 1990. 10. R. C. Wolbers, Slides for workshop on new methods in the cleaning of paintings. Laboratory exercises”, University of Delaware, Newark, DE, USA without year, unpaginated 11. R. C. Wolbers, Cleaning painted surfaces. Aqueous methods, Archetype Publications, London, 1–198, 2000. 12. P.-B. Eipper, Museum aktuell, Vol. 93, 4012, 2003. 13. Y. Ashani and G. N. Catravas, Anal. Biochem., Vol. 109, 55, 1980. 14. P.-B. Eipper, Die Reinigung von Gemäldeoberﬂächen mit Tensiden. Der Einsatz von modiﬁzierten Polyvinylacetaten zur Konservierung von textilen Bildträgern, Paul Haupt Verlag, Bern 1993. 15. P.-B. Eipper, Vier Künstlerfarbenhersteller zwischen 1900 und 1970. Die Reinigung von Gemäldeoberﬂächen mit wässrigen Systemen“, Paul Haupt Verlag, Bern 1997. 16. K. Mansmann, Zeitschrift für Kunsttechnologie und Konservierung, Vol. 2, 220, 1998. 17. S. Hackney, J. Townsend, and N. Eastaugh, (Eds.), “Dirt and Pictures Separated”, The U.K. Institute for Conservation, London 1990. 18. P.-B. Eipper and G. Reiss, Mitteilungsblatt 49, Museumsverband Niedersachsen und Bremen e.V., Hannover 51, 1995.

56 Cleaning of Painted Surfaces and Examination of Cleaning

485

19. P.-B. Eipper and G. Reiss, The Picture Restorer, Vol. 10, 5, 1996. 20. P.-B. Eipper and J. Welzel, Mitteilungsblatt 54 Museumsverband Niedersachsen und Bremen e.V., Hannover, 69, 1997. 21. P.-B. Eipper and J. Welzel, The Picture Restorer, Vol. 15, 5, 1999. 22. P.-B. Eipper, G. Frankowski, H. Opielka, and J. Welzel, Ölfarbenoberﬂächenreinigung. Die Überprüfung gereinigter Ölfarbenoberﬂächen durch das Raster-Elektronen-Mikroskop, das Niederdruck-Raster-Elektronen-Mikroskop, die Laser-Proﬁlometrie und die 3D-Messung im Streifenprojektionsverfahren, Verlag Dr. Christian Müller-Straten, München 2004, 1–152. 23. P.-B. Eipper, Malmittel und Maltechnik August Deussers. Zwei Beiträge zu grundlegenden Fragen der Restaurierung Düsseldorfer Malerei der Jahrhundertwende, Wienand Verlag, Köln 1995, 1. 24. P.-B. Eipper, Museum aktuell, Vol. 91, 3898, 2003. 25. G. Hedley, M. Odlyha, A. Burnstock, J. Tillinghast, and C. Husband, A study of the mechanical and surface properties of oil paint ﬁlms treated with organic solvents and water, op. cit., 98–105; S. Michalski, A physical model of varnish removal from oil paint, op. cit., 85–92; A. Phenix, and A. Burnstock, The deposition of dirt: a review of the literature, with scanning electron microscope studies of dirt on selected paintings, “Dirt und Pictures Separated”, UKIC, London 11–18, 1990. 26. P. M. S. Romao, A. M. Alarcao, and C. A. N. Viana, Studies in Conservation, Vol. 35, 153, 1990. 27. R. R. Lehmann, Dental Magazin, Vol. 1 , 102, 1992. 28. B. Ramsay-O’Hoski, An investigation into the composition and properties of saliva in relation to surface-cleaning of oil paintings, National Gallery of Canada, Ottawa 1976. 29. J. Girling, The use of human saliva in conservation, MA Thesis. Institute of Archaeology. University College, London (1992) 30. T. B. Nelsen and J. A. Reynolds, Methods in Enzymology, Vol. 48, 4, 1978. 31. H. Stegemann, SDS-gel-electrophoresis in polyacrylamide, merits and limits, Righetti, Ossand, Vanderhoﬀ (eds.) Electrokinetic Separation Methods 1979, 313. 32. H. Stegemann and G. Pietsch, Seed Proteins, Biochemistry, Genetics, Nutritive Value, Vol. 45, 1983. 33. R. Saur, U. Schramm, R. Steinhoﬀ, and H. H. Wolﬀ, Der Hautarzt, Vol. 42, 499, 1991. 34. M. Rohr and K. Schrader, Euro Cosmetics, Vol. 24, 1994. 35. M. Rohr and K. Schrader, Skin Care Forum, Vol. 12, 12, 1995. 36. J. Efsen, H. N. Handsen, S. Christiansen, and J. Keiding, Laser proﬁlometry. In: J. Serup and G. B. E. Jemec, (eds.), Handbook of Non-Invasive Methods und the Skin, CRC Press Boca Raton, Ann Arbor, London, Tokyo 97, 1995. 37. U. Müller, Roughness Measured by Proﬁlometry: Mechanical, Optical and Laser, E. Berardesca, P. Elsner, K. P.Wilhelm, and H. I. Maibach, (eds.), Bioengineering of the Skin: Methods and Instrumentation, CRC Press Boca Raton, Ann Arbor, London, Tokyo 41, 1995. 38. R. Saur, Computergestützte Laser-Proﬁlometrie Entwicklung eines neuen dermatologischen Verfahrens zur Objektivierung der Rauheitsstruktur der Haut, medical dissertation, Lübeck 1993. 39. P. B. Eipper and G. Frankowski, The Picture Restorer, Vol. 26, 5, 2004.

57 Applicability of Optical Coherence Tomography at 1.55 µm to the Examination of Oil Paintings A. Szkulmowska1 , M. Góra1 , M. Targowska2 , B. Rouba2 , D. Stifter3 , ∗ E. Breuer3 , and P. Targowski1 1

∗ 2

3

Institute of Physics, Nicolaus Copernicus University, ul. Grudziądzka 5, 87-100 Toruń, Poland [email protected] Institute for the Study, Restoration and Conservation of Cultural Heritage, Nicolaus Copernicus University, ul. Gagarina 9, 87-100 Toruń, Poland Upper Austrian Research GmbH, Hafenstrasse 47-51, 4020 Linz, Austria

Summary. With 47 samples of commercially available oil paints, the applicability of Optical coherence tomography to noninvasive tomography of paint layers was examined. Two diﬀerent instruments, utilizing near-infrared light with central wavelength of 823 nm and, for the ﬁrst time, 1.55 µm, were used to obtain cross-sectional images. Example tomograms are given; a ray tracing correction of images is also discussed. The tests revealed that applicability of OCT is limited to certain pigments and the longer wavelength is better suited for this application.

57.1 Introduction Optical coherence tomography (OCT) is a technique used for noncontact and nondestructive examination of the internal structure of various objects. It relies on the detection of light of low temporal coherence scattered at various interfaces within the object. It is mostly applied in medicine but it has also been successfully used in materials science [1]. In addition to varnish thickness determination [2], it is tempting to use this technique for analysis of paint layers. The main drawback is usually high absorption of light by these layers, especially in the visible region. In the infrared, wavelengths of 0.81 µm [3] and 1.3 µm [4] have already been used. If a pigment absorbs the light moderately, the cross-section of the paint layer is visible. The aim of this chapter is to assess the applicability of OCT technique by direct evaluation of many commonly used paints, since the absorption data for pigments are usually available only for wavelengths longer than 2.5 µm.

488

A. Szkulmowska et al.

Fig. 57.1. Exemplary samples used for examination. From the left: Carmine alizarin, Madder lake light, Brownish madder alizarin, Scarlet alizarin. Line at sample #29 indicates the place where OCT tomograms, shown in Figs. 57.2a, c, were taken

57.2 Experimental Methods In the present study the longest wavelength available for OCT (1.55 µm) has been utilized in comparison with 0.82 µm. In both systems light scattered from the structural elements of the object examined was brought to interference in a Michelson interferometer, detected electronically and then analyzed by computer. The 0.82 µm Spectral OCT system is described elsewhere in this volume [2]. Brieﬂy: the “Broadlighter” from Superlum (Russia) was used as a light source with λcentre = 823 nm and ∆λ = 74 nm. This leads to a depth resolution of 6 µm. The lateral resolution of this instrument can be estimated to 15 µm. The power of light illuminating the sample was about 600 µW. The 1.55 µm time domain OCT device is described in detail elsewhere [1]. As a light source, a 3 mW superluminescence diode with a center wavelength of 1 550 nm and a spectral width of 51 nm is used and provides a depth resolution of 20 µm with a similar lateral resolution. The incident power on the sample for the system is 580 µW. From both instruments, the cross-sectional image (the OCT tomogram) of the examined structure is obtained in a noninvasive way. This image is shown in gray scale corresponding to the intensity of light scattered from the given location within the cross-section. Refractive Index Correction As follows from the description of the OCT method, all measured in-depth distances are optical ones. Therefore, the vertical scale of the image is diﬀerent in air (above the sample) than inside the sample. Moreover, if the scattering interfaces are not ﬂat, a signiﬁcant distortion to the shape of the underlying layer may appear. The simplest solution to overcome this disadvantage is to readjust the vertical scale in the image by multiplying the length of the vertical scale bar by the group refractive index of the media penetrated by the scanning light, as in Fig. 57.2, equivalent to contracting the height of the picture by the same factor. This standard approach permits us to properly measure thickness of ﬂat layers essentially perpendicular to the penetrating beam, but fails otherwise because of refraction. Such eﬀect may be seen in

This page intentionally blank

57 Applicability of Optical Coherence Tomography at 1.55 µm

489

820 nm

1550 nm

Fig. 57.2. The OCT tomograms obtained of Brownish madder alizarin (#333): (a, c) and Asphaltum extra (#414): (b, d). In both cases, two wavelength were applied: 820 nm (a, b) and 1,550 nm (c, d). Samples were scanned across the edge of the spot of paint with light incident from above

a

b

Fig. 57.3. The OCT tomogram obtained from Madder lake light (#327): (a) image not corrected, and (b) after ray tracing procedure. Both images are shown in the same scale. The second layer, visible inside the ground, is caused by oil from paint penetrating the ground layer

Fig. 57.2a,c. To properly assess structure of the sample, the ray tracing procedure is necessary. In this case the whole image is numerically corrected in such a way that for each position of penetrating light beam in turn, the refraction angle at the air-sample interface is calculated. Then a new point of incidence of this ray to the next intrasample interface is calculated. This procedure is repeated for all structural layers in the sample and for all positions of the scanning beam. Knowing the real optical paths of all light beams, a new, undistorted tomogram may be generated. A representative result is given in Fig. 57.3. The refractive index data may be taken from the literature [5] or, in the case of a varnish layer, also with aid of the OCT technique [4].

490

A. Szkulmowska et al.

Samples for examination were prepared by applying commercial oil paints, Rembrandt (R) or Van Gogh (VG) from Talens (The Netherlands), from Rowney (UK), or from Maimeri (Italy), to cardboard supports primed with chalk-and-animal-glue ground. There were 47 paints tested altogether (Fig. 57.1, Table 57.1).

57.3 Results and Discussion The OCT tomograms of paint layers In Fig. 57.2, examples of cross-section images (OCT tomograms) of two pigments obtained with both wavelengths used in this study are presented. Similar measurements were performed on all paint samples. Results are summarized in Table 57.1. Each sample was examined with both wavelengths, at the same place. Resulting images have been evaluated qualitatively and the samples divided into three categories on the bases of this evaluation: Table 57.1. The capability of OCT in paint layer imaging Pigment

Source

Cat. No.

Transparency

Paynes grey Transparent brown Transparent oxide yellow Vermilion Carmine alizarin Madder lake light Brown madder (Aliz.) Burnt umber Asphaltum Stil de grain brun Rembrandt brown Ultramarine deep Cobalt blue deep Scarlet alizarin Olive green Indian yellow Gold ochre Raw sienna Green earth Cassel earth Indigo extra

Rowney Rowney Talens (R)

65 260 265

o t 4

G G G

P G G

Talens (VG) Talens (R) Talens (VG) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (VG) Talens (R) Rowney Talens (R) Talens (VG) Talens (R) Talens (R) Rowney Maimeri Talens (R)

311 319 327 333 409 414 418 419 506 515 569 620 244 231 234 380 490 533

2 4 4 4 3 4 4 3 4 2 t 4 4 3 3 1 4

G G G G G G G G G G G G G M M M M M

P G G M M M M M G G G G G P P M P P

OCT applicability 1,550 823 nm nm

Continued

57 Applicability of Optical Coherence Tomography at 1.55 µm

491

Table 57.1. Continued Pigment

Source

Cat. No.

Transparency

Cinnabar green light extra Flake (lead) white Zinc white Titanium white Cerulean blue Naples yellow deep Yellow ochre Brilliant yellow light Aureoline Cadmium red deep English red light Caput mortuum violet Venetian red Pozzuoli earth Brown Vandyke Brown ochre light Raw umber Greenish umber Burnt sienna Sepia extra Prussian blue (phthalo) Cadmium green light Cobalt green deep Cadmium yellow deep Emerald green Chromium oxide green

Talens (R)

642

1

M

P

Talens (R) Talens (R) Talens (VG) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (R) Talens (VG) Talens (R) Talens (R) Rowney Talens (R) Talens (R)

101 104 105 194 223 227 239 242 306 340 344 349 365 403 405 408 410 411 416 566 604 612 613 615 668

1 2 1 1 1 1 3 1 1 1 1 1 2 2 2 3 3 2 3 1 2 1 1

P P P P P P P P P P P P P P P P P P P P P P P P P

P P P P P P P P P P P P P P P P P M P P P P P P P

OCT applicability 1,550 823 nm nm

Transparency of paints is described in scale 4. . . 1 (transparent, semitransparent, semiopaque, opaque) as indicated by Talens or t/o (transparent, opaque) as by Daler–Rowney. The applicability of the OCT technique is marked as G – good, M – medium, or P – poor

G whole paint layer is distinguishable: the paint-ground interface is well visible (Fig. 57.2c) M the paint-ground interface is visible for thin paint layer only (Fig. 57.2a) P the paint-ground interface is not visible due to high scattering/reﬂectivity (Fig. 57.2a, left hand side) or high absorption (Fig. 57.2b) in a paint layer. As expected, the longer wavelength is better suited for this application: for 1 550 nm, the OCT capability for 16 pigments may be qualiﬁed as Good and 6

492

A. Szkulmowska et al.

as Medium while, for 823 nm, these numbers are 9 and 7, respectively. There is no direct correlation between paint transparency indicated by the producer and the depth of penetration of light used for OCT examination. However, paints marked as most transparent are also more suitable for OCT examination. Refractive Index Correction Detailed inspection of Fig. 57.2a and c reveals a distortion to the paint-ground interface under a thick lump of paint. This eﬀect is also illustrated in Fig. 57.3 where the paint-ground interface is more visible. In the ﬁgure, a nodule of paint acts as a lens distorting the shape of layers below. The ground layer under the paint layer seems not to be ﬂat. This is an obvious artifact. After ray-tracing procedure, the ground layer ﬂattens, as expected.

57.4 Conclusions As may be seen from Table 57.1, the applicability of OCT for the examination of paint layers is limited to certain classes of pigments. There is no doubt that the longer wavelength penetrates the paint layer better making this method more useful. However, the depth resolution is four times lower for the system utilizing a light source of twofold longer wavelength and of the same spectral width [2]. Then the application of OCT to stratigraphy of oil paintings demands the OCT instrument with a very broad and long-wavelength light source. It is expected that rapid improvements in sweep-wavelength lasers may fulﬁl this requirement. Acknowledgments This work was supported by Polish Ministry of Science with Grant 2 H01E 025 25 and by the Austrian Science Fund FWF for Project P16585-N08.

References 1. D. Stifter, P. Burgholzer, O. Höglinger, E. Götzinger, and C.K. Hitzenberger, Appl. Phys. A 76, 947 (2003). 2. I. Gorczyqska, M. Wojtkowski, M. Szkulmowski, T. Bajraszewski, B. Rouba, A. Kowalczyk, and P. Targowski, in this volume. 3. P. Targowski, B. Rouba, M. Wojtkowski, and A. Kowalczyk, Stud. Conserv. 49, 107 (2004). 4. H. Liang, M. Gomez Cid, R. Cucu, G. Dobre, B. Kudimov, J. Pedro, D. Saunders, J. Cupitt, and A. Podoleanu, Proc. SPIE 5857, 2005 (in press). 5. R. de la Rie, Stud. Conserv. 32, 1 (1978).

58 Varnish Thickness Determination by Spectral Optical Coherence Tomography I. Gorczyńska1 , M. Wojtkowski1 , M. Szkulmowski1 , T. Bajraszewski1 , ∗ B. Rouba2 , A. Kowalczyk1 , and P. Targowski1 1

∗ 2

Institute of Physics, Nicolaus Copernicus University, ul. Grudziądzka 5, 87-100 Toruń, Poland [email protected] Institute for the Study, Restoration and Conservation of Cultural Heritage, Nicolaus Copernicus University, ul. Gagarina 9, 87-100 Toruń, Poland

Summary. The applicability of spectral optical coherence tomography (SOCT) for noninvasive and noncontact assessment of varnish layer thickness and structure on easel paintings is discussed. The SOCT tomograms of such objects are presented.

58.1 Introduction Knowledge of the thickness and structure of varnish layers is important for both conservation and historical reasons. It is desirable to use noninvasive and noncontact techniques for this task in order to be able to apply it as often as necessary anywhere in the picture, including the sensitive region of the artist’s signature. This technique could also make possible the continuous monitoring in situ of varnish removal during laser ablation. In addition to being nondestructive, the method must be fast and simple to use, and its in-depth resolution should be better then 5 µm. All the above conditions are fulﬁlled by spectral domain optical coherence tomography (SOCT) implemented with a broadband light source. SOCT is a rapidly developing new modality of imaging based on interference of partially coherent light: optical coherence tomography (OCT). The OCT technique utilizes light sources of high spatial but low temporal coherence, such as superluminescent diodes (SLD) or femtosecond lasers. It relies on the detection of broadband light scattered from the internal structure of a semitransparent object. The main applications of OCT remain in medicine, mainly in ophthalmology [1], but it may also be used for artworks [2, 3].

494

I. Gorczyńska et al.

Fig. 58.1. The set-up for a spectral OCT instrument: LS, light source; OI, optical isolator; DC, directional coupler; PC, polarization controller; NDF, neutral density ﬁlters; RM, reference mirror; X-Y, scanners; L, lenses; DG, diﬀraction grating; CCD, linear CCD camera

58.2 Experimental Methods The SOCT instrument is based on a ﬁber-optic Michelson interferometer setup (Fig. 58.1). The broad-band light source LS (a Broadlighter from Superlum, Russia) is based on two coupled superluminescent diode modules with slightly shifted central wavelengths. The central wavelength of the source is λcentre = 823 nm with ∆λFWHM = 74 nm. The 5 mW output power light source is launched into one of the single mode ﬁbers of the 50:50 ﬁber coupler DC through an optical isolator OI. The optical isolator separates the light source from the light back-reﬂected from the elements of the interferometer. The light is split by the coupler into two arms: the reference and object. The former is equipped with a polarization controller PC, collimator, and open air delay line with a reﬂective mirror RM kept in a ﬁxed position. The latter (an object head) consists of a collimator, transversal scanners X-Y and lenses L and L1 . The lens L1 is placed between the scanner and an object so the distances, lens-to-object and from pivot point of the scanner to the lens, are equal to the focal length of the lens. The optics produces a narrow beam of light of high spatial but low temporal coherence. This beam penetrates the object, scatters from elements of its structure, then is collected by the same optics L and L1 back to the coupler DC. Then it interferes with the light returning from the reference arm and the interference signal is directed into a custom-designed spectrometer. It consists of a volume phase holographic grating DG with 1 200 lines/mm and achromatic lens L (f = 150 mm), which focuses the spectrum on a 12-bit line scan CCD camera. The spectral fringe patterns registered by this detector are then transferred to a personal computer COMP. The resulting signal, i.e., spectral fringe patterns, is Fourier-transformed into a single line of a cross-sectional

58 Varnish Thickness Determination

495

Fig. 58.2. (a) The SOCT image of a cross-section of painting layer covered by varnish. The dark bands correspond to strong light scattering at (58.1) the airvarnish and (58.2) varnish-paint layer interfaces. The light penetrates from the top, the structures below the varnish layer are not visible due to high absorption of paint. (b) The microscopic image of the cross-section of a sample taken from the same object. The following layers are visible: varnish (v), paint layer (p), imprimatura (i), ground (g), wood (w). The bar scales are 100 µm in both directions in both images

Fig. 58.3. The SOCT cross-section image taken across a white-painted pipe

image (A-scan). In order to obtain either 2D slice (B-scan, Figs. 58.2a and 58.3) or 3D volume tomogram, the beam is scanned transversally by galvanometric scanners X-Y. The system operates in the shot-noise-limited detection mode (the intensity of light in the reference arm of interferometer is controlled by neutral density ﬁlter NDF) with sensitivity of 90 dB. The exposure time per A-scan is equal to 50 µs, thus one 2D slice (composed usually of 2 000–5 000 A-scans) is collected in a fraction of second. The numerical processing of the data, besides the Fourier transformation essential to the SOCT method, includes subtraction of noninterference background, spectral shaping [4], and numerical dispersion correction [5]. The signal may be visualized as a cross sectional view (B-scan), or converted into a variety of 3D images, like a map of varnish thickness (Fig. 58.4). In OCT generally, the axial resolution ∆z is decoupled from the lateral resolution ∆x. The ﬁrst depends on the properties of the light source: ∆z =

1 2 ln 2 λ2centre , nR π ∆λFWHM

while the second is determined by the properties of the optical head:

(58.1)

496

I. Gorczyńska et al.

Fig. 58.4. The eﬀect of refraction correction. (Left) an uncorrected image, bars indicate 0.2 mm in varnish (n = 1.5). (Right) a corrected image, bars indicate 0.2 mm of geometrical distances. The picture examined was tilted to eliminate specular reﬂection from the varnish layer

a

b 150

m]

6 8

4

4

6

6

[m

m]

8

] m

m

2

[m

[µm]

4

4 [m

0 2

]

0 2

2

100 50 0

[m

[µm]

0 −50 −100 −150

6

Fig. 58.5. The surface map of a painting layer (a) under the varnish and (b) a varnish thickness map as revealed by SOCT examination

∆x =

1 4λcentre nR π

f , d

(58.2)

where nR is the refractive index of the examined media, f is the focal length of the lens L1 , and d is the beam diameter at this lens. For the system described here: ∆z = 6 µm in a layer of varnish and ∆x = 15 µm. The power of infrared light illuminating the sample is about 600 µW. The object chosen for examination was an eighteenth century painting on wood panel (The Netherlands school). A sample for microscopic analysis of cross-section has been taken from the picture in an area of destruction.

58.3 Results and Discussion In Fig. 58.2, an example of a 2D slice obtained with SOCT is presented and compared with microscopic image of the cross-section of a sample taken nearby. The direct microscopic analysis of cross-section (Fig. 58.2b) obtained from the same picture indicates the presence of about 70 µm thick varnish layer. The same thickness is obtained from OCT measurement. The chemical assay indicated the oil-and-resin varnish, not tinted. In Fig. 58.3, a diﬀerent area of the same picture is analyzed. Here, a strong impasto of a white paint is elevated over the varnish surface. With OCT imaging it is possible to precisely access this detail of picture structure.

58 Varnish Thickness Determination

497

It is worthwhile to emphasize that all in-depth distances at OCT tomograms are optical ones. To adjust for this eﬀect, a refraction correction [6] is sometimes necessary. In Fig. 58.4, an example of this eﬀect is presented. If the surface is scanned in both (X, Y ) directions, the information on volume structure of the object becomes available. From every 2D slice, both interfaces (marked 1 and 2 in Fig. 58.2a) have been automatically recognized and then linked together into two surface maps. The lower one represents a painting layer surface under the varnish while the upper one demonstrates a ﬁnal picture surface. A point-to-point map of distances between both interfaces shows the varnish layer thickness. To prepare images presented in Fig. 58.5, a set of 54 subsequent, parallel slices (B-scans), each composed of 2 000 A-scans, was collected. Then both interfaces were recovered and varnish thickness map was calculated. In Fig. 58.5a, for clearer presentation, only the paint layer surface is shown.

58.4 Conclusions The study presented conﬁrms that, under deﬁned technical conditions, SOCT can be used for determination of varnish layer thickness. The major limitation is the axial resolution of the method, demanding use of expensive, very broad, but spatially coherent light sources. However, progress in developing multiSLD sources promises rapid reduction of their price. Then, in the future, this technique may be established as a routine method in art examination, as it is at present in medicine. Since the method is nondestructive, noninvasive, and instantaneous, it may be used in particular in combination with laser ablation devices for the on-line control of varnish removal processes. Acknowledgments This work was supported by Polish Ministry of Science with Grant 2 H01E 025 25.

References 1. J.S. Schuman, C.A. Puliaﬁto, and J.G. Fujimoto, in Optical Coherence Tomography of Ocular Diseases, 2nd edn., Slack, Thorofare, NJ, 2004. 2. P. Targowski, B. Rouba, M. Wojtkowski, and A. Kowalczyk, Stud. Conserv. 49, 107 (2004). 3. H. Liang, M.G. Cid, R.G. Cucu, G.M. Dobre, A.G. Podoleanu, J. Pedro, and D. Saunders, Opt. Express 13, 6133 (2005). 4. M. Szkulmowski, M. Wojtkowski, P. Targowski, and A. Kowalczyk, Proc. SPIE 5316, 424 (2004). 5. B. Cense, N.A. Nassif, T.C. Chen, M.C. Pierce, S-H. Yun, B.H. Park, B.E. Bouma, G.J. Tearney, and J.F. de Boer, Opt. Express 12, 2435 (2004). 6. A. Szkulmowska, M. Góra, M. Targowska, B. Rouba, D. Stifter, E. Breuer, and P. Targowski, in this volume.

59 Multidimensional Data Analysis of Scanning Laser Doppler Vibrometer Measurements: An Application to the Diagnostics of Frescos at the US Capitol ∗

J. Vignola1 , J. Bucaro1 , J. Tressler1 , D. Ellingston1 , A. Kurdila2 , G. Adams3 , B. Marchetti4 , A. Agnani4 , E. Esposito4 , and E.P. Tomasini4 1

∗ 2

3 4

Physical Acoustics Branch Code 7136, Naval Research Lab, 4555 Overlook Ave., SW Washington, D.C. 20375, USA [email protected] Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville, FL 32611-6250, USA Cunningham-Adams, 216 Third St. NE, Washington, DC 20002, USA Università Politecnica delle Marche, Dipartimento di Meccanica, Via Brecce Bianche, 60131 Ancona, Italy

Summary. A large-scale survey (∼700 m2 ) of frescoes and wall paintings was undertaken in the US Capitol Building in Washington, DC to identify regions that may need structural repair due to detachment, delamination, or other defects. A common approach for post-processing time series called Proper Orthogonal Decomposition, or POD, was adapted to frequency-domain data in order to extract the essential features of the structure. We present a POD analysis for one of these panels, pinpointing regions that have experienced severe substructural degradation.

59.1 Introduction In the 1850s the accomplished Italian artist Constantino Brumidi [1] was hired to decorate walls and ceilings of the US Capitol building with a series of frescos extolling the virtues and history of this ﬂedgling nation. While the frescos themselves have proven to be eminently durable, the same may not always be said of the brick and mortar buildings that house them. To accurately assess the extent of a possible structural plaster damage, the Physical Acoustics Branch of the US Naval Research Laboratory in cooperation with other partners carried out a broad survey (∼700 m2 ) of the frescos and wall paintings in the US Capitol Building in 2003 using a Scanning Laser Doppler Vibrometer (SLDV), which measured the surface vibration in response to broadband acoustic or mechanical excitation applied to the structure. Areas of high mobility tend to correlate with the presence of underlying faults, and the direct observation of these high velocity areas has been useful

500

J. Vignola et al.

Laser Beam LDV

Data Aquisition Computer

Shaker Art Work Trigger Source Wave Form Generator Power Amplifier LDV Controller

Fig. 59.1. SLDV experimental setup

for such assessments. In this paper we examine the use of a proper orthogonal decomposition (POD) technique applied to the SLDV data. The objective is to provide a more sophisticated, objective approach than direct visualization of the velocity scans for identifying fault areas from the large data sets. The POD analysis is expected to reduce the broadband data to a minimal set that accurately represents the dominant structural response, and to do so without relying on subjective judgments. In addition, before applying the POD technique, we implement a denoising algorithm. The SLDV technique on frescos was ﬁrst employed using acoustic speaker excitation by the Mechanical and Thermal Measurement Group at the Università Politecnica delle Marche in Ancona, Italy [2]. This SLDV technique was implemented in the Capitol using both acoustic and electromechanical excitation. The survey encompassed eight preselected spaces where roughly 60% of the area surveyed was domed or vaulted ceilings, the rest being walls. Approximately 250 scans were done ranging in size from 1 to 4 m2 . The typical mesh density was 400 scan points per square meter.

59.2 Instrumentation The SLDV apparatus consists of two parts: an excitation mechanism and a measurement system, see Fig. 59.1. Excitation may be accomplished either acoustically by means of a speaker or electromechanically by means of a point shaker, the latter the preferred source, the former useful when the artwork is inaccessible, e.g., on the ceiling, or known to be fragile. A linear FM chirp was used for both acoustic and mechanical excitation with a band extending from just under 100 Hz to 1 kHz. In the acoustic method, the chirp excita-

This page intentionally blank

59 Multidimensional Data Analysis of SLDV Measurements

501

y ∆f = 2 Hz

450 bin

s

50 bins Excitation Ba

nd

nF = 50 X

100 Hz 200 Hz 298 Hz frequency 1000 Hz 1600 Hz

Fig. 59.2. Graphic depiction of a 3D data array

Fig. 59.3. Magnitude of the force-normalized velocity decomposed by frequency. The scale on the right is normalized velocity in units of dB referenced to 1 m s−1 N−1

tion repeatedly insoniﬁes the artwork by a set of loudspeakers operating at ∼90 dB. Because they have greater mobility, defects respond with greater velocity amplitudes than typical intact regions of the wall, especially at particular resonant frequencies. The SLDV instrument is used to measure these velocities, allowing precise identiﬁcation of regions of suspected instability in the structure. In the mechanical method, the artwork is excited electromechanically by a locally placed actuator rather than by a sound wave. This is done by means of a shaker with a Teﬂon-coated tip that is pressed against the fresco and introduces mechanical energy directly into the wall without the

502

J. Vignola et al.

intervening medium of air. The main advantages over the loudspeaker method are that it transfers energy more eﬃciently into the artwork and it is less disruptive to the surrounding workplace. Velocity data was collected by means of a Polytec SLDV using a laser emitting ∼1 mW of power at wavelength 6,328 Å, harmless for artworks and humans alike. Velocities are derived from the measured Doppler shifts of light rays reﬂected oﬀ the vibrating surface. Once the velocity data has been collected it is normalized to a reference signal, a force cell response in the case of shaker excitation or the local acoustic pressure for the case of the speaker excitation. The normalized data has dimensions of m s−1 N−1 for the former or m s−1 Pa−1 for the latter. This normalized velocity data then undergoes a further post-processing. The result is a graphic overlay that may be superimposed on an image of the artwork to give a precise map of the location of structural defects. In this paper “defects” refers to all manner of cracks, delaminations, detachments, and inhomogeneities. See Drain for further reading in the SLDV technique [3].

59.3 Data The data in an SLDV scan ﬁle consist of a set of complex numbers, representing force-normalized velocities, in a three-dimensional array. The ﬁrst two dimensions are the spatial coordinates, labeled m and n. The third coordinate is frequency, labeled k. Explicitly, the SLDV assigns to each point in {m, n, k}space a velocity amplitude V(m,n),k . The subscript “(m, n), k” emphasizes that the velocity is associated with physical grid point (m, n) and excitation frequency k. Fig. 59.2 shows the array of data graphically. In Fig. 59.3, we take frequency sections of the data array and display them serially as a mosaic. Each individual data frame in the mosaic displays the characteristic response, within a 2 Hz frequency bin, of each point scanned on the artwork. Direct observation of these narrowband velocity maps obtained through Fourier decomposition can be used by one skilled in the application of this technique to identify potential fault areas. By comparing diﬀerent data frames one can identify locations where the velocity response is persistently high across several frequencies. One practical method for accomplishing this is to compute the average over a number of frames of the product VV∗ where V∗ is the complex conjugate of the velocity amplitude [4]. With either method, the persistent locations over a ﬁnite bandwidth of unusually high mobility typically correspond to points where the substructure may have been compromised. 59.3.1 Proper Orthogonal Decomposition As an alternative to Fourier decomposition, we use a technique called proper orthogonal decomposition (POD) [5, 6]. The POD technique generates eigenvalues in descending order, from most energetic to least, that correspond to eigenfunctions constitutes of the original frequency domain data. A subset of

59 Multidimensional Data Analysis of SLDV Measurements

503

the associated dominant eigenfunctions are singled out, displaying the significant behavior of the original vibration data set. → − To apply the POD technique to our data, let us suppose that { V k |1 ≤ k ≤ nF } represents a stochastic data set, where the index k refers to frames of data. Each of the frames is the collection of scanned velocity data indexed by the frequency bin, where nF is the total number of such bins. Each of the random variables V k is taken to be an nX × 1 column vector whose individual components are scalar random variables denoted Vj,k , where 1 ≤ j ≤ nX and 1 ≤ k ≤ nF . The covariance matrix Ci,j is given by n F (Vi,k − E (Vi )) (Vj,k − E (Vj )) /nF . (59.1) Ci,j = k=1

In this expression, E(V ) is the expected value of the stochastic process {V k }, hence it is an nX × 1 column vector. 59.3.2 Two-Dimensional Spatial Data: Mapping Frequency Bin Scans to Vector Random Variables The next step is to map the two-dimensional arrays obtained from the SLDV scans, which are indexed by frequency bin, to the framework outlined earlier. Suppose that for each frequency bin k ∈ {1, 2, 3, . . ., nF }, we have a twodimensional array Ak whose entries are denoted A(m,n),k , where m is the row index and n is the column index of the array. It is assumed 1 ≤ m ≤ nR and 1 ≤ n ≤ nC , where nR and nC are the numbers of rows and columns, respectively, in the array. In our case these two indices correspond to a twodimensional spatial location on a two-dimensional grid. By unpacking the two-dimensional arrays Ak (in either row major or column major format), we deﬁne the vector of random variables V k . Introduce the correspondences Vj,k ⇔ A(m,n),k , j ⇔ m × n,

(59.2) (59.3)

and deﬁne nX = nR × nC . The covariance matrix deﬁned in the last section then takes the form n F

A(m,n),k − E A(m,n) V(s,t),k − E A(s,t) nF . C(m,n),(s,t) = k=1

(59.4) Once the covariance matrix has been computed, it can be used to identify coherent patterns appearing in the collection of two-dimensional arrays {Ak |1 ≤ k ≤ nF }. To accomplish this, ﬁrst the nX × nX eigenvalue problem {C − λk I} Ψk = 0,

(59.5)

504

J. Vignola et al.

is solved for nX eigenvalues and eigenvectors {(λk , Ψk )|k = 1 . . . nX }. Again, each eigenvector will have length nX and there will be nX real eigenvalues, since the covariance matrix is real and symmetric. A physical interpretation of the one-dimensional eigenvector Ψk is achieved by renumbering the entries into matrices Φk in exactly the same fashion that the one-dimensional vectors V k are associated with two-dimensional arrays Ak . Thus, we deﬁne the correspondences Ψj,k ⇔ Φ(m,n),k , j ⇔ (m, n).

(59.6) (59.7)

Maps of coherent image structures are obtained by ordering the eigenvectors according to the magnitude of their corresponding eigenvalues. The contribution of a particular eigenvector-image to the subspace spanned by all nF frames of data is directly related to the magnitude of its corresponding eigenvalue. Thus, for some small integer nP of kept modes that satisfy nP ≥ nX and nP ≤ nF , we retain the images { Φk | k = 1 . . . nP } ⊂ { Φk | k = 1 . . . nX }

(59.8)

as good representations of the coherent features in our original data set. The set {Φk |k = 1. . .nP } forms an optimal basis that nearly spans the space of all modes in the original data set. Moreover, the size of the eigenvalue provides an indication of the relative strength of that eigenfunction. In practice we ﬁnd that nP is seldom more than two or three. 59.3.3 Denoising It is well known that noise levels vary signiﬁcantly over diﬀerent locations in an SLDV measurement. This is due to random phase and amplitude ﬂuctuations in the back-scattered light known as speckle [3], which cause some locations to be much noisier than the rest. To address this problem, prior to using the POD algorithm, the SLDV data is subjected to a denoising algorithm. This algorithm uses the portion of the power density spectrum at a location falling above the excitation band to infer the noise level within the measurement band. If the out-of-band signal is unusually high, the data at that point is replaced with spatially averaged data from the adjacent spatial locations.

59.4 Results This section summarizes the performance of combining the denoising and POD methodologies for identifying coherent patches of high mobility in plaster frescos that we scanned. As an example we will examine the data associated with the North Lunette fresco in the House Appropriations Committee Room (Brumidi’s ﬁrst work in the Capitol, painted circa 1855.) Figure 59.4a depicts one

59 Multidimensional Data Analysis of SLDV Measurements

505

Fig. 59.4. North Lunette fresco overlay of the result of the VV∗ algorithm (a) without denoising and (b) with denoising

Fig. 59.5. First four POD modes derived from 50 frequency data frames, each of width 2 Hz, from 200 to 298 Hz

of the surface scan graphics (the lighter region) overlaid on a photograph of the fresco. This surface plot is obtained by averaging the square of the magnitude of the normalized velocities (referred to as VV∗ [4]) over a collection of frames from the SLDV survey. In the overlay graphic we see a large region of high mobility (irregular, darkened area) in the lower right corner, which we have found to be closely correlated to data collected by an art conservator in tap tests. The graphic is generated by averaging the normalized velocity magnitudes and interpolating the result to obtain a smooth representation. No denoising has been used in this post-analysis. Figure 59.4b depicts another graphic overlay obtained from the same SLDV survey data over the North Lunette Fresco. In this case the denoising technique described in the previous section has been applied to the SLDV survey data to eliminate anomalous spikes in the spatial plots. In contrast to Fig. 59.4a, the “ringing” or oscillatory phenomenon around local peaks is virtually eliminated using this method. Figures 59.5a–d depict the ﬁrst four POD modes derived from the same sequence of frequency frames. Note that the gray scales of these diagrams are dictated naturally by the range of data values. So, for example, the blackened frame in the left column portrays a very narrow range of values with a small peak along the right edge. Thus we can deduce the energy of a given POD mode by simply inspecting the dynamic range of values. We see that, in this

506

J. Vignola et al.

case, only one POD mode would suﬃce to identify the signiﬁcant coherent structure in this sequence of SLDV survey frames.

59.5 Conclusions The SLDV technique is already well-established as an eﬀective approach to art conservation work. The US Capitol plaster assessment project provides a useful proving ground for new SLDV analyses that we have made to improve the reliability of the ﬁndings of the scans and remove subjectivity from the analysis. The laser apparatus is nonintrusive and safe for both operator and subject. It is easily portable and provides objective results that are reproducible. This allows for baseline studies that can chart progressive changes in the artwork over time, e.g., before-and-after studies and studies on the eﬃcacy of restoration eﬀorts. Our study here focuses on innovations in data analysis. Speciﬁcally, we ﬁnd that the preliminary use of a denoising algorithm reduces the subsequent number of false positives in either the VV∗ or POD algorithms. We also ﬁnd POD to be a useful tool in reducing a large array of broad-band data down to a very compact representation, often consisting of a single eigenfunction. This is achieved by drastically reducing the basis set of eigenfunctions while highlighting the prominent features of the data. In addition, the POD algorithm showed itself capable of removing subjective judgement from this process. In particular, the gray scale of all POD plots is determined naturally by the range of values in the data frame. Finally, we have found that POD consistently produces results that are in good agreement with tap tests performed by experienced art conservators. Acknowledgements This work was partially supported by the Oﬃce of Naval Research.

References 1. B.A. Wolanin, Constantino Brumidi, Artist of the Capitol, U.S. Government Printing Oﬃce, Washington DC, 1998. 2. P. Castellini, N. Paone, and E.P. Tomasini, in Proceedings of the First International Conference on Vibration Measurements by Laser Techniques: Advances and Applications, SPIE 2358, 70, 1994. 3. C.B. Scruby and L.E. Drain, in Laser Ultrasonics, Adam Hilger, Bristol, 1990. 4. J.F. Vignola, J.A. Bucaro, B.R. Lemon et al., “Shaker-based laser Doppler approach for locating delaminations in wall paintings at the United States Capitol”, APT Bulletin: The Journal of Preservation Technology, 2005. 5. J.L. Lumley, in Coherent Structures in Turbulence, Ed. by R.E. Meyer, Academic Press, 1981, 215. 6. P. Holmes, J.L. Lumley, and G. Berkooz, Turbulence, Coherent Structures, Dynamical Systems and Symmetry, University Press, Cambridge, 1996.

60 Spectral Domain Optical Coherence Tomography as the Proﬁlometric Tool for Examination of the Environmental Inﬂuence on Paintings on Canvas ∗

T. Bajraszewski1 , I. Gorczyńska1 , B. Rouba2 , and P. Targowski1 1

∗ 2

Institute of Physics, Nicolaus Copernicus University, ul. Grudziądzka 5, 87-100 Toruń, Poland [email protected] Institute for the Study, Restoration and Conservation of Cultural Heritage, Nicolaus Copernicus University, ul. Gagarina 9, 87-100 Toruń, Poland

Summary. The changes of surface proﬁle of oil paintings on canvas caused by rapid humidity are examined by spectral OCT. The speed, resolution, and long-time stability of SOCT make this technique appropriate for quantitative determination of surface proﬁle changes.

60.1 Introduction Oil paintings on canvas are exposed to periodic changes of air humidity and moisture exchange with the surrounding air, which can cause deformations of canvas support and lead to progressive destruction of the paint layer. The knowledge about the range and dynamics of these processes is important for preservation of artworks. Therefore, there is a need for precise proﬁlometry of the painting surface. Optical methods are an obvious choice because they are noncontact and noninvasive. Since the deformations to be traced are relatively small and tend to develop during a long period of time, the proﬁlometric method needs to be insensitive to the displacements of the whole object between the sequential images. Thus, all methods based on mutual interference between optical waves generated at original and altered surfaces may not be practical since they require object stability within a fraction of wavelength. Also, if alterations to the object surface are big, distances between optical fringes become very small and diﬃcult to resolve. Therefore, it is desirable to use a method that avoids double exposures separated by a considerable time period. Additionally, the chosen method should be fast enough to record the surface proﬁle in seconds to ensure that obtained image will represent a snap-shot, not blurred by changes in object structure during the

508

T. Bajraszewski et al.

data acquisition. Optical coherence tomography (OCT), originally designed for recovering cross-sectional images of semi-transparent objects, especially in ophthalmology, may be also used for surface proﬁlometry [1]. In this contribution we demonstrate the method to isolate the shape of the object surface from the three-dimensional OCT data.

60.2 Experimental Methods 60.2.1 The Instrument Optical Coherence Tomography is a noncontact and noninvasive technique of recovering cross-sectional views of semi-transparent objects. In this method the position of the scattering interface along the penetrating beam is obtained from the interference of the scattered light with the light propagating in the reference arm of the Michelson interferometer. The detailed description of the instrument used in this study is given elsewhere [2]. If OCT is used as a proﬁlometric tool, only the position of the air–paint interface is recovered. In this case the method is not limited to transparent medium. The penetrating beam of infrared light (with power of about 600 µW) scans the examined surface two-directionally to sample it at 424 × 450 points. To collect such amount of data in a reasonable time, a spectral version of OCT is particularly well suited because acquisition time of getting information about one surface point is as short as 50 µs. The whole 3D tomogram contains a sharp image of air–paint interface (picture surface) followed by diﬀused image of underneath layers [3]. Then the position of the surface is precisely retrieved by cross-correlation with a gauss-shape model function. In Optical Coherence Tomography, axial and lateral resolutions are decoupled. The axial resolution depends on the spectral width of utilized light, usually produced by a superluminescent source (λcentre = 823 nm, ∆λ = 70 nm, ∆z = 6 µm). The lateral resolution, deﬁned as size of the light spot at examined surface, depends mostly on numerical aperture of the lens used in object arm of the Michelson interferometer [4]. In our instrument this spot size was about 15 µm. The lateral scanner step was 2.2 µm. Due to oversampling, the quality of the image increases because of signal-to-noise ratio improvement. 60.2.2 The Sample The aim of the study was to investigate the applicability of SOCT to monitor changes on surface of an oil painting on canvas caused by a rapid humidity growth. For the experiment a sample of an oil painting on linen canvas was chosen (Fig. 60.1). This painting sized with animal glue and primed with emulsion ground was specially made for the purpose of further experimentation twenty years ago. The metal frame was used to stabilize the sample. The sample

60 Spectral Domain OCT as the Proﬁlometric Tool

509

Fig. 60.1. The surface of oil painting on canvas used in the study. Rectangular (1 × 1 mm2 ) marks a scanned area

was dried under moderate vacuum at room temperature and then placed in a chamber where the relative humidity was rapidly increased to 100%. Due to the way of mounting the water vapor could access the sample from the front side only. The SOCT data were collected every 1.3 minute during 60 minutes after the humidity jump.

60.3 Results and Discussion In Fig. 60.2a a perspective view of surface proﬁle (near the crack) of the sample of oil painting on linen canvas is presented. Similar data may be obtained in altered environmental conditions, for example, after a humidity jump (Fig. 60.2b) and directly compared with the initial one. This kind of study may be performed in a laboratory, using a climate chamber or in situ. Because the method is fast, it is also possible to trace changes of the surface shape as a function of time after humidiﬁcation or other environmental changes. The obvious ﬁnding is a translation of whole painting surface about 170 µm towards the front side. However, more detailed analysis is also possible. As an example one may consider the examination of behavior of surface along some arbitrarily chosen direction. In Fig. 60.3a such direction is marked with a dash-dot line. In Fig. 60.3b a set of 2D proﬁles along this line is presented. First proﬁle is obtained just before a rapid increase of relative air humidity to 100%. Next proﬁles are shown for time intervals of 6.5 min. The change of surface position is again evident. Moreover, one may notice no signiﬁcant change in

510

T. Bajraszewski et al.

Fig. 60.2. Direct comparison of the same area of sample from Fig. 60.1 before (panel a) and after (panel b) 1 h exposition to 100% relative humidity

260 600 500 400 300 200 100

0,2

Y [mm]

0,3 0,4 0,5 0,6 0,7

c

250 240

Width [µm]

a

0,1

Z [µm]

0,0

230 220 210

0,8 200

0,9 1,0 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0

190 0,0

900

13,0

26,0

39,0

52,0

65,0

Time [min]

X [mm]

b

800

Z [µm]

700 0 min 6.5 min 13 min 19.5 min 26 min 32.5 min 39 min 45.5 min 52 min 58.5 min

600 500 400 300 200 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1

[mm]

Fig. 60.3. (a) Elevation map of the sample surface, a dash-dot line indicates location of the examined proﬁle; (b) proﬁles along a line shown in the panel (a) measured after a humidity jump at indicated times; c: changes in time of the width of the crack shown in panel (b), measured at half depth

60 Spectral Domain OCT as the Proﬁlometric Tool

511

a surface shape, namely a lack of additional elevation of the crack edges. However, detailed inspection of the crack width, shown in Fig. 60.3c, reveals its signiﬁcant decrease.

60.4 Conclusions As it follows from the example presented, Spectral OCT may be used for proﬁlometry of painting surfaces. The method is fast enough and has necessary lateral and longitudinal resolution. Since every surface is acquired separately and does not relay on phase information, it is not necessary to maintain all distances within a set-up with an optical wavelength precision. This permits this technique to be used in place of artwork exhibition to monitor the inﬂuence of its usual environment. Acknowledgements This work was supported by Polish Ministry of Science with Grant 2 H01E 025 25.

References 1. P. Targowski, B. Rouba, M. Wojtkowski, and A. Kowalczyk, in Studies in Conservation, Vol. 49, 107, 2004. 2. I. Gorczynska, M. Wojtkowski, M. Szkulmowski, T. Bajraszewski, B. Rouba, A. Kowalczyk, and P. Targowski, in this volume. 3. A. Szkulmowska, M. Góra, M. Targowska, B. Rouba, D. Stifter, E. Breuer, and P. Targowski, in this volume. 4. P. H. Tomlins and R. K. Wang, in J. Phys. D; Appl. Phys., Vol. 38, 2519, 2005.

61 Polish Experience with Advanced Digital Heritage Recording Methodology, including 3D Laser Scanning, CAD, and GIS Application, as the Most Accurate and Flexible Response for Archaeology and Conservation Needs at Jan III Sobieski’s Residence in Wilanów P. Baranowski1 , K. Czajkowski2 , M. Gładki2 , T. Morysiński2 , R. Szambelan3 , and A. Rzonca4 1 2 3 4

www.palac-wilanow.art.pl www.kobidz.pl www.asp.waw.pl www.dephos.com

Summary. Review of recent critical points for introduction of laser technology into the ﬁeld of heritage documentation, management, conservation, and archaeology will be discussed. The relationship of beneﬁt versus cost of 3D laser scanning technique for complex multitask heritage recording project at Wilanow is presented. Deﬁnition of basic criteria for the successful use of such heritage detailed record as laser scanning is given.

61.1 Experience Collected During Research and Conservation Works at the Palace in Wilanów The royal summer residence in Wilanów is a palace and garden complex consisting of 40 ha of park and buildings area, among them the main building with a decorated façade. The site preserves numerous historic work of art collections, mostly hosted by the unique decorated interiors from eighteenth and nineteenth century. Recently, a program was launched for reconstructing the Baroque appearance (“rebaroquization”) of the Palace and its surroundings with the objective to emphasize the original look of the preserved parts and remains of the historic architecture in the context of their development during the subsequent epochs. Additionally, there is a strong tendency developing not to forget the potentially positive aspects of future social impact of this project.

514

P. Baranowski et al.

The alarming state of preservation of some important parts of the historic complex led to the formulation of a general conservation and renovation plan. Important factors in decision making here seem to be connected with Poland joining the European Community. This enabled the local government institutions to obtain ﬁnancial support for the conservation program’s eligible costs. There are three partner institutions involved in the conservation and research process at this site: – Museum Palace in Wilanów (Muzeum Pałac w Wilanowie) – National Center of Research and Documentation of Historical Monuments (Krajowy Ośrodek Badań i Dokumentacji Zabytków or KOBiDZ) – Interacademy Institute of Conservation and Restoration of Art (IAICR) (Międzyuczelniany Instytut Konserwacji I Restauracji Dzieł Sztuki) Additionally there is a cooperation between IAICR and DEPHOS (a Company from Cracow specialized in recording of cultural heritage), which was responsible for preliminary work of preparation of base maps for mappings of the façades. In 2003, a conservation program commenced after initial continuous work on developing the ﬁeld of investigation. During the restoration works, more than half a thousand of decorative elements on the facades like sculptures, reliefs, portals, window borders, and so forth were found to be in need of treatment by art restorers according to their state of preservation, which should be previously registered in a form of digital documentation. Such an uncommon documentation eﬀort requires prudent selection of methods and optimization of the selected technique. All this has obligated us to achieve an important goal, namely, to be as eﬀective as possible with the use of rather limited funds.

61.2 Accurate and Flexible Response for Archaeology and Conservation Needs All the managers know of the vanishing miracles in the vortex of the Bermuda Triangle: GOOD, CHEAP, and FAST. The dream of a perfectly executed work disappears if one suddenly considers that only two of the three positive parameters will be the characteristic features of his ﬁnished creation. The same is valid for digital recording. Depending on the necessities and constrains, regardless of strong ambition, one is able to produce only documentation, which cannot be three in one, even in laser scanning. But on the other hand, there is no other technique providing us so quickly with such an accurate and densely recorded data of spatial information. It is a state-of-the-art technique. The reason for choosing an expensive method is our hope that the result of the work will be good and fast to obtain. Accuracy and speed are the

This page intentionally blank

61 Polish Experience with Advanced Digital Heritage Recording

515

indisputable parameters of laser recording. Lucky those, who with a speed of light were able to obtain appropriate subsidies and turn a dream into reality. The accuracy, which is related to mentioned “GOOD” parameter, can be outlined as follows: – Through the choice and description of instruments used, their parameters, and performance – Properly organized plan for the shots taken during a measuring session, for which in our case the geodesy specialists from DEPHOS Company have been responsible – Preliminary geodesic survey not omitted prior to laser scanning; – And last, but not least, the density of the record, which is able to cover the visible surface of an object with a points grid and, if repeatedly scanned, the grid precision can go up to a certain limit and, of course, the shadowed areas will be ﬁlled Enormous ﬂexibility can be achieved through the following: – A choice of instruments designated for various ranges of the recording scale – Short duration of a recording session at the site. Easy to organize power supply and on-site preparation works – The instrument can be operated under almost any weather conditions perhaps with the exclusion of only heavy rain or snow – Ability to precisely determine suﬃcient amount of work Number of shotstakes, their time, and post recording edition in stages – Ability of post-recorded CAD editing to meet a client’s speciﬁc needs The ﬂexibility ﬁeld is also signiﬁcantly expanded with the following: – Interoperability of data set in a software environment. The ability of linking spatial geometry with digital photo images as well as the usefulness of laser recorded base map for digital surface condition assessment with CAD and GIS. – Editing operation and data manipulation, which can be transferred to an oﬃce – Modern data storage procedures regarding their future usefulness – Postponed process of completing the maps up to the ﬁnal printing Also some organizational improvements should be mention such as the following: – The possibility of diverting the conservators and other specialists back to their work, relieving them from time-consuming preparation of base maps – Possibility to work in an out-source cooperative mode with a specialized team responsible only for measurements or the qualiﬁed in-house persons who are able and delegated to operate the scanner.

516

P. Baranowski et al.

Elaborated information from laser scanning provides us with data which are itself a spatial skeleton to input other information, a digital structure which is able to match the traditional form of photo and drawing documentation, providing it in a complementary additional value of exactness. Properly constructed methodology including laser scanning, together with CAD and GIS application, enables a continuous collection of relevant virtual data coming in from the responsible restorers and archeology investigators. Since the time lapse is often a critical point on the scaﬀolding or on the excavation site, the preparation for mapping procedures in documentation process should be performed as quickly as possible. This can also be a point where introducing laser scanning helps to eliminate obstacles and work stoppage causing incohesive investigation. There are of course security reasons and some speciﬁc site diﬃculties that should be taken into account as they can prevent the recording at times from being optimal in terms of completeness and precision. In our experience, the process of generating base maps for digital mapping during condition recording has been signiﬁcantly shortened, through some improvements based on the use of specialized software originally owned by DEPHOS Company. Recording a surface shape with laser scanning produces point clouds, after that an adequate approach model of a chosen building part being of a particular interest has had to be formed. In this instance, for the ﬁrst approach to the architectural scale record of the Palace’s façade, it was enough to obtain a front view – a 2D digital image in orthogonal projection, based on the collected information from several scanner shotstakes. The basic accuracy of the image obtained equals 1 pixel = 2 mm. For practical reasons, because of the capacities of the hardware used, the original accuracy has been delimited to 4–8 mm pixel−1 during the editing process. Instead of a considerable eﬀort put into the preparation of the vector data through photogrammetry, just in the early phase of documentation, one can automatically generate a support image for overlays, which is originated directly from the information in the point cloud. The image can be produced in two ways: 1. Using the reﬂection parameter of a laser shot during measurement 2. Transforming part of the original point cloud with its coordinate data into a hypsometric image in a local CAD coordinate system Both methods will provide correctly prepared 2D ortho-picture in terms of measurement correctness. In some ways the synthetic images will show visually signiﬁcant visible signiﬁcant motif details. It is only the scope of details one would like to concentrate on which determines which picture has to be chosen.

61 Polish Experience with Advanced Digital Heritage Recording

517

61.3 Technical Notes For the archaeology investigation purposes, the documentation is prepared using Multitask Systems of Documentation and Information elaborated in National Centre of Research and Documentation of Monuments (KOBiDZ ). The systems consist of 3D laser scanning, close-range photogrammetry, and CAD drawings. For scanning, they use laser scanner Cyrax 2500 with Cyclone software (modules: Scan, Register, Model). The scanner is supported by laptop Fujitsu Siemens P4 with Cyclone 5.1 software. Works on 3D models are prepared using desktops PC (P4 2.6 GHz) with the Autodesk Land Desktop software and Autodesk CloudWorks upgrades. Also the CAD drawings, cuts, and sections are prepared in this environment. Additionally, depending on needs, the photogrammetry records are being prepared on the basis of object’s photographs. The photographs are usually taken with a Canon EOS 20D camera with 8 megapixel CCD, with 18–55 mm lens. Supporting photo documentation is made using a camera series of Canon Powershot Pro 1, G6, G3, and G2. The ortho-photos are created according to the current needs using three independent applications – PhotoModeler 5 Pro, ImaGold i, and WiseImage for AutoCad. The color scheme, white balance, and mosaic ortho-photographs are made in Adobe Creative Suite environment. The second group of heritage recording specialists (DEPHOS Company) have worked on the architecture measurements equipped in another set of instruments. During the ﬁrst session of laser scanning for the Wilanów façade at the southern wing of the palace, the surface recording was executed with the use of Z+F Imager 5003 − 53500. ShotsTakes have been executed on a special scaﬀolding ca. 3.5 m above the ground. The control was measured by reﬂectorless total station Trimble 3305 DR. All point clouds were ﬁltered in Z+F software and orientated in a common coordinate system using the original DEPHOS-owned software. All reﬂectometric ortho-images were generated after additional ﬁltering in DEPHOS Mapper software process, especially created in-house for the elaboration of point clouds for 3D conservation documentation purposes. Other products like vector ﬁles and hypsometry images were also obtained, thanks to DEPHOS specialized software. Figure 61.1 shows a result of laser scanning transformed into 2D orthopicture based on reﬂectometric parameter (intensity of backscattered light) recorded during the measurements on the upper part of the north façade of south wing in the Palace in Wilanów. Figure 61.2 shows examples of the recorded data. It depicts a 2D gray-scale image of reﬂectometry as a base map, for example, of damage mapping on the façade compatible with existing archive documentation material on paper. It also shows an overlay with observed and manually registered surface damages (cracks layer), all positioned together in a common coordinate system and organized as a digital map in GIS (Map Info Corp.)

518

P. Baranowski et al.

Fig. 61.1. Museum Palace in Wilanów, southern wing façade, laser scanning. Bottom left: 2D gray scale reﬂectometry ortho-image based on points cloud from laser scanning (DEPHOS). Right: hypsometric ortho-images – color and gray scale

Fig. 61.2. Selected part of the recorded data

61 Polish Experience with Advanced Digital Heritage Recording

519

61.4 As an Example of Executed Works, a Detailed Description of Geodetic Works Conducted in Wilanow, Northern Front Façade of the South Palace Wing (2005) The objective of geodetic works carried out in Wilanów by the DEPHOS Company team was to generate orthogrametric images of the southern wing’s north façade of the Royal Palace in Wilanów. The applied scanning technology allows us to create orthophotograms in gray scale of the view shots of the façade. The completed materials constitute a base mapmap backing for conservation inventory. The work encompasses a general scan of the mortar surface on the entire façade for the objective of a later vector work and scanning of all decorative details indicated by the art conservationist: reliefs, sculptures, portal decorations. The ﬁnal product is a black and white orthophotogram of the entire façade, the resolution or picture density of which increases in the areas of the location of the decorations, thanks to the input of information obtained during the detailed scanning process. 61.4.1 Method Description The scanner applied during the above said works, Z+F LARA 53500, generates during the measurement a point cloud ﬁle, in which aside of the point placement in space, its fourth coordinate of the returning signal (intensity) is saved. Its value depends on the distance from which the object is being scanned, type of material the object is made of, and the angle at which the laser beam falls during the measurement. Also signiﬁcant is the lighting of the building or object and its temperature. Thanks to the recording of this information, a picture of the point cloud can be generated in which each pixel is a point of the said cloud and its shade from 0 to 255 according to the value of the reﬂection intensity. Such an image may play a role of an ortho-photomap and constitutes the map’s basis, which is successfully used by art conservationists in creating databases of the condition of an object before and after restoration. 61.4.2 Field Work Three basic actions may be distinguished within the framework of ﬁeldwork related to laser scanning. Prior to the measurements, the shot plan of laser scanning takes has been developed by the DEPHOS team. The conservationist’s expectations were discussed. The scanner stations were established based on dimensions of the object and scanner characteristics. During the next step, the control points were marked with black and white chessboards, selecting their placement in

520

P. Baranowski et al.

such a way that each position of the scanner permitted registering of at least six marks. The remaining two activities, the geodetic measurement and scanning, were done simultaneously. A TRIMBLE 3305 DR Series reﬂectorless tachimeter (without a prism) was used to measure the coordinates of the scanned control points. All 16 control points (8 at a height of approx. 1 m above ground, 8 at a height of 5 m) were measured from two ends of the parallel base to the façade. Also a measurement of two points of the existing control was taken with the objective of enabling a possible transformation of the local base layout to the layout of the already existing control on the building. The following pattern describes the conducted scanning process: First full horizon from the palace courtyard was scanned. This was followed by three general scenes for the purpose of registering the entire façade using high resolution. Locations of the scanner placed on a tripod on the ground have been selected in such a manner so as to minimize the number of blind spots and shadows also, so that the distance and the angle of the laser beam falling on the surface of the façade plaster was optimal. Then measurement of details using super high scanning resolution was conducted. To ensure the product’s average resolution to be at a level of 2 mm, the details were scanned from a distance of 5 m. First 8 scenes were scanned from the ground encompassing sculptures in niches and busts of emperors. After that 7 scans were taken by an instrument standing on a tripod placed on an approx. 3 m high scaﬀold. The objective of the above was the registration and capture of details of the portal and the relief on the attic above the portal (ﬁrst of 7 scans) and six semicircular reliefs over the windows. 61.4.3 Elaboration of the Collected Data Working out the measurement results was conducted as follows: The coordinates of the scanner matrix points in the local base layout were calculated. From both locations, the points location was marked using the polar method. Next, the obtained results were averaged out from the calculations of both location positions. Deviations from the XY coordinate’s average oscillated around 1 mm, in one case the deviation totalled 2 mm. Deviations of the set heights did not exceed 0.5 mm. Based on the results’ compatibility, one may estimate the precision of the appointed point location below ±2 mm and, therefore, less then the desired resolution of the photomap. Editing work on the scanner data was executed in several stages: The ﬁrst consists of ﬁltering point clouds in software made available by the scanner’s producer, just to remove points being the scanner’s instrument errors. (The objective of the said process is the removal of points from the cloud which are the result of the scanner’s instrument errors). Next, clouds in an ASCI format are imported to the DEPHOS Mapper software of the SCANVIEW module. The said module can transform and edit point clouds. The following stage in processing the point clouds was their orientation – transformation to a layout

61 Polish Experience with Advanced Digital Heritage Recording

521

deﬁned by 16 control points. Each scene was oriented by ﬁve control points or more. The obtained ﬁtting errors maximally neared ±3 mm. Next, the point clouds were subjected to editing. Each cloud encompassed a given area of the object of the work but also contained a great deal of expandable information exceeding the said area (e.g., view of the interior measured through window panes, ground). Also a view of the work object seen from a distance and from a very sharp angle is registered and not always needed. Such information is also unnecessary and is deleted because the scanner’s positions are spaced in such a way that almost every area of the object was scanned as orthogonally as possible and from a distance which ensured an appropriate resolution of the ﬁnal product. The jointly prepared mutually supplementary point clouds are projected simultaneously in the DEPHOS Mapper program’s front view and, by a special command, screen shots are then done. After making a mosaic of the said shots, a scanner ortho-photomap is obtained. The ﬁnal product obtained within the work area has a ﬁeld resolution within 1.5–3 mm range. The remaining elements of the building exceeding the scope of the work (attic) have been presented with a resolution of approx. 10 mm. The precision of the ortho-photomap’s pixel placement should be treated at a level of ±3 mm.

61.5 Conclusions Laser images in architectural recording means time saved spare time and improve the precision of the record. It costs as much as the usually executed tiresome photogrammetric routine and so, at this point, the methods are comparable but the advantages of laser images are the following: 1. The density of recorded data (they can be widely used in post-recording process, e.g., for the positioning and calibration of photos, vectorizing, and rendering) 2. The speed of work, which shortens the time usually spent for photogrammetric drawing The qualities of such visual approach are suﬃcient for immediately using the picture as support for manual damage mapping overlays. Acknowledgements We would like to thank Dyr. Pawel Jaskanis and Prof. Andrzej Koss for intensive support and introduction of new technologies into documentation works executed in Wilanów.

62 Evaluation by Laser Micro-Proﬁlometry of Morphological Changes Induced on Stone Materials by Laser Cleaning C. Colombo1 , C. Daﬀara2 , R. Fontana2 , M.Ch. Gambino2 , M. Mastroianni2 , ∗ E. Pampaloni2 , M. Realini1 , and A. Sansonetti1 1

∗ 2

CNR-ICVBC, Istituto per la Conservazione e la Valorizzazione dei Beni Culturali, sezione di Milano “Gino Bozza”, Piazza Leonardo da Vinci 32, 20133 Milano, Italy [email protected] CNR-INOA, Istituto Nazionale di Ottica Applicata, Largo E. Fermi 6, 50125 Firenze, Italy

Summary. This work, aimed at quantifying the morphological variations induced by laser cleaning on stone materials, represents a very ﬁrst attempt for a new quantitative means to reveal changes in stone surfaces. It is a three-dimensional survey carried out by means of laser micro-proﬁlometry and completed by statistical data processing. The acquisition and successive computation is performed on small selected areas both before and after the cleaning process.

62.1 Introduction Roughness measurement of an artwork is important to document the surface condition, to assess changes either due to restoration intervention or surface decay due to wearing agents, and to monitor the evolution with time in terms of shape variations. Roughness measurements are the basis of many industrial quality controls: in that case, roughness arises from the working process and its characteristics are deﬁned in standard protocols (UNI, ISO). When dealing with craftworks, roughness computation is complicated by the lack of rules deﬁning its calculation, due to the peculiarity of artworks that does not allow for a straightforward application [1]. In this work we present a new quantitative approach for roughness computation, based on a high-resolution, three-dimensional survey. Roughness measurement deriving from a shape survey is a new application in the Cultural Heritage ﬁeld, where little attention has been paid to this kind of diagnostic.

524

C. Colombo et al.

62.2 The Micro-proﬁlometer A micro-proﬁlometer, realized by Art Diagnostic Group of INOA, was used for the three-dimensional survey. This device combines a high resolution with the capability to work on diﬀerent surface materials, being thus suitable for this kind of investigations. The instrument, composed of a commercial conoscopic probe [2] mounted on two motorized high-precision linear stages, has a height resolution of 1 µm, an accuracy better than 6 µm, and a transverse resolution of 20 µm. It works at a standoﬀ distance of 40 mm with a measurement range of ±4 mm. The maximum acquisition rate is 400 points s−1 .

62.3 Results and Discussion A set of ﬁve red Verona limestone specimens were cleaned with diﬀerent laser ﬂuences and with two diﬀerent cleaning devices: specimens V2, V3 and V4 were cleaned by Quanta System Michelangelo with handtool manual control; specimens V11 and V12 were cleaned by Syl 201. In this latter case a X-Y galvanometric scanner delivered the laser beam on the specimen surface with a regular pattern. [3]. Three-dimensional measurements were carried out on both the cleaned and the uncleaned surfaces on areas 1 cm2 wide. Surface shape was removed by least square ﬁt and statistical surface parameters [4] (Sq and λq ) were calculated on the conditioned surfaces. Figure 62.1 shows a plot of the two parameters for the ﬁve red Verona limestone samples. Red Verona limestone Sample V2, V3, V4, V11, V12

Red Verona limestone Sample V2, V3, V4, V11, V12

6.4

650

6.2 600 6.6 5.8 lambdaq [micron]

550

sq [micron]

5.6 5.4 5.2

500

450 5 4.8

400

4.6 4.4

350 1

2

3 Sample

4

5

1

2

3 sample

4

5

Fig. 62.1. Root Mean Square roughness Sq and Root Mean Square wavelength λq for the cleaned (black circle) and uncleaned (gray diamond ) areas. Samples V2, V3, V4, V11 and V12 correspond to laser ﬂuences of 1.25, 1.50, 2.0, 2.5 and 3 J cm−2 , respectively. The last two samples were cleaned using SYL 201

This page intentionally blank

62 Morphological Changes Induced by Laser Cleaning

525

Laser cleaning seems to level the intrinsic asperities of the sample due to the calcareous materials, whereas a cut-oﬀ seems to be present in the wavelength picture: probably the increase in laser ﬂuence induces a regular pattern. These data represent a surface texture which includes the feature characteristics of both the material and the laser cleaning. To discriminate those eﬀects, the former (roughness) at a higher frequency than the latter (waviness), a Gaussian ﬁlter was used with four diﬀerent cutoﬀ lengths (800, 400, 200 and 100 µm). Increasing the cut-oﬀ length means that more widely spaced features are included into the roughness computation, and the choice of the cut-oﬀ depends on the nature of the surface texture. Figure 62.2 shows the results of roughness and waviness computation for the cleaned part of V11 sample, according to the above mentioned cut-oﬀs. The right cut-oﬀ seems to be around 400 µm, where there is the best separation between waviness and roughness. After ﬁltering, for the cut-oﬀ Waviness Cut off 800 µm

Waviness Cut off 400 µm 10 10 5

5

0 0

−5

−5

−10 −15

roughness

15

roughness

15

10 10 5 5 0 0

−5

−5

−10

−10

−15

−15

Waviness cutoff 100 µm

Waviness cutoff 200 µm 10

10 0

0

−10

−10

−20

−20

−30

−30

roughness

−40

roughness 15

15

10

10

5

5

0

0

−5

−5

−10

−10

Fig. 62.2. Pairs showing waviness (upper ) and roughness (lower ) computation for the cleaned part of Sample V11 at 800, 400, 200 and 100 µm

526

C. Colombo et al.

Fig. 62.3. Waviness (left) and roughness (right) for the red Verona limestone samples calculated at 400 µm

at 400 µm, the results of waviness and roughness computation are shown in Fig. 62.3. For the highest ﬂuences (Samples V11 and V12), the eﬀect of laser cleaning is dominant and induces a variation in the morphology of the sample generating a regular pattern.

62.4 Conclusions This work represents a very ﬁrst attempt for a new quantitative means to reveal changes in stone surfaces. A set of red Verona limestone samples were surveyed by means of laser micro-proﬁlometry and roughness computation was carried out on the 3D data sets. By properly choosing the width of a Gaussian ﬁlter, waviness to roughness separation is possible, outlining, 5 thus, the eﬀects of laser cleaning.

References 1. R. Fontana, M. C.Gambino, M. Greco, L. Marras, M. Materazzi, E. Pampaloni, and L. Pezzati, in Proc. SPIE, Vol. 5146, 236, 2003. 2. G. Y. Sirat and D. Psaltis, in Opt. Comm., Vol. 9(65):243-245, 1988. 3. C. Colombo, E. Martoni, M. Realini, A. Sansonetti, and Valentini, in Proceedings of Lacona VI, 2005. 4. D. J. Whitehouse, Handbook of Surface Metrology, 1994.

63 A Mobile True Colour Topometric Sensor for Documentation and Analysis in Art Conservation ∗

Z. Böröcz1 , D. Dirksen2 , G. Bischoﬀ1 , and G. von Bally1 1

∗ 2

University of Münster, Laboratory of Biophysics, Robert-Koch-Straße 45, 48129 Münster, Germany [email protected] University of Münster, Institute of Prosthetics Dentistry, Waldeyerstraße 30, 48129 Münster, Germany

Summary. Drawings and photographs are often insuﬃcient for documentation in the ﬁeld of art conservation. A reliable documentation requires the acquisition of geometrical as well as colour information. For this purpose a mobile true colour topometric measurement system has been developed.

63.1 Introduction A 3-D coordinate measurement system based on the fringe projection technique is presented that uses colour CCD cameras with digital interfaces which allow a simpliﬁed photogrammetric calibration as well as a true colour reproduction. Combined with a standard video projector for fringe projection and controlled by a laptop, a very compact and mobile sensor is implemented. The system oﬀers a non-destructive and non-contact method for a safe investigation of fragile objects. The result of each measurement is a cloud of several hundred thousand 3-D points. These data samples are the basis for digital documentation, image processing, database management, restoration and analysis of cultural assets and archaeological samples [1–6].

63.2 Experimental Methods The basis of the system for 3-D coordinate measurements is a topometric sensor head consisting of two CCD cameras and a fringe projector ﬁxed on a tripod (Fig. 63.1). Standard colour-capable IEEE 1394 CCD cameras (SONY DFW-X700) with a resolution of 1, 024 × 768 and a DLT projector with a resolution of 1, 024 × 768 are used. The system is controlled by a mobile computer (Fig. 63.2). During the measuring process a sequence of four phase-shifted,

528

Z. Böröcz et al.

Fig. 63.1. Experimental setup for mobile 3-D data acquisition. C1,2 : CCD cameras, P: fringe pattern projector (video projector)

Fig. 63.2. Laptop with software for control and analysis

quasi-sinusoidal fringe patterns and seven grey code patterns are projected onto the object and recorded as stereo images by the CCD cameras (Fig. 63.3). After calculating the phase distribution, the stereoscopic images are evaluated by photogrammetric techniques and a 3-D coordinate is calculated for each valid pixel. Each calculated 3-D coordinate is stored together with the 2-D coordinates of the CCD camera pixel in a database.

63 A Mobile True Colour Topometric Sensor

529

Fig. 63.3. Measurement of a colour object. For the alignment of single measurements, control points are arranged around the object

By the use of control points which are placed around the object (Fig. 63.3), data from diﬀerent viewpoints can be acquired, integrated and processed automatically. A single measurement is performed in 2–3 s. Due to eﬀects like electronic noise and lens distortions, each 3-D coordinate shows some inaccuracy and the resulting surface exhibits a certain roughness (Fig. 63.4). Thus, a smoothing algorithm has been implemented which operates on a 5 × 5 pixel array and calculates the mean value of the coordinate which is orthogonal to the surface of the model. The result is shown in Fig. 63.5. For the reproduction of realistic colours of the 3-D model, calibrated colour CCD cameras and standard halogen studio lamps for the illumination of the object are employed. Using the obtained coordinate points (point cloud), the object surface is reconstructed by connecting sets of three neighbouring points to a triangle. The object surface is now covered with a grid of triangles. This technique is also called triangulation (Figs. 63.6, 63.9). By use of colour CCD cameras, it is also possible to map colour textures onto these triangles (Fig. 63.7). As the investigated objects are quite complex, an approach has been developed that uses the database mentioned above for triangulation of the point cloud. For each selected pixel the adjacent pixels in horizontal and vertical directions are obtained from one CCD image and the corresponding 3-D coordinates are connected by a triangle.

530

Z. Böröcz et al.

Fig. 63.4. Rough object surface

Fig. 63.5. Smoothed object surface

The achievable measurement accuracy depends on the angle between the cameras (which is called triangulation angle and should not be mixed up with the technique of surface triangulation, see above), the image ﬁeld size, as well as on the number of camera pixels. The accuracy of the 3-D scanner can be obtained by measuring a skittles ball several times from the same position. Afterwards the mean for each coordinate is calculated. With a triangulation angle of about 40◦ , a measuring distance of 50 cm, 25 mm Cosmicar/Pentax lenses and an image diagonal of approximately 180 mm, for example, the height resolution is < 25 µm for a single measurement. If CCD cameras with 1, 024 × 768 pixels are employed, the lateral resolution is about 0.17 mm. The inaccuracy increases slightly if several point clouds are integrated into a common coordinate system.

63.3 Results and Discussion 63.3.1 Analysis An important application of the 3-D measurement technique is the analysis of objects. It is possible to compute sections and proﬁles of the measured 3-D object (Fig. 63.6) to investigate, for example, the similarity with other objects

63 A Mobile True Colour Topometric Sensor

531

Fig. 63.6. Triangulated surface, section and proﬁle of an iron tool

of this type which were found at diﬀerent places. Furthermore, this technique could substitute for the drawing of objects from diﬀerent views. 63.3.2 Documentation The developed mobile 3-D measurement system, including colour acquisition of the object’s surface, has a wide range of applications in the ﬁeld of art conservation as well as archaeology. In Fig. 63.6, an image of the colour 3D reconstruction of a measured Madonna statue is shown. The object has a height of about 1.1 m. For this measurement we used 12 mm Cosmicar/Pentax lenses and a distance of about 1.8 m to the object. A total of 11 measurements were carried out. The lateral resolution is about 5 mm. The use of studio lamps for additional illumination results in more realistic colours than illumination with the video projector (see above). This example illustrates the possibility of a documentation of the actual state of the object before the restoration. Another application of the 3-D measurement technique is the analysis of a surface structure (Fig. 63.8). The triangulated object allows the examination from diﬀerent viewing angles as shown in Fig. 63.9. After computing a range

532

Z. Böröcz et al.

Fig. 63.7. Example of 3-D model of a statue

Fig. 63.8. Photo of a stone in a megalithic grave

image, usually visualised in false colours, image processing routines may be applied to enhance, e.g. shallow pictographs, and thereby support their analysis (Fig. 63.10).

63 A Mobile True Colour Topometric Sensor

533

Fig. 63.9. Screenshots of a detail on the reconstructed stone from diﬀerent viewing angles

Fig. 63.10. False colour representation of the detail in Fig. 63.9

63.3.3 Virtual Exhibitions An important advantage of the presented method compared to a conventional drawing is achieved by the fact that documentation and distribution may not only be accomplished in the classical way, i.e. through print media, but also in the form of digital 3-D models, such as via Internet, which furthermore allows exhibitions. The digital 3-D models are not only suitable for documentation and analysis of objects but also for a virtual presentation of digitised objects e.g. in a museum. Recent developments in auto-stereoscopic display technologies here oﬀer the possibility to reproduce the objects in a true three-dimensional way.

534

Z. Böröcz et al.

63.4 Conclusions The developed 3-D mobile and true colour measurement system provides a wide range of applications for documentation and analysis of cultural assets and archaeological samples. The system oﬀers a non-destructive and noncontact method for a safe investigation of fragile objects. A triangulated and true colour digital 3-D model of an object provides, in particular, a more detailed documentation than photographs. Converted to range images, the existing tools of image processing can be applied for various needs. Acknowledgements We would like to thank EUREGIO for supporting this project within the INTERREG IIIA-program (project-no. 2-EUR-II-2-30).

References 1. F. Bernardini, I. Martin, J. Mittleman, H. Rushmeier, and G. Taubin, IEEE Comp. Graph. Appl. 22(1), 59 (2002). 2. W. Böhler, M. Bordas Vicent, G. Heinz, A. Marbs, and H. Müller. Proceedings of the FIG Working Week 2004, May 22–27, Athens, Greece. Publ. by FIG. 2004. 3. Z. Böröcz, C. Thomas, D. Dirksen, and G. von Bally, in Rapport annuel d’activité scientiﬁque, Bibracte 2000. 4. D. Dirksen, Y. Kozlov, and G. von Bally. Cuneiform surface reconstruction by optical proﬁlometry. in: D. Dirksen, G. von Bally (eds): Optics Within Life Science (OWLS IV): Optical Technologies in the Humanities, Springer, Heidelberg, 257–260, 1997. 5. D. Dirksen, G. von Bally, and F. Bollmann. Automatic acquisition and evaluation of optically achieved range data of medical and archaeological samples. in: C. Fotakis, T. Papazoglou, C. Kalpouzos (eds): Optics Within Life Science (OWLS V): Biomedicine and Culture in the Era of Modern Optics and Lasers, Springer, Heidelberg, 147, 2000. 6. R. Sablatnig and C. Menard. 3D Reconstruction of Archaeological Pottery Using Proﬁle Primitives, 1997. in N. Sarris, M.G. Strintzis, Proc. of International Workshop on Synthetic-Natural Hybrid Coding and Three-Dimensional Imaging, 93.

64 Reconstruction of the Pegasus Statue on Top of the State Opera House in Vienna using Photogrammetry and Terrestrial and Close-Range Laser Scanning C. Ressl Christian Doppler Laboratory for “Spatial Data from Laser Scanning and Remote Sensing”, Institute of Photogrammetry and Remote Sensing, Vienna University of Technology [email protected] Summary. This chapter describes the surveying work and the creation of a 3D model of a Pegasus statue, which builds the basis for a static analysis. The supporting legs of the statue were surveyed with the close-range laser scanner, Minolta VIVID 900. Approximately 45 individual scans were required to cover each leg. The rest of the statue was surveyed with the terrestrial laser scanner, Riegl LMS-Z420i with eight individual scans. The surveying of the statue using these two laser scanners is particularly interesting because the statue represents a rather “small” object for a terrestrial laser scanner, whereas for a close-range laser scanner it represents a rather “huge” object. With the aid of photos of the statue, the relative orientation between the diﬀerent laser data was determined in the course of a hybrid bundle block adjustment. Finally, a “waterproof” 3D model of this complex statue was derived.

64.1 Introduction In the nineteenth century, on top of the State Opera House in Vienna, two Pegasus statues of size 4 × 3 × 4 m were built. Because of their age and the disadvantageous stabilization (each horse stands on three legs, which are also almost arranged on one line, see Fig. 64.1), some concern about the stability arose and so a static expertise was commissioned for one of the two statues. Prerequisite for such a static investigation is a 3D model of the entire statue. Because of the static nature of this expertise, the accuracy for modeling the supporting legs is most important. The rest of the statue’s body can be modeled with lower accuracy. The Institute of Photogrammetry and Remote Sensing (IPF) of the Vienna University of Technology was commissioned to provide such a 3D model.

536

C. Ressl

Fig. 64.1. Left: The selected Pegasus statue on top of the State Opera House in Vienna. Right: The laser scanner Riegl LMS-Z420i on site (Photo: Riegl LMS)

The Vienna University of Technology owns two laser scanners: the terrestrial laser scanner, Riegl LMS-Z420i (range: 2–800 m with an accuracy of approx. ±1 cm), [1], and the close-range laser scanner, Minolta VIVID 900 (range: 0.6–2.5 m with an accuracy of approx. ±1 mm), [2]. Both were used for reconstructing the selected Pegasus statue. Because of the high importance of the supporting legs for the static expertise, the Minolta scanner was used for surveying these supporting legs. The Riegl scanner was used for the rest of the statue. Further photos with the digital camera Kodak DCS 460c were acquired around the statue. By identifying corresponding points in these photos and the two laser data sets, all data could be transferred into the same system of coordinates. This paper describes all steps necessary to reconstruct the statue – from the survey work and the registration of all data to the creation of the ﬁnal 3D model.

64.2 The Surveying Work The following Table 64.1 gives a short overview of the properties of the laser scanners used to survey the Pegasus statue. 64.2.1 The Close-Range Laser Scanner MINOLTA VIVID-900 The Minolta scanner [2] is primarily conceived for indoor-surveying of objects with size of a few dm. Because of the triangulation light block method applied, the usage of the Minolta scanner requires certain lightning conditions (i.e., constancy and not too bright). Scanning of an object is usually done in the

64 Reconstruction of the Pegasus Statue

537

Table 64.1. Overview of the two laser scanners used within the project Minolta VIVID 900 type close-range laser scanner distance measurement triangulation light block method range using the wide-angle lens: range: 0.6–2 m (depending on focus) ﬁeld of view (hz × v) : 33◦ × 25◦ accuracy < ±1 mm (at 1 m) points per scan 680 × 480

recorded intensity

RGB: immanent

measurement time

2.5 s

Riegl LMS-Z420i terrestrial laser scanner time of delay range: 2–800 m horizontal: 0–360◦ ±40◦ 50–130◦

vertical:

< ±1 cm (at 3 m) up to 280,000,000 (depending on selected angular range and angular increment) intensity of backscattered laser pulse: immanent RGB: optional ca. 24,000 points s−1

following way: The object is placed on a rotary stage and is scanned for diﬀerent rotations with a certain overlap by the Minolta scanner, which is mounted on a ﬁxed tripod. From the viewpoint of the object, each of these scans represents a point cloud in a diﬀerent system of coordinates. Consequently for representing the entire object, all the scans in their diﬀerent systems have to be transformed into a common system of coordinates. This transformation is often termed “registration.” Usually, in the individual scans no corresponding points can be identiﬁed because the scanning method delivers diﬀerent (and unpredictable) points on the object every time. Therefore the shifts and rotations of the registration cannot be computed in a simple way using some corresponding points. Instead, some form of the so-called ICP algorithm (“iterative closest point” [3]) is applied. The registration of two point clouds A and B with a suﬃcient overlap by means of the ICP algorithm is done in the following way. Point cloud A is kept ﬁxed and the other is shifted and rotated iteratively so that the distance between the points of A to the closest points of B (in the overlap) is minimized. The ﬁnal registration of the two point clouds will not be optimal in general, because the iteratively assigned point pairs never really correspond. A reﬁnement of the original ICP algorithm is the so-called SDM algorithm (“squared distance minimization”; e.g. [4]). With this algorithm, not the distance between the points of cloud A and the closest points of cloud B is minimized, but the distance to the closest tangent plane in B. In this way not only a better registration can be achieved but also the SDM algorithm converges much faster than the ICP algorithm (quadratic vs. linear).

538

C. Ressl

The data acquired with the Minolta scanner can directly be processed directly with the software Raindrop Geomagic Studio 5 [5], which computes the registration of two or more scans using the SDM algorithm. 64.2.2 The Terrestrial Laser Scanner RIEGL LMS-Z420i The LMS-Z420i, [1], is equally suited for in- and outdoor projects. It has an active laser source and therefore does not require certain lighting conditions. Opposed to the Minolta scanner, the LMS-Z420i is used for objects with extents of a few meters up to some hundred meters. Independent of the size of the object, more than one scan will usually be required to cover the entire object. In principle the registration of several Riegl scans can also be done with the ICP or SDM algorithm. Usually in practice, however, before using the laser scanner retro-reﬂecting targets are stuck on the object or in its vicinity. After scanning the entire object together with the targets, the latter can be identiﬁed in diﬀerent scans and therefore allow for a registration of the diﬀerent scans by means of really corresponding points. Furthermore, by using such targets, the diﬀerent scans can be made with a lesser overlap than if the registration would be done by ICP or SDM. Controlling the scanner and processing the data is done with the software RiSCAN-Pro [1]. 64.2.3 Surveying the Selected Pegasus Statue Because the supporting legs had to be modeled with much more detail than the rest of the statue, these legs were scanned with the Minolta scanner. Due to the small ﬁeld of view of the Minolta scanner, each supporting leg (height approx. 1.5 m) had to be composed of a lot of many scans. Each single Minolta scan covers approx. 30 × 30 cm with a typical point sampling distance of 1 mm on the object; see Fig. 64.2 (middle). In order to register all the scans of one leg in Geomagic with the SDM algorithm, the diﬀerent scans were acquired with an overlap of 30–40%. Some parts of the legs were very diﬃcult to access, e.g., the hoof parts, the bended bent heel, or the inner sides of the legs, which were only accessible from under the statue. Because of these circumstances about 200 diﬀerent Minolta scans were needed to completely survey all three supporting legs (duration: approx. 10 hs). Surveying the entire statue with the Riegl scanner turned out to be much easier. In 6 h the entire statue was covered with eight scans (including the stomach, omitting a few non visible higher regions and a total number of points of ≈2.4 million). With the aid of 23 retro-reﬂecting targets, all scans could be registered on site. The stress analysts made radiographic and ultrasound measurements at certain spots points of the statue. These spots points were marked with white tape strips, see Fig. 64.2 left, whose coordinates should be determined in 3D also. For this purpose eight images were acquired with a digital camera Kodak

64 Reconstruction of the Pegasus Statue

539

DCS 460c and a 15 mm lens. Later during the orientation step (see Sect. 64.3), these spots together with the Riegl targets build the frame to which all data will be referred to. In order to model the Pegasus statue correctly with respect to the gravity, the Riegl targets were surveyed with a level. All surveying was accomplished from beginning to middle of November 2003. The usage of the Riegl and Minolta scanner at the Pegasus statue is depicted in Figs. 64.1 and 64.2, respectively.

64.3 Orientation of all Data Due to space limitations the description of this work can only be summarized, a more detailed one can be found in [6]. First, the Minolta scans were processed in Geomagic. After removing the cross errors from the scans, examples shown in Fig. 64.2 (right), about 45 individual scans were registered semi-automatically for each supporting leg using the SDM algorithm; see Fig. 64.4 (left) at the end of this paper. Then all data – the eight images, the eight Riegl scans, the combined Minolta scans for each of the three supporting legs from the ﬁrst step and the level data – were simultaneously orientated in a hybrid bundle block adjustment using the program ORIENT [7]. The main result of this bundle block adjustment are the improved positions and rotations of the eight Riegl scans

Fig. 64.2. Left: The laser scanner Minolta VIVID 900 on site. Middle: One selected Minolta scan (25 × 35 × 20 cm3 ) on the bended bent leg (height ca. 1.3 m). For modeling this particular leg, 42 scans in total were used. This selected individual scan is depicted with the originally recorded RGB information and superimposed on all 42 registered scans, which are shown in green. Right: Gross errors (highlighted in red, max. error 4.5 cm) of one original Minolta scan (shown in the recorded RGB texture, size 30 × 15 × 8 cm3 )

540

C. Ressl

Fig. 64.3. Visualization of the orientation of the Riegl scans and Minolta legs resulting from the hybrid adjustment. Left: the eight Riegl scans shown in diﬀerent colors. Right: the orientation of the Minolta legs with respect to the Riegl scans, the latter being thinned out for visualization purposes

and of each Minolta leg with respect to a properly deﬁned vertical and northward oriented coordinate system. Such a system was necessary because measurements for wind direction and wind strength by the oﬃcial weather station should be considered in the static expertise, particularly to assess their inﬂuence on the wings of the Pegasus statue. Figure 64.3 depicts the adjustment result by superimposing the point clouds of the eight Riegl scans with the Minolta legs.

64.4 3D Model of the Statue The aim of this survey was a 3D model of the entire statue. Because of the subsequent static modeling, the 3D model had to be free of holes. It had to represent a “waterproof” volume. Furthermore the 3D model had to be represented as compact as possible, because for the static modeling the geometric volume model would be replaced by ﬁnite elements. For these reasons (holes and large data volumes) the registered and thinned out point clouds could not be simply triangulated. Instead a so-called NURBS model was created in Geomagic. NURBS is an abbreviation for, non-uniform rational B-splines’. Creating an NURBS model in Geomagic is done in several steps (the Minolta and Riegl data are dealt with separately and are combined only at the very end): 1. Transformation of the diﬀerent data into the same system of coordinates: Using the Geomagic tool “transform” the translations and rotations determined in the adjustment can be applied to the diﬀerent data sets (eight Riegl scans, and the three Minolta legs).

64 Reconstruction of the Pegasus Statue

541

2. Thin out of the original point clouds (Geomagic tool “uniform sample”): The Minolta data of the supporting legs were thinned out with a point distance ranging from 3 mm (hoof) to 15 mm (upper end of leg). Finally each Minolta leg was represented by approx. 150,000 points. The Riegl data were thinned out with point distances of 15–20 mm and ﬁnally contained approx. 450,000 points in total. 3. Triangulation of the data (Geomagic tool “wrap”) including triangulationcorrections (Geomagic tool “clean”) and one time minimal noise reduction (Geomagic tool “noise reduction”): It turned out to be beneﬁcial to apply “wrap” and “clean” several times one after the other: in doing this the triangulation has to be broken up (Geomagic tool “modify current points”) before each rerun to access the adapted point cloud. In this way the number of holes decreases a lot, whereas the number of points decreases only slightly; e.g., in the beginning the Riegl data contained 100,000 holes and, after running “wrap” and “clean” four times, just 200 holes remained. These remaining holes were centered at three regions (the upper side of the wings, the passage from the saddlecloth’s inner side to the horse’s

Fig. 64.4. Left: The combined Minolta scans for the rear stretched leg. Each individual scan is shown in a diﬀerent color. Right: the 3D model of the entire Pegasus statue as NURBS representation. Finally only 165,000 points were used for the NURBS creation – originally each of the three Minolta legs consisted of ≈ 4.6 million points and the eight Riegl scans consisted of 2.4 million points in total. Because of the smaller point density many details vanished at the upper part of the statue during the NURBS-modeling (e.g., harp or right hand). For subsequent statistic analysis, however, these lost details were of no concern

542

C. Ressl

stomach, the passage from the rider to the horse’s back) and could be ﬁlled manually. 4. Combining all triangulated data (Geomagic tool “merge”) and automatic NURBS creation: The four data sets (three Minolta legs and the Riegl body) must be combined as triangulations (and not as point clouds), because of their very diﬀerent point densities. If the data sets were combined as point clouds many new holes would be created at the transitions. The NURBS creation of the entire statue can then be derived rather simply from the combined triangulation and is controlled only by a few parameters (e.g., surface accuracy and detail richness); see Fig. 64.4.

64.5 Summary In this paper we described the reconstruction of a Pegasus statue using two diﬀerent laser scanners (Minolta VIVD 900 and Riegl LMS-Z420i) and the generating of a NURBS-based 3D model for static analysis. The registration between the Riegl data and the Minolta data (the latter being combined for each supporting leg) was determined in a hybrid bundle block adjustment involving also images of the statue. The posterior accuracy of the combined Minolta data determined at individual points turned out to be approx. ±3 mm. This accuracy level was far suﬃcient for the static problems to be dealt with on the basis of this 3D model. Acknowledgments Parts of this project were supported by the innovative projects “The Introduction of ILScan Technology into University Research and Education” and “3D Technology” of the University of Technology, Vienna.

References 1. Riegl Laser Measurement Systems: http://www.riegl.co.at/ 2. Konica Minolta Photo Imaging: http://kmpi.konicaminolta.us/vivid/ 3. P.J. Besl and N.D. McKay, in IEEE Trans. Pattern Analysis and Machine Intelligence. Vol. 14(2), 239, 1992. 4. H. Pottmann, S. Leopoldseder, and M. Hofer, in Proceedings PCV ’02, Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXIV, Part 3A, Commission III, 2002. 5. Raindrop Geomagic, Inc.: http://www.geomagic.com/products/studio/ 6. C. Ressl, in: Optical 3-D Measurement Techniques VII – Vol. I, A. Grün and H. Kahmen (eds.), 2005. 7. ORIENT – An Universal Photogrammetric Adjustment System, Product Information, Institute of Photogrammetry and Remote Sensing, TU Vienna, http://www.ipf.tuwien.ac.at/products/produktinfo/orient/html_hjk/orient_e. html

All http-links were accessed in June 2005.

65 Some Experiences in 3D Laser Scanning for Assisting Restoration and Evaluating Damage in Cultural Heritage ∗

L.M. Fuentes , J. Finat, J.J. Fernández-Martin, J. Martínez, and J.I. SanJose DAVAP Research Group, Lab 2.2, Ediﬁcio I + D, Campus Miguel Delibes, University of Valladolid, 47011 Valladolid, Spain ∗ [email protected] Summary. The recent incorporation of laser devices provides advanced tools for assisting the conservation and restoration of Cultural Heritage. It is necessary to have as complete as possible understanding of the object state before evaluating or deﬁning the reach of the restoration process. Thus, a special eﬀort is devoted to surveying, measuring and generating a high-resolution 3D model prior to restoration planning. This work presents results of several experiments performed on damaged pieces for evaluation purposes in Cultural Heritage. Some software tools are applied for carving-work analysis, conservation-state monitoring, and simulation of weathering processes for evaluating temporal changes. In all cases considered, a high resolution information capture has been performed with a laser scanner, the Minolta 910. Our approach is ﬂexible enough to be adapted to other kinds of pieces or Cultural Heritage artefacts, in order to provide an assessment for intervention planning in conservation and restoration tasks.

65.1 Introduction Restoration constraints impose an approach that disallows direct contact with Cultural Heritage pieces in the surveying stage. The availability of a 3D global model allows multiple views and a virtual manipulation without risk. High resolution photogrammetry has provided a standard for 3D surveying based on the identiﬁcation on an increasingly high number of control points for generation of a 3D model of the piece to be restored or surveyed. The generation and manipulation of 3D models associated ﬁles is an invaluable tool for assisting restoration, even in a remote way, thanks to the possibility of export to diﬀerent formats and Web navigation. Links between geometric and radiometric properties relative to the shape can be symbolically represented as script nodes of a graph to simplify query processes by shape properties, extending usual search based on keywords or numerical data. The main goal is to provide

544

L.M. Fuentes et al.

graphical tools for experts in restoration without previous knowledge in Computer Graphics so that they can perform essays on the 3D models, including (semi) automatic virtual assembling of sherds in pottery modelling, and to elaborate diﬀerent proposals via Web on a common 3D ﬁle without damaging or risking the piece to be surveyed. Once the software tool is available, the next step involves the generation of a 3D database of pieces and artefacts with Cultural Heritage contents for evaluation, follow-up and testing purposes. Such database should be available using Web services. This goal has been achieved with diﬀerent strategies for pilot cases, including Web accessibility issues to an extension of GIS analysis for architectural environments and archaeological sites [1], but their extension to museum ﬁles has received less attention; see, however [2] for a survey and references. It is also desirable to provide a 3D support for comparative studies with diﬀerent resolutions, a remote consultation via Web for restoration experts, and the virtual replacement of the original physical object by almost identical copies for inspection and measuring, between other applications. In this work some intermediate goals are developed related with an accurate 3D surveying of pieces of diﬀerent materials (stone, wood, copper, concrete) and simulation of degradation processes. The contribution of laser scanners is the cornerstone for the improvement of image-based techniques in accurate surveying of Cultural Heritage by greatly simplifying the creation of a 3D model. Indeed, the generation of a 3D model from photogrammetric pairs of stereo views with control points provided by a total station is a tedious task and requires a very high expertise. Computer Vision techniques have contributed along the 1990s to visualising complex shapes from a “suﬃcient number” of views, in the calibrated and noncalibrated case [3], as much in the appearance- or geometric-based approaches. An important bottleneck for Computer Vision approaches is the generation an accurate 3D model, which requires a permanent update of epipolar constraints with increasing constraints for complex objects, or the generation of dense information for each view of the object. Finally, classic 3D modelling tools in Computer Graphics are often inappropriate for representing the very high complexity of Cultural Heritage objects due to shape irregularities (lack of geometric primitives), partial self-occlusions and the discrete nature of photogrammetric information. Hybrid approaches between close-range photogrammetry and 3D laser scans are currently acknowledged as an optimal solution for Cultural Heritage surveying. An evaluation of hybrid approaches can be read in [4]. The accurate, textured, three-dimensional model arising from a short-range laser scanner provides complete information from iconic to metric level, avoiding the high cost processing linked to standard techniques of Computer Vision. In this work a special attention is paid to the support, but corresponding to diﬀerent kinds of pieces. It is very diﬃcult to identify optimal criteria for Cultural Heritage objects. In a similar way to pattern recognition strategies, syntactic or structural

65 Some Experiences in 3D Laser Scanning

545

approaches; the former provides criteria following successive reductions from complex to increasingly simpler primitives, whereas the latest pays more attention to meaningful prototypes where pilot experiences have been successfully performed. The lack of a suﬃcient number of cases, and the very large casuistic suggest a structural approach with some prototypes for paints and sculptures, including bas- and high-relief in wood, stone and concrete. Similarly, following an Image Processing inspired approach, there is a dichotomy between top-down versus bottom-up approaches. The large shape diversity in Cultural Heritage suggests a bottom-up approach, taking advantage of the dense information captured with laser. The future availability of large 3D databases for objects surveyed with a common methodology will allow the validation and, in the aﬃrmative case, the extension of results presented in this work. Little more than 30 years after Brooks and Taylor discovered carbonaceous mesophase [1], there has been a great number of studies on mesophase formation related to the production of various carbon materials, ranging from anode coke (electrode for aluminum production) [2], needle coke for graphite electrodes used in electric-arc furnaces for iron and steel production [3], and carbon ﬁbres [4, 5].

65.2 Relation Between 2D and 3D Geometric Information Digitisation of large 3D objects presents some diﬃculties linked to geometric (e.g. shape irregularities), and radiometric (e.g. pigmentation irregularities, variable reﬂectivity) characteristics. The traditional geometric approach is based on partial views arising from perspective models, including orthogonal and conical. Currently, the use of dedicated software on 3D ﬁles captured with laser scanners provides linear systems of sections or isosurfaces of a potential function linked to geometric (e.g. depth) or radiometric (e.g. optical characteristics) properties. The selection of a viewpoint imposes strong constraints on the object to be analysed. The discriminant locus is the variety of points of the 3D support where the tangent plane to its surface passes through the viewpoint. These points are well deﬁned from the mathematical viewpoint, but they are not well determined for capture devices; they are in the issue of noise or holes linked to capture. However, they are crucial for the visual perception of the object, because they make part of the apparent contour which is the key for volumetric segmentation. To avoid problems linked to the lack of deﬁnition of apparent contours, several views or scans must be taken from complementary viewpoints. The basic functionalities of a model scanned with Minolta 910 and processed with 3D software packages like PolyWorks or RapidForm: – It provides a 3D geometric support for vector data linked to a dense cloud of points.

546

L.M. Fuentes et al.

– It allows the integration of raster data arising from 2D views, including radiometric properties, graphical attributes, and textures. – Simplify the management and visualisation of superimposed structures (triangular meshes and textured surfaces) with adaptive behaviour linked to geometric constraints (e.g. linked to curvature variations) and radiometric constraints (linked to allow number of colours in dynamic/static display). – Multi-resolution for making easier a fast remote access and manipulation based on data reduction with contraction/expansion behaviour linked to navigation and exploration tasks. Currently, the combination of laser scanners and dedicated software allows the visualisation of systems of points or curves on planar or volumetric objects as intersections by a sheaf of planes. In our case, the selection of sections and projections on a selected plane allows the identiﬁcation of: – Shape characteristics in terms of the geometry of discriminant locus (boundaries, rims, creases, etc.) – Contours linked to sections by a sheaf of parallel planes with a step selected by the user – Topological properties (inclusion, adjacency, connectivity, order) of planar regions for each view by applying standard processing tools of Computer Vision, and their symbolic representation by means of graphs, strings or tables. The evaluation of metric and topological properties on the whole object or its projections and sections is the ﬁrst step for the design and development of relational databases supporting contents based on recognition. Currently, 3D recognition based on volumetric segmentation is still at its beginnings. Most of the available tools for automatic query, retrieval and classiﬁcation of vector and raster data are still only valid for the 2D case.

65.3 Laser Surveying in Conservation and Restoration Assessment: Some Case Studies The application of the data obtained with a laser scanner in conservation and restoration assessment is linked to the capability of software tools for diﬀerent aspects of data structures such as: identiﬁcation of geometric and radiometric properties; query by properties or following additional information; interactive visualisation and manipulation. 65.3.1 Damaged Painting on Almost-Planar Metal Surface Digitisation of 2D Cultural Heritage is a mature ﬁeld. The application of 3D laser scanners for assessment and tracking restoration of paintings on diﬀerent supports is a subject which is receiving an increasing interest due to the

65 Some Experiences in 3D Laser Scanning

547

Flatness 0

+ Name : Flatness 0 + Flatness : 5.99266 mm + Plane Centre : 0.25152, 0.72516, -3.63219 + Plane Normal : 0.00199, 0.00919, 0.99999 + Average Height : 3.46946 + Standard Deviation : 3.50842 + Region Type : User Selected vertices

Fig. 65.1. Original state, identiﬁcation of impact and evaluation of planarity

Fig. 65.2. Level curves showing diﬀerent depths on the painting in Fig. 65.1 and textured model of the Sayon’s head (National Museum of Sculpture, Valladolid, Spain)

curved nature or deformation of the support. In this section, the analysis of a pilot case of a damaged paint on a metallic support is developed. Damage is diverse and includes pigment discolouration and an impact involving strong alterations in the support. The origin of the surface damage seen in Fig. 65.1 is unknown, and it is only possible to identify some kind of small impact made perhaps with a pointed object which has produced a meaningful deformation in the nearest zones.Furthermore, a very precise evaluation of the lack of planarity appearing in the ﬁgure at the right of Fig. 65.1, perhaps the most meaningful information concerns the treatment of the support as preparation before painting performed by the artist which appears in the central ﬁgure. An inspection section through the impact has suggested a more careful examination of the magnitude of 2D deformation around the critical zone. The puncture in the painting is 3.3 mm deep. Finally, the one-dimensional information relative to damage on the support can be summarised in a map of level curves showing the deformation, left of Fig. 65.2.

548

L.M. Fuentes et al.

65.3.2 Polychromated Sculpture in Wood Polychromated sculpture in wood has a long tradition in Spain, with a predominance of religious iconic representations. Original statues usually take part in parades around cities in Easter, and very often they suﬀer serious damages due to incorrect manipulation or atmospheric agents. Three-dimensional surveying based on range scanning provides tools for assisting a conservation policy helping to identify, evaluate and track the current state and its evolution linked to weather degradation, see right of Fig. 65.2. Some pigmentation loss and small cracks can be easily identiﬁed and quantiﬁed on the 3D model generated from data captured with Minolta 910. Unfortunately, we have had no opportunity of tracking the whole restoration process, due to some administrative problems. 65.3.3 Romanesque Capital of Monastery’s Cloister in Silos The analysis of damage produced by atmospheric agents on weathered stone is a topic which has been receiving some interest recently. A 3D surveying with rendering applications has been developed in [5]. Three-dimensional surveying of high- and bas-reliefs corresponding to sculptures [6] and capitals pose some interesting problems due to deterioration arising from abrupt changes in temperature, pollution and some stone-illness. In this work, we restrict ourselves to some geometric aspects which concerning shape evaluation corresponding to a selected capital from the cloister of the Romanesque monastery of Silos (Burgos, Spain). The intricate geometry of multiply linked ornaments or arabesque details makes very complicated the generation of complete model of the capital based on range-scanning, Fig. 65.3. The resulting 3D surveying has provided a very accurate model with diﬀerent levels of detail (LoD), including sections along longitudinal and transversal planes for identifying material losses or additional depositions arising from atmospheric or animal interventions. Currently, we are working on a variant of volumetric diﬀusion method [7] for ﬁlling the holes corresponding to braids or tresses, or some another self-occluded elements. The comparison with resulting 3D model of future scans will allow a comparative analysis for tracking the evolution of the scanned capital as a pilot case. 65.3.4 Evaluation of Intentional Damage on a Concrete Model Unfortunately, the lack of ﬁnancial support for previous analysis makes the evaluation and prediction for assessment conservation and restoration tasks very diﬃcult. To obtain precise information about the performance of the laser scanner, Minolta 910, for acquiring small geometric variations, we have performed an experiment on a model made in concrete. The main diﬀerence

65 Some Experiences in 3D Laser Scanning

549

Fig. 65.3. Picture, model and view of a Romanesque capital

Fig. 65.4. Damage sites on a concrete model and their graphical evaluation

between the non modiﬁed parts of the model was 0.008 mm, all non-modiﬁed parts of the model were enclosed in the range [−0.06, 0.06], making differences greater than 0.06 mm detectable. As a matter of fact, the deposition of the dust coming from sanding part of the created damage (labelled as material erosion) was also detected. This particular example shows the advantages of using 3D surveying in damage evaluation and restoration (Fig. 65.4).

550

L.M. Fuentes et al.

65.4 Conclusions and Future Work We have described some results relative to geometric aspects of a high resolution digital model for small pieces in Cultural Heritage involving metal, wood and stone for their support. Our analysis allows identifying diﬀerent kinds of pathologies which can be analysed for adapting optimal strategies for conservation and restoration tasks. A tracking along several years by comparing 3D surveying of the same piece is still not possible. An experiment has been designed for evaluating the performance of a laser scanner device. Controlled damage sites have provided very accurate information allowing us to identify sub-millimetric losses or additions of material. Unfortunately, the most painful process is linked to the lack of sensibility of responsible persons for conservation and restoration tasks about the interest of the application of IST for 3D surveying of Cultural Heritage, giving as a result some diﬃculties in taking scans in the most appropriate optical conditions. Thus, in this work, we have restricted ourselves to geometric properties of the support to be restored. Therefore, the main aspect to be improved is how to achieve a better control of radiometric properties. Variations in colour require a more careful study; diﬀuse, specular and roughness parameters require the combination of range and colour measurements following the pioneering work of Sato [8]. The projection of colour on the mesh modiﬁes the original visual perception, and it would be necessary to take advantage of a neutral illumination to avoid colour alterations. Acknowledgements The acquisition of the laser device, Minolta 910, has been supported by EU research funds for regional development (FEDER), by the Spanish Ministry of Science and Technology, and regional institutions (JCYL) in the Project DELTAVHEC (Dispositivos para el Escaneo Láser Tridimensional, Adquisición y Visualización de la Herencia Cultural), Research Group Responsible, Prof. Javier Finat. This work has been partially ﬁnanced by the Spanish Ministry of Culture, CICYT Research Project MAPA (Modelos y Algoritmos para visualización del Patrimonio Arquitectónico) Research Group Responsible Prof. Juan José Fernández Martin).

References 1. M. Gaiani, E. Gamberini, and G. Tonelli, VR as Work Tool for Architectural and Archaeological Restoration: The Ancient Appian Way 3d Web Virtual GIS. Proc. 7th Intl Conf. On Virtual Systems and Multimedia, VSMM’01, IEEE 2001. 2. J. Taylor, J.A. Beraldin, G. Godin, L. Cournoyer, R. Baribeau, F. Blais, M. Rioux, and J. Domey, in J. of Visualization and Computer Animation, 14(3), 121, 2003.

65 Some Experiences in 3D Laser Scanning

551

3. R. Hartley and A. Zisserman, Multiple View Geometry in Computer Vision. Cambridge University Press, 2000. 4. J. A. Beraldin, Integration of Laser Scanning and Close-Range Phtogrammetry: The Last Decade and Beyond. XXth ISPRS Congress, Comm. VII, Istanbul, Turkey, 972, 2004. 5. J. Dorsey, A. Edelman, H.W. Kensen, J. Legakis, and H. K. Pedersen, in Proceedings SIGGRAPH’99, ACM Press, 225, 1999. 6. J. Martinez, J. Finat, L.M. Fuentes, M. Gonzalo, and A. Viloria, in CIPA Intl Symposium, Torino, Italy, 2005. 7. J. Davis, S. Marschner, M. Garr, and M. Levoy, Proc. 1st Intl Symp on 3D Data Processing, Visualization and Transmission, 2002. 8. Y. Sato, M.D. Wheeler, and K. Ikeuchi, in Proceedings SIGGRAPH’97, ACM (1997) 379–387.

66 Monitoring of Deformations Induced by Crystal Growth of Salts in Porous Systems Using Microscopic Speckle Pattern Interferometry ∗

G. Gülker1 , A. El Jarad1 , K.D. Hinsch1 , H. Juling2 , K. Linnow3 , M. Steiger3 , St. Brüggerhoﬀ4 , and D. Kirchner4 1

∗ 2 3

4

Applied Optics Group, Institute of Physics, Carl von Ossietzky University Oldenburg, 26111 Oldenburg, Germany [email protected] Stiftung Institut für Werkstoﬀtechnik, Paul-Feller-Str.1, 28199 Bremen, Germany Institut für Anorganische und Angewandte Chemie, University Hamburg, 20146 Hamburg, Germany Deutsches Bergbau-Museum Bochum, 44787 Bochum, Germany

Summary. Electronic speckle pattern interferometry (ESPI) has been used to monitor microdeformations and surface microstructure changes produced by crystallization or hydration pressure of magnesium sulfate in a porous material. Samples of fritted glass were chosen as a standard porous substrate because of its mean grain size, its porosity distribution, and its negligible humidity expansion. The glass samples, soaked with salt solution, were exposed to changes in relative humidity of the surrounding air. The full-ﬁeld ESPI measurements were combined with cryogenic SEM visualizations. Results from these investigations were partly not expected theoretically and give new insight in the underlying salt phase transition processes.

66.1 Introduction It is generally recognized that growth of crystal salts in the porous structures of materials such as stone, brick, plaster, ceramics, and concrete is a major cause of damage limiting the durability of our architectural and cultural heritage. Salt weathering has been found to be responsible for signiﬁcant damage which is apparently caused by the generation of high crystallization and hydration pressures. Experimental evidence that growing crystals can exert pressure was already recognized 150 years ago [1]. Although known for a long time, the basic mechanisms of salt deterioration are not completely understood and the situation is suﬀering from a lack of fundamental insight in the underlying processes. There is still a controversial debate about the signiﬁcance of diﬀerent damage mechanisms and agreement among investigators has not been achieved. In order to provide a better understanding,

554

G. Gülker et al.

an investigation was initiated considering both theoretical aspects and novel experimental strategies. One approach to study the onset and the dynamics of salt-induced pressure is to measure microdeformations and surface microstructure changes of a porous substrate during the crystallization or hydration of salt. For this purpose electronic speckle pattern interferometry (ESPI) was used. ESPI is a noncontact and full-ﬁeld method, which is able to detect deformations in the submicron range and in video real-time. This has the big advantage that deforming parts of the object’s surface can be immediately identiﬁed and further examined in detail, e.g., in our investigations with scanning electron microscopy (SEM). In several situations we found that a magniﬁed imaging of the object was necessary to resolve deformations on a small lateral scale. Thus, an ESPI system with a small and well-adapted ﬁeld of view was realized by using a microscope objective. In this chapter, ﬁrst results obtained with this ESPI setup are presented and discussed on the basis of the present theories.

66.2 Standard and Microscopic ESPI In the ﬁeld of optical metrology, ESPI has been found to be a very useful technique for measuring the deformation or displacement of an object surface. As noted earlier, ESPI is a full-ﬁeld method meaning that the displacements within the ﬁeld of view can be detected spatially resolved. The size of the ﬁeld of view is deﬁned by the imaging objective and can widely be adapted to the demands of the investigations. The method is based on holography with the main diﬀerence being that a CCD camera is used (instead of a holographic ﬁlm) to capture the interference patterns, which are then digitally stored in a computer. Therefore, in many reports, the technique is also called TV-holography or digital holography. Our group has been using this technique for a long time to investigate deterioration processes in works of art and to develop procedures for their preservation [2]. The underlying principles of ESPI are widely known and comprehensively described in literature, e.g., in [3]. In Fig. 66.1, a scheme of the setup used is shown. The test specimen is illuminated by the coherent light of a laser. Since sensitivity and spatial resolution of the method increase with decreasing wavelength of the light, a HeCd-laser with a short wavelength of 442 nm was used. The object beam illuminates the specimen at a small angle to the viewing direction. Thus, the system is almost solely sensitive to out-of-plane deformations. To perform humidity changes during the measurements and produce climatic conditions where phase transitions of salts should occur, the objects under investigation were placed in a small computer-controlled climatic chamber. Here objects could be exposed to well-deﬁned temperature and relative humidity (RH).

This page intentionally blank

66 Monitoring of Deformations Induced by Crystal Growth of Salts beamsplitter

HeCd-laser (λ= 442 nm)

objectbeam

555

mirror

mirror referencebeam

fiber

specimen CCD objective or microscope objective

20 mm

Fig. 66.1. Scheme of ESPI setup used

To create the image of the object, a CCD-camera (Adimec MX 12P) with standard objectives was used. It has a dynamic resolution of 8 bit and 1,024 × 1,024 pixels. The eﬀective pixel size is about 10 µm. If high resolution is necessary in object space, a microscope objective with a numerical aperture of 0.42 can be used for imaging. The Mitutoyo inﬁnity corrected objective M Plan APO 20 has been chosen because of its long working distance of 20 mm. This leads to a minimum ﬁeld of view of about 230 µm2 and to a lower spatial resolution limit in object space of about 3 µm. Further details can be found in [4]. As in conventional holography not only the intensity of the light scattered by the object under investigation but also its phase has to be recorded. Therefore, a part of the laser light is separated by a beam splitter and acts as a reference beam. It is superimposed onto the object image thus forming an image plane hologram directly on the CCD array. The source point of this reference beam is located in the plane of the aperture with a small but well-deﬁned horizontal shift out of the center of the aperture. This so-called “spatial phase shift method” permits the calculation of the saw tooth image of the object wave modulo 2π [5]. For means of illustration, a typical result of ESPI is shown in Fig. 66.2. In this example a standard system instead of a microscopic ESPI was used to measure the tiny denting of a metal plate from the rear by a screw, which is outlined in the left part of Fig. 66.2. The phase maps of the interferograms representing the two states of the object, before and after denting, are calculated. In the middle of Fig. 66.2, these phase maps encoded in gray values are shown. The subtraction of the two phase maps then leads to the diﬀerence

556

G. Gülker et al.

Fig. 66.2. Example of a dented metal plate measured by ESPI. Left: sketch of loading the object. Middle: 0 light phase maps, before and after denting. Upper right: phase diﬀerence. Lower right: 3D representation of deformation. Maximum deformation is about 1 µm

phase modulo 2π, also known as saw tooth image which, in the case of the denting, is a concentric fringe system. The fringes can be interpreted as contour lines of deformation. Knowing the deformation diﬀerence corresponding to adjacent fringes, the fringe system can be demodulated and the amount of deformation can be determined. This is shown in a 3D representation in the lower right of Fig. 66.2. In this example the fringe spacing is about 350 nm, thus yielding a maximum deformation of about 1 µm.

66.3 Experimental Results To investigate the dilation caused by phase transitions of salts in our experiments magnesium sulfate (MgSO4 ) was used. This salt is known to be deleterious based on crystallization tests and is subject to a number of diﬀerent phase changes depending on temperature and relative humidity, which can clearly be deduced from its phase diagram shown in Fig. 66.3. In this temperature– humidity diagram, only the stable phases in the temperature range from about 0 to 100◦ C are considered. As a porous substrate, fritted glass with a very narrow pore size distribution was chosen. This material produced by Schott, Germany, has a mean pore radius of about 6 µm. It has the big advantage that its chemical composition is simple and that its humidity expansion coeﬃcient is known to

66 Monitoring of Deformations Induced by Crystal Growth of Salts

557

100

solution

80 70

epsomite

60 50 40

he xa hy dr ite

relative humidity (%)

90

kieserite

RH variation during experiment

30 20 0

20

40

60

80

100

temperature (˚C) Fig. 66.3. Temperature–humidity diagram of the MgSO4 − H2 O system. Variation in RH during experiment marked

be extremely small. This guarantees that, in our experiments, any material expansion detected is exclusively caused by crystallization or hydration pressure and is not a response of the material itself. In our experiments small cubes of fritted glass (about 1 cm3 ) were soaked until saturation with 19% (w/w) MgSO4 solution. The cubes were then coated with a thin gold layer to prevent the illuminating laser light from penetrating into the bulk material. Thereafter, the cubes were freeze dried and in addition dried 3 weeks at 200◦ C in an oven to obtain low-hydrated MgSO4 (kieserite). In a ﬁrst experiment a contaminated specimen was placed in the climatic chamber at room temperature and low RH of about 20%. While increasing the humidity in one step to about 80% (vertical arrow in Fig. 66.3), ESPI images are recorded. In Fig. 66.4, left, an ESPI saw tooth image is shown resulting from a time period of 60 min after reaching 80% RH. The investigated area shown here is about 1 cm2 . On the left side of this saw tooth image, about one saw tooth fringe can be identiﬁed due to a small deformation of this area. On the right side of the saw tooth image, no deformation pattern can be seen. Instead, we clearly could observe a de-correlated area with increasing dimension. Measurements with high magniﬁcation showed the same loss in correlation and proved thus, that this highly de-correlated region was not produced by deformations of small subareas. Instead, it indicates signiﬁcant surface changes, possibly liquid on the surface. For clariﬁcation the experiment was stopped at this point and the sample was investigated by cryogenic SEM. Cryopreparation means that the cube was frozen with a cooling rate of more than 1,000 K s−1 by putting it into melting nitrogen (so-called SlushN2 ). Any existing liquid on the surface or in the pores is frozen without ice crystallization or drying artifacts and can be analyzed conﬁdently under

558

G. Gülker et al.

Fig. 66.4. Left: ESPI saw tooth image from a period of 60 min after reaching 80% RH shows de-correlation due to surface changes, area ca. 1 cm2 . Right: visualization of cross section by cryo-SEM exactly from the transition region identiﬁes unexpected solution (arrows)

vacuum conditions of the microscope. Glassy frozen samples are ﬁxed on a cooling tablet inside the microscope and examined at a temperature of about −150◦ C. The method is described in more detail in [6]. The cube was cut inside the microscope along the line indicated in Fig. 66.4, left. In Fig. 66.4, right, an SEM micrograph is presented showing a cross section exactly from the transition region where de-correlation stopped. And indeed SEM could identify the formation of liquid on and directly beneath the surface (see arrows in Fig. 66.4, right). Such liquids appear darker in the micrographs due to the lower atomic number of the solution compared with the surrounding solid material. The formation of a solution on the surface due to this humidity change was not expected theoretically, as can be deduced from Fig. 66.3. It shows that the thermodynamic model, which is the basis of Fig. 66.3, has to be revised. Instead of surface changes in the next experiments, we wanted to measure salt-induced deformations of the surface. In order to be able to measure these deformations, diﬀerent attempts were made to minimize the above-mentioned de-correlation eﬀects. After several unsuccessful approaches we found that by simply changing the order of sample preparation and using a platinum coating, the surface of the specimen could be made hydrophobic. In practice the samples ﬁrstly were coated and thereafter soaked with salt solution. With this procedure speckle de-correlations were nearly completely avoided and deformation measurements could be performed during the whole RH variation. In the following experiment, the RH was increased stepwise from nearly 0% to about 80% (Fig. 66.6). As a ﬁrst result we saw that high magniﬁcation imaging was not necessary since no small scale deformations occurred. Further on, due to the salt phase changes, the surface of the sample deformed nonhomogeneously. This is

66 Monitoring of Deformations Induced by Crystal Growth of Salts

559

1,0

100

0,8

80

strain x 10−4

RH 0,6

60

0,4

40

rel. humidity/%

Fig. 66.5. (Left) ESPI saw tooth image showing inhomogeneous deformation during a time period of about 70 min after reaching 70% RH, area ca. 1 cm2 ; (right) visualization of the deformation in a 3D representation

strain 0,2

20

0,0 0

200

400

600

800

1000

1200

0 1400

time/min

Fig. 66.6. Local strain of the center of the specimen vs. measuring time and RH. Increase of strain rate at about 60% RH corresponds to the deliquescence of kieserite

illustrated in Fig. 66.5, left, where an ESPI saw tooth image of the complete surface area of about 1 cm2 is shown representing the deformations during a time period of about 70 min at 70% RH. In Fig. 66.5, right, a 3D representation of the deformation is shown. It can be seen that the edges of the specimen deform more than the central part. SEM investigations showed that a compression of salt in the outer regions of the specimen was responsible for this behavior, which again was unexpected and is content of further investigations. Secondly, the local strain of the center of the specimen was evaluated by temporal phase unwrapping [7]. In Fig. 66.6, the measured strain is plotted

560

G. Gülker et al.

vs. time and RH. It can be seen that immediately after increasing RH to about 30% the sample started to deform signiﬁcantly. Moreover, it showed that the strain rate increases considerably at about 60% RH, which is the deliquescence of kieserite at about 20◦ C. In contrast to theoretical predictions, these results again prove that the direct transition from kieserite to hexahydrite or epsomite is inhibited. Instead, kieserite ﬁrstly dissolves and recrystallizes immediately to hexahydrite or epsomite, leading to pronounced dilations.

66.4 Conclusions An ESPI setup for detecting surface changes, deformations, and strain was presented. It was shown that unexpected results were obtained suggesting the adaptation of the basic thermodynamic model. The investigations proved that full-ﬁeld information gained by ESPI measurements combined with cryogenic SEM could lead to new insight in salt-weathering processes, which could hardly be obtained by commonly used tactile dilation measurement. Acknowledgment Financial support by Deutsche Forschungsgemeinschaft (DFG) for this research is gratefully acknowledged.

References 1. 2. 3. 4.

M. Lavalle, in Compt. Rend. Vol. 36 (1853) 493. K.D. Hinsch and G. Gülker, in Phys. World 37, 2001. R.S. Sirohi, Speckle Metrology, Marcel Dekker, New York, 1993. A. El Jarad, G. Gülker, and K.D. Hinsch, in Proc. Speckle Metrology, SPIE Vol. 4933, 335, Bellingham, 2003. 5. D.C. Williams, N.S. Nassar, J.E. Banyard, and M.S. Virdee, in Opt. Las. Tech. Vol. 23, 147, 1991. 6. M. Langenfeld, H. Juling, and R. Blaschke in Wiss. Z. Hochsch. Archit. Bauwes.–Weimar 40, 23, 1994. 7. J. Huntley and H. Saldner, in Appl. Opt. Vol. 32.17: 3047, 1993.

67 Cultural Heritage Documentation by Combining Near-Range Photogrammetry and Terrestrial Laser Scanning: St. Stephen’s Cathedral, Vienna F. Zehetner1 and N. Studnicka2 1

2

Masons’ Lodge of St. Stephen’s Cathedral, 1010 Vienna, Austria, [email protected] RIEGL Laser Measurement Systems GmbH, 3580 Horn, Austria

Summary. A powerful sensor system providing both high-resolution textures and highly accurate 3D geometry information is created by combining near-range photogrammetry and terrestrial laser scanning. As both sensors are integrated closely into a single system, the textures can be applied to the 3D data automatically and with high precision. These sensors have proven as extremely valuable tools in applications of cultural heritage, architecture, and archaeology. We demonstrate the capabilities of the RIEGL LMS-Z420i system with an integrated high-resolution camera by presenting the work ﬂow of data acquisition and postprocessing performed for modeling St. Stephen’s Cathedral in Vienna, Austria, with an emphasis on the construction of CAD models.

67.1 Objectives of the Project It turns out more and more necessary to record restoration work accurately not only to prove the quality of restoration but also to provide long-term detailed documentation. To obtain the indispensable exact plans for some monuments, expensive photogrammetric maps (e.g., for St. Stephens in a scale of 1/50) were commissioned. As unscalable analogue drawings, they can hardly be used by modern technologies and are not likely to be sustainable. Therefore, a research project was initiated aiming at an economic way to produce an exact three-dimensional basis for damage recording and general mapping on monuments. The restorer/stonemason should be able to handle the 3D-based damage recording system to do the regular documentation of restoration work. The system should be compatible with commercial databases providing information not only on damages and restorations, but also on history and art (CIS cathedral information system). In times of short public money, the cost-factor becomes increasingly important; high accuracy is, however, necessary to watch development of damage and to analyze the nature of the building. The mapping team too should be capable to control accuracy, as mistakes happen in photogrammetric measurements. Irregularities serve as

562

F. Zehetner and N. Studnicka

indicators for construction steps astray from original plans, static problems or later changes as, for example, the reconstruction after the Second World War [1–10]. 67.1.1 2D-Recording on Paper or Computer with Enhanced CAD-Programs Paper-based mapping is limited to a ﬁxed scale, allowing only raw description and accuracy of groups of phenomena, due to limitation of discernable hatchings to signify the aﬀected area. Computer aided mapping provides a diﬀerentiated glossary to describe the phenomena and a wide range of scale, where even small, but critical damage can be shown even in overview-scale. Damage and measures of restoration can be located and described easily. Problems of 2D-Recording, however, are the invisibility of some areas on the map masked by other pieces, and the impossibility to show spatial correlations and curved surfaces such as vaults or the exact shape of pillars. The novelty therefore consists of an expansion of this method to 3D mapping.

67.2 Data Acquisition 67.2.1 System Description The whole system is battery powered and portable, but yet robust and operable in a wide range of environmental conditions. Data acquisition, sensor conﬁguration, data processing, and storage are eﬀectuated by the companion software RiSCAN PRO (Table 67.1). The characteristics of the most powerful instrument of the RIEGL LMS scanner series, the LMS-Z420i, are its narrow beam divergence, its wide operating range, and its excellent single-shot accuracy. As a consequence, the raw scan data provide a precious basis for various postprocessing techniques. The scanning system is perfectly complemented by the photogrammetric method Table 67.1. Key system speciﬁcations hybrid sensor measuring range ranging accuracy beam divergence measuring rate scan range scan resolution camera resolution camera lenses used specs of lighting

RIEGL LMS-Z420i with Canon EOS 1Ds up to 1.000 m at target with 80% reﬂectivity 10 mm (single shot), 5 mm (averaged) 0.25 mrad 8,000 points per s 0–80◦ vertically, 0–360◦ horizontally up to 0.004◦ approx. 12 megapixels 20 mm focal length sunlight color/3 lamps, 2.5 kW each

67 Cultural Heritage Documentation

563

(high pixel density, hence deﬁnition and vectorization of edges and high accuracy of details). The hybrid sensor provides geometry data from the laser scan and image data from the camera. These data can be automatically processed to generate products such as textured triangulated surfaces or orthophotos with depth information. Additionally, all tools developed for image analysis, such as edge detection, can be used for direct extraction of 3D content from the combined image data and scan data. The collected data therefore provide a solid and large database for any measurements related to construction, destruction, and decay of the scanned object. 67.2.2 Strategy For scanning a complex building, a strict workﬂow strategy should be respected. In general it is helpful to create a logical, hierarchic dissection structure of the building (Fig. 67.1). At St. Stephens, the established dissection structure deﬁning pillars and bays was adapted to the needs of IT and helps to structure even the archives. For the interior of the cathedral, 37 vaults, 18 freestanding pillars, 36 wall-pillars, and 42 wall-parts are deﬁned. N09, e.g., assigns the ninth bay counted from west in the north nave. It is necessary for postprocessing to name each scan position according to the dissection structure of the building. In a single scan, one construction part should be recorded as completely as possible under constant illumination – “object-oriented scanning.” To follow this guideline, we scanned thrice in every bay, i.e., two times in diagonal direction (to record the pillars) and once parallel to the axis of the church to record the walls. To stitch all scans immediately after the recording, the approx. 300 retroreﬂecting markers were measured previously with a total station in the national coordinate system, thus permitting a permanent check on accuracy (Fig. 67.2).

Fig. 67.1. Preparation of the scanning mission: map of the dissection structure

564

F. Zehetner and N. Studnicka

Fig. 67.2. Checking the data during acquisition: Point cloud and dissection structure

67.2.3 Data Acquisition Due to special circumstances of a cathedral, all data acquisition had to be done at night in order not to bother services and for tourists not to disturb the measurement proceedings. Besides, professional artiﬁcial light at night guarantees regular illumination regardless of ambient conditions. Compared with the advantages of night work, the fact of not recording the remaining mediaeval stain-glassed windows is a minor disadvantage. The speciﬁc challenge, however, was to supply constant illumination of the photographed area. The distance between recording system and object varied from 5 to 25 m. We used three ﬂoodlights to achieve homogeneous brightness. It took seven nights from 8:00 p.m. to 6:30 a.m. to acquire all necessary data of the interior of the cathedral (Figs. 67.3 and 67.4).

67.3 Postprocessing 67.3.1 Point-Cloud/Merging/Data Reduction We took 130 positions to record the entire cathedral (Fig. 67.5). At every position, 2 M points and ten photos were taken. So the ﬁnal data collection consisted of 1,300 photos and a total amount of raw data of approx. 15 GB. To enable the representation and processing of the whole model, data were resampled with a resolution of 5 cm. The resulting point cloud of the cathedral (20 M pixel) can be split into single elements, layers, or sections. Of course, for detailed processing of every step of the workﬂow, the original data can

67 Cultural Heritage Documentation

565

Fig. 67.3. Photograph of the system in daylight in front of the cathedral

Fig. 67.4. Photograph of the whole system in operation at night within the cathedral

566

F. Zehetner and N. Studnicka

Fig. 67.5. Segmentation in about 120 basic elements of the interior

Fig. 67.6. Framework of the cathedral’s irregular coordinate system

be retrieved. However, with regards to the present capacity of computers, it cannot be considered as useful to load more than 5–10 scans simultaneously (Fig. 67.6). 67.3.2 Triangulation The result of scanning, the so-called “point cloud,” does not form a surface, but only isolated points. To gain a surface useful for further processing, the points have to be triangulated or “meshed.” There are two ways to achieve a triangulated surface: – All points are connected to a more or less complete mesh. For vaults or walls this is a good method, in other cases – as in the pillars – the relevant structure is covered by sculptures or altars. Compound pillars have many areas hard to be reached by the laser beam, because they are intrinsically shadowed. Triangulation of our model was eﬀectuated by the software “RiSCAN PRO” (http://www.riegl.com). – Another method is the so-called “Monoplotting.” A corpus is modeled by CAD based on scan- and photo-data (to a so-called “low poly model”).

67 Cultural Heritage Documentation

567

Fig. 67.7. Triangulated polygonal model of the ceiling

Fig. 67.8. Low polygonal CAD model within Phidias for MicroStation

Surfaces can easily be identiﬁed in the point cloud, edges in the congruent photos and thus a CAD drawing is generated. A compound pillar, for example, can be extruded along its elongation out of a section. For vaults, the caps are approximated by automatically generated segments of cylinders or cones. This simpliﬁcation showing cylinders and their axes gives important information about irregularities of the building and of damage and static problems. (The high steeple caused displacements in the foundation, but also in the vaults, subsidences caused by underground work can be shown, etc.). Of course, this method approximates the shapes, but missing details can easily be inserted in the overall complete model when required. Monoplotting is implemented in the software Phidias (see [11], http://www.phocad.de) (Figs. 67.7–67.9).

67.3.3 Textured Triangulated Surfaces To use a 3D surface model for damage recording, calibrated photos are mapped on the mesh. Both monoplotted and triangulated models were texturized in Riscan PRO. Previously, the CAD-model had to be converted into the

568

F. Zehetner and N. Studnicka

Fig. 67.9. Distorted axes of vault-caps in the choir bay next to the steeples

STL-format. For using the textured 3D surface model in the mapping-software “aSPECT 3D” (see [12], http://www.arctron.de); it was exported into WRLformat.

67.4 3D Mapping/Damage Recording 67.4.1 Applied 3D-Recording For 2D recording of St. Stephens, a catalog of approx. 6,000 phenomena – describing the status of the building and the activities of preservation – has been established. The catalog consists of 12 basic groups – materials (stone, plaster, metal, etc.), damage, and measures of restoration. This is a quite complex system of description, but it has proved very useful for our purpose and has been applied in a 2D system by craftsmen and restorers of the lodge for several years and, besides, it is a tool that can be used even by nonspecialists. Due to the size of the monument and the manifold attributes it is necessary to establish a powerful database in the system to execute queries for combinations of speciﬁc phenomena of the whole cathedral (e.g., sandstone that was treated with a method that have turned out problematic). The handling of 3D objects turned out to be easier than expected. For architectural objects it is not necessary to turn the object and to draw on it at the same time. 67.4.2 Recording in CAD for Low Polymodels Presently, no CAD application providing a convenient function for mapping and drawing on a 3D surface is available on the market. However, it would be possible to combine a number of diﬀerent functions mapping drawn objects to a 3D surface. This is only a ﬁrst step, but due to existing database engines

67 Cultural Heritage Documentation

569

Fig. 67.10. Detail of 3D mapping/damage recording of wall element “N09”

of CAD software, it could be possible to develop a functional mapping tool based on 3D CAD. 67.4.3 Recording in aSPECT 3D for Triangulated Models For triangulated surfaces, the software “aSPECT 3D” with the necessary options was developed. The advantage of this solution is the ability of exact drawing on the surface – the lineation of the mapping expert is independent of triangles. For the deﬁnition of the parts of the surface (e.g., a particular piece or damage), the triangles will be cut on the drawing line. The surface is textured and can be drawn like a vaulted paper. A database connection is already established, further developments are foreseen. Our recommendations for the future functionalities are: – Possibility to handle several overlapping mapping objects on a 3D surface (e.g., stone/damage/measure) – Possibility to deﬁne a more complex database, which can handle several ﬁles of 3D mappings that a complex building has to be divided into for better handling (Fig. 67.10) – Query to ﬁnd any combination of phenomena

67.5 Conclusion and Outlook There are systems to: – Organize a “virtual cathedral” – Describe the phenomena and the condition of a monument in 2D

570

F. Zehetner and N. Studnicka

– Easily produce 3D data in both mesh and low poly CAD models – Do mapping in 3D for speciﬁc topics What we need and are working on is to combine all these topics to allow craftsmen, restorers, and not-computer-skilled users to carry out accurate documentation, using standard laptops for a 3D model that can be easily generated and provides suﬃcient accuracy. The system should be able to combine models of diﬀerent sources and in diﬀerent accuracy according to the nature of the object (e.g., architecture/sculpture = laser scanner/stripe light scanner, etc.). The system should be regarded as a tool to be handled as easily as a “pencil on a curved surface.” Apart from the “classic” 3D model application like showing the cathedral or visualize projected adaptations (pictures, new altars, acoustic adaptations, new stain-glassed windows, etc.), the exact documentation provides a precious tool for preservation itself and for its sustainability. Acknowledgments The project was sponsored and equipped by RIEGL Laser Measurement Systems (http://www.riegl.com). The team of the Technical University was ﬁnanced by the Masons’ Lodge of St. Stephen’s Cathedral. For the ﬁeld work and postprocessing work, we thank Hanna Liebich (building archaeology), Robert Kalasek (head of scanning team), Michael Wimmer (head of postprocessing team), Christian Kurtze (surveying team), Josef Tschannerl (surveying team), Christoph Getzner (general assistance), scanning and postprocessing team: Martin Carnogursky, Michal Idziorek, Anton Frühstück, and Hannes Witzmann.

References 1. A. Ullrich, N. Studnicka, and W.A. Neubauer, in Workshop 9, Archäologie und Computer, Vienna 2004. 2. A. Ullrich, N. Studnicka, and J. Riegl, in Proceedings of SPIE Vol. 5412, 241, 2004. 3. N. Studnicka, U. Riegl, and A. Ullrich, in Photogrammetrie, Laserscanning, Optische 3D-Messtechnik, Beiträge der Oldenburger 3D-Tage 2004, Wichmann, 2004. 4. A. Ullrich, R. Schwarz, and H. Kager, Using hybrid multi-station adjustment for an integrated camera laser-scanner system, Optical 3-D Measurement Techniques IV, Volume I, 298, Zürich, 2003. 5. A. Ullrich, R. Schwarz, and H. Kager, in Österreichische Zeitschrift für Vermessung & Geoinformation, Vol. 91. (4), 281, 2003. 6. St. Groh, W. Neubauer, N. Studnicka, and A. Ullrich, in Jahresberichte des Österreichischen Archäologischen Instituts, 2003.

67 Cultural Heritage Documentation

571

7. A. Ullrich, N. Studnicka, J. Riegl, and W. Neubauer, in Workshop 7 – Archäologie und Computer, Vienna, 2002. 8. A. Ullrich, et al., Optical 3-D Measurement Techniques V, Conference Proceedings, 2, 2001. 9. RIEGL Laser Measurement Systems GmbH, www.riegl.com 10. Masons’ Lodge of St.Stephen’s Cathedral, www.dombauwien.at 11. PHOCAD GmbH, www.phocad.de 12. ARCTRON GmbH, www.arctron.de

68 Laser Engraving Gulf Pearl Shell – Aiding the Reconstruction of the Lyre of Ur ∗

C. Rawcliﬀe1 , M. Aston1 , A. Lowings2 , M.C. Sharp1 , and K.G. Watkins1,3 1

∗ 2 3

Lairdside Laser Engineering Centre, Campbeltown Road, Birkenhead, CH41 9HP, UK [email protected] 15 Church Street, Northborough, Peterborough PE69BN, UK Laser Group, Department of Engineering, University of Liverpool, Brownlow Street, Liverpool L69 3GH, UK

Summary. The Lyre of Ur was one of two bull’s-headed harp instruments that British archaeologist Sir Leonard Woolley recovered from a mass grave during his highly publicised excavations of Ur’s Royal Cemetery in the 1920s and 1930s in Iraq. Recreated unplayable models of the Lyre are held in Pennsylvania, London and, until recently before it was vandalised, the Baghdad Museum. It is believed that the original Lyre is approximately 4,750 years old. An attempt to recreate an authentic playable version is being spearheaded by Mr. A. Lowings of Stamford. The instrument is to be ornamented by a golden-sheeted bull’s head and geometrically inlaid lapis lazuli as well as engraved gulf pearl shell plaques depicting Sumerian images, which are to be placed on the front of the lyre. Authentic cedar wood from Baghdad was used to create the main framework of the lyre. Due to the intricate designs required for the shell plaques, an investigation to laser engrave them was undertaken. The main objective when laser engraving the pearl shells (which are chieﬂy calcium carbonate) was to achieve suﬃcient depth without compromising image quality so that the plaques could be backﬁlled with bitumen in order to replicate the originals. Moreover, it was imperative to recreate the images without damage to the surrounding pearl either by scorching or re-depositing removed material. Experiments were carried out to engrave the shells utilising a Synrad CO2 laser which has galvanometer controlled mirrors to direct the beam which enables it to be scanned in an XY direction across the surface of the material. An image software package was used in conjunction with the laser so that the images could be manipulated in terms of size and positioning. It was necessary to remove excess material produced between successive laser passes using a soft brush to allow full penetration of the laser beam to the newly created lower surface level. Successful engraving was achieved using multiple laser passes and an identiﬁed optimum-processing window.

574

C. Rawcliﬀe et al.

68.1 Introduction The world of art conservation has for some time now embraced the use of laser technology to optimum eﬀect due its versatility, controllability, selectivity and environmental compatibility [1,2]. These unique characteristics have been demonstrated eﬀectively in the treatment of many works of art, in particular, laser cleaning [3–8]. Traditional conservation techniques have in the past utilised solvent and mechanical removal techniques, which have caused lasting damage [9]. Hence steps were taken to develop novel laser systems capable of tackling everdemanding conservation challenges as well as preventing damage to the precious artefacts. This chapter discusses the use of laser technology for engraving mother-ofpearl shell plaques as part of the reconstruction of a 4,000-year-old Sumerian Treasure.

68.2 Background Amongst the estimated 170,000 valuable antiquities that ﬁlled Iraq’s Baghdad Museum prior to its tragic looting in April 2003, some of civilization’s oldest musical instruments were proudly exhibited. One such instrument was The Golden Lyre of Ur which was one of more than a dozen Sumerian stringed instruments discovered at the ancient site of Ur in 1929. Headed by British archaeologist, Sir Leonard Woolley, a multi-national expedition excavated the instruments at the “Royal Graves of Ur”. These burial sites yielded some of Mesopotamia’s most cherished artefacts providing an astounding wealth of new information about Sumerian culture and the origins of music in civilization [10, 11]. The Golden Lyre was found in the grave of “Queen” Pu-Abi along with 74 bodies, presumed to be sacriﬁcial victims, and numerous ﬁne jewellery, stone and metal vessels. One body was found draped over the Lyre, the bones of her hands were placed where the strings would have been (Figs. 68.1 and 68.2). The wooden parts of the Lyre had decayed in the soil, but Woolley poured Plaster of Paris into the depression left by the vanished wood and so preserved the decoration in place. The front panels were made of lapis lazuli, Gulf shell and red limestone originally set in bitumen and helped determine the instrument’s shape. The gold mask of the bull decorating the front of the sounding box had been crushed. Strips of waxed cloth were place over the Golden Lyre to remove it from the ground. The cloth acted like a sheet of glue, holding the instrument’s sound box, arms, cross bar and inlay in place. The instrument was then taken from the excavation to be restored. When the waxed cloth dried, sketches and accurate measurements of the Golden Lyre were made. New wooden parts were crafted and the ancient mosaic borders

This page intentionally blank

68 Laser Engraving Gulf Pearl Shell

575

Fig. 68.1. Great death pit sketch, depicted by Sir Leonard Woolley

Fig. 68.2. Excavation site of Golden Lyre of Ur in 1929

and decorations were transferred from the cloth to the new wooden reconstruction [12–14]. The lyre remained in the Baghdad Museum until it was looted and destroyed. This led to the English harp enthusiast, A. Lowings, setting about recreating this piece of history.

68.3 Method Original Sumerian images, which were depicted on the Lyre, were obtained in a bitmap format courtesy of Loughborough University. The images were transferred onto a graphics drawing canvas, which is part of a laser marking software package. The software was used in conjunction with a Synrad CO2 laser marker. The laser delivers a near collimated beam of 10.6 µm wavelength at a maximum power in the order of 25 W with a beam diameter of approximately 200 µm. The beam is delivered to the workpiece using a scanning head system,

576

C. Rawcliﬀe et al.

which contains two galvanometer driven orthogonal mirrors. These mirrors direct the beam through a special focusing optic, which converts the angular deviation of the beam into a linear displacement of the beam within a planefocussing ﬁeld. The controllable laser parameters are: laser power (10−100%), the traverse speed (mm s−1 ) and the resolution. In eﬀect the resolution, given in DPI (dots per inch), determines the overlap of adjacent passes. Systematic changing of each individual laser parameter was undertaken in order to establish a processing window which not only clearly ablated surrounding shell to create each engraving but also to prevent re-depositing of ablated material and scorching of the underlying shell. Light brushing across the surface of the shell was undertaken between each successive laser pass to remove ablated material and prevent the laser interacting with it. The ablated material was collected and observed using optical microscopy. Each shell plaque engraved was 5 cm × 5 cm. Plaque thickness was 2.8 mm and a depth of 1.2 mm was required in order for bitumen backﬁll to be successfully applied.

68.4 Results and Discussion Successful engraving of the shell was achieved operating the laser system at full power with a scan velocity of 500 mm s−1 and a DPI of 600. The shell was positioned at the focus of the laser approximately 242 mm below the laser scanning head. Six consecutive passes of the laser were used and removal of ablated debris was necessary between successive passes by lightly brushing the surface. The etch depth of the shell for each laser pass and removal rates were calculated and are shown in Figs. 68.3 and 68.4. One complete laser pass took approximately 185 s. Figure 68.3 shows that in order to achieve the required etch depth of 1.2 mm, six passes of the laser were necessary. It was assumed that the same

Fig. 68.3. Etch depth achieved with each laser pass

68 Laser Engraving Gulf Pearl Shell

577

Fig. 68.4. Ablation rate of shell with time

volume of material was ablated with each laser pass and on further investigations this was proven, as accurate measurements of etch depth were taken between each pass using vernier calipers on test pieces of shell. Figure 68.4 shows the shell ablation rate with time. It is seen to steadily increase with time. The graph covers the time period of the ﬁrst laser pass. From known dimensions of the shell plaques, required etch depth and time taken to complete one pass, a simple calculation was made to determine ablation rate. This was found to be 2.8 mm3 s−1 . Figures 68.5 and 68.6 show the ﬁnished laser engraved shell plaques. Figure 68.5 is prior to the backﬁlling of bitumen and Fig. 68.6 is with bitumen backﬁll added in order to replicate the originals. Figure 68.7 shows the shell inserts on the replica Lyre. Figure 68.8 shows the completed replica Lyre. A heat conduction analysis is presented to characterize the ablation of the shell by the laser radiation. A one-dimensional thermal model is considered. Let us suppose that: 1. A uniform heat ﬂux (I0 ) impinges on a planar semi-inﬁnite surface during time t. 2. The heat is conducted inward into the material so that the temperature varies with depth (z). 3. The thermal properties of the material are constant and do not vary with temperature. 4. No overlapping takes place in the scanning, hence a pulse length ∼1 ms was calculated using spot size, distance covered and time to complete scan. For a stationary system and heat ﬂow in one dimension, the temperature is given by: A(z, t) 1 ∂T (z, t) =− , (68.1) ∇2 T (z, t) − k ∂t K

578

C. Rawcliﬀe et al.

Fig. 68.5. Laser engraved shell plaques before backﬁll

Fig. 68.6. Laser engraved shell plaques with bitumen backﬁll

68 Laser Engraving Gulf Pearl Shell

579

Fig. 68.7. The replica mother-of-pearl shell inserts in situ

where T (z, t) is the temperature at distance z after time t; A(z, t) the heat produced per unit volume and per unit time as a function of position and time; K the thermal conductivity; and k is the thermal diﬀusivity. If a constant ﬂux I0 is absorbed at the surface (z = 0) and there is no phase change in the material, the solution of the above equation is: z 2αI0 √ √ kt ierfc T (z, t) = , (68.2) K 2 kt where ierfc is the integral of the complimentary error function. At the surface (z = 0): 2αI0 kt , T (0, t) = K π

(68.3)

where α = absorptance (= 1− reﬂectance) and I0 is the incident ﬂux. In order to calculate temperatures on the surface of the shell with consecutive laser passes, it is assumed that due to surface roughening eﬀects and hence multiple reﬂections, the absorptivity of the material will increase. All constants used are for that of typical minerals. Predicted temperature rises, calculated using (68.2) based on absorptivity increases between 0 and 85%

580

C. Rawcliﬀe et al.

Fig. 68.8. The replica Lyre with its gold bullhead and other decorative features

were between 423 and 650◦ C, which is well below the melting temperature of CaCO3 , which is 825◦ C. Hence it becomes apparent that the mechanism in which the material is removed involves little or no vaporisation, which was evident from the need to constantly remove solid ablated material from the shell surface between each successive laser pass. Figures 68.9 and 68.10 show optical micrographs of the ablated material. It is evident from the image that no melting has taken place due to the irregularity in shape of the particulate material. For organics and geological materials such as mother-of-pearl, the important mechanism is due to photomechanical fracture (laser induced stress) which induces stresses in the material in excess of the material strength causing fracturing and rupturing of the surface. Due to the pulse length being of the order of milliseconds, this “particle ejection” mechanism is consistent with other work in this ﬁeld, which suggests that mechanisms at shorter pulse lengths involve melting, vaporisation, ionisation and shock waves.

68 Laser Engraving Gulf Pearl Shell

581

Fig. 68.9. Optical micrograph of ablated material, magniﬁcation ×10

Fig. 68.10. Optical micrograph of the ablated material, magniﬁcation ×10

68.5 Conclusions The use of a scanning, low power CW CO2 laser is an extremely useful tool for the engraving of diﬃcult materials such as mother-of-pearl. Depending on chosen parameters, it is possible to remove small volumes of material and hence allow the user greater scope to engrave intricate and ﬁne designs by reducing the threat of scorching or destroying the workpiece. The longer pulse lengths suggest that laser-induced stresses on the surface are responsible for ejection of the shell in particle form, which was consistent with microscopic observations of the material ablated. Overall, it can be stated that the laser technology used aided the reconstruction of the Lyre of Ur in a positive way.

References 1. J. M. Lee and K. G. Watkins, in The Laser Industrial User Vol. 18, 29, 2000. 2. J. M. Lee and K. G. Watkins, in In-process monitoring techniques for laser cleaning, Optics and Lasers in Engineering, Vol. 34, 429, 2000.

582

C. Rawcliﬀe et al.

3. K. G. Watkins, in Lasers in the Conservation of Artworks, (W. Kautek, Ed), Wien 1997, 7. 4. K. G. Watkins, C. Curran, and J. M. Lee, in Journal of Cultural Heritage Vol. 4, 59, 2003. 5. M. I. Cooper, D. C. Emmony, and J. H. Larson, in J. Photographic Science Vol. 40, 55, 1992. 6. J. F. Asmus, in Interdisciplinary Science Reviews, Vol. 12, 170, 1987. 7. J. F. Asmus, in Light cleaning, laser technology for surface preparation in arts and conservation, Vol. 13, 14, 1978. 8. C. Curran, Laser Cleaning Microelectronic components for the Semiconductor Industry, PhD Thesis, University of Liverpool, UK, 2002. 9. R. Reed, Ancient skins,parchmnets and leathers, Seminar Press, London and New York, 1972. 10. V. Greene, Publications of the joint expedition of the British Museum and the Museum of The University of Pennsylvania to Mesopotamia, Vol. 9, 47, 1963. 11. Proceedings of American Institute for Conservation of Historic and Artistic Works, 37, 1979. 12. D. Colon, in British Archaeological Reports, Oxford 1982. 13. P. Benn, Excavations at Ur, 1968. 14. L. Woolley, Discovering the Royal Tombs at Ur, 1969.

69 Fluorescence Lidar Multispectral Imaging for Diagnosis of Historical Monuments, Övedskloster: A Swedish Case Study ∗

R. Grönlund1 , J. Hällström2 , S. Svanberg1 , and K. Barup2 1 ∗ 2

Atomic Physics Division, Lund Univeristy, P.O. Box 118, 221 00 Lund, Sweden [email protected] Department of Architecture and Built Environment, Lund University, P.O. Box 118, 221 00 Lund, Sweden

Summary. A ﬂuorescence lidar measurement has been performed on the castle Övedskloster in Sweden. A mobile system from the Lund University was placed at ∼40 m distance from the sandstone façade. The lidar system, which uses a frequencytripled Nd:YAG laser with a 355-nm pulsed beam, induces ﬂuorescence in each target point. Areas were studied by using whisk-broom scans. The possibility of detecting biodeteriogens on the surface and characterization of materials was conﬁrmed. The method can be a tool for conservation planning and status control of the architectural heritage where ﬂuorescence light can point out features that are not normally visible under natural illumination.

69.1 Introduction The possibility of in situ optical remote investigations is an important instrument for the ongoing maintenance and care for our architectural heritage. Also, for future conservation and restoration projects, the non-destructive methods of investigation are vital. The ﬂuorescence technique is well known in areas such as remote sensing of marine oil spills and medical diagnostics. The use of laser-induced ﬂuorescence (LIF) for building investigation was introduced in ﬁeld studies of, for example, the Lund Cathedral in Sweden and the Cathedral and Baptistery in Parma, Italy [1, 2]. The technique allows us to reveal aspects that are not evident to the naked eye or photography, and makes it possible to extend the application of ﬂuorescence spectroscopy to the outdoor environment with sometimes large distances and uncontrollable background light and with no need for scaﬀolding or taking samples. The aim of the ﬂuorescence lidar method is to use both point monitoring and scanning of areas (whisk-broom technique) to provide multispectral images in order to distinguish and identify diﬀerent materials and biodeteriogens and other substances not yet visible to the human eye.

584

R. Grönlund et al.

These colour-coded images could then form a basis for the decision-making and analysis of the historical building façade. The possible application of the technique for the cultural heritage ﬁeld includes: detection and identiﬁcation of biodeteriogens on the façade, characterization and identiﬁcation of stone materials, and identiﬁcation of treatments performed on the surface. Biodeteriogens can be identiﬁed through the presence of chlorophyll that has strong absorption bands in the UV–visible region and ﬂuoresces in the near infrared. The ﬂuorescence spectral features from stone materials can be used when identifying diﬀerent lithotypes on a building surface and also for localizing past treatments. By processing of the spectra, the speciﬁc materials and their distribution can be pointed out in thematic maps or in images accessible for the cultural heritage sector. A study of Övedskloster was performed during periods between May and October 2004. In May 2004, ﬂuorescence lidar measurements were made with the mobile unit from Lund University over a period of 3 days and, in October 2004, an additional survey and measurements of the areas with traditional methods were carried out (Fig. 69.1). The study was divided into three main areas: the courtyard portal, the balustrade between the west pavilions and parts of the main façade facing the courtyard. The primary aim of the study was the possibility of identifying materials and phenomena on the surface, their origin, biodeteriogens present and their extent and possible treatments used on the sandstone in previous conservation projects.

Fig. 69.1. The mobile system of Lund Institute of Technology parked in the Övedskloster courtyard for the ﬁeld study in May 2004

69 Fluorescence Lidar Multispectral Imaging

585

The Swedish Rococo style castle Övedskloster with its surroundings is a listed and entailed estate. It is situated in the province of Scania in the south of Sweden. A monastery from the Premonstratenser order, which was founded in the twelfth century was previously on this site [3]. During the eighteenth century the estate came into the Ramel family possession and a new castle was built during 1763–1776, under the architects C. Hårleman and J. E. Rehn, of French model and with magniﬁcent Rococo interior. The castle, its pavilions and the courtyard portal are built of the red Swedish sandstone called Öved. The stone was quarried at a nearby opencast mine, which is now closed. The deposits form a 5–10-m thick bank of red and soft stone of Upper Siluric sediment, with a porosity at about 10% [4]. By Swedish standards this is relatively young sandstone. The light red colour derives from oxidized iron compounds and, in addition to quartzite mineral, the stone contains feldspar, mica and calcite. The stone has a somewhat uneven quality and is vulnerable to environmental exposure. The sandstone shows signs of algal growth, weathering, exfoliation and water rising through the capillary systems with salt eﬄorescence and, on some areas, greater damage, causing parts of the stones to fall oﬀ. On some parts, the old Öved stone, being much weathered, has been replaced with new stones of Orsa sandstone, a similar red Swedish sandstone [5].

69.2 Methods The technique used in these measurements was laser-induced ﬂuorescence, where an ultra-violet laser pulse is directed to the point of interest. The laser excites atoms and molecules in the material which then relax and send out ﬂuorescence light. The exciting wavelength is not critical in the sense that the atoms and molecules in the solid target have smeared-out energy levels, but the results may be diﬀerent with diﬀerent excitation wavelengths, and a combination of wavelengths may give extra information [6]. With an excitation pulse of 3–4 ns, the ﬂuorescence is temporally conﬁned which makes it possible to detect using a gated detector. In this way background light can be suppressed. The measurements were performed using the mobile lidar system of Lund University (Fig. 69.2). The laser radiation from a tripled Nd:YAG laser at 355 nm was sent to the point of interest through a rooftop dome with computer-controlled mirrors that can direct the light to the desired point. The laser light was used to induce ﬂuorescence in the stone at ∼40 m distance and the ﬂuorescence radiation was collected using an on-axis Newtonian telescope. The radiation was ﬁltered with a coloured glass ﬁlter which suppresses the elastic reﬂex. An optical multichannel analyser (OMA) system with a timegated image intensiﬁer and a CCD camera captured the ﬂuorescence radiation in the approximate wavelength range 400–800 nm for each measurement point,

586

R. Grönlund et al.

Fig. 69.2. The mobile lidar system used in the study. Modiﬁed from [7]

averaging over 40 laser shots. The laser beam, with a diameter of about 3 cm, was then swept over the area of interest to gather the spectrum in each point.

69.3 Measurements The measurements performed on Övedskloster concerned three areas of the castle, where the material and analyses from two speciﬁc parts will be discussed in this chapter, namely the south side of the portal and one of the sandstone urns on the Attica on the main façade. The ﬁeld study with the ﬂuorescence lidar system from Lund took place during 3 days with relatively warm and sunny weather. The mobile unit was parked on the inner courtyard where the possibility to move the dome vertically and horizontally enabled the measurements to be performed on the points and areas chosen. The distance to the main façade was 44 m and, to the portal, 38 m. The main portal chosen in the study has a core of bricks and is covered in Öved sandstone with a family coat-of-arms and two medallions surrounded by rich ornaments. The portal’s north side is often in shadow from surrounding tall trees causing algal growth on the sandstone. The brick core draws the water high up through capillary action, which also causes many problems to the stone, for example, eﬄorescence and exfoliation. The portal has undergone conservation projects, the most recent in the middle of the 1990s including

69 Fluorescence Lidar Multispectral Imaging

587

Fig. 69.3. To the right, a survey drawing and a sketch of the damage classiﬁcation of the portal after ocular inspection indicating biodeteriogens, chromatic alteration, weathering and eﬄorescence

cleaning of the stone, consolidation and stone repair work [5]. Figure 69.3 shows the present status, after ocular inspections with indications of biodeteriogens, weathering and eﬄorescence on the portal. The detection of chlorophyll from algae was shown with the ﬂuorescence lidar technique and also the intensity of the growth (Fig. 69.4). The presence of a copper ledge on the portal was also detected, which indicates the distinction that can be made of

588

R. Grönlund et al.

Fig. 69.4. To the left, a ﬂuorescence image of the algae presence, indicating the chlorophyll and its intensity, the white spots have the highest intensity and the dark, lower

Fig. 69.5. Below, ﬂuorescence of the copper ledge shown in grey-scale, and an indicated area showing presence of a mortar mending and its spectra together with a reference spectrum from the nearby stone in dotted line

69 Fluorescence Lidar Multispectral Imaging

589

Fig. 69.6. Areas indicated refer to red-shifted spectra (while the reference stone spectrum is shown in a dotted line). The spectral anomaly is from earlier mending with cement

Fig. 69.7. An image of the far right urn on the Attica visualising the much lower spectral signal from cleaned stone that also could be found in an associate experiment of cleaned and non-cleaned sandstone performed in Lund. The stronger, clear signal above the festoon is due to impurities and substances on the surface which emit ﬂuorescence

diﬀerent material due to their diﬀerent spectra. The presence of earlier mending with mortar where the stone had fallen oﬀ or in joints between stones was also visible with the ﬂuorescence lidar (Fig. 69.5). Some old mending with cement was found to have red-shifted spectra and the areas are indicated in Fig. 69.6. The ﬂuorescence lidar technique could also indicate the ﬂuorescence from chlorophyll on the urns on the main façade (Figs. 69.7 and 69.8). While the clean sandstone has a low ﬂuorescence signal, the weathered stone with both visible and non-visible substances attached to

590

R. Grönlund et al.

Fig. 69.8. The urn on the main façade visualizing the presence of chlorophyll with ﬂuorescence. A stronger algae presence is observed on the upper part and around the festoon while the stone is still clean below

the surface give a stronger spectral signal. The ﬂuorescence lidar could not easily distinguish between the red Öved and Orsa sandstone in this study, the data processing has shown a low intensity in the signal from both clean Öved and Orsa sandstone which may derive from the little ﬂuorescence emitted from the dominant part of quartz in the stone.

69.4 Conclusions The Övedskloster ﬁeld study demonstrated that the use of ﬂuorescence lidar makes it possible to perform non-destructive, in situ, remote monitoring of stone building façades. It conﬁrms previous results on identifying and remote mapping of biodeteriogens. It also gives the possibility of identifying the materials present, i.e. stone or previous mending with mortar and cement. Also, to some extent, it shows the sandstone condition after cleaning in a conservation project regarding the diﬀerences of a clean sandstone surface and the one aﬀected by algae and particles present on the surface due to environmental

69 Fluorescence Lidar Multispectral Imaging

591

conditions. The research on ﬂuorescence lidar has shown good potential for the architectural conservation ﬁeld concerning ﬂuorescence identiﬁcation and mapping, and could be a useful application supporting the status control of the monument and in planning for future interventions. Very recently, the remote ﬂuorescence imaging technique has been extended to remote imaging laser-induced breakdown spectroscopy (LIBS). Remote ablative cleaning of statues using the laser spark was also demonstrated [8,9]. This opens the possibility of remote cleaning under spectroscopic control and guidance. Acknowledgements This work was supported by the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS) and the Knut and Alice Wallenberg Foundation.

References 1. P. Weibring, T. Johansson, H. Edner, S. Svanberg, B. Sundnér, V. Raimondi, G. Cecchi, and L. Pantani, in Applied Optics Vol. 40, 6111, 2001. 2. D. Lognoli, G. Cecchi, I. Mochi, L. Pantani, V. Raimondi, R. Chiari, T. Johansson, P. Weibring, H. Edner, and S. Svanberg, in Applied Physics B Vol. 76, 453, 2003. 3. G. Upmark and Övedskloster, in Svenska slott och herresäten vid 1900-talets början, Skåne, Stockholm, 266, 1909 (in Swedish). 4. Natursten i byggnader, Stenen i tiden, Stockholm, 95, 1996 (in Swedish). 5. M. Johansson, Restaureringsrapport Övedskloster portal, Dnr 5025, RAÄ, Stockholm, 1994 (in Swedish). 6. R. Grönlund, J. Hällström, A. Johansson, K. Barup, and S. Svanberg, Remote multicolor excitation laser-induced ﬂuorescence imaging, submitted to Laser Chemistry, 2006. 7. P. Weibring, H. Edner, and S. Svanberg, in Applied Optics, Vol. 42, 3583, 2003. 8. R. Grönlund, M. Lundqvist, and S. Svanberg, in Optics Letters, Vol. 30, 2882, 2005. 9. R. Grönlund, M. Lundqvist, and S. Svanberg, in Applied Spectroscopy, Vol. 60, 853, 2006.

70 OptoSurf Measurement Technology for Use on Surfaces of Historic Buildings and Monuments Cleaned by Laser R

W.P. Weinhold1 , A. Wortmann1 , C. Diegelmann1 , E. Pummer2 , N. Pascua3 , Th. Brennan4 , R. Burkhardt5 , and L. Goretzki6 1 2 3 4 5 6

Innowep GmbH, Haugerring 6, 97070 Würzburg, Germany, [email protected] Atelier Pummer, Nr. 165, 3600 Rossatz, Austria CLAR Rehabilitación S. L., Cakidad 70, 28906 Madrid, Spain TKB & Associates LTD, Conservation House, Monaghan, Ireland IBW, Industriestrasse 1a, 99427 Weimar, Germany Bauhaus University (BUW), Coudraystrasse 13c, 99421 Weimar, Germany

R Summary. A documentation and analysing instrument with the name OptoTop was developed in order to measure the quality of laser cleaning on site and to document the individual stages of virgin surface area, reference area and the cleaned area on the building. The instrument comprises a fully automated optical documentation unit, which is portable and can be used without external power supply. Several high precision images of the area of interest are taken during the automated routine. A special newly developed software calculates ﬁrst the 3D topography and also speciﬁc structure and macro- and micro-roughness parameters. As documentation for the visual impression of the human eye, an additional direct image of the area is taken and also analysed. This new instrument was applied in Austria, Spain, Ireland and Germany at several sites by several organisations and restoration companies. The results are shown and analysed. The beneﬁts and limitations of this technology are demonstrated and discussed.

70.1 Introduction Building and monument preservation of cultural heritage in Europe is a current problem due to the decay of surface work. So far, the micro- and macrovisual impression and the micro-topography of a building’s surface workings could be measured only with highly sophisticated scientiﬁc instruments used in laboratories, such as auto-focus testing devices for topography. To achieve this, samples had to be removed from the building and measured in a laboratory. Furthermore, to date there are no mobile, aﬀordable and easy-to-apply methods or devices available for use by small and medium enterprises (SMEs) providing the surface cleaning of buildings and monuments. For their use on

594

W.P. Weinhold et al.

site, methods and devices that generate easy to interpret values and numbers in order to quantify results in an objective and reproducible way are needed. The objective of the proposed project was to develop a new measurement technology for mobile and non-destructive evaluation of the eﬀectiveness of laser surface cleaning of historic stone surfaces. The technology has to be capable of determining the surface topography and, at the same time, the micro- and macro-visual impression of the surface. The measured data must be processed in a suitable way to assess the performance of the surface treatment in terms of conservation and restoration. The new technology will enable the SMEs providing the cleaning services to decide which cleaning parameters to utilise in a particular situation for an optimised renovation result. In addition, it will minimise the consumption of time and energy for the cleaning process and reduce production of hazardous dust, which is of special importance for the health of the workers. Another goal connecting to this project is the generation of a sound basis for a European standard that helps to assure a consistent quality of surface cleaning work performed throughout Europe [1–4].

70.2 Applied Methodology, Achievements and Deliverables R The OptoSurf device takes surface images of monuments/buildings with historical relevance to evaluate their status before, during and after laser cleaning. An image processing software calculates various parameters of data for determination of the micro- and macro-topography as well as visual impression of natural stone surfaces. The instrumental developer is Innowep. In particular the hardware part (lab model) and the research work for software tools have to be carried out together with the Bauhaus University Weimar (BUW). Mainly, up to now, little experience exists in stone topography analysis and R and that is why reference measurements with the stationary devices, Optotop R UST (Innowep) were carried out to specify measurement ranges (Fig. 70.1).

Profilometry Lateral Resolution (x/y-axis): 1µm/10µm Axial Resolution (z):60 nm Measuring Range: ±125 µm Normal Force:1,0 mN (+) high resolution (-) time intensive; not portable R Fig. 70.1. Reference device – UST

This page intentionally blank

R 70 OptoSurf Measurement Technology

595

R Fig. 70.2. Function principle of OptoSurf technology

BUW was responsible for measurements in the laboratory and on site. BUW R introduction and training for the SME-partners. The gave the OptoSurf R OptoSurf development is of highest relevance for a laser cleaning process. To achieve a high quality result for cleaning processes in the conservation based on a carefully cleaning method and eﬀect a slower deterioration, it is crucial to create a norm for optimal and careful cleaning results. Therefore, R , which gives quick feedback users need a quality control device like OptoSurf as to cleaning results while on site. 70.2.1 Function Principle Collimated light sources from deﬁned directions produce shadings, which are separated, conditioned and evaluated automatically by the software. This technique makes it possible to distinguish between structure and colour patterns, which have no three-dimensional elongation (Fig. 70.2). 70.2.2 Laboratory Model The essential parts of the build up of the laboratory model are the Nikon “Coolpix 990” camera and a telecentric light source (Fig. 70.3), which is used from three sides (0◦ , 120◦ and 240◦ ). This lab model is integrated in a darkbox (dimension 630 × 800 × 1,000 mm) to optimise illumination properties. Preliminary tests were accomplished to determine optimal conditions to collect images. Diﬀerent light sources and geometrical speciﬁcations were applied and evaluated. Tests were done on diﬀerent stone surfaces with variation of inclination, distance, intensity and divergence (use of LED and telecentric light source).

596

W.P. Weinhold et al.

Fig. 70.3. Nikon camera and telecentric illumination light source

70.2.3 Image Processing Software The software has a Windows-based graphical user interface. The primary goal of the program is to help evaluate the quality of laser cleaned stone surface by analysing its topographical and optical properties using a nondestructive optical method suitable for ﬁeld measurement. The software uses several images of the same surface area – some with a collimated, directional illumination and some illumination using white diﬀused light – acquired by R device which serves as the primary source of data. In the heart the OptoSurf of the method is a combination of Shape from Shading and Photometric Stereo methods, which is based on a relation between intensity of reﬂected light and the angle between direction of illumination and the surface normal, allowing deduction of slopes at every point of the surface. Once the slopes are determined for every point of the surface, the program procedures reconstruct the surface topography as a 3D height distribution using a modiﬁed version of propagation method. The software allows its user to evaluate, visualize and save topographical, roughness, colour and lightness data and other related properties in the form of images, datasets and reports. R Device 70.2.4 Portable OptoSurf R , was developed in The documentation and analysing instrument, OptoSurf order to record the quality of laser cleaning on site and to document the individual stages of virgin surface area, reference area and the cleaned area at the building. The instrument comprises a fully automated optical documentation unit, which is portable and can be used without external power supply. Several high precision images are taken during the automated routine of the area of interest. R hardware (Fig. 70.4) consists of three parts integrated into The OptoSurf one handy apparatus: (1) shading illumination using new specially designed

R 70 OptoSurf Measurement Technology

597

R Fig. 70.4. OptoSurf prototype

mini-light sources to illuminate the surface; (2) image capture of high resolution images with a macro-CCD camera (image capturing is done with an integrated electronic control board) and (3) synchronisation electronics for switching between the diﬀerent light sources. Its relevance and potential can be summarized as follows: due to its unique, non-contact and non-scanning principle (“one shot”), the measuring time is decreased by several orders; due to its intuitive control elements, measurements can be obtained by everybody; and, due to its robustness, R can be used even in dirty environments (e.g. construction site) OptoSurf where other measurement techniques fail.

70.3 Measurement Results In the second project phase, measurements on virgin and laser cleaned sandstones (several laser ﬂuences were applied step by step) were carried out with the prototype at the same surface position. Furthermore, common measurements with the SME-partner could show advantages and disadvantages of R technology based on diﬀerent material applications (sandstone, the OptoSurf limestone, granite, coated concrete). Optimal results are given by limestone samples. Measurement results on sandstone and granite have to be considered carefully, especially when the reﬂection or transmission properties are very high. The recorded diﬀerences using the collected images (alone) and the calculated topography parameter on sample surfaces, measured before and after the cleaning process, are realized in every case. Important application ﬁelds are the evidence documentation of the cleaning quality and the quality assurance. Using all calculated topography parameters and diagrams, a lot of tests and experience are still necessary. Tests on several types of stones (e.g. diﬀer in grain sizes, bond, sculptured in several

598

W.P. Weinhold et al.

Fig. 70.5. Photos, calculated height proﬁles and 3D surface evaluation of stone samples

ways, various types of weathering) in combination with several cleaning methods and cleaning grades have to be carried out. Statements, like an increasing of applied laser ﬂuence on sandstones is connected with an increasing topography parameter (e.g. Ra, Rz), could be formulated. Limits could be ﬁxed. Depending on the sandstone grain sizes, the optimal topography range is to be determined through tests. The selection of measurement areas on the inhomogeneous stone samples has to be done carefully. A calibration of the device is still necessary, e.g. through correlation with measurement results of reference devices (Fig. 70.5).

70.4 Conclusions R The new OptoSurf measurement technology will enable the workers performing the laser cleaning of historic stone surfaces to quickly and properly adjust the laser cleaning parameters in such a way to achieve the optimised cleaning result with minimum treatment time and minimum removal of stone surface material.

R 70 OptoSurf Measurement Technology

599

R Direct beneﬁts of the new OptoSurf technology for the SMEs are:

– Increased productivity due to a faster adjustment of the laser cleaning parameters for each individual historic building or monument surface condition – Lower costs by reduction of working time, reduced consumption of electrical energy and less expensive repair work for unintended damage – Increased quality, consistency and reliability due to the avoidance of unintended damage during the cleaning process by wrongly adjusted laser parameters or over long treatment time – Improved functionality by simpliﬁcation of the surface documentation and evaluation technology – Increased market share as a consequence of the above-mentioned cost reductions and quality improvements R will be available in spring 2006. OptoSurf

References 1. W. P. Weinhold and L. Goretzki, Cleaning Techniques for Historic Buildings, ISBE Int. Society for Building Environment, Trinity College, Dublin, September 2003. 2. H. Siedel and G. Wiedemann, Laserstrahlreinigung von Naturstein, Fraunhofer IRB-Verlag Stuttgart 2002. 3. L. Goretzki, Bewertung von Werkstein-Oberﬂächen, WTA-MB 3-9-95/D. 4. L. Goretzki and W. P. Weinhold, in International Journal for Restoration of Buildings and Monuments, Vol. 2, 3, 223, 1996.

71 Multi-Tasking Non-Destructive Laser Technology in Conservation Diagnostic Procedures ∗

V. Tornari1 , E. Tsiranidou1 , Y. Orphanos1 , C. Falldorf 2 , R. Klattenhof 2 , E. Esposito3 , A. Agnani3 , R. Dabu4 , A. Stratan4 , A. Anastassopoulos5 , D. Schipper6 , J. Hasperhoven6 , M. Stefanaggi7 , H. Bonnici8 , and D. Ursu9 1 ∗ 2 3 4 5 6 7 8 9

FORTH/IESL, [email protected] BIAS UNIVPM NILPRP Envirocoustics S.A. Art Innovation b.v LRMH MMRI ProOptica

71.1 Introduction Laser metrology provides techniques that have been successfully applied in industrial structural diagnostic ﬁelds but have not yet been reﬁned and optimised for the special investigative requirements found in cultural heritage applications. A major impediment is the partial applicability of various optical coherent techniques, each one narrowing its use down to a speciﬁc application. This characteristic is not well suited for a ﬁeld that encounters a great variety of diagnostic problems ranging from movable, multiple-composition museum objects, to immovable multi-layered wall paintings, statues and wood carvings, to monumental constructions and outdoor cultural heritage sites. Various diagnostic techniques have been suggested and are uniquely suited for each of the mentioned problems but it is this fragmented suitability that obstructs the technology transfer. Since optical coherent techniques for metrology are based on fundamental principles and take advantage of similar procedures for generation of informative signals for data collection, then the imposed limits elevate our aim to identify complementary capabilities to accomplish the needed functionality. In particular, structural diagnosis by optical coherent techniques in art conservation intends to visualise the structural and mechanical condition of the cultural object of interest in the form of output signals (distorted wave-

602

V. Tornari et al.

fronts, for example) as compared to well-deﬁned input ones (plane wavefronts, for example). The purpose of such visual information may serve as a tool for prioritising restoration strategy or assessing interventions made during the restoration or even comparing the artwork’s condition at a later time for deterioration assessment of handling, transportation, storing and so forth. Conventional conservation practices rely on lengthy point-by-point visual observation through stereoscopic microscopes or manual contact on the exposed surfaces which are used to subjectively diﬀerentiate the acoustics of sound. Costly, ambiguous and time-consuming manual investigation may be replaced by remote, non-contacting and non-destructive, as well as standardised optical inspection procedures, which can be identically repeated. The commonly used tool incorporated only for movable items is provided by X-ray imaging and thermography cameras with known limitations in resolving capabilities and defect detection. Lasers provide their wavelength as the unit of length and the method of laser interferometry employed for metrology applications increases optical path length measurement accuracy far into the fractional wavelength regime. In fact, recent advances in modern optical metrology may oﬀer better suited alternatives providing remote, non-contacting and repeatable procedures for fast visualisation of sub-surface topography with transportable, safe, objective and sensitive instrumentation. The aim of the presented research project is based equally on the development of instrumentation allowing complementary operation of techniques as well as of a diagnostic methodology to classify an integrated approach to the variety of conservation problems of movable and immovable cultural heritage. 71.1.1 Partner Contribution The consortium is based on the close collaboration of nine partners with complementary know-how ranging from optical coherent metrology techniques, to laser design and construction, to art conservation specialists. Each partner proﬁle describes a speciﬁc and non-overlapping role in the project work packages. In Table 71.1 is presented the project partnership indicating the high degree of complementary expertise and potential interaction necessary for the accomplishment of multi-task system integration which would be the ﬁnal output of the project. The partnership was divided in three main groups: (a) Laser metrology: To perform the investigation of techniques that could become basic functional modules for hardware integration and construction. It provides output of results to the other partners. (b) Design and software/hardware construction: to design optics, laser, hardware and software integration. (c) End users: Industry and conservation institutes to provide samples, support study cases and assess the use of laser technology for art conservation structural diagnosis.

This page intentionally blank

71 Multi-Tasking Non-Destructive Laser Technology

603

Table 71.1. LASERACT Project Partnership CO. Institute Electronic Structure and Laser/Foundation for Research and Technology – Hellas, (GR), Heraklion – Crete, Greece, REC CR Bremen Institute of Applied Beam Technology/Optical Metrology, BIAS (D) Klagenfurter Strasse, 228359 Bremen, Germany, REC CR Universita Politechnica Delle Marche/Department of Mechanical Engineering, UNIVPM (I), Via Brecce Bianche, 60131 Ancona, Italy, REC CR National Institute for Laser Plasma and Radiation Physics/Solid-State Laser Laboratory, NILPRP (RO), Atomistilor 111, 76900 Bucharest-Magurele Romania, REC CR Envirocoustics, S.AEnvirocoustics (GR), EL. Venizelou 7 & Delfon, 4452 Athens, Greece, IND/ENDUSER CR Art Innovation b.v (NL), Westermaatsweg BW Hengelo, Netherlands, IND CR Laboratoire de Recherche des Monuments Historique, LRMH (FR), 29 rue de Paris, (F)- 77420 Champs sur Marne, France, ENDUSER CR Ministry of Resources and Infrastructure/Works Division-Restoration Unit, EM/WD (MT), CMR02 Floriana, Malta, ENDUSER CR Societatea Comerciala PRO OPTICA S.A/ Research and Development Department, ProOptica (RO), Aleea Gh. Petrascu, Bucharest Romania, IND CO: coordinator, CR: contractor, REC: Research center, IND: Industry/SME

71.1.2 Rationale: Laser Metrology Properties in Art Conservation The optical coherent techniques of laser metrology employed in the project justify well their use for art conservation demands since they enjoy properties of extreme importance which are rarely found in other conventional or industrial techniques. These properties and the complementary advantages which they provide in the conservation of artworks are summarised in Box 1. • Main Common Characteristics • Non-destructive-Non-invasive-Non-contacting • Non-penetrating irradiation-Surface illumination-sub-surface observation • Safe irradiation for artwork and operator • No surface preparation-No sample removal • High information content-High resolution • Fast-Repeatable-Objective • Complementary Functions & Advantages • Remote on-ﬁeld inspection-Laboratory-based research • Whole ﬁeld imaging-Scanning matrix selective mode • Qualitative assessment-Quantitative evaluation

604

V. Tornari et al.

Fig. 71.1. From left to right the internal of the modules for digital holography, shearography, vibrometry modules as developed in laboratory prototype scheme for complementary use in the multi-task sensor

71.1.3 Rationale: Use of Laser Metrology in Art Conservation Another important issue signifying the implementation of optical coherent techniques in the investigation of artworks is derived from the numerous and diverse applications potential which, by proper establishment, can be oﬀered to the ﬁeld. The application range varies from tracing defects in the subsurface and their potential propagation, to monitoring of interventive restoration actions. A summary of some potential applications is shown in Box 2. • Diagnostics • Defects detection and identiﬁcation-location, size, structure Detachments, cracks, inhomogeneities, stress concentration • Monitoring of conservation actions • Environmental studies-Materials kinetics • Estimation of conservation condition-Existing defects and propagation • Preventive Conservation • Long term alteration/aging-environment • Sudden alteration/accident-handling-environment

71.2 Integrated Structural Diagnosis Multi-task diagnosis is the integration of equally important methodology and instrumentation aspects. The basis of the parallel development is the connection of three distinct but inter-related developments: (a) Techniques development based on complementary advantages. Holographyrelated techniques tested to provide diverged object beams allowing large ﬁeld of view (≥ 50 cm) for artworks of moderated dimensions (small-tomedium scale museum and movable artworks); high resolving power for complex micro-defect detection (defects smaller than, and in depth less

71 Multi-Tasking Non-Destructive Laser Technology

605

Fig. 71.2. Left to right are shown photos from custom-made optics, the internal of the micro-chip laser source and an excitation mechanism

than 1 cm) and parametrical analysis (artwork response to external perturbations) whereas the complementary use of laser vibrometry can permit remote access to distant objects (≥ 500 cm) of extended dimensions (large-scale movable or immovable cultural items or monuments) and large deteriorated areas [1–3]. Laboratory prototype modules as were developed for transportable and alternate use by the multi-task sensor are shown in Fig. 71.1. (b) Optics, Laser and Peripheral Instrumentation. The construction and operation of modules is supported by custom-made development of specialised optics and components according to the requirements of each speciﬁc module. The development of a highly coherent, TEM00 nanosecond laser of 3 mJ output energy at 532 nm was done based on ampliﬁed harmonic generation of a micro-chip laser. The high quality of optics and lowest order Gaussian beam mode allow for the maximum use of the available energy in the expansion of beam for holography and shearography purposes. In addition, the 5 ns pulses allow use under out-of-laboratory conditions. A vibrometry module is also supported by specially developed exciters for accurate acoustic excitation. Photos of the constructed instrumentation are shown in Fig. 71.2. (c) Interconnected Software. Software development is aimed to using a common logic for system performance on which all modules and thus complementary functions are operated from. A user-friendly pleasant window environment provides the operator interface. Operation of multi-task system foresees three modes for the operator to choose from: a) Standard mode of operation depended on artwork description after which the system automatically decides the inspection process. b) Deliberate mode of operation depended on the operator knowledge. c) Interactive mode of operation where the system suggests according to the description input of the operator but the operator need not follow the inspection procedure. The environment of the developed integrated software is shown in Fig. 71.3. (d) Art classiﬁcation table vs. defect pathology. Despite the broad ﬁeld of objects and materials constituting cultural heritage, two structural problems seem to dominate deterioration. These are detachments and

606

V. Tornari et al.

Fig. 71.3. From left to right: Exemplary windows for introduction screen to operation guide and choosing a category

Fig. 71.4. Part of the artwork classiﬁcation vs. defect potential and technique suitability that forms the “decision-making brain” of the system

cracks which form in a plethora of artworks (Fig. 71.4) with multi-layered structure and inhomogeneous materials. The experimental work was based on detection, from simulated samples and real artworks, of the major conservation problems. Suitability of techniques assigned to the art classiﬁcation table permits the standardisation of an inspection sequence according to artwork characteristics and their potential pathology. In Fig. 71.5 is shown the standardised operation mode for a variety of artwork dimensions from defect detection, to strain components visualisation for defect mapping, to vibration amplitude and velocity signal for estimation of structural condition.

71 Multi-Tasking Non-Destructive Laser Technology

607

1st Step: INSPECTION SSS DSHI→DSS* SSS: Defect map a. Image Processing b. Defect Detection c.Indicate a suggested type of defect

MSS SLDV*→DSS→ DSHI 2nd Step:ANALYSIS MSS Defect map-strain *plus vibration threshold

LSS SLDV→DSS→DSHI**

LSS Vibration threshold *plus defect map-strain

3rd Step: EVALUATION Fig. 71.5. Operational sequence for development of integrated diagnosis. (SSS: Small Scale Structures, M: Medium and L: Large). The technique underlined is suggested; in italics is the optional sequence dependent on the operator decision; with asterisk is optional in the case of research requirements Operating system

INPUT

Working procedure

OUTPUT

Operating system EXPERT

Art class database

SLDV DSS DSHI

Lasers Exciters E/O

Operational procedure

Fig. 71.6. Principle of operation for the multi-task prototype system

71.3 Results The feasibility tests were concluded with the simultaneous development of interchangeable transportable modules based on the techniques of digital speckle holographic interferometry (DSHI), digital speckle shearography (DSS) and scanning laser Doppler vibrometry (SLDV) to constitute one compact prototype. For the DSHI, a custom 5 ns pulsed laser at 532 nm, based on micro-laser pumping, was developed with green output energy adjustable in 50 steps using a 532 nm half-wave plate (rotated by a stepping motor controlled by computer) and a Glan polariser. Software integrating the artworks classiﬁcation table with the operational parameters of modules drives the operator, by an interactive user-friendly interface, to perform the investigation and con-

608

V. Tornari et al.

Fig. 71.7. DSHI on marquetry sample for worm tunnelling defect detection (upper left), DSS on detachment defect detection of wood sample (upper right) and SLDV on Maltese samples stone for deﬁnition of quality (lower centre)

clude the diagnosis, as shown in Fig. 71.6. Preliminary laboratory results from simulated samples, retrieving the complementary use of the developed techniques, are shown in Fig. 71.7. The complementary alternate use of the integrated multi-task system would thus allow investigations ranging from multi-layered, organic and inhomogeneous artworks to solid, inorganic and structural materials. The prototype system was constructed and in-ﬁeld use is shown in Fig. 71.8. The compact dimensions allowed for transportation under extreme out-of-laboratory conditions. The system was transported to various interesting places (from cultural point of view) and a systematic inspection according to delivered working procedures and system operation standards was attempted. Initial ﬁeld trials were satisfying and characteristic results are shown in Figs. 71.9 and 71.10.

71 Multi-Tasking Non-Destructive Laser Technology

1. 2. 3. Output Parameters of the Laser 4. 5. 6.

609

Wavelength: 532 nm Pulse energy: > 10 mJ Standard deviation: < 5% Pulse duration: 5–8 nsec Rep. rate: Adjustable, max. 5 Hz Coherence length: > 1.5 m

Fig. 71.8. Photograph of system during in-ﬁeld investigation and laser parameters table

Fig. 71.9. Age diﬀerentiation inspection of Maltese stone by multi-task diagnosis: (left) 50 years old, (right) 400 years old, based on diﬀerences in interferometry Place: Floriana Fortiﬁcation (Malta) Aim: Age diﬀerentiation of stone

71.4 Discussion The key requirements for a technique to be qualiﬁed by conservation community as suitable are: non-destructive; non-invasive; non-contacting; acquire sub-surface information and visualise defect presence; be capable for remote access and on-ﬁeld transportation; be applicable to a variety of artworks/shapes/materials and provide objective and repeatable results. It was suggested that these requirements can be met by integrating complimentary capabilities available from laser techniques existing in optical metrology, together with development of an artwork classiﬁcation database and user-friendly integrated software.

610

V. Tornari et al.

Fig. 71.10. Age diﬀerentiation inspection of Maltese stone by multi-task diagnosis, (left) 50 years old, (right) 400 years old, based on diﬀerence in pulse propagation velocity

The main features of the LASERACT project are presented here and were targeted at satisfying art conservation requirements by the integration of complementary advantages. It has succeeded in accordance to the project work packages to deliver a multi-task instrumentation that, in its current state, can be regarded as a pre-industrial demonstrative prototype capable of performing inspection on a plethora of movable and immovable artworks. Most of the visible results are evidential without requiring extensive operator training. There is remaining work on optimisation and standardisation during the ﬁnal phase of the project. Therefore the project is still on progress and the ﬁnal, conclusive results with regard to the diagnostic procedures should be expected in a year after this presentation. Nevertheless, considerable progress has been made through the recent work since introducing laser structural diagnosis in the conservation community is no longer based on fragmented approaches. As a consequence, an exploitation strategy is in the consortium’s future plans.

References 1. V. Tornari, V. Zaﬁropulos, A. Bonarou, N.A. Vainos, and C. Fotakis, J. Opt. Las. Eng. 34, 309 (2000). 2. P. Castellini, E. Esposito, N. Paone, and E.P. Tomasini, SPIE 3411, 439 (1998). 3. V. Tornari, A. Bonarou, E. Esposito, W. Osten, M. Kalms, N. Smyrnakis, and S. Stasinopulos, SPIE 2001, Munich Conference, Vol. 4402, 2001.

72 Time-Dependent Defect Detection by Combination of Holographic Tools ∗

E. Tsiranidou1 , V. Tornari1 , Y. Orphanos1 , C. Kalpouzos1 , and M. Stefanaggi2 1

∗ 2

Foundation for Research and Technology-Hellas/Institute of Electronic Structure and Laser, Vassilika Vouton, Voutes, 71 110 Heraklion, Crete, Greece, part of an on-going study performed for MSc under the FW of EC funded project LASERACT [email protected] Laboratoire de Recherche des Monuments Historique, 29 rue de Paris, F-77420 Champs sur Marne, France

Summary. The most widely known application of holographic and speckle interferometry (termed HINDT, ESPI, or DSHI) is in the sensing of invisible structural ﬂaws, which are represented visually as discontinuous interference patterns. This property raised laser metrology techniques as the candidates best suited for nondestructive detection of subsurface defects in qualitative assessment of artworks [1–6]. Nonetheless, in the case of multilayered structures, an underlying imperfection may not always generate directly the expected visible discontinuity in the interference pattern since its inﬂuence may not reach the top of the illuminated surface which witnesses its presence, size, and magnitude. Hence eﬀectiveness of the sensing method depends on the investigation procedure and experimental tactics are developed to warranty results standardisation [7, 8]. In this context, the authors have performed inspection of known construction samples with excitation and relaxation time variants [9]. It conﬁrmed that the routine procedure of inspection for artworks should foresee the need for in-time observation of defect kinetics; if a misleading assessment due to mismatch between surface deformation eﬀect and the time of image capture is to be avoided. Herein, part of the ongoing study on defect detection dependencies [8,9] is presented and strong evidence is provided for the time-dependent capturing of the spatial evolution of deformation based on a combination of interferometric measurement techniques.

72.1 Introduction Laser techniques that can assess the conservation condition of individual artifacts have gained increasing interest among the conservation specialists. For the last twenty years, coherent optical metrology trends are being revised to confront speciﬁc art conservation problems in order to provide ﬂexible structural diagnostic tools [10–13]. Holographic-related techniques have been

612

E. Tsiranidou et al.

Fig. 72.1. Photo of sample used. Note that there is no visible evidence of defect existence

welcome candidates since they are completely nondestructive and can acquire high spatial resolution information about the shape and displacement without any dependence on surface roughness or texture, therefore not requiring any prior object preparation as the holographic interference principles imply [14, 15]. These techniques, based either in wave ﬁeld or speckle ﬁeld superposition [16], oﬀer the complementary advantages (a) of in-time monitoring through digital recording every 150 ms and (b) with high spatial resolution up to 1,600 lines mm−1 . This study is based on this complementary use in which a hybrid combination of optical and digital holographic interferometry is used to investigate the time dependence of alteration visualization and recording. In order to achieve this task, prototype marquetry samples, shown in Fig. 72.1, were constructed (among other type of construction materials not shown here) by the “Laboratoire de Recherche des Monuments Historique,” in the framework of the European project LASERACT [13]. The samples were altered by worm tunneling (no topography map is available for this type of alteration) purposefully induced for out-of-plane deformation, which is the main sensitivity direction of both interferometric techniques.

72.2 Experimental Methods The technique used is double exposure holographic interferometry, which requires that two optical waves are recorded with a time interval between them and later can be reconstructed simultaneously [14–16]. Between the two sequential holograms, a deformation on the object surface occurs under an induced thermal load. During the simultaneous reconstruction of the two holograms during the reconstruction process, visual interference fringes are formed. The reconstruction regenerates the three-dimensional wavefront wit-

72 Time-Dependent Defect Detection

613

LA SE R

Beam Splitter Variable Beam Splitter Reference beam 1

SAMPLE

Mirror

Mirror

Spatial Filter Object beam Reference beam 2

CCD ES PI SE T

FILM Divergent Lens

UP

Mirror

Fig. 72.2. Hybrid experimental setup for simultaneous holographic recording by optical (ﬁlm) and digital (CCD array) sensor in optical holographic (HI) and digital speckle holographic interferometry (DSHI) geometries

nessing the deformation of the object surface propagating from object surface toward the observer. The interference fringes comprise a quantitative measurement of phase diﬀerence caused by the surface deformation between the two time-diﬀerential exposures, expressed in intensity values, due to constructive and destructive interference. 72.2.1 Experimental Arrangement and Procedure First, the sample is monitored by digital speckle holographic interferometry (DSHI) and, after an induced thermal load is applied and an appropriate time delay has passed, internal alterations appear and can be recorded. For the speciﬁc time delays of interest, optical holographic interferograms follow, in order to acquire a high quality, reduced noise measurement. The experimental setup used for simultaneous holographic recording is seen illustrated in Fig. 72.2. The sample is placed in an oﬀ-axis transmission geometry setup and, after laser illumination of the object surface, some of the diﬀusely reﬂected light reaches the photosensitive media. In this experiment, the photosensitive media are (a) a green-sensitive silver halide emulsion on

614

E. Tsiranidou et al. Table 72.1. Features of the two techniques on the experimental setup

Sensor

Exposure time

Sensor type

Sensor size, mm

Obj.Sensor Distance, cm

RB-OB angle

Spatial recording frequency, lines mm−1

HI

1.5 s

63 × 63

54

45◦

1,330

DSHI

1.2 ms

Light sensitive ﬁlm CCD

10.2 × 8.3

70

< 2, 7◦

89

Table 72.2. Experimental data Illumination angle

Thermal loading duration, s

Time interval, s

< 10◦ < 10◦

1 9

5–420 5–420

Ratio: object beam /reference beam 1/5 1/5

Initial temp. To , ◦ C 21.2 21.3

Final temp. T1 , ◦ C 22.9 29.4

Temp. diﬀerence ∆T = T1 −To min + 1.7 max + 8.1

glass reaching spatial resolutions up to 5,000 lines mm−1 , by fy = sin θmax /λ, allowing recording of beams up to 180◦ between the reference and object beams, and (b) a high resolution CCD array with corresponding maximum resolvable spatial frequency calculated by fmax = 1/2∆x. For a distance ∆x between neighboring pixels of 5 µm, the maximum resolvable spatial frequency is only about 100 lines mm−1 and recorded at 2.7◦ . Accordingly, the CCD recording usually requires a small angle between recording beams due to this limiting factor. The laser used is a CW diode pumped Nd:YAG emitting a highly coherent TEM00 beam at 532 nm. In the scheme is shown mainly the division of the initial beam as it is directed to generate a simultaneous record of object deformation by optical and digital photosensitive media. The laser beam is separated into two parts by a variable beam splitter in order to illuminate the sample while one beam is split again into two parts to be used as a reference beam for both sensors. The ﬁrst exposure is performed at the sample in its initial (relaxed) state and then the thermal load is induced to excite the sample surface. A time interval is allowed to pass in order for the sample to reach a higher temperature and then a second exposure is performed. The employed CCD array consists of 1, 392×1, 040 square pixels of 6.45 µm size. Other features of the two techniques on the experimental setup are shown in Table 72.1. Experimental data, including the loading of the sample in terms of temperature increase, are shown in Table 72.2. The uniform thermal loading was induced by 50 W halogen lamps assembly and measured via an IR thermometer. All peripheral instrumentation was computer controlled and driven.

72 Time-Dependent Defect Detection

615

Fig. 72.3. Eﬀect of insuﬃcient thermal loading. (a) The internal alterations induced by worm tunneling may not be detected if the thermal loading is insuﬃcient, while in (b) the same alterations are very well deﬁned at higher loading

72.3 Results and Discussion Internal alterations are visualized as aberrations on the fringes homogeneity from the overall reaction of the sample. The detection of alterations, as authors have previously shown [6, 8, 9, 17], depends on parameters such as the direction and degree of loading and the interval time (time delay) of recording after loading. An arbitrary induced alteration may not always visualize the desired subsurface detection of defects. This is demonstrated in Fig. 72.3 where is shown an arbitrary thermal alteration to be insuﬃcient and the subsurface alterations do not all have a visible eﬀect on the surface. The overall displacement is small and only big defects are seen. Thus suﬃcient induced thermal loading is essential for both detection and good deﬁnition of internal alterations. Photos of some simultaneously acquired interferograms are shown in Fig. 72.4. Figure 72.4 shows the alterations at time intervals depicted in (a), (d), after the sample has undergone the excitation load and thus is considered to be at the position of maximum displacement. At ﬁrst sight, the raw data looks very similar. However, in Fig. 72.5, graphs 1 and 2 show the intensity proﬁles after processing by histogram equalization and spin ﬁlter functions to enable intensity variations in the diﬀerent pixel positions to be clearly described. So graphs 1 and 2 trace some diﬀerences in object point-displacement as visualized and captured by DSHI and HI. The results may signify the geometry of sensors is equally as important as the time interval parameter for accurate diagnosis. In addition, measurements taken using DSHI geometry exhibit “noisier” images compared to those acquired by HI. This eﬀect is due to a variety of noise sources not explicitly analyzed in this paper but strongly depended on electronic noise as well as the readout and dark noise of the CCD without underestimating also the lens imperfections. In spite of the diﬀerences of the two techniques (Table 72.1), the detected discontinuities appear to be small

616

E. Tsiranidou et al.

Fig. 72.4. Time dependence of alterations visualization. At (a) and (d), alterations are visualized in great detail at short time intervals after the loading (+5, 4◦ C), while the sample is in highest displacement and, at (b), (c), (e), (f), many alterations disappear at longer time interval when the sample is in a lower excitation state and thus the surface displacement is much less or ceases altogether. In even longer time intervals, alterations are no longer traceable as the eﬀect of underloading shows in Fig. 72.3

but important when the time dependence of defect detail topography is taken into account. Moreover, the diﬀerences in the results acquired by the two techniques could be attributed to the much lower spatial recording frequency of DSHI as well as to the diﬀerent sensitivity vector based on the geometry of the experimental setup. If we consider the sample as ﬂat, the illumination angle is ∼9◦ and the resulting sensitivity vector for HI, KHI , is perpendicular to the sample, while for DSHI, KDSHI has a deviation of 15◦ . The vector scheme is shown in Fig. 72.6. The sample temperature and relative displacement of the sample surface as it cools down for the suﬃcient loading of +5.4◦ C is shown in the diagram of Fig. 72.7. On the diagrams, the time interval in which all traceable alterations appear is marked in ***red. Consequently, if the acquisition time is not suﬃciently synchronized, the structural diagnosis of an artwork could be misleading, since certain alterations might be overlooked. There are some alterations

72 Time-Dependent Defect Detection

617

(row_80 with DSHI) (row_80 with HI) 250

Pixel Intensity

200

150

100

50

0 -20

0

20

40

60

80

100

120

140

position at x axis along raw_80 (column_62 with DSHI) (column_62 with HI) 250

Pixel Intensity

200

150

100

50

0 -20

0

20

40

60

80

100

120

140

160

position at y axis along column_62

Fig. 72.5. Graphs 1 and 2 superimposed proﬁles along same pixel coordinates visualizing diﬀerences in displacement as captured by DSHI and HI

though, such as detachments, which have been visualized throughout the cooling time of the sample. This fact has been noticed not only on wooden samples but also on fresco wall painting samples with induced detachments where the alteration is traceable from the beginning until even 10 min later. In this case,

618

E. Tsiranidou et al.

o Sample Temperature, C

Fig. 72.6. Sensitivity vectors based on oﬀ-axis geometry of the setup 32 31 30

(a )

29 28 27 26

Relative Surface Displacement, µm

0

100

200

300

400

500

Time, sec

6

(b ) 4

0

100

200

300

400

500

Time, sec

Fig. 72.7. (a) Sample temperature and (b) relative displacement of the marquetry sample surface as it is cooling down. The red marked time interval is illustrating the time where all traceable alterations appear, while the green mark shows the time period where detachments appear, i.e., throughout the cooling time of the sample

there is a time period where the alteration is better deﬁned and after which, though it can still be detected, it cannot be mapped in much detail as shown in Fig. 72.8. Further results of marquetry and wall painting, while their raw data is not shown in this paper, have shown that the time period in which all

o Sample Temperature, C

72 Time-Dependent Defect Detection 30

29

(a ) 28 27

T ime, s ec 26 0

Relative Surface Displacement, µm

619

50

100

150

200

250

9

(b)

6

3

T ime, s ec 0 0

50

100

150

200

250

Fig. 72.8. (a) Sample temperature and (b) relative displacement of the fresco wall painting sample surface as it is cooling down. The red marked time width is illustrating the time interval where the detachment is better deﬁned, while the blue mark is the time period where the alteration is detected but cannot be mapped in much detail

variable alterations are traceable lies within a time interval of less than 1 min. The interval conﬁrms the critical time-window of observation after loading.

72.4 Conclusions The hybrid combination shown demonstrates a powerful tool for the fast identiﬁcation of the optimum time when accurate defect detection is concerned. It also conﬁrmed the importance of the time interval parameter, as well as the sensor parameters and setup geometry, for accurate structural visualization and diagnosis. It is recognized that in order to avoid a misleading diagnosis in structural alterations of artworks, it is critical to investigate the time interval after loading in which all traceable, altered areas are visualized. The speciﬁc time period may vary depending on the kind and depth of the existing alteration, the construction material of the object, its age, and preservation condition but the time interval should be determined in all cases for accuracy. Also the capturing sensor involves further deﬁnition since it aﬀects the ﬁnal output. In this context, standardized operational procedures should investigate

620

E. Tsiranidou et al.

defect alteration time dependencies before assessing the structural condition. Further work is required in the ﬁeld of structural diagnosis with regard to the dynamic behavior of the object and possible modeling that would beneﬁt perspective automatic inspection procedures. Acknowledgements The present study was feasible through the provision of facilities at IESL/FORTH and the EC project LASERACT EVK4-CT-2002-00096.

References 1. S. Amadesi, F. Gori, R. Grella, and G. Guattari, Appl. Opt. 13, 2009 (1974). 2. S. Amadesi, A.D. Altorio, and D. Paoletti, Appl. Opt. 21, 1889 (1982). 3. G. Gulker, K. Hinsch, C. Holscher, A. Kramer, and H. Neunaber, in Proc. SPIE/ Laser Interferometry: Quantitative Analysis of Interferograms, Vol. 1162, 156, 1990. 4. P.M. Boone and V.B. Markov, Stud. Conserv. 40, 103 (1995). 5. V. Tornari and K. Papadaki, Proc. of the 1st International Congress on Science and Technology for the Safeguard of Cultural Heritage in the Mediterranean Basin, Catania – Siracusa, Italy 1995. 6. V. Tornari, V. Zaﬁropulos, N.A. Vainos, C. Fotakis, S. Stassinopoulos, and N. Smyrnakis, in Restauratorenblätter, Sonderband – LACONA II, Vienna, 2000. 7. V. Tornari, A. Bonarou, E. Esposito, W. Osten, M. Kalms, N. Smyrnakis, and S. Stasinopulos, in SPIE 2001, Munich Conference, Vol. 4402, 2001. 8. V. Tornari, E. Tsiranidou, and Y. Orphanos, International Conference on Experimental Mechanics ICEM 12, McGraw-Hill, Bari, 2004. 9. V. Tornari, E. Tsiranidou, Y. Orphanos, and C. Kalpouzos, submitted CLEO, 2005. 10. D. Paoletti, G. Schirripa Spagnolo, M. Facchini, and P. Zanetta, Appl. Opt. 32–31, 6236 (1993). 11. P. Castellini, E. Esposito, N. Paone, and E.P. Tomasini, in Proc. of the VI International Conference on Non-destructive Testing and Microanalysis for the Diagnostics and Conservation of the Cultural and Environmental Heritage, Rome, 1999. 12. V. Tornari, V. Zaﬁropulos, A. Bonarou, N.A. Vainos, and C. Fotakis, J. Opt. Las. Eng. 34, 309 (2000). 13. V. Tornari, C. Falldorf, E. Esposito, R. Dabu, K. Bolas, D. Schipper, M. Stefanaggi, H. Bonnici, and D. Ursu, CLEO, 2005. 14. C.M. Vest, Holographic Interferometry. Wiley, N.Y., 1979. 15. R.K. Erf, Holographic nondestructive testing, Academic Press, London, 1974. 16. R. Jones and C. Wykes, Holographic and Speckle Interferometry, Cambridge Studies in Modern Optics, 1989. 17. E. Tsiranidou, Y. Orphanos, Y. Yingjie, and V. Tornari, in SPIE the International Society for Optical Engineering, AIVELA, 2004.

Part VII

Safety and Miscellaneous

73 Health Risks Caused by Particulate Emission During Laser Cleaning ∗

R. Ostrowski1 , St. Barcikowski2 , J. Marczak1 , A. Ostendorf2 , M. Strzelec1 , and J. Walter2 1

∗ 2

Institute of Optoelectronics, Military University of Technology, 2 Kaliskiego Str., 00-908 Warsaw, Poland [email protected] Laser Zentrum Hannover e.V., Hollerithallee 8, 30419 Hannover, Germany

Summary. Air contaminants which emerge during laser ablation often cause health risks if released in the workplace and decrease laser cleaning eﬃciency if redeposited at the material surface. In addition, ultra-ﬁne particles are generated if short pulses are applied. Consequently, a description of the nano-particle aerosol generation and the inﬂuence of laser parameters and material surface on the nano-particle size distribution are given in this paper. The high respirability of such particles can pose health risks, so suitable capture systems near the processing zone or personal protective equipment such as respiratory masks are required.

73.1 Introduction All historical objects and artworks require special care and attention during their renovation and conservation. The main problem connected with traditional cleaning methods is the erosion of original substrate placed near encrustation, often causing irretrievable loss and defects of ﬁne detail. Due to numerous advantages, laser cleaning oﬀers an opportunity to overcome these drawbacks and, in eﬀect, ﬁnds more and more widespread use. However, it gives rise to potential hazards. Especially workplace safety and environmental aspects are becoming more relevant due to the awareness of these aspects by modern society. A systematic approach to laser safety must not be limited to hazards caused by radiation and protection against it, but must include all aspects of a laser installation. Potential hazards related to the use of lasers can generally be divided into primary and secondary hazards [1]. The laser beam itself represents the primary potential hazard, as it can aﬀect humans or objects, in the form of raw, focused, reﬂected or scattered radiation. Secondary potential hazards are further subdivided into direct and indirect hazards. The former are caused by technical components of the laser installation (high voltage, high

This page intentionally blank

624

R. Ostrowski et al.

current, excitation radiation, gas supplying system, etc.), and the latter are generated by the interaction of the laser beam with matter. This includes the UV-radiation caused by plasma formation, emerging hazardous substances, potential ignition of explosive materials and the danger of ﬁre. During laser cleaning, laser-generated air contaminants (LGAC) emerge in the form of dust clouds, which often cause health risks if released in the workplace and may decrease the process quality if redeposited at the material surface. The formation of LGAC results from the interaction of laser beam, assisting gas and material. The physical process begins with the absorption of laser radiation in the interaction zone. Thus, laser cleaning may involve evaporation, melting, sublimation, smoke-generating pyrolysis and/or spallation of the layer to be removed. Rapid increase in temperature and a corresponding high pressure can occur, resulting in the LGAC usually leaving the interaction zone at high velocities. The size and weight of aerosol particles determine their behaviour in the ambient air. For usual air speeds in the range of 0.1–3 m s−1 , aerosol particles smaller than 10 µm can be regarded as airborne while larger fragments, due to their size and weight, immediately settle out in the processing zone. The smallest particles have geometric diameters of about 5–10 nm and because of high respirability (more than 80%), they are the most hazardous [1]. Correlating the particle size distribution with deposition rates for inhaled particles in the human lung, it can be calculated that, in the worst case, 40–60% of them remain in the alveoli, where they can aﬀect lung cells [1]. Many LGAC are known in industrial medicine as irritating, toxic, allergic or carcinogenic and some of them can be the cause of diseases after many years of exposure. Both the chemical composition and amount of LGAC depend on the process parameters and the material being processed. Excessive laser ﬂuence leads to higher ablation rates, and consequently to higher emission rates. LGAC are partially composed of removed material and substrate. Encrustations consist of a mixture of particles of both natural (quartz, feldspars, clays, pollens, etc.) and industrial origin (ﬂy ash, soot, coal particles, iron oxides, etc.) [2,3]. Fragments of the substrate itself (quartz, calcite) may also be found within these crusts. All these particles are bound together by gypsum. Additionally, black crusts contain various hydrocarbons and other organic compounds, some of which are known to be hazardous if inhaled. Often surfaces are covered with polychrome containing lead [4]. Until now, only sporadic investigations and analyses have been performed in order to assess the health risks connected with laser cleaning application [4] and especially, concerning particulate emission during laser ablation of encrustations [5,6]. This contrasts with the large number of studies performed on health hazards linked to laser use for medical and industrial applications, where there are well established standards governing the safe usage of laser devices and systems [1, 6]. The main part of the LGAC emitted during laser cleaning and relevant to risk assessment is inorganic particulate matter. Permanent and harmful

73 Health Risks Caused by Particulate Emission During Laser Cleaning

625

gases such as carbon monoxide, ozone and nitrogen oxides also emerge, but in low emission rates, which cause concentrations below the workplace threshold limit. It is known that the mean aerodynamic diameter of the particles becomes smaller if short pulses are applied, but no information on the inﬂuence of the laser parameters and material surface on the particle size distribution during laser cleaning is available up to now. Therefore, a characterization of the nano-particle aerosol generation is carried out in this paper.

73.2 Experimental Set-Up All measurements were conducted in the experimental set-up schematically shown in Fig. 73.1, arranged at the Laser Center of Hanover (LZH). The laser system ‘ReNOVALaser 5’ utilized during the experiments was operated at a constant repetition rate of 10 Hz. It was able to emit Q-switched pulses in energy up to 900 mJ and 15 ns in duration, at fundamental wavelength of 1,064 nm. The output beam size was 8 mm. Additionally, a highly multi-mode operation of the laser allowed us to obtain the output beam with the top hat pattern. The ﬂuence level was adjusted by movement of the tested samples and the deﬂecting prism relatively to the focusing lens. During the experiment, the ﬂuences were varied up to 8.5 J cm−2 . It should be pointed out that such high energy density levels are not applicable in practice. Instead, for the safety of the cleaned object, ﬂuences up to 1 J cm−2 are used. During the laser cleaning process, the particulate emissions were characterized using an online-measurement system. The particulate matter (PM) was captured using a mobile suction nozzle at a volume ﬂow rate of 10 l min−1 . The aerosols were guided to the online-measuring unit using an anti-static ﬂexible tube. The characterization of the particle size distribution was carried out by

ELPI

Computer

Prism

ReNOVALaser 5 Lens

Cleaned sample

Fig. 73.1. Experimental set-up

626

R. Ostrowski et al.

an automatic 12-level low pressure cascade impactor (Dekati Inc.; Type ELPI), which additionally allows post-analysis of the particle morphology by scanning electron microscopy (SEM). The aerodynamic particle size measurement range was 0.03–7 µm. Because of the natural sample non-homogeneities, averaging of the single measurement intervals (10 s) was necessary. Therefore, a single particle size distribution measurement presented in this investigation is a result of averaging thousands of laser pulses (typically 18,000) to obtain a suﬃcient particle count in all ELPI stages. In all experiments, samples of sandstone, basic building material of the Sigismund Chapel (referred to as ‘Myslenicki stone’), as well as samples of gypsum, utilized as a ﬁlling material in reconstruction processes, were irradiated with high peak power laser pulses. Examples of such gypsum elements of decoration before and after partial laser cleaning are presented in Fig. 73.2. Figure 73.3 shows, in turn, the general view of the sandstone sample with and without encrustation. The Sigismund Chapel, built by Bartolomeo Berrecci of Florence in the years 1519–1533, is an outstanding achievement of Renaissance architecture and sculpture in Poland and in the whole of northern Europe. Every inch of its stone walls and dome is covered with superbly sculptured, ﬁne ﬂoral

Fig. 73.2. The gypsum sample of Sigismund Chapel decoration with encrustation before (on the left) and after partial laser cleaning (on the right)

Fig. 73.3. The sandstone sample referred to as ‘Myslenicki stone’ (original stone after mechanical cleaning (on the left), encrustation (on the right))

73 Health Risks Caused by Particulate Emission During Laser Cleaning

627

arabesques, grotesque creatures and mythological scenes. The greenish colour of the walls heightens the deep red of marble statues of St. Peter, St. Paul, St. Venceslas, St. Florian, St. John the Baptist and St. Sigismund, as well as tombstones of the Kings Sigismund I the Old and Sigismund Augustus and Queen Anne Jagiellon. The recent renovation was executed using laser ablation, in order to reconstruct the original appearance of the chapel. The application of laser cleaning technique in renovation processes was connected with the realization of tasks within the frame of the ﬁnished EUREKA E!2542 ‘RENOVA LASER’ project.

73.3 Results and Discussion Particulates emitted during laser ablation have a considerably small median diameter. It has been demonstrated that shorter laser pulses (femto- and nano-second) cause generation of smaller particles. In order to evaluate the dependence of the particle size distribution on the material surface roughness, laser energy and ﬂuence, characterization of ﬁne particles emission during laser cleaning is presented in this section. The decoration samples of Sigismund Chapel consist of sandstone and gypsum covered with crusts. The sandstone surface was cleaned dry only in areas which were either raw (rough) or polished. Figure 73.4 presents the inﬂuence of surface type on the particulate size distribution. The laser ablation of the rough surface material results in a size distribution with a main maximum of the relative concentration at a particle diameter of 100 nm and a secondary maximum at 1,100 nm. In contrast, laser treatment of the polished surface produces a normal size distribution with a maximum at a smaller diameter of 50 nm, as shown by means of the hatched bars in Fig. 73.4. The tests were carried out at a ﬂuence of 8.5 J cm−2 . For the reasons mentioned above, ﬂuences up to 1 J cm−2 are used in practice. Most lasers, however, in contrast to our case of top-hat pattern, operate in the fundamental transverse mode, where the ﬂuence achieves its maximum on the axis of the beam. In the latter, therefore, the value of the ﬂuence should be considered as an averaged value. Comparison of both types of laser beams possessing the same energy and the beam diameter according to “π” criterion shows [7], that on-axis ﬂuence in the ﬁrst case is about ﬁve times greater than in the second case. Thus, such high ﬂuences used in our experiment may not diﬀer from real practical values. The laser cleaning of gypsum sample required lower laser ﬂuence in the range of 1.8–3.8 J cm−2 . The ﬂuence was varied by changing the distance between the sample and the focusing lens, which allowed us to control the laser beam spot size. The particle size distribution dependence on laser ﬂuence is shown in Fig. 73.5. In general, at a ﬂuence of 3.8 J cm−2 (spot size of 4.5 mm), the average particle diameter is smaller as compared to the results with a ﬂuence of 2.1 J cm−2 (spot size of 6 mm). At the higher ﬂuence, about 78% of

628

R. Ostrowski et al.

frequency [%]

45% 40%

raw surface

35%

polished surface

30% 25% 20% 15% 10% 5% 0% 0,03

0,05

0,1

0,2

0,3

0,4

0,7

1,1

1,8

2,7

4,3

7

aerodynamic diameter [µm]

Fig. 73.4. Inﬂuence of the character of object surface on the ultra-ﬁne particle size distribution during Nd:YAG laser cleaning of sandstone at ﬂuence of 8.5 J cm−2 (spot size of 3 mm)

60% Spot: 4,5 mm

50%

frequency [%]

Spot: 6 mm

40%

30%

20%

10%

0% 0,03

0,05

0,1

0,2

0,3

0,4

0,7

1,1

1,8

2,7

4,3

7

aerodynamic diameter [µm]

Fig. 73.5. Inﬂuence of the ﬂuence (spot size) on ultra-ﬁne particle size distribution during Nd:YAG laser cleaning of gypsum at a constant pulse energy of 600 mJ

the particulates are nano-particles with diameters in the range of 30–100 nm, whereas at the lower ﬂuence, the number of nano-particles decreases to around 65% in the same range. At the higher ﬂuence, 55% of the overall particles have a diameter of 30 nm whereas only 40% at the lower ﬂuence. This clearly reveals that the increased ﬂuence shifts the median particle diameter towards smaller values.

73 Health Risks Caused by Particulate Emission During Laser Cleaning

629

60% Energy: 510 mJ

frequency [%]

50%

Energy: 600 mJ

40%

30%

20%

10%

0% 0,03

0,05

0,1

0,2

0,3

0,4

0,7

1,1

1,8

2,7

4,3

7

aerodynamic diameter [µm]

Fig. 73.6. Inﬂuence of the pulse energy on sub-ﬁne particle size distribution during Nd:YAG laser cleaning of gypsum at a constant beam spot size of 4.5 mm

Obviously, a higher laser ﬂuence at the surface leads to generation of ﬁner particles. The ﬂuence may be changed by varying laser beam spot size at constant pulse energy and vice versa, varying pulse energy at constant spot size. In the former, however, larger beam spot size could result in a sideeﬀect, relying on an averaging of any inhomogeneities of the surface being cleaned with larger spot sizes and, in eﬀect, result in quite diﬀerent particle size distributions. Therefore, to cancel out this eﬀect and to check only the inﬂuence of increasing pulse energy on particle size distribution, the measurements with constant spot size of 4.5 mm and variable energy have been done. The results are shown in Fig. 73.6. It can be clearly seen that lower pulse energy (510 mJ) causes emission of less nano-particles, about 54% within the range of 30–100 nm and therefore, a higher average particle size at the workplace. More than half the particles have a size range of 30 nm if 600 mJ is applied, whereas less than 30% of them fall into this size range if the pulse energy is 510 mJ. As in the case of the rough sandstone, laser ablation of gypsum results in a size distribution with two maxima: the main placed at a particle diameter of 30 nm and the secondary at 2,700 nm. The position of these maxima is not dependent on laser ﬂuence or pulse energy, which can be seen in Figs. 73.5 and 73.6.

73.4 Conclusions Despite many advantages, laser cleaning causes particulate matter (PM) emission, which must be captured to ensure occupational safety. At the highest laser ﬂuence applied in the experiments, 78% of the PM is in the nano-particle

630

R. Ostrowski et al.

size range of 30–100 nm. Lower ﬂuence or pulse energy results in a lower emission of nano-particles. In addition, a rough surface produces a particle size distribution with a larger average size than during laser treatment of the polished surface. In consequence, the assessment of secondary hazards is important with regard to the safety of machinery and workplace. Both the manufacturers of laser cleaning systems and the operators have to be aware of the hazards. Therefore it is necessary to know that the characteristics of the PM depend on the laser and material parameters. The detailed characteristics of the particles generated during laser cleaning at high ﬂuences show a remarkable shift of the mean aerodynamic diameter of the aerosols to ultra-ﬁne diameters (<100 nm), in contrast to material processing with conventional lasers. Since the ultra-ﬁne particles are highly respirable, hazards occur especially when material processing releases fumes containing carcinogenic or toxic substances. Therefore, suitable capture systems near the processing zone or respiratory masks are required to avoid possible health risks. Since the latter often constrain renovation work by restriction of the ﬁeld of view and an operational head space, suction nozzles will often be the best choice of achievable hazard control techniques. Acknowledgements A part of the presented studies were carried out within the frame of the EUREKA RENOVA LASER project (E!2542).

References 1. K. Schulmeister, T. Püster, M. Green, and R. Henderson, Handbook of Industrial Laser Safety, Edited by K. Schröder, ARC Seibersdorf Research GmbH, 2003. 2. R. Van Grieken, F. Delalieux, and K. Gyseis, Cultural Heritage and the Environment, Pure & Appl. Chem, Vol. 70, No. 12, 2327, 1998. 3. R.M. Kabbani, in, The Chemical Educator, Vol. 2, No. 1, 1, 1997. 4. V. Verges-Belmin, G. Wiedemann, L. Weber, M. Cooper, D. Crump, and R. Gouerne, in J. Cult. Heritage Vol. 4, 33, 2003. 5. M. Laboure, P. Bromblet, G. Orial, G. Wiedemann, and C. Simon-Boisson, in J. Cult. Heritage Vol. 1, 21, 2000. 6. H-G. Kusch, T. Heinze, and G. Wiedemann, in J. Cult. Heritage, Vol. 4, 38, 2003. 7. A.E. Siegman, Lasers, University Science Books, Mill Valley, California, 1986.

74 Generation of Nano-Particles During Laser Ablation: Risk Assessment of Non-Beam Hazards During Laser Cleaning ∗

St. Barcikowski , N. Bärsch, and A. Ostendorf Laser Zentrum Hannover e.V., Hollerithallee 8, 30419 Hannover, Germany ∗ [email protected] Summary. Nano-particles are released from the material during short-pulse and ultrashort-pulse laser ablation. It is well known that nano-particles can cause adverse health eﬀects. Therefore, the aim of the presented study is to assess the risks caused by nano-particulate emission during nanosecond (ns) and femtosecond (fs) laser cleaning. The size distribution of particulate matter (PM) in the fumes which emerge during Nd:YAG and Ti:Sapphire laser ablation is characterized using an automatic 12-level low pressure cascade impactor. It is shown that the mean diameter and the dispersion of the PM strongly depend on the laser parameters (ﬂuence, pulse energy) and material (gypsum, graphite and paper). The particulate matter size which emerges during Nd:YAG laser cleaning signiﬁcantly decreases to the ultraﬁne particle diameter range (up to 75% with a diameter <100 nm) if the ﬂuence or the laser pulse energy is increased. Using shorter laser pulse duration, femtosecond laser ablation tends to the emission of even ﬁner particles. Though emission rates during ns and fs laser ablation are remarkably lower than during conventional laser processes such as cutting or welding, the high respirability and number concentration of ultraﬁne particles (<100 nm) can pose serious adverse health eﬀects. In consequence, the assessment of secondary hazards during laser cleaning is done with regard to safety of machinery and workplace. Both the manufacturer of a laser cleaning machine and the operator must be aware of these hazards. The investigations contribute to the recommendation of suitable emission capture and ﬁltration systems.

74.1 Introduction There has been a substantial increase in publications on the application of laser ablation for the conservation of artworks. In many cases, laser cleaning of various materials such as paper, metals and stone has been investigated not only in laboratory but also in ﬁeld experiments and even in restoration of complete building segments [1]. Depending on the geometry and surface composition of such objects, the laser beam is guided automatically by galvanic scanners or manually by a handpiece with focusing optic. In the ﬁrst case, possible process by-products emitted at the material surface are often

632

St. Barcikowski et al.

not suﬃciently captured by ﬁltering equipment. In the latter case, the fumes are often not captured at all. Even if the emission rates (the ablated mass per time) are low compared to high power laser applications such as laser cutting or laser welding, it has to be taken into account that particulate emissions generated during laser ablation show a median diameter which is considerably smaller. It has been shown that particles generated with nanosecond or femtosecond pulses show smaller particle diameters than those generated during long-pulse or cw laser material processing [2–4]. This particulate matter (PM) size distribution is signiﬁcant for risk assessment of laser ablation since reports published by various groups show that the interaction of ultra-ﬁne particles (UFP) or nano-particles (size <100 nm) with various biological systems may directly or indirectly aﬀect human health [6, 7]. In Fig. 74.1, the fractional deposition of particles during nasal breathing (ICRP) or breathing through the mouth (WHO) is drawn. Particular attention should be paid to the deposition of UFP, which increases with smaller particle diameters. For example, more than 50% of particles with a diameter of 20 nm are deposited in the alveoli. In experiments with rats, these particles induced chronic inﬂammation, ﬁbrosis and lung tumors. Recent publications have shown that nano-particles with sizes below 50–100 nm can have toxic eﬀects and that these eﬀects are, in certain cases, greater than those caused by larger (>1 µm) particles [8]. Moreover, epidemiological studies have conﬁrmed these ﬁndings, estimating that per 10 µg m−3 increase in the concentration of particles <2.5 µm, overall mortality increases by 0.9%, while deaths from respiratory-speciﬁc diseases can increase by as

Fig. 74.1. ICRP and WHO model of fractional deposition of inhaled particles ranging from 0.6 nm to 20 µm. Within the ultraﬁne particle size range (<100 nm), the maximum alveolar, tracheo-bronchial and nasopharyngeal deposition fraction is observed [5]

74 Generation of Nano-Particles During Laser Ablation

633

much as 2.7% [9]. In consequence, the particulate matter, in particular nano-particles released during laser ablation, is highly relevant to the risk assessment of laser cleaning.

74.2 Experimental The experimental details of the laser cleaning of gypsum/sandstone were recently reported [1]. The laser process was carried out by the Institute of Optoelectronics, Military University of Technology, Warsaw, Poland, using its laser system ReNOVA 5 laser, operated at a constant repetition rate of 10 Hz. The basic technical parameters of the system are 900 mJ per pulse at 1,064 nm, 15 ns pulse duration and a beam diameter of 8 mm. The highly multi-mode operation of the laser provides an output beam proﬁle with a top-hat intensity distribution. The laser cleaning process of paper has been carried out using a femtosecond laser system. The laser system, Spitﬁre-Pro from Spectra-Physics, was operated at a constant repetition rate of 1 kHz. The basic technical parameters are a centre wavelength of 800 nm, pulse duration of 120 fs, and a maximum laser output power of 1 W. During both Nd:YAG and femtosecond laser ablation, the particulate emissions were characterized by an online-measuring, electrical, low pressure ﬁlter with a volume ﬂow rate of 10 l min−1 and 12 separation stages. The aerodynamic particle diameter detection range was 7 nm–3.9 µm or 30 nm–7 µm. Figure 74.1 shows the system set-up for the experiments. Figures 74.2 and 74.3 shows the laser system characteristics.

74.3 Results and Discussion 74.3.1 Nanosecond (Nd:YAG) Laser Ablation The ablation of material with short laser pulses in the nanosecond regime, that are emitted by solid state lasers typical for micro-machining, always involves the emission of micro- and even nano-particles. When investigating the eﬀect of increasing pulse energies at constant geometric focusing conditions on the surface, it is observed that the frequency of smallest particles increases with higher laser ﬂuence (i.e. higher pulse energy). When applying pulses of 600 mJ, the frequency of particles detected on the ﬁnal stage was twice as high as when applying pulses of 510 mJ. It is therefore obvious that the size of particulate matter emerging during Nd:YAG laser (Fig. 74.4) cleaning signiﬁcantly decreases to the ultraﬁne particle range (of diameters below 100 nm) when higher ﬂuences are applied. In addition, a polished sandstone surface also leads to smaller particles than the laser processing of untreated sandstone.

634

St. Barcikowski et al.

Focussing optics Nanoparticles Sample Electric low pressure cascade impactor (ELPI)

Separation of particles from 10µm (A) to 30 nm (B) Constant gas flow: 10 l/min Vacuum pump Close

Fig. 74.2. Set-up of beam delivery and particle collection systems used for experiments

CPA 2001 Clark-MXR λ = 780 nm τ = 150 fs PRF = 1 kHz Ep ≤ 800 µJ

Spitfire PRO Spectra-Physics λ = 800 nm τ = 120 fs PRF = 1 kHz Ep ≤ 1000 µJ

Fig. 74.3. Summary of fs laser system characteristics

74.3.2 Femtosecond (Ti:Sapphire) Laser Ablation Ablation with ultrafast lasers, oﬀering pulse durations in the femto- or picosecond regime using low ﬂuences, is another approach that has been successfully applied for the cleaning of stained objects. Particle emissions during surface cleaning processes with femtosecond laser pulses have also been shown to be

74 Generation of Nano-Particles During Laser Ablation 60%

frequency [%]

50%

Energy: 510 mJ

Material/Process Parameter

Energy: 600 mJ

Material: Laser: Type: Wavelength: Energy: Feed Rate: Spot Diameter:

40% 30%

635

gypsum Renova 5 Nd:YAG 1064 nm various 15 m/s 4.5 mm

20% 10% 0% 0.01

0.1

1

10

Fig. 74.4. Inﬂuence of varying pulse energy on the size distribution of microparticles and nano-particles during Nd:YAG laser cleaning of gypsum, applying a ﬂuence of 0.80 J cm−2 (blue) and 0.94 J cm−2 (red)

60 50 µJ

frequency [%]

50 40

300 µJ

Graphite Repetition Rate: 1 kHz Wavelength: 780 nm Feed Rate: 10 mm/min Pulse Duration: 150 fs

30 20 10 0

0.007 0.03 0.06 0.1 0.16 0.25 0.39 0.63 0.98 1.59 2.43 3.93 aerodynamic diameter [µm]

Fig. 74.5. Inﬂuence of the energy on the micro- and nano-particle size distribution during femtosecond laser ablation of graphite, applying a ﬂuence of 40 J cm−2 (blue) and 75 J cm−2 (red stripes)

in the micro- and nanometre range. However, the behaviour for changes of ﬂuence is signiﬁcantly diﬀerent from that during ablation with lasers in the nanosecond regime. The size distribution of particles that are released during the ablation of graphite using a fs Ti:sapphire laser is shown in Fig. 74.5. When applying a higher pulse energy of 300 µJ, the fraction of nano-particles smaller than

636

St. Barcikowski et al. 60 Short-fibred Paper Paper after 1820

frequency [%]

50

Energy: Repetition Rate: Wavelength: Feed Rate:

100 µJ 1 kHz 780 nm 10 mm/min

40 30 20 10 0 0.007 0.03 0.06

0.1

0.16 0.25 0.39 0.63 0.98 1.59 2.43 3.93 aerodynamic diameter [µm]

Paper after 1820

20

60

< 100 nm

Short-fibred paper

< 60 nm

Frequency of particles

Fig. 74.6. Micro- and nano-particle size distribution of fs laser cleaning of two types of paper, applying a ﬂuence of 50 J cm−2

0

40

%

100

Fig. 74.7. Frequency of particles in the nanometre size range (<100 nm) during femtosecond laser cleaning of paper at 50 J cm−2

100 nm is over 90%, which is a much higher fraction of the emitted particles than in the case of lower pulse energies. Figure 74.6 demonstrates the size distribution of micro- and nanoparticles that were emitted during fs laser cleaning of two diﬀerent types of paper. The mean particle diameter for the cleaning process of “Paper after 1820” is in the 7 nm size level. The short-ﬁbred paper generates particles with a mean diameter in the range of 7 to 30 nm using similar parameters. Cleaning “Paper after 1820” causes a higher frequency of particles smaller than 100 nm than cleaning the other paper, which is also summarized in Fig. 74.7: The frequency of small particles is higher in the case

74 Generation of Nano-Particles During Laser Ablation

637

of “Paper after 1820” for particles both below 60 nm and below 100 nm. The dependence of particle sizes on paper type has therefore been shown clearly.

74.4 Conclusion and Risk Assessment Laser ablation is applied for a high variety of objects and materials in laser cleaning. Despite the unique advantages of this technique, process by-products have to be considered with regard to safety and health regulations. During Nd:YAG laser cleaning, particulate matter (PM) emerges. At the highest laser ﬂuence applied in the presented laser cleaning investigation, 78% of the PM is in the nano-particle size range (30–100 nm). Lowering the ﬂuence or pulse energy results in a lower frequency of nano-particles in the PM. It is notable that femtosecond laser ablation causes ﬁner particles than nanosecond laser ablation, which may be due to e.g. higher energy density compared to Nd:YAG laser ablation. Another signiﬁcant diﬀerence of both ablation techniques is the emission rate which is one order of magnitude higher in the case of Nd:YAG laser ablation during laser cleaning [10]. Taking the further development of laser cleaning systems towards higher ablation rates into account, risk assessment and protection against aerosols (nano-particles) is becoming more and more important. Since the ultraﬁne particles are highly respirable, hazards occur especially if materials are processed which release fumes containing carcinogenic or toxic substances. It is obvious that the operator (such as the restorer) has to be eﬃciently protected from inhaling aerosols. The fraction precipitation rate of the respiratory masks or ﬁlters must be designed to meet these requirements. Given that respiratory masks often constrain renovation work by restriction of the operational head space, suction nozzles will often be the best choice of achievable hazard control techniques. But ﬁne particles have a negligible settling velocity in air, so that eﬃcient capture close to the source of emission is necessary to avoid contamination of the workplace [11–14]. Equation (74.1) shows the formula for the nominal hygienic air requirement limit values (NHL) where m ˙ G is the emission mass rate, TLV is the threshold limit value and A, B are the source components involved.

NHL = m ˙ G 3,600

A B + + ··· TLVA TLVB

[m3 h−1 ].

(74.1)

To quantify the technical measures to be applied to a workplace, the NHL may be calculated by (74.1). Based on these data and typical emission rates during Nd:YAG (2 mg h−1 ) and fs-laser cleaning (14 µg h−1 ), the air exchange rate can be calculated for emission main components for which an example is given in Table 74.1.

638

St. Barcikowski et al.

Table 74.1. Nominal hygienic air requirement limit values and air exchange rates during Nd:YAG and fs-laser cleaning Emission Main Component

GypsumE GraphiteE CarbonA Copper-II-OxideE Copper-I-OxideE

Nd:YAG Femto-Ti:Sapphire TLV [mg m−3 ] Air Air NHL NHL exchange [m3 h−1 ] exchange [m3 h−1 ] rate [1 h−1 ] rate [1 h−1 ] 6 3 1.5 1 0.1

0.3 0.7 1.5 2.1 21

1,200 2,400 4,800 7,000 70,000

0.002 0.005 0.01 0.02 0.15

8 17 33 50 500

E = Inhalable Fraction, A = Alveolar Fraction

For example, in a workplace with an air volume of 3,300 m3 (same size as the LACONA VI lecture hall), a nominal air exchange rate of 0.3–21 l h−1 or 0.002–0.15 l h−1 is required to meet the TLV during Nd:YAG or fs-laser cleaning, respectively. This means a volume ﬂow of fresh air of up to 70,000 or 500 m3 h−1 would have to be fed into the workplace if copper-I-oxide emerges during laser cleaning. But as seen from Table 74.1, the TLVs for most of the substances given as an example for emission main components in the workplace air will not be exceeded, if some 100–1000 m3 h−1 are fed to the work area, which may be easily realized by blowers. But of course, the maximum achievable control technique (MACT) for this process is to apply an emission capture module at the emission source (the ablation process). An exhaust volume ﬂow of 50–200 m3 min−1 is appropriate to assure suﬃcient capturement. As a result, safety precautions regarding non-beam hazards will contribute only marginally (<10%) to the operating costs of a laser cleaning system. From the legal point of view, it is important to consider that even if the emission source (which is the laser cleaning system) is not subject to an EUstandard or international standard in detail, it is obligatory to apply the “lowest achievable emission rate” [15, 16]. This liability applies to the owner or operator of the laser cleaning system to meet the recommended maximum achievable emission limitation [17–19]. Since laser users may like to purchase a fully equipped laser cleaning system, it may be advantageous to supply (and at least to recommend) a built-in fume extraction. A punctually applied suction (e.g. in technical alignment with the beam guidance) would be the adequate fume extraction technique. Making such systems “safer” by providing data for the improvement of the laser machine periphery, the presented studies may increase the acceptance of Nd:YAG laser cleaning and femtosecond laser cleaning.

74 Generation of Nano-Particles During Laser Ablation

639

Acknowledgements The authors would like to thank Patricia Engel (University of Applied Sciences and Arts Hildesheim/Holzminden/Göttingen, Germany) for the paper substrates. We thank Roman Ostrowski, Jan Marczak and Marek Strzelek (Institute of Optoelectronics, Military University of Technology, Warsaw, Poland) for the Nd:YAG laser installation and its operation at our institute within the EUREKA E!2542 “RENOVA LASER” project.

References 1. S. Barcikowski, R. Ostrowski, J. Marczak, M. Strzelec, J. Walter, and A. Ostendorf, in 4th International Workshop on Laser Cleaning (IWLC4) & 4th New Trends in Laser Cleaning (NETOLAC IV), Macquarie University, Sydney, Australia, 2004. Published by World Scientiﬁc Publishing Company, Singapore, as “Laser Cleaning II”, Ed. D M Kane. In press. 2. J. Bunte, S. Barcikowski, T. Puester, T. Burmester, M. Brose, and T. Ludwig, Chapter 9.2 In: Femtosecond Technology for Technical and Medical Applications (Ed.: Dausinger, Lichtner, Lubatschowski), Springer Reihe “Topics in Applied Physics”, Vol. 96/2004, 309–321. 3. H. Haferkamp, M. Goede, S. Barcikowski: Laser Generated Air Contaminants – Incorporating environmental aspects in quality control of laser material processing. In: Proceedings Laser 2001 World of Photonics – Lasers in Manufacturing. December 2001, pp. 716–726. AT-Fachverlag GmbH, Stuttgart. 4. S. Barcikowski, N. Bärsch, M. Hustedt, R. Sattari, A. Ostendorf: Continuous Production and Online-Characterization of Nanoparticles from Ultrafast Laser Ablation and Laser Cracking. In: Proceedings of 23rd International Conference on Applications of Lasers and Electro-Optics ICALEO 2005, 31.Oct.-03.Nov, Miami, CA, USA. Paper M707. 5. ICRP. 1994. Human respiratory tract model for radiological protection. Ann. ICRP 24 (1–3). ICRP publication 66. 6. J.S. Brown, K.L. Zeman, W. D. Bennett, in Am Journal Respir Crit Care Med., Vol. 166, 1240, 2002. 7. S.P. Faux, C.-L. Tran, B. G. Miller, A.D. Jones, in Health and Safety Executive, Research Report 154. 2003. 8. G. Oberdörster, J. Ferin, and B.E. Lehnert, in Environ Health Perspect Vol. 102, 173, 1994. 9. P.J.A. Borm and W. Kreyling, in J. Nanosci. Nano tech., Vol. 4(5), 521, 2004. 10. T. Burmester, M. Meier, H. Haferkamp, S. Barcikowski, J. Bunte, and A. Ostendorf, in Lasers in the Conservation of Artworks. LACONA V Proceedings, Osnabrueck, Germany, Sept. 15–18, 2003. Springer Proceedings in Physics, Band 100 (Hrsg.: K. Dickmann, C. Fotakis, J.F. Asmus). 61, 2005. 11. H. Haferkamp, F. von Alvensleben, D. Seebaum, M. Goede, T. Püster, in Proceedings: International Laser Safety Conference, 17.-20.03.97 Orlando. 12. H. Haferkamp, M. Goede, T. Püster, et.al., AILU workshop, “Processing plastics with lasers”, 15. Nov. 2000, Warwick, Coventry. 13. ISO/TR 12100-1:1992, Safety of machinery – Part 1: Basic concepts, general principles of design – basic terminology, methodology.

640

St. Barcikowski et al.

14. DIN EN :1997-01, Safety of machinery – Principles for risk assessment. 15. N.N.: Council Directive 1999/30/EC of 22 April 1999 relating to limit values for sulphur dioxide, nitrogen dioxide and oxides of nitrogen, particulate matter and lead in ambient air. Oﬃcial Journal L 163, 29/06/1999 P. 0041–0060. 16. N.N.: Environmental Data Germany 2002. Federal Environmental Agency (Umweltbundesamt). Berlin 2003. 17. 40 CFR Part 63: Part II Environmental Protection Agency. Federal Register: March 23, 2001 (Volume 66, Number 57) Proposed Rules, Page 16317–16360. Revised as of July 1, 2000. From the Federal Register Online via GPO Access: DOCID:fr23mr01-36. 18. CAA90: Clean Air Act 1990, Sec. 112 and Title 42 – The public health and welfare. Chapter 85 – Air pollution prevention and control. Subchapter 1 – Programs and activities. Part A – Air quality emission limitations. Hazardous air pollutants. From the U.S. Code Online via GPO Access :CITE: 42USC7412. (Laws in eﬀect as of January 27, 1998). 19. 29CFR1910: Occupational safety and health standards. Title 29 – Labor chapter 17 – Occupational safety and health administration, department of labor (continued) Part 1910. 29 CFR 1910, 7–543.

75 A Novel Portable Multi-Wavelength Laser System ∗

A. Charlton and B. Dickinson Lynton Lasers Ltd, Lynton House, Manor Lane, Holmes Chapel, CW4 8AF, United Kingdom ∗ [email protected] Summary. There is an established need for a portable and aﬀordable Q-switched laser system for use in studio conservation and small scale ﬁeld use. The ideal system would be capable of producing a variety of wavelengths ranging from the ultraviolet to the infrared with suﬃcient energy per pulse to treat a wide range of materials including stone, marble, terracotta, wood, organic materials, bone, parchment, textiles, and metals. In this paper we report on such a system which is capable of delivering Q-switched output at 1,064 nm in excess of 300 mJ per pulse and at repetition rates of up to 25 Hz. Additional outputs are also reported at 266 nm, 355 nm, 532 nm, and 2.94 µm. Preliminary cleaning results on a small range of objects using the Q-switched 1,064 nm output are presented.

75.1 Introduction There are several diﬀerent laser manufacturers supplying equipment to the conservation market [1], and almost without exception the design and development is a by-product of existing technology for completely diﬀerent applications ranging from medical to industrial. For this reason, the laser products which are available to conservators are generally expensive to buy, and diﬃcult to use and maintain. This is especially true when exposed to the working environment found on a typical conservation worksite. Factors which should be considered at the design stage include portability both within a studio environment and on scaﬀolding; ﬂexibility of the delivery system; 110 V operation for ﬁeld use; 110 V or 240 V single phase operation for studio use; usability in hostile environmental and weather conditions; low susceptibility to dirt, contamination and even rain; ease of maintenance; and ﬁnally cost. Gradually manufacturers are taking these issues into account and it is hoped that this will eventually bring the beneﬁts of laser technology to an ever increasing number of conservators around the world.

642

A. Charlton and B. Dickinson

75.2 Experimental Methods A system has been developed using established medical technology and incorporating a microprocessor controlled, variable pulse width, high current power supply to drive a range of intelligent laser handpieces. The handpieces consist of a gain medium [Nd:YAG or Er:YAG] and optical cavity together with a Q-switch to generate short [5 ns] pulses. Harmonics of the fundamental wavelength are produced using appropriate non-linear crystals. Each handpiece is uniquely identiﬁed by means of a programmable integrated chip to enable failsafe operation of the power supply and store essential usage information. A schematic of the system design is shown in Fig. 75.1. The system comprises a DC power supply and capacitor which can store up to 3 kJ of energy; a user interface which allows the operator to control the standard output parameters of the laser (energy per pulse and repetition rate); a cooling system with water-to-air heat exchanger rated at 800 W; and a laser handpiece weighing less than 1 kg which is connected by approximately 2 m of umbilical to the main power system. The whole of the main power unit is managed by a microprocessor unit which controls the output and monitors various interlocks to ensure safe operation of each laser handpiece. The laser handpiece comprises a gain medium (Nd:YAG or Er:YAG) housed in ceramic or barium sulphate pumping chamber together with a pulsed Xe ﬂashlamp. The laser cavity consists of a high reﬂectivity plane mirror and output coupler together with a Q-switch which allows short pulses to be generated. Harmonics are generated using KTP and BBO non-linear crystals to give useful output at 532 nm, 355∗ nm, and 266∗ nm [∗ not yet fully tested]. As the system operates a number of diﬀerent laser handpieces with varying operating conditions, the set up parameters are stored on a programmable

Fig. 75.1. Layout of compact Phoenix laser system

75 A Novel Portable Multi-Wavelength Laser System

643

integrated circuit (PIC) which is contained in the detachable connector shown in Fig. 75.1. The PIC contains information on the minimum and maximum ﬂashlamp duration (ranging from 100 µs to 10 ms) and the number of pulses ﬁred by each laser handpiece. The dimensions of the prototype unit shown in Fig. 75.2 are approximately 50×40×30 cm. The whole system weighs under 30 kg and requires an electrical supply at either 110 V or 240 V to operate making it ideal for use either within a conservation studio or on site. In general the higher output powers were obtained using the ceramic pumping chamber as the barium sulphate degraded rapidly with the high input powers.

Fig. 75.2. Photograph of the compact Phoenix prototype Table 75.1. Results obtained from various laser handpieces Handpiece

Wavelength

Energy In

Energy Out

Rep. Rate

Q-switched Nd:YAG Frequency doubled Q-switched Nd:YAG Frequency tripled or quadrupled Q-switched Nd:YAG Erbium YAG Long pulse Nd:YAG

1,064 nm 532 nm

10–20 J 10–20 J

100–300 mJ 35–100 mJ

1–25 Hz 1–25 Hz

266 nm or 355 nm

10–20 J

5–10∗ mJ

1–10 Hz

2.94 µm 1,064 nm

10–30 J 10–20 J

5–175 mJ 120–500 mJ

1–15 Hz 1–25 Hz

∗

The quoted results for 266 nm and 355 nm handpieces are based on previous experimental data. However measured outputs from this handpiece are not available at time of printing. The main results are shown graphically in Fig. 75.3.

644

A. Charlton and B. Dickinson 600 Energy Out (mJ) at 25Hz

1064 nm

500

532 nm 2940 nm

400

1064 nm LP

300 200 100 0 9

14

18

Energy In (J)

Fig. 75.3. Energy Out per pulse v. Energy In for the Q-switched Nd:YAG, frequency doubled Nd:YAG, Erbium YAG and long pulse (LP) Nd:YAG laser handpieces

75.3 Results and Discussion The laser system has been tested and is capable of producing output ranging from the UV to the infrared. A total of ﬁve handpieces were used for the testing as follows: The output power of the Nd:YAG Q-switched laser handpieces was measured as a function of input power (Fig. 75.4) by varying the repetition rate, and exhibits an almost linear relationship with a maximum power achieved of 8 W at 25 Hz. This suggests that thermal eﬀects are not too severe and it may be possible to increase the performance further still. Results for the Erbium YAG at 2.94 µm however show the strong thermal lensing eﬀect in this medium which makes the laser cavity become unstable and causes the output to drop rapidly above a repetition rate of 10 Hz. Further work is planned to scale the output of the erbium YAG handpiece at higher average powers through improved laser cavity optimization. The laser has been used with the Q-switched Nd:YAG handpiece operating at 1064 nm to clean a variety of samples provided by the City of London, National Museums of Liverpool (NML) and Lincoln Cathedral. In each case the laser successfully removed the pollution layer without causing any visible damage to the patina of the underlying object. Examination by optical microscopy at the NML Conservation Centre conﬁrmed the quality of the cleaning and a selection of objects cleaned are presented in Figs. 75.5–75.8.

Power Out (W) at 18.4 J lamp energy

75 A Novel Portable Multi-Wavelength Laser System

645

9 8 7 6 1064 nm

5

532 nm

4

2940 nm

3 2 1 0 0

100

200

300

400

500

Power In (W)

Fig. 75.4. Output of laser handpieces as a function of input power with increasing repetition rate. N.B. output at 2,940 nm scaled ×10

Fig. 75.5. A piece of early twentieth century glass [approximately 5 × 5 cm] which has been cleaned using the laser operating at 300 mJ per pulse at 5 Hz and 100 mJ per pulse at 10 Hz. No discernible diﬀerence was observed between the results and in each case an area of around 5 cm2 was cleaned in approximately 30 s

Fig. 75.6. A small sample of sulphated Portland stone cleaned at The Monument to the Great Fire in London. The laser was operated at 100 mJ per pulse at 20 Hz and an area of approximately 10 cm2 was cleaned in slightly less than 1 min

646

A. Charlton and B. Dickinson

Fig. 75.7. A piece of Lincoln limestone [approximately 10 cm high] from Lincoln Cathedral cleaned using the laser operating at 300 mJ per pulse at 10 Hz. The sample is heavily sulphated and it took several minutes to clean the top half of the sample representing an area of approximately 10 cm2

Fig. 75.8. A plaster frieze from the National Museum of Liverpool with an area approximately 5×5 cm cleaned in the lower left hand corner. The laser was operated at 300 mJ per pulse at 10 Hz and a total of 2,000 pulses was required to clean the area

75.4 Conclusions We have successfully designed and produced an aﬀordable, portable multiwavelength laser system which will allow a large number of conservators access to technology which has previously been exclusive in its nature due to the high capital cost of the equipment. The system is capable of delivering Qswitched output energy of 300 mJ per pulse at a repetition rate of 25 Hz and a wavelength of 1,064 nm. Outputs in excess of 100 mJ per pulse at 532 nm and 2.94 µm have also been measured, and outputs at 266 nm and 355 nm are currently being investigated. Measured cleaning rates vary from approximately 100–600 cm2 per watt per hour giving a realistic average cleaning time of approximately 2–4 h per m2 with the system operating at maximum

75 A Novel Portable Multi-Wavelength Laser System

647

output power. The system can be used in a conservation studio or on site, with interchangeable handpieces to generate various wavelengths allowing the conservator to clean a wide variety of objects.

Reference 1. Artwork Conservation by Laser in Europe Database, Programme of the European Commission: Cost Action G7.

springer proceedings in physics 70 Computational Approaches in Condensed-Matter Physics Editors: S. Miyashita, M. Imada, and H. Takayama

83 Computer Simulation Studies in Condensed-Matter Physics X Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler

71 Amorphous and Crystalline Silicon Carbide IV Editors: C.Y. Yang, M.M. Rahman, and G.L. Harris

84 Computer Simulation Studies in Condensed-Matter Physics XI Editors: D.P. Landau and H.-B. Sch¨uttler

72 Computer Simulation Studies in Condensed-Matter Physics IV Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler 73 Surface Science Principles and Applications Editors: R.F. Howe, R.N: Lamb, and K. Wandelt 74 Time-Resolved Vibrational Spectroscopy VI Editors: A. Lau, F. Siebert, and W. Werncke 75 Computer Simulation Studies in Condensed-Matter Physics V Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler 76 Computer Simulation Studies in Condensed-Matter Physics VI Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler 77 Quantum Optics VI Editors: D.F. Walls and J.D. Harvey 78 Computer Simulation Studies in Condensed-Matter Physics VII Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler 79 Nonlinear Dynamics and Pattern Formation in Semiconductors and Devices Editor: F.-J. Niedernostheide 80 Computer Simulation Studies in Condensed-Matter Physics VIII Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler 81 Materials and Measurements in Molecular Electronics Editors: K. Kajimura and S. Kuroda 82 Computer Simulation Studies in Condensed-Matter Physics IX Editors: D.P. Landau, K.K. Mon, and H.-B. Sch¨uttler

85 Computer Simulation Studies in Condensed-Matter Physics XII Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 86 Computer Simulation Studies in Condensed-Matter Physics XIII Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 87 Proceedings of the 25th International Conference on the Physics of Semiconductors Editors: N. Miura and T. Ando 88 Starburst Galaxies Near and Far Editors: L. Tacconi and D. Lutz 89 Computer Simulation Studies in Condensed-Matter Physics XIV Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 90 Computer Simulation Studies in Condensed-Matter Physics XV Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 91 The Dense Interstellar Medium in Galaxies Editors: S. Pfalzner, C. Kramer, C. Straubmeier, and A. Heithausen 92 Beyond the Standard Model 2003 Editor: H.V. Klapdor-Kleingrothaus 93 ISSMGE Experimental Studies Editor: T. Schanz 94 ISSMGE Numerical and Theoretical Approaches Editor: T. Schanz 95 Computer Simulation Studies in Condensed-Matter Physics XVI Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler