METALLOGRAPHY

AND

MICROSTRUCTURE

OF

ANCIENT

AND

HISTORIC METALS

METALLOGRAPHY

AND

MICROSTRUCTURE

OF

ANCIENT

AND

HISTORIC METALS DAVID A. SCOTT

THE GETTY CONSERVATION INSTITUTE IN THEASSOCIATION J. PAUL WITH GETTY ARCHETYPE MUSEUM BOOKS Front and back cover: Photomicrograph of a Wootz Credits: Figures 73-74: Courtesy of the American steel prill from the Deccan region ofIndia. The steel is Society for Testing and Materials; Figures 106, 145, hypereurectoid and cast, and was made in a crucible 148, 162: Peter Dorrell, Photography Department, process. Voids appear dark. The pearlite appears in a Institute of Archaeology, London; Figures 1-8,12-20, variety of colors due to differences in spacing. The 26-40,55,198-212: redrawn by Janet Spehar white needles are cementite; the occasional lighter Enriquez; Figures 198-212: Courtesy of the Interna­ patches are rich in phosphorus. This crucible steel is a tional Copper Research and Development Association; high-quality product of ancient historic significance. Figures 75-80: Dennis Keeley; Cover, Plates 1-20, Color interference tint etched in selenic acid. x420. Figures 9-11, 21-25, 41-54: David A. Scott.

Publications Coordinator: Irina Averkieff, GCI

Editor: Irina Averkieff

Technical Drawing: Janet Spehar Enriquez

Design: Marquita Stanfield Design, Los Angeles, California

Typography: FrameMaker / Adobe Garamond and Gill Sans

Printing: Tien Wah Press, Ltd.

© 1991 The J. Paul Getty Trust All rights reserved

Published in association with Archetype Books which acknowledge a grant from the Commission of European Communities

Printed in Singapore

Libraty of Congress Cataloguing-in-Publication Data

Scott, David A.

Metallography and microstructure of ancient and historic metals/

David A. Scott.

p. cm.

Includes bibliographical references and index.

ISBN 0-89236-195-6 (pbk.)

1. Metallography. 2. Alloys--Metallography. 3. Metallographic

specimens. 4. Art objects--Conservation and restoration.

I. Tide. TN690.S34 1991

669'.95--dc20 91-19484

CIP THE GETrY CONSERVATION INSTITUTE

The Getty Conservation Institute, an operating program of the J. Paul Getty Trust, was created in 1982 to address the conservation needs of our cultural heritage. The Institute conducts world­ wide, interdisciplinaty professional programs in scientific research, training, and documentation. This is accomplished through a combination of in-house projects and collaborative ventures with other organizations in the USA and abroad. Special activities such as field projects, interna­ tional conferences, and publications strengthen the role of the Institute.

TABLE OF CONTENTS

Foreword IX Preface XI List of Color Plates and Figures Xlli Color Plates xv

The Nature of Metals

2 The Microstructure of Ancient Metals 5

3 Two-phased Materials 11

4 The Microstructure of Tin Bronzes 25

5 Notes on the Structure of Carbon Steels 31

6 Martensite in Low-carbon Steels 33

7 The Tempering of Martensite 35

8 Structure and Properties of Cast Iron 37

9 Corroded Microstructures 43

10 Reflected Polarized Light Microscopy 49

11 Grain Sizes of Ancient Metals 51

12 Metallography and Ancient Metals 57

13 Metallographic Sampling of Metals 61

14 Mounting and Preparing Specimens 63

15 Recording Results 67

16 Etching and Etching Solutions 69

17 Mounting Resins 75

18 Microhardness Testing 77

A Appendix: Common Microstructural Shapes 79

B Appendix: Microstructure of Corroded Metals 81

C Appendix: Microhardness Values fo r Diffe rent Alloys and Metals 82

D Appendix: Alloys Used in Antiquity 84

E Appendix: Terms and Techniques in Ancient Metalworking 85

F Appendix: Metallographic Studies 86

G Appendix: Phase Diagrams 121

Glossary 137

Bibliography 147

Index 151

FOREWORD

Information about the structure of materials and technology of manufacture of ancient and his­ toric objects and artifacts can provide insight into their date, place of origin, and probable use. Investigations of structure may also be very important fo r documentation, preservation strat­ egy, or conservation treatment. For many decades objects from a variety of materials and dating from prehistoty to the present have been scientif­ ically studied, many of them by metallographic techniques. Most of the results are scattered throughout the international literature, are some­ times inaccessible, and often not published at all. Frequently the need has been expressed, in national and international meetings, to collect information on certain types of objects or classes of materials in one place and to make it available as a database or publication. This volume is an attempt to provide a measured amount of infor­ mation regarding the techniques of metallogra­ phy as they apply to ancient and historic metals. It is illustrated with many examples of different types of microstructure, drawn from David Scott's many years of experience in this field of study. We hope that the present volume, developed with the guidance of Dr. Frank Preusser, Associ­ ate Director, Programs, GCl, will be a useful book fo r students, conservators, conservation sci­ entists, and workers in the area of metallography, especially those seeking to understand the nature of microstructure as it applies to ancient materi­ als. The book is the first in a series of reference works that the Getty Conservation Institute is publishing on materials used in conservation and technology. The Getty Conservation Institute and the ]. Paul Getty Museum have been involved collaboratively with this work and present this volume as copublishers.

Miguel Angel Corzo Director The Getty Conservation Institute Marina del Rey, California

PREFACE

This book began as a series of laboratory notes of samples fo r metallographic study. The quanti­ and the author hopes that in the process of rewrit­ tative interpretation of alloy phase diagrams has ing and integrating, the original text has been ren­ not been included here, and in general, mathe­ dered more accessible. There are many studies of matical content has been kept to a minimum. ancient and historic metalwork published in the The practical information in the text also includes literature, but it is more difficultto find a general details on etching solutions and short accounts of account of metallographic techniques and an microhardness and the grain size of metals. There interpretation ofmicrostructure written primarily is a lengthy appendix (F) in which examples of fo r the conservation scientist and conservator. different rypes of alloys and microstructures are This book attempts to fill this gap by ptoviding a given, drawn from studies carried out by the guide to the structure of metals. From the mate­ author. This appendix is not comprehensive, but rials science perspective, it is also useful to explore it is hoped that the reader will find it interesting the ways in which alloys have been used in and informative. ancient metalwork. The analytical data that have been presented There are many reasons fo r studying the in the book are quoted without a discussion of structure of metals. The proper conservation of how the results have been obtained. There are objects requires or sometimes enables the conser­ many accounts of analytical methods and tech­ vator to observe microstructure. Investigative niques, such as electron microprobe analysis, studies may be necessary in order to assess the atomic absorption spectrophotometry, induc­ degree of corrosion or embrittlement of an object. tively coupled plasma mass spectrometry, and x­ A new conservation treatment may have implica­ ray fluorescence analysis, and one or more of tions fo r the preservation of metallographic struc­ these techniques are the principal methods by ture. Cyril Stanley Smith states that the hierarchy which the results quoted in the text have been of structure can be examined at many different obtained. It was not the aim of the present text to levels ofaggregation and that the incorporation of enter into detail concerning the chemical analysis empirical experience of materials into a theoreti­ of metals. Similarly, although corrosion and cor­ cal framework has enabled materials science to rosion products are often essential components of appreciate the effects of structure on properties ancient metals, there is no detailed discussion of and even the artistic qualities of materials. It is the nature of corrosion products given, since to certainly true that metallographic structures do so would add substantially to the length of the themselves are often visually compelling both in a book. The smelting, casting, and working of met­ scientific and an artistic sense. Metals are interest­ als is also not covered in detail by the text, ing materials since their properties can be manip­ although the glossary does provide some informa­ ulated in many ways. By combining metals, by tion and common terms used in describing metals heating and quenching, by making them liquid and metalworking processes. and casting them, or by working them to shape with a hammer or a lathe, they allow a plasticiry Acknowledgments of movement while being shaped and a finaliry of The author is very grateful to the staff of the fo rm when that process is completed. Getry Conservation Institute Publications The structure of the book should have a word Department fo r seeing the manuscript through of explanation here. The approach that has been from editing to printing, in particular to Irina taken is to describe briefly what metals are and to Averkieff fo r her thoughtful and dedicated edito­ discuss phase diagrams and the kinds ofstructur es rial work. Janet Enriquez was responsible fo r to be fo und in different and relevant alloys, before redrawing the original figures, Dennis Keeley proceeding to deal with the practical application took the photographs in Chapter 11, and Mar­ of this knowledge: the sampling and preparation quita Stanfield directed the overall design. Nota- bly, Frank Preusser, Associate Director fo r Programs, and Irina Averkieff, Publications Coordinator, must be thanked for their enthusi­ asm and support. The author is also grateful to Summer Schools Press fo r assistance with the publication of the first version of the text. Several of the photomicrographs taken by the author would not have been possible without the help and assistance provided by those who have generously devoted samples or time to the cause. In particular I would like to thank Dr. Nigel See­ ley, fo rmer Head of the Department of Conserva­ tion and Materials Science, Institute of Archae­ ology, London, currently Surveyor of Conserva­ tion, National Trust, London; James Black, International Academic Projects, London; Dr. Rodney Clough, fo rmerly Research Associate, Department of Conservation and Materials Sci­ ence, London, and fo rmer students of the Department, Noel Siver, Heather Burns, Bob Haber, Dr. Warangkhana Rajpitak, Naylour Ghandour, Dr. Abdulrasool Vatandoost­ Haghighi, and Jane Porter. I would like to give thanks to the following members of the staff at the Institute of Archaeology: Dr. Warwick Bray, Reader in South American Prehistory; Peter Dor­ rell, Head of the Photography Department; and Stuart Laidlaw, Senior Photographic Technician. At the Getty Conservation Institute I would like to thank, in addition, my secretary Ruth Feldman, who has carried out many retyping and reformatting jobs in connection with the prepara­ tion of the manuscript; Dr. Neville Agnew, Director of Special Projects; and Michael Schill­ ing, Associate Scientist. From the J. Paul Getty Museum I am most grateful to Jerry Podany, Head of the Department of Antiquities Conser­ vation and Linda Strauss, Associate Conservator, Department of Decorative Arts and Sculpture Conservation.

Dr. David A. Scott Head, Museum Services The Getty Conservation Institute COLOR PLATES AND FIGURES

Plate I. Section from a Lurisran Fig. 7. Progressive movement of an in Ag-Cu. axe. dagger handle. edge dislocation. Fig. 39. Cu-Au phase diagram. Fig. 71. Section of a corroded Plate 2. Fragment of corrosion crust Fig. 8. Dendrite atms. Fig. 40. Cu-Pb phase diagram. fragment from an Ecuadorian gilded from a Chinese cast iron lion. Fig. 9. Polished and unetched view Fig. 41. Cast toggle pin from Iran. copper ceremonial axe. Plate 3. Section from an outdoor of a section through a "Darien"-sryle Fig. 42. Chinese cast-bronze incense Fig. 72. Nomograph for grain size. bronze sculpture. pectoral from Ancient Colombia. burner. Fig. 73. Typical standard for Plate 4. Fragment of a brass Fig. 10. Polished section of a small Fig. 43. A small mirror of beta­ estimating the (austenitic) grain size medallion from the La Perouse cast frog from the T airona area of quenched bronze from Sumatra. of steel. shipwreck off the coast of Australia. Colombia. Fig. 44. High-tin bronze mirror Fig. 74. Typical standard for Plate 5. Roman mitror from Fig. I la-f. Some microstructural from Java. estimating the (austenitic) grain size Canterbury. features in solid solution FCC Fig. 45. Cast high-tin leaded bronze of annealed nonferrous materials Plate 6. Corroded section of a metals. of 22% tin,6% lead, and 72% such as brass, bronze,and nickel Roman incense burner. Fig. 12. Relationship between single­ copper. silver. Plate 7. Cast iron fragment from phase strucrures in FCC metals. Fig. 46. Laborarory quenched alloy Fig. 75. Mounting small specimens. 19th-century scales. Fig. 13. Section through copper alloy of 24% tin,76% copper. Fig. 76. Grinding mounted samples. Plate 8. Section from a high-tin axe from Iran showing twinned Figs. 47-50. Japanese sword blade Fig. 77. Polishing mounted samples. bronze vessel from Thailand. grains. fragment. Fig. 78. Sample storage. Plate 9. Section from a Greek Herm Fig. 14. Twinned grains of gold­ Fig. 51. Partially Widmanstatten Fig. 79. Examination by polarized of about 120 B.C. copper alloy sheet. steel. light microscopy. Plate 10. Iranian Iron Age dagger Fig. 15. Twin planes in Indian zinc Fig. 52. Grain boundary structure Fig. 80. Use of inverted stage hilt. co 111. with subgrain features. metallurgical microscope. Plate II. Section of an Islamic Fig. 16. Twin planes in zinc. Fig. 53. Grain size of knife edge. Fig. 81. Drawing of an axe showing inkwell. Fig. 17. Twin planes in zinc. Fig. 54. Banded structure of a ideal location of sample cuttings. Plate 12. Section of a corroded cast Fig. 18. Phase diagram for the gold­ quenched sword blade. Fig. 82. Two samples of mounted iron cannonball from the Tower of silver system. Fig. 55. Part of the Fe-Fe3C phase wire or rod. London. Fig. 19. Eutectic diagram of silver­ diagram. Fig. 83a-c. Holding small samples. Plate 13. Section of a cast iron copper alloy. Fig. 56. Steel prill from lid of a Fig. 84a-d. Embedding small cannonball hom Sandal Castle. Fig. 20. Eutectic-rype Wootz crucible, Deccan area of samples. Plate 14. Section of a high-tin bronze microstructures. India. Fig. 85. Shapes of ferrite in low­ mirror from Java. Fig. 21. Dendritic a in 60% Ag 40% Fig. 57. Medieval knife blade from carbon steels. Plate 15. Section from a small Indian Cu cast alloy. Ardingley,Sussex, England. Fig. 86. Common descriptive Wootz ingot. Fig. 22. 60% Ag 40% Cu etched in Fig. 58. Photomicrograph of kris microstructural terms. Plate 16. Section from a Japanese potassium dichromate. from India. Figs. 87-89. Base silver-copper alloy swotd blade. Fig. 23. 60% Ag 40% Cu cast alloy Figs. 59,60. French cut-steel bead. coin from western India. Plate 17. Late Bronze Age sword illustrating eutectic infill. Fig. 61. Part of the Fe-Fe3C phase Figs. 90-93. Islamic inlaid inkwell from Palestine. Fig. 24. Wootz steel ingot from India diagram for cast iron. cast in a copper-tin-zinc-Iead alloy. Plate 18. Polished and etched section Figs. 25a, b. a and � Fig. 62a-f. Flake graphite in cast Figs. 94-96. Cast bronze arrowhead of a late medieval Indian zinc coin. microstructures. Iron. from Palestine. Plate 19. Section of a small Luristan Fig. 26. Eutectic a and �. Fig. 63. 18th-century cast iron scales. Figs. 97-99. Palestine bronze sword. ceremonial axe from Iran. Fig. 27. Eutectic and dendrites. Fig. 64. Cast iron cannonball from Fig. 100. Roman wrought iron. Plate 20. Section from inside the Fig. 28. Fibrous structure in worked the Tower of London. Figs. 101-103. Colombian gold­ base of a Greek Herm. two-phase alloy. Fig. 65. Typical variations in the copper alloy sheet. Fig. !. Close-packed hexagonal unit Fig. 29. a and Iieutectoid. preservation of surface detail in Figs. 104,105. Ecuadorian copper­ cell structure. Fig. 30. Iron-carbon phase diagram. ancient metallic artifacts. alloy nose ornament. Fig. 2. Face-centered cubic unit cell. Fig. 31a-d. Breakdown of 0 grains. Fig. 66. Mounted and polished Figs. 106-108. Cast arsenical copper Fig. 3. Body-centered cubic unit cell. Fig. 32. Cementite and pearlite. section through a bronze rod axe from Ecuador. Fig. 4. Graph of relationship Fig. 33. Copper-tin phase diagram. fragment. Figs. 109, 110. Chinese bronze between stress and strain. Fig. 34. a + � peritectic. Fig. 67. Drawing of the cross section incense burner. Fig. 5a, b. Stress and strain relation Fig. 35. E phase grains. of a bronze rod fragment. Fig. Ill. Thai bronze cast bell. for FCC,BCC, and CPH; stress and Fig. 36. Copper-zinc phase diagram. Fig. 68a-d. Examples of corrosion of Figs. 112,113. Luristan dagger strain for interstitial materials. Fig. 37. � grains in copper-zinc. gold-copper alloys. handle. Fig. 6. An edge dislocation. Fig. 38. Discontinuous precipitation Figs. 69,70. Luristan ceremonial Figs. 114, 115. Fragment of a Thai bronze container. Fig. 173. Section of a circular globular region of pearlite. Figs.116,117. Columbian cast Siml bracelet from Thailand. Fig. 194. White cast iron etched in ear ornament. Fig. 174. Structure of the circular Murakami's reagent and picral. Figs. IIS-120. Cast iron cannonball bracelet from Thailand after etching Figs. 195,196. Tin ingot from from Sandal Castle. in alcoholic ferric chloride. Cornwall. Figs. 121,122. Bronze Age copper Fig. 175. High magnification of Fig. 197. Tensile properties, impact ingot from Hampshire, England. circular bracelet from Thailand value,and hardness of wrought Figs. 123-125. Roman brass coin. showing redeposited copper, copper copper-tin alloys. Figs. 126,127. Thai bronze sulfide inclusions, and bronze metal. Fig. 198. Copper-tin system. container fragment. Fig. 176. Section through a Roman Fig. 199. Part of the copper-tin Figs. 128-130. Gold necklace bead bronze bowl,lightly etched in diagram under different conditions. from Colombia. alcoholic ferric chloride. Fig. 200. Copper-arsenic system. Figs. 131-133. Gold alloy nail and Fig. 177. Early medieval brass sheet Fig. 201. Copper-lead binary system. gold ear spool. etched in alcoholic ferric chloride Fig. 202. Copper-iron binary system. Figs. 134,135. Tang and blade and potassium dichromate. Fig. 203. Copper-gold binary sectlon. Fig. 17S. Recrystallized and heavily system. Figs. 136,137. Iron knife. worked grain structure of brass stud. Fig. 204. Copper-antimony binary Fig. 138. Roman copper alloy coin. Fig. 179. Microstructure of system. Figs. 139, 140. Roman iron nail. Romano-Greek iron arrowhead. Fig. 205. Copper-silver binary Figs. 141-144. Native copper from Fig. 180. Unusual corrosion pattern system. the Great Lakes region in North through Byzantine bronze blade. Fig. 206. Copper-nickel binary America. Fig. lSI. Heavily etched view of system. Figs. 145-147. Head of a roggle pin Byzantine leaf-shaped blade. Fig. 207. Copper-zinc binary system. from Iran. Fig. 182. Corroded iron knife blade Fig. 208. Iron-carbon system. Figs. 148-151. Javanese iron blade. from medieval Britain. Fig. 209 a. Lead-tin system Figs. 152, 153. Bronze axe fragment Fig. 183. Iron knife blade showing a (pewters). b. Gold-silver system. from Iran. weld where different pieces of iron Fig. 210. Copper-silver-gold ternary Figs. 154,155. Laboratory cast have been joined togerher. liquidus. 60:40 brass. Fig. 184. Low-carbon steel area of Fig. 211 . Copper-silver-gold ternary Figs. 156-159. Gilded silver earring medieval iron knife blade. solidus. from Jordan. Fig. IS5. Section of panpipes etched Fig. 212. Copper-tin-Iead ternary Figs. 160, 161. Fragment of a brass in cyanide/persulfate. system. medallion from Australian Fig. IS6. Fragment of Byzantine iron shipwreck. dagger blade showing part of the Figs. 162-164. Fragment of a small edge. Siml ear ornament. Fig. 187. Overall view of the Roman Fig. 165. Roman mirror fragment. coin of Victor en us etched in Fig. 166. Roman bronze figurine. alcoholic ferric chloride. Fig. 167. Roman btonze Fig. lS8. Grain structure of the microstructure. Roman coin of Victoren us. Fig. 168. Renaissance silver basin Fig. IS9. Microstructure of grey cast from Genoa. iron of the early 20th century. Fig. 169. Part of solder blob from a Fig. 190. Grey cast iron showing repair to the outer radial panel of the graphite flakes and pearlitic infill. Renaissance silver basin. Fig. 191. Grey cast iron showing the Fig. 170. Section through the cast structure of the iron-carbon Renaissance silver basin. alloy. Fig. 171. Overall view of a core Fig. 192. Graphite flakes in a ferrite drilled plug from the silver basin. matrix with an infill of pearlite Fig. 172. Part of the silver sheet containing some steadite patches. etched in acidified potassium Fig. 193. White cast iron showing dichromate. long cementite laths and small

1 THE NATURE OF METALS

Metals are an aggregation of atoms that, apart metal is cooled very rapidly, as in splat cooling, from mercury, are solid at room temperature. the normal crystalline structure can be sup­ These atoms are held together by "metallic pressed. In splat cooling, metal droplets are bonds" that result from sharing available elec­ cooled very quickly between chilled metal plates trons. A negative electron bond pervades the and the structure that develops is similar to structure, and heat and electricity can be con­ glass-a random arrangement of atoms rather ducted through the metal by the free movement than a crystalline array. In the usual crystalline of electrons. The negative electron bond sur­ state, metal will consist of a number of discrete rounds the positive ions that make up the crystal grains. The metal is then referred to as being poly­ structure of the metal. There are three common crystalline. An important property of metals is unitFigure cell I. structure. Close-packed Atomic hexagonal packing types of lattice structure that metals belong to: that they undergo plastic deformation when 1 close-packed hexagonal, face-centered cubic, and stretched or hammered. This is illustrated by a factor of 0.74. ' body-centered cubic. stress-strain diagram using Young s Modulus (YM): Close-Packed Hexagonal (CPH) load applied Models of crystal structures can be made up of stress cross-sectional area YM= spheres stacked in close-packed layers. Two strain change in length arrangements are possible, one being hexagonal original length and the other cubic in basic structure. In the close-packed hexagonal system the spheres repeat the same1). position every second layer (ABABAB ...; Fig. low-carbon steel FCCFigure structure 2. Face-centered has four atomsunit. An per Face-Centered Cubic (FCC) ------'---pure copper unit cell and an atomic packing Layers can be built up so that the third layer of factor of 0.74 in most elemental spheres does not occupy the same position as the crystals. spheres in the first row; the structure repeats every third layer (ABCABCABC. .. ). FCC metals tend ----t_ 2 strain to be ductile (i.e., can be mechanically deformed, drawn out into wire, or hammered into sheet). Examples are lead, aluminium, copper, silver, Figure 4. Relationship between stress and strain. gold, and nickel (Fig. 2). Before plastic deformation occurs, materials Body-Centered Cubic (BCC) deform by the elastic movement of atoms that Another common type fo und in many metals, the hold the structure together. This elastic deforma­ body-centered cubic structure, is less closely tion occurs in metals and in other materials, such packed than the FCC or CPH structures and has as glass, which have no ability to deform at room Figure 3. Body-centered cubic unit atoms at the corners and one atom at the center temperature. Glass will stretch by elastic deforma­ cell. A BCC metal has two atoms of the cube. The atoms at the corners are shared tion and then break. Metals can deform plasti­ per unit cell and an atomic packing with each adjoining cube (Fig. 3). cally because planes of atoms can slip past each factor of 0.68 in elemental crystals. Other metals important in antiquity have other to produce movement. This kind of move­ BCC metals are both ductile and entirely different lattice structures; fo r example, ment cannot take place in a glassy structure. strong (e.g., iron, chromium, arsenic, antimony, and bismuth are rhombohe­ When metals such as pure copper or iron are tungsten, and molybdenum). dral, and ordinary tin is body-centered tetragonal. stretched they will break or fracture, but only Metals are crystalline solids under normal after a certain amount of plastic deformation has conditions ofworking and melting. However, if a occurred (see Figs. 4, Sa, Sb). Th e Nature of Metals 2

Hardness Dislocations The hardness of a metal is measured by its resis­ It is rare fo r crystals to have a perfect atomic struc­ tance to indentation. The metal is indented under ture; there are usually imperfections present. In a known load using a small steel ball (as in the metals, edge dislocations and screw dislocations Brinell test) or a square-based diamond pyramid are the most important faults (see Figs. 6, 7). (as in the Vickers test). In the Vickers test the These crystal fa ults enable deformation to take result is given as the Diamond Pyramid Number place at lower applied stress by slip than would be DPN (or Hv). possible if the lattice structure was perfect. When relatiFigureonsh Sa, ipright. for FCC,Stress BCC, and strain and I CPH metals compared with glass fracture points I andin in aterat typicalomic polymer. spacings An is increase I whresponile plasticsible for movement elastic behavior, results BCC (e.g.I Fe) fromWhen dis thislocations plastic givingmovement rise to can slip . typical brittle material (e.g., glass) I no longer occur fracture will take Q)V> I I CPH (e.g., Zn) I place. �.... II typical amorphous polymer (e.g., polystyrene) Figure Sb, below. Stress and strain I ----�_/�:---�FCC I! (e.g., Cu) foelemr inentsterstitial are smallmaterials. in size, In terstitialsuch as II carbon,the lattice and spacing can be of inser someted metals, into lI alloysuch asof steel,iron andwhich carbon. is an interstitial Inter­ stitials tend to reside at the base of strain E Whendislocations slip occurs, and anchor carbon them. is left --plastic deformation ---I4 elastic4 � 1-4__ -I FCC, BCC� behind and the dislocation is held 1deformation Fracture I Whenuntil some the stresshigher factorstress is reached.reached an enormous number of disloca­ upper yield point tionslower occur stress that than can originally now move at a points.required-hence the two yield Cu substitutionally alloyed, e.g., CuSn e.g.,interstitial Fe3C alloy,

Cu unalloyed

strain f Th e Nature of Metals 3

a metal is deformed then slip takes place until a Notes tangle of dislocations builds up which prevents 1. The atomic packing factor is a measure of the any further working (i.e., dislocation entangle­ space actually used by the atoms in the lattice. ment). As metal is worked, slip planes become ePH and FCC lattices are both 0.74 while the thicker and immobile. If the metal is worked fur­ Bee metals have a lower factor of 0.68. A com­ ther, it must be annealed (heated up to bring plete infill of the lattice would be 1.00. about recrystallization). Before annealing, the 2. Ductility is often used in the sense of tensile metal can be said to be work-hardened. Work­ movement (i.e., stretching) and malleability is hardening is accomplished by, fo r example, ham­ used fo r ability to be worked (i.e., hammering). mering at room temperature. This increases the metal hardness value, but decreases ductility. Figure 6. This figure shows an extra plane of atoms inserted in the edgeoriginal dis locationlattice. The and plane its direction of the of movementslip plane. are perpendicular to the edge dislocation crystal lattice Figure 7. Progressive movement of an edge dislocation showing slip. r __ r __

crystal lattice afterslip edge dislocation movement through crystal

2 THE MICROSTRUCTURE OF ANCIENT METALS

There are two basic means of manipulating met­ phenomenon that often arises in impure metals or als: they can be cast or worked. All the various alloys because one of the constituents usually has methods by which casting and working are car­ a lower melting point than the other. For exam­ ried out cannot be examined here in detail, but ple, consider the cooling of an alloy of copper and the different types of structures are described. tin. Copper melts at 1083 °C and tin at 232 °C. When the alloy cools and begins to solidifyby Casting dendritic segregation, the firstpart of the dendrite There are essentially three types of microstructure arms to fo rm are richer in copper since this con­ that can arise during the casting and cooling of a stituent solidifiesfirst, while the outer parts of the melt in a mold, regardless of the exact nature of arms are richer in tin. The result is that there is a the technology involved. Most ancient metals are compositional gradient from the inner region of a impure or are deliberate alloys of two or more dendritic arm to the outer surface. Such dendrites metals, such as copper and tin (bronze) or copper are usually referred to as cored. Coring is a com­ and zinc (brass). The fact that they are impure is mon feature in castings of bronze, arsenical cop­ an important one, for the kind of crystal growth per, debased silver, etc. It is usually necessary to that can occur is ro a large extent dependent on etch a polished section of the metal to investigate the purity of the metal. This is one reason why the whether coring is present or not. Depending on great majority of ancient castings show a den­ the amount and nature of the alloying constituent dritic structure. Dendrites look like tiny fernlike present, the remaining fluid in the interdendritic growths scattered at random throughout the channels or spaces will then solidifyto fo rm a dif­ metal. They grow larger until they meet each fe rent phase of the particular alloy system. A other. Sometimes outlines of grains fo rm between phase is any homogeneous state of a substance them, and the rate at which the metal is cooled that has a definite composition. In practice this influences their size. Usually a microscope must definition must be interpreted a little loosely be employed to make dendrites visible, but on because, very often, ancient metallic systems are objects that have cooled slowly, the dendrites not fully in equilibrium conditions, which means have also fo rmed slowly and may be visible to the that the proportion and even the composition of naked eye or under a binocular bench microscope the individual phases that are present in an alloy at low magnification (xlO or x20). The faster the may not match the precise values that can be rate of cooling, the smaller the dendrites. It is pos­ determined from a phase diagram. The subject of tertiary sible to measure the spacing between dendrite phases and phase diagrams will be taken up later arms if they are well fo rmed and to compare the in this section. Dendrites, then, dominate the secondary arm � spacings obtained from those from known alloys world of ancient castings (see Figs. 9, 10), but cast in different molds or under different condi­ there are occasions when other types of segrega­ tions. Arms of dendrites are usually referred to as tion occur in addition to dendritic segregation, or primary, secondary, or tertiary (Fig. 8). when cooling conditions give rise to completely primary arm It may be of interest to record dendritic arm different structures. spacing fo r comparative purposes, even if condi­ The other principal types of segregation are Figure 8. Dendrite arms. tions are not precisely known or there is a lack of normal segregation and inverse segregation. Nor­ background information in the metallurgical lit­ mal segregation occurs when the lower melting erature. Dendrites may be rather fuzzy or point constituent is concentrated towards the rounded in outline or quite sharp and well­ inner parr of the mold, while inverse segrega­ defined, depending on the nature of the alloy and tion- often associated with alloys of copper con­ the cooling conditions of the melt. Dendritic taining arsenic, antimony, or tin-can push the growth is actually one fo rm of segregation that alloying element to the exterior of the surface of can occur during casting. It is a segregation the mold. Inverse segregation may be responsible The Microstructure of Ancient Metals 6

unetchedFigure 9, right.section Polished through and a ancient"Darien"-style Colombia. pectoral Magnif fromication of(x 16the0) dendshowsrite selective arms. Note corrosion the verydendritic rounded shapes. impression The alloy of is the an alloy18% gold,cast by4% the silver, los t-wax68% copper process. Figure 10, far right. Polished sectionTairona of area a small of Colombia cast frog from showing the differentdendritic structure. Here fo r some of the silvery coatings occasionally castings) or causing gas porosity in the metal. The the magnification is x80 and the reported in the literature, such as the antimony third type of structure, which is particularly asso­ section has been etched with coatings on some cast Egyptian copper objects ciated with chill castings, is columnar growth. potassium cyanide/ammonium (Fink and Kopp 1933). Copper, lead, or gold Chill castings are fo rmed when metal cools persulfate etchant. castings can occasionally be relatively free of quickly on being poured into a mold. In this type impurities and on slow cooling no dendrites may of structure, long narrow crystals fo rm by selec­ be visible. Under these circumstances, the metal tive growth along an orientation toward the cen­ may cool and produce an equi-axial, hexagonal ter of the mold. They may meet each other and grain structure. An equi-axed hexagonal crystal thus completely fill the mold. It is rare to find this structure, in which all the grains are roughly the type of structure in ancient metals, although some same size, randomly oriented, and roughly hexag­ ingots may show columnar growth. onal in section, corresponds to an ideal model of a metallic grain or crystal. It is the arrangement of Working separate growing crystals that meet as they grow Working refers to a method or combination of that gives the hexagonal nature to the ideal struc­ methods fo r changing the shape of a metal or an ture, since this results in the least energy require­ alloy by techniques such as hammering, turning, ment. It is an equilibrium structure fo r this raising, drawing, etc. A list of useful terms is given reason, which the dendritic structure is not (see in Appendix E. Further details can be fo und in Fig. 1 1). One result of this is that it may be pos­ many of the texts mentioned in the bibliography, sible to obtain an equi-axed hexagonal grain especially those by U ntracht (1975) and Maryon structure by extensive annealing of the original (1971). dendritic structure. On the other hand, a den­ The initial grain structure of a homogeneous dritic structure cannot be obtained by annealing alloy can be considered as equi-axed hexagonal an equi-axed grain structure. Cast metals that do grains. When these grains are deformed by ham­ not show a dendritic structure can be quite diffi­ mering they become flattened (their shape is cult to etch and it may be difficultto develop any altered by slip, dislocation movement, and the structure apart from the visible inclusions and any generation of dislocations as a result of working) porosity in the metal. Cast metals often display until they are too brittle to work any further. At characteristic spherical holes or porosity, which this point, the grains are said to be fully work­ can be due to dissolved gases in the melt or to hardened. If further shaping or hammering of the interdendritic holes and channels that have not metal is required then the metal must be annealed been kept filled with metal during solidification. in order to restore ductility and malleability. Fur­ AI;the metal cools, the dissolved gases exsolve, ther deformation of the metal by hammering may creating reactions with the metal itself to fo rm then lead to work-hardening again and, if further oxides (for example, the production of cuprous shaping is required, then another annealing oper­ oxide [Cu20], the copper eutectic in ancient ation can be carried out. Many objects have to be The Microstructure of Ancient Metals 7

shaped by cycles of working and annealing in Schweizer and Meyers 1978). order to achieve sufficient deformation of the Cold-working and annealing can be com­ starting material, which may be a cast blank or bined into one operation by hot-working. The ingot of metal that must be cut or shaped into object to be worked is heated to near red heat and individual artifacts. Typically, annealing temper­ then immediately hammered out. The two pro­ atures would be in the range of 500-800 °C fo r cesses, namely cold-working fo llowed by anneal­ copper-based alloys, iron, and steel. If the metal is ing and hot-working, will give essentially the an alloy, then, strictly, the type of annealing oper­ same microstructure of worked and recrystallized ation should be specified: process anneal, stress­ grains, so it is always not possible to know if cold­ relief anneal, solid solution anneal, etc. Time is an working and annealing has been used in a partic­ important factor as well: too lengthy an anneal ular case, although there may be other indications may lead to grain growth and a weakening of the that have a bearing on the interpretation of the structure of the artifact; too short an anneal, and resulting structure (see Figs. 11a-f, 12). Some heterogeneity and residual stresses may not be metals, such as iron, usually must be worked into eliminated sufficiently. There are other practical shape while they are red hot. The fo rging of problems associated with annealing depending on wrought iron, which contains slag globules as the metal concerned; fo r example, when debased impurities, produces a worked structure in which silver alloys (usually silver-copper alloys) are the slag gradually becomes elongated or strung annealed by heating in air, they are liable to out into slag stringers along the length of the undergo internal oxidation. A black skin of cupric object. It is important to note that most inclusions oxide fo rms (CuO), overlying a subscale of in ancient metals do not recrystallize as a result of cuprous oxide (Cu20), while oxygen can diffuse hot working or working and annealing: they into the alloy, attacking the readily oxidized cop­ either are broken up into smaller particles or they per-rich phase and producing internal cuprite are flattened out as the working process proceeds. embedded in a silver-rich matrix (for further Face-centered cubic metals, except fo r details see Charles and Leake 1972; Smith 1971; aluminium, rectystallize by a twinning process. Figure IIa-f. Some microstructural featumetals.res in solid solution FCC dendritica. Cored grainsstructure. with remnant c.b. FullyCold-worked annealed coredhomogeneous grains. hexagonald. Cold-worked equi-axed annealed grains. me tal e.with Annealed flattened af tergrains. cold-working f.showing Cold-worked twinned af grains.ter annealing strainshowing lines distorted in the grains. twin lines and The Microstructure of Ancient Metals 8

Figure 12. Relationship between original cast material showing single-pmetals usedhase instructures antiquity. in FCC dendritic segregation )

extensive annealing will dendrcold-workedites distorted coredremove structure the segregated and

deflinesormed evident grains on heavywith some working strain equi-axed hexagonal grains ------�) �

hot-working annealing annealing showingworked defbentormed twins grains and strain now lines ��� �1??S�

New crystals that grow fo llowing the annealing of fectly straight: they may not run completely cold-worked face-centered cubic metals such as through every grain, but they are straight. If the gold, copper, silver, and their alloys, produce the grains are subsequently deformed, then the twin effect of a mirror reflection plane within the crys­ lines will also be deformed. In polished and tals, with the result that parallel straight lines can etched specimens, they appear as slightly curved be seen in etched sections traversing part or all of lines. In heavily worked metals, slip of crystal the individual grains ofthe metal (Fig. 12). After planes can occur in individual crystals resulting in annealing, the twin lines in the crystals are per- a series of parallel movements that can be seen in The Microstructure of Ancient Metals 9

etched sections as a series of fine lines in some of of cold-working. the grains. These lines are called slip bands or The percentage of deformation, or degree of strain lines. Another feature that may show that cold-work, is usually expressed in terms of reduc­ metal has been heavily worked is the presence of tion of height of the specimen, h. a fibrous morphology in the microstructure. This h h may occur when grains have been flattened out by = initial worked % cold-work x 100 h . .. hammering, producing an elongated structure rmtra I commonly described as a "textured effect." When this metal is annealed to produce a recrystallized It is often difficult to remove the segregation grain structure, the recrystallized material may that occurs during the casting operation, and still show a preferred orientation or a fibrous many ancient objects that were worked to shape nature. This is an extreme example of the fact that from a cast ingot still show some coring or rem­ recrystallized grains are not fo und to be com­ nant dendritic structure, even though the object pletely randomly oriented. Depending on the was subsequently worked to shape. Variations in amount of cold-working that the metal has composition are not confined by the new, worked received it will be able to recrystallize at succes­ grain structure and appear superimposed upon sively lower temperatures, which are markedly the grains when the specimen is etched (Figs. 13, different fo r particular metals. For example, 14). heavily worked pure copper is capable of recrys­ The kind of twinning occurring in CPH met­ tallizing at 120 0c, iron at 560 °C, zinc at 10 0c, als is shown by the structure of the Indian zinc tin at -12 0c, and lead at -12 0C. coin in Figure 15. While slip and twinning are the This is why metals like pure lead can be bent main methods by which zinc crystals accommo­ back and fo rth at room temperature without date plastic deformation, there are situations in work-hardening: they are effectively being hot­ which neither of these events can occur directly. worked at room temperature. Alloying elements This can happen when zinc crystals are com­ and impurities will, of course, affect recrystalliza­ pressed parallel to the basal plane or where tion temperature, as will grain size and the degree crystals are restrained from movement, such as in

Figure 13. Photomicrograph of a section through a copper alloy axe from Iran showing twinned grains fromFigure pre-Hispanic 14. Twinned Colombia. grains of gold-copper district of Nariiio. alloy sheet afx300).ter etching Note that with the ferric porosity chloride in the solution metal has(approx. not been Corrosion. which appears grey here. outlines twin eliminated by working. grainsthe section of the are crystals due to without selective etching. corrosion Striations as a resultthrough of tosegrega unequaltion distribution in the worked of copperalloy. The and segregat gold in ionthe is due thathammered annealing sheet. was Annealing the final stage twins in are manuf straightacture. showing x 120. The Microstructure of Ancient Metals 10

Figure 15, right. Twin planes in Indian zinccoin. Figure 16, far right. Twin planes in zinc.

basal planes bend plane/ bend plane the usual polycrystalline solid. Under these con­ tWin In ZinC C rystal ditions, stress can be relieved by the movement of '" the basal plane axis. The axes of thebend are con­ � tained in a bend plane whichbisects the included 7Vf\L angle of bend and which has no definedcrystallo­ 7./:/ graphic indices. The bending mechanism i/ involves slip, since individual basal planes must II move relative to each other fo r bending to occur. , Although in some cases the volume of metal / ______involved in a series of bends is sufficientto pro­ basal plane duce macroscopic kinks, bending often occurs Figure 17. Twin planes in zinc. around a very large number of closely spaced par­ allel axes giving the effect of a curve. When such deformed structures are annealed close to the melting point, a coalescence of the fine bend seg­ ments into coarser units occurs (Figs. 16, 17). 3 TWO-PHASED MATERIALS

Apart from compositional variations produced by interpreted with caution and, preferably, after segregation or by inclusions, the varieties of experience with ancient alloys gained by micro­ microstructure produced by the presence of two scopical examination. or more phases in metal have not been discussed. When two metals are mixed together there are Some reasons why more than one phase can be three main possibilities. The first is a solid alloy present are discussed in the fo llowing sections: showing complete solid solubility of two metals. An example is the range of alloys fo rmed between ' 1. Eutectic structures silver and gold. Gold is soluble in silver and silver 2. Eutectoid structures is soluble in gold. The phase diagram that results 3. Peritectic structures from this kind of alloy shows just one solid phase 4. Widmanstatten transformations present at all temperatures up to the solidus. The 5. Discontinuous precipitation solidus line separates the region of solid metal 6. Intermetallic compound fo rmation from the pasty stage of solidification at tempera­ 7. Immiscible structures tures above this line. As the temperature rises, the alloys pass through a pasty, semisolid region in When two (or more) metals are mixed which some liquid is present in equilibrium with together to fo rm an alloy there are a number of some solid. This continues to the liquidus line different possibilities regarding their mutual solu­ and above this line the alloy is present as a liquid bility. Usually a phase diagram (or equilibrium melt. The liquidus line separates the pasty stage diagram as it is also known) is employed. It is from the liquid melt above (Fig. 18). essentially a map that can be used to predict what In cooling from the molten state, the alloy phases should be present in the alloy at equilib­ undergoes the fo llowing transitions: rium. It is important to stress that diagrams usu­ ally only refer to very slowly cooled melts which L --7L + a -----'3> a rarely occur in archaeological materials. This does = = not mean that it is useless to consult phase dia­ where L liquid and a solid. grams: It means that the information must be Figure 18. Phase diagram for the gold-silver system. Note that both Toe ofmelting pure goldpoint metalseach other. are completely soluble in " 1050 1000

950 solid solution fJ.

0%100% gold silver 50% goldsilver 0%100% silver gold Two-phased Materials 12

If no coring or other fo rms of segregation are varies, of course, depending on the nature of the present, then the microstructure will be a collec­ alloying constituents-the liquid melt passes to tion of equi-axed hexagonal grains of uniform solid, which is two-phased and consists of fine composition-there will be only one phase plates of alpha phase and beta phase interspersed present. in each other (see Figs. 21-23, 25a, b). The second possibility is that a solid alloy can There is a large area where alpha phase coex­ show only partial solubility of the metals in each ists with liquid and a similar region where beta other. One example is silver and copper. There phase coexists with liquid (Fig. 19). If an alloy of are three principal types of phase diagrams that composition B is cooled down from the melt, can arise from this situation. The most common then the fo llowing transitions will occur: is the eutectic type, second is the eutectoid, and L (t ) L ex l ----'7(tJ + third is the peritectoid. The third possibility is that the two metals are completely immiscible in L (ex+ � (tJ ---7> ) each other. = where t temperature. Eutectic Structures Silver-copper alloys are examples of the eutectic exThe(ex + final solid structure therefore consists of type and have the fo llowing characteristics: the + �). The original alpha will be present as solubility of copper in silver and of silver in cop­ either grains of alpha solid solution or as dendrites per falls as the temperature falls (this is a general of the primary alpha, which will probably be cored. characteristic fo r most alloys), and there is one The infilling around the alpha grains will then con­ temperature at which the liquid melt can pass sist of the alpha + beta eutectic as a fine inter­ directly to solid: The eutectic point. At this par­ spersed mixture and under the microscope in ticular composition and temperature-which etched section will look something like Figure 20a. ofFigure silver-copper 19. Eutectic alloys. phase diagram B+L eutectic point ! alloy C alloy B composition Two-phased Materials 13

Figure 20a. b. Eutectic-type infill of alpha + beta eutectic alpha + beta eutectic microstructures.

dendrites of alpha grains of alpha a usually cored b

Figure 21. middle left. The feathery nature of dendritic Figure 22. middle right. The same alloy as Figure 21. exin inpot a assium 60% Ag dich 40%romate. Cu alloy x35. that While has been the cast.dend Etchedrites The60% exAg+ 40% � eutectic Cu etched phase in potais justssium begi dichrnningomate. to be x I 00. can be seen clearly at this magnification. theex eutectic resolved into a fine series of lines in the section. infill of ex+ � cannot be seen. Figure 23. bottom left. 60% Ag 40% Cu alloy illustrating showingFigure 24. fragments bottom right.of cementite Wootz needlessteel ingot with from an infill India of thephase nature etches of the dark eutectic and theinfill si inlver-rich which thephase copper-rich etches eutectoid pearlite (ex+ Fe3C, x300). etched in nital. exlight. Electron-probe analysis of one of the� dendritic Wootz steel ingots from India were high-carbon steels arms that appear here as dark globules. gives an analysis the(ofte manufn overacture 0.8% ofcarbon). sword cast blades in crucibles. and other and qual usedity for compositionof 92% Cu 8% to Ag. the corresponding first solid formed very from well thein melt in products. The eutectoid may be similar to the eutectic. andthe phasepotassi diumagram dichr foromate. Cu-Ag x 16etched0. in alcoholic FeCI3 Two-phased Materials 14

The cooling rate determines whether the orig­ At the eutectic composition, the liquid melt inal alpha phase is present as dendrites or as hex­ passes directly to solid and ideally will consist of a agonal grains. Usually in archaeological materials fine, intermixed matrix of alpha and beta phase the primary alpha will be dendritic and cored; (Fig. 26). later working and annealing may remove den­ A feature of the microstructure of eutectic­ dritic segregation and grains ofalpha may become type alloys is that there may be a depletion of part more apparent. It is beyond the scope of this text of the eutectic phase near the grains or dendrites. to provide a quantitative interpretation of the For example, suppose the original dendrites are phase diagram, but what can be said is that as the alpha phase, with an infillof alpha + beta eutectic. eutectic point is approached, there will be corre­ Some eutectic alpha constituent can migrate and spondingly less initial phases of alpha or beta. join the dendritic alpha, which will leave a fri nge As the area of alpha is approached there will surrounding the dendrites appearing to contain a be less eutectic present and more alpha. As the more homogeneous zone before eutectic infilling alloys approach the beta side of the diagram the is reached (Fig. 27). same variation is fo und: the alloys are progres­ One of the interesting changes that can occur sively richer in beta and have less eutectic. In two­ when a two-phased alloy is worked is that either phase alloys where dendritic segregation has one or both phases can become elongated or occurred, the proportion of the twO phases will strung out, much like slag stringers in wrought not be quite what it should be at full equilibrium. iron, along the direction of the working of the The alloy at composition A (Fig. 19) will have alloy. Slag stringers are the broken-up remnants a slightly different composition in terms of the of slag inclusions in wrought iron that become distribution of the two phases. As it cools down elongated upon hammering the iron to shape. In the fo llowing sequence should occur: theory, one would expect the process of working L L and annealing to remove any original dendritic (t1 ) ----7(t2) + (J.. L segregation and to produce worked and recrystal­ (t-) -----7>� lized grains with a eutectic infill,depending on The resulting structure will then be alpha the composition of the original raw ingot. How­ grains with a thin film of beta surrounding them, ever, it is often very difficultto remove the initial or alpha dendrites with a fringe of beta (Fig. 25). dendritic structure and instead the microstruc- Figure 25a.b. a and � micro­ alpha dendrites usually cored structures.microstructure. A typical eutectic alloy

a some alpha + beta eutectic because of nonequilibrium cooling alpha grains beta films between grains Figure 26. Eutectic a and �.

eutectic mix onlywill consist of alpha + beta Two-phased Materials 15

Figure 27, right. Eutectic and dendrites in a typical eutectic alloy. eutectic Figure 28, far right. Fibrous workedstructure two-phase in a typical al loy.heavily eutectic phase infill eutecticalpha near deple dendtedrites in elongated dendritic remnants bronzeFigure 29. alloy. a and 8 eutectoid in a tin cored alpha dendrites (phaselighta+ blue8) has eutectoidthe composition where deltaCU31 Sna

ture tends to consist of elongated ribbons of one not the same. In bronzes, the eutectoid constitu­ phase with the eutectic in-between. The length­ ent is made up of the two phases, alpha (the to-breadth ratio of these elongated stringers then copper-rich solid solution of tin in copper) and gives some idea of the extent to which the alloy delta (an intermetallic compound of fixed com­ has been hammered out-very thin stringers sug­ position, CU3JSnS). This eutectoid phase begins gesting more severe deformation during manu­ to appear in the microstructure between about facture. Sometimes alloys also have a fibrous 5% to 15% tin (and above), depending on the quality fo r the same reason (Fig. 28). cooling conditions of the alloy. It is a light blue, Common examples ofsimple eutectic systems hard and brittle material that often has a jagged in ancient metals are those of debased silver arti­ appearance. The structure is often shaped by facts, which are usually silver-copper alloys and grain boundaty edges and the blue delta phase soft solders, which are lead-tin alloys. often contains small islands of alpha phase dis­ persed through it (Fig. 29). If there is a lot of this Eutectoid Structures eutectoid phase present, the bronze is difficult to The eutectoid phase is similar to the eurectic work. Proper annealing of bronzes with up to structure, the principal difference being that the about 14% tin will result in a homogeneous solid eutectoid reaction occurs when an already exist­ solution of alpha grains that can then be worked ing solid solution transforms into two distinct to shape much more readily because the hard and phases. The types of phase diagrams that give rise brittle eutectoid has been eliminated. to eutectoid-type transformations are necessarily Eutectoid in the bronze system originates more complex because there are series of changes from a complex series of changes that are not in the solid as it cools to room temperature. There delineated in detail here, but are summarized as are two important eutectoid transformations in fo llows: archaeological metals: those in tin bronzes and in carbon steels (for one example, see Fig. 24). The 1. The alloy passes through the alpha + liquid fo rm the eutectoid takes in bronzes and steels is region as it cools. Two-phased Materials 16

2. It reaches a transition at about 798 °C and two constituents as a fine collection of small a peritectic transformation occurs. plates is called pearlite. The name fe rrite is given 3. A beta intermediate solid solution results. to the pure iron alpha phase grains, while cement­ 4. On cooling to about 586 dc, the beta phase ite (Fe3C) is another very hard and brittle constit­ transforms to gamma. uent, a compound of fixed proportions between 5. At 520 °C the gamma solid solution trans­ iron and carbon. fo rms to the final solid mixture of alpha + Consider the cooling of an alloy from above delta eutectoid. 900 °C, in the austenitic region of the phase dia­ gram with an average content of carbon, repre­ Because the eutectoid in the copper-tin sys­ sented by the line fo r alloy A in Figure 30. As tem is rather difficult to fo llow, most textbooks cooling proceeds, austenite (gamma-phase) grains on the subject introduce the idea of eutectoid will separate out, and as the temperature falls, fe r­ transformation by looking at the phases fo rmed rite begins to separate from the austenite at the when carbon is added to iron to produce steels, grain boundaries. Also as the temperature falls, and, .as the carbon content increases, cast irons. the gamma phase becomes richer in carbon, and Most ancient steels were made from iron con­ the fe rrite loses carbon until it reaches a low of taining up to about 1 % carbon, although not 0.03% carbon, while the austenite reaches the only is the carbon content of many ancient arti­ eutectoid composition at 0.8% carbon. Asthe facts very variable in different parts of the same temperature falls below 727 °C, the austenite object, but many of them only contain about decomposes by a eutectoid reaction into ferrite + 0.1-0.5% carbon. These low-carbon steels were, cementite. The changes can be represented as however, very important products and could be symbols: used to produce excellent edged tools. The eutectoid is fo rmed when the austenite Alloy A: Y ------7Y + a (a t about 820 ° C) solid solution (gamma phase) decomposes at C) 72rC) Y + --7(a Fe3 + a (beLow about 727 °C to fo rm the two new solid phases, a + ferrite and cementite. The combination of these Figure 30. Significant region of the iron-carbon phase diagram. J alloyY A 1 austentitel'

y austentite + cementite

_1------pearlite + cementite ------i.._ o 2 % carbon Two-phased Materials 17

Figure 31a-d. Breakdown ofy grains in the iron-carbonsystem. y grains (above 850°) �"'''''-::��--'�'''''''''''''''a(about grains 727 °C)

a at y grains a Fe)C+ eutectoid y grains (about 800°) a (below 727°C)

or as a series of drawings at different temperatures The final structure will usually consist of cementite on the way to room temperature (Fig. 31a-d). films or a continuous cementite network between The cooling of alloy B, shown on the portion the pearlite regions (Fig. 32). of the iron-carbon phase diagram (Fig. 30), fo l­ lows a line leading through the eutectoid point with a composition of 0.8% carbon. For this par­ ticular alloy the microstructure, if cooled slowly, would consist of an intimate mixture of pearlite (alpha + Fe3C). In the case of the iron-carbon eutectoid, the initial appearance is very similar to pearlite eutectoid that of the eutectic mixture drawn previously (Fig. 26). If the rate of cooling of alloys contain­ ing pearlite as a constituent increases, then the Figure 32. Cementite and pearlite. spacing between eutectoid constituents becomes progressively finer. If the cooling rate is very fast, then the true nature of the phases that might fo rm Peritectic Structures on a phase diagram cannot be shown because Peritectic structures arise from a type of transfor­ nonequilibrium cooling conditions would be mation that may seem rather peculiar at first involved. What happens in steels in fast cooling is sight. It is unusual in the sense that a liquid reacts very important, and new phases, such as marten­ with an existing solid phase to fo rm a new solid site, can fo rm, which has an extremely hard and phase. An example can be taken from part of the brittle needlelike structure (see Chapters 6 and 7). copper-tin phase diagram (Fig. 33) to illustrate The cooling of alloy C (Fig. 30), whose posi­ the typical shape of the phase system in peritectic tion is shown on the phase diagram, produces a alloys. different but analogous series of transformations AlloyA cools down from the melt with about fr om the austenitic region: 18% tin content. Ignoring fo r the moment the complications produced by coring, as alloy A Alloy C:Y ----7 Y + Fe3 C cools down, initially an alpha-phase solid solution Fe3 C Fe3Cj C Y + ----;;.(a + + Fe3 of tin in copper separates out, while the liquid Two-phased Materials 18

diagram.Figure 33. Copper-tin phase

20% Sn

that is left gets progressively richer in tin. A reac­ completely converted to a new grain structure of tion now occurs at about 800 °C between this liq­ beta grains: uid and the alpha phase, which produces a new Alloy B: liquid � a + liquid (tin-rich) phase, beta. Since alloy A occurs in the alpha + + a liquid (tin-rich) ------7>� beta region of the diagram, all of the tin-rich liq­ uid will be used up before the alpha phase is com­ final structure: gra� ins pletely dissolved, and the alloy will then consist of One of the difficulties of the peritectic reac­ alpha + beta crystals: tion is that it is rarely possible to get complete Alloy A: liquid -----37a + liquid (tin-rich) conversion of alpha grains into beta because alpha liquid (tin-rich) -----7 grains become covered with a coating of beta as + a � finalstructure: a + � grains they transform, and the film ofbeta then prevents the diffusion of tin-rich liquid to the alpha grains. Often the peritectic reaction gives rise to precipi­ The result is that there is very often a core ofalpha tation of the new beta phase both within the grains left, even if the phase diagram suggests that alpha crystals and also at the grain boundaries so all the material should have been converted to the that the alpha has rather rounded contours second phase. A complicated example is given by (Fig. 34). the finalstructure resulting from an alloy contain­ An alloy of composition B on the phase dia­ ing 70% tin and 30% copper. Some of these gram (Fig. 33) starts to solidifYby production of alloys were used, usually with about 40% tin, as alpha phase grains, but these grains then react the alloy sp eculum, a white alloy used since with the remaining liquid at about 800 °C and are Roman times fo r the production of mirrors. In Figure 34, right. a and � peritectic. phaseremaining grains alpha Figure 35, far right. E phase grains. eutectic (11+Sn)

beta produced by peritectic 11(eta) phase coating (CuJSn) transformation preventing peritectic reaction Two-phased Materials 19

theory, this alloy should simply be a mixture of A solid alloy melts as a result of cooling during the eutectic (eta and tin); however, these alloys these phase changes-a rather unique occurrence. often show a nonequilibrium structure with epsi­ An alloy system of interest in which a series of (£) peritectic transformations can occur is the lon phase grains (Cu35n) coated with eta (11) copper-zinc system (brass; see Fig. 207). Most phase grains (Cu35n3)' in a eutectic mixture of (11 + 5n; Fig. 35). copper-zinc alloys of antiquity were made by a Many of the mirrors used in Roman times cementation process that had, as an upper limit, a were either made using high-tin leaded bronze, zinc content of about 28%. Zinc ore was mixed wi th tin contents of 20-24% and lead variable with copper ore and the two were smelted (typically 5-12%), or they were made with a together direcrly so that the zinc was absorbed more common low-tin bronze alloy, which was into the copper during reduction, thus avoiding then tinned on the surface to produce the desired loss of zinc, which boils at 907 °C. Most ancient color. zinc alloys, therefore, possessed an alpha phase or There are other unusual fe atures in the cored dendritic structure. However, metallic zinc copper-tin system (see Figs. 198, 199), such as was also produced; fo r example, an alloy contain­ that shown by the cooling of an alloy with about ing 87% zinc was reputedly fo und in prehistoric 41 % tin. During cooling of this alloy, gamma (y) ruins in Transylvania, while in ancient India and phase crystals start to separate out at a tempera­ China, metallic zinc was produced. ture of about 715 °C. At slighrly below 700 °C, The brasses are generally divided into three freezing is complete and all the liquid is trans­ categories depending on the phase type: alpha fo rmed to gamma solid phase. At about 650 0C, brasses with up to about 35% zinc; alpha + beta the eta (11) phase starts to fo rm from the gamma brasses with between 35% and 46.6% zinc; and (y) until a temperature of 640 °C is reached. At beta brasses with between 46.6% and 50.6% zinc. about 640 °C the residual gamma (y) decomposes As zinc content increases the britrle y phase to fo rm simultaneously the liquid + eta (11)phase. begins to appear and thus alloys with more than diagram.Figure 36. Copper-zinc phase T DC liquid 1000 alloy A alloy B 900 800 700

40% Zn 100%0% ZnCu 20% Zn Two-phased Materials 20

50% zinc are generally avoided. Beta (�) phase plunging it into water or oil from a red heat. In brasses are very much harder than the alpha and contrast, the Widmanstatten precipitation is the can withstand very little cold-working. The beta result of one solid phase at a high temperature phase begins to soften at about 470 °C (as the lat­ decomposing into two solid phases at a lower tice changes from an ordered to a disordered temperature. This precipitation usually occurs at state), and at about 800 °C it becomes much the grain boundaries of the initial crystals and as easier to work. The alpha brasses, which include plates or needles within the grains themselves, most of the ancient specimens, are much better which have a particular orientation depending on when they are cold-worked and annealed rather the crystallographic structure of the original crys­ than hot-worked because, if hot-worked, impuri­ tals. ties tend to segregate at the grain boundaries and In the case of alloy B (Fig. 36), a mixture of make the brass very weak. about 58% copper and 42% zinc, we can fo llow These types of structures are essentially simi­ the precipitation of the alpha solid solution from lar to the possibilities given fo r the section of the the beta high temperature region. copper-tin diagram examined brieflyearlier (Fig. In Figure 37a the beta grains are shown as 33). An alloy of composition A will, having they would appear at about 800 °C, or if the alloy passed below the liquidus line, begin to precipi­ was suddenly quenched in water, which would tate out alpha grains, which are then partially prevent it from decomposing into the alpha + attacked and converted to beta during solidifica­ beta region. The appearance of the grains is just a tion so that the resulting structure consists of homogeneous solid solution of beta grains. Figure alpha + beta grains (Fig. 36). 37b shows the nature of the Widmanstatten pre­ cipitation upon cooling to room temperature. If Widmanstatten Transformations the structure is annealed or heated to about 600 The copper-zinc alloys may apply to a brief dis­ °C, then it can become quite coarse and the alpha cussion of the Widmanstatten transformation. phase may grow into large crystals with the back­ The Widmanstatten structure results from the ground becoming a fine mixture of alpha + beta. precipitation of a new solid phase within the Widmanstatten structures also occur in grains of the existing solid phase. It is thus quite ancient steels as a result of the working process or different from the martensitic transformation, deliberate heat treatments used during manufac­ which is essentially a single-phased structure usu­ ture. Very often Widmanstatten precipitation is ally occurring as a nonequilibrium component of only partially carried through the grains so that a quenched alloys. Martensite is a collection offine jagged effect is produced. It is useful to return to intersecting needles that can fo rm in alloys cooled the iron-carbon diagram (Fig. 30) at this stage in very quickly. Usually the alloy is quenched by order to define a few common terms. Steels Figure 37. � grains in copper-zinc. a original � grains at 800°C original � grain boundaries matrix probably mix of a + � Two-phased Materials 21

containing less carbon than the amount needed to Very often the silver was debased to some extent make the eutectoid structure complete are called with copper, partly to make the alloy harder and hypoeutectoid steels, whereas those containing car­ also to reduce the amount of silver. Debased silver bon in excess of the eutectoid composition (and objects then often consist of silver-rich grains in up to l.7% carbon) are usually called hypereutec­ which the copper has not yet begun to separate toid steels. The eurectoid composition itself occurs out as it should according to the phase diagram. at 0.8% carbon. In hypoeutectoid steels there will The solution of copper in the silver grains is generally be more fe rrite than is required, and this therefore in a metastable state and can be precip­ is called the proeutectoid (or free) ferrite.In hyper­ itated slowly with time at the grain boundaries. eutectoid steels there will generally be too much Precipitation of this nature is called discontinu­ cementite to form a complete eutectoid, and this ous when it occurs at the grain boundaries. The is called proeutectoid cementite. essential part ofthe phase diagram is shown below Proeutectoid fe rrite occurs in several different (Fig. 38). shapes. In the lower carbon steels of antiquity it is A typical, slightly debased silver alloy is shown characteristically found as extensive areas among by alloy A on the silver-copper phase diagram. scattered islands of pearlite. This, according to Note that it cuts across the (J. + � phase region Samuels (1980), should be called massedferrite. In where it cools down to room temperature. It can steels nearing eutectoid composition, the ferrite is exist as a homogeneous solid solution alpha phase usually found as thick films located at what were between temperatures tl and t2' When the alloy originally the austenitic grains. This is called gets to t2' the decomposition of part of the solid grain-boundary ferrite. Ferrite may also be fo und solution into beta may not occur and instead a in the fo rm of broad needles, which can be sec­ metastable solid solution will result. The copper­ tions of plates of fe rrite occurring as a Widman­ rich phase may precipitate out very slowly at room statten pattern within the pearlite. A descriptive temperature, and Schweizer and Meyers (1978) scheme for some of the various fo rms of fe rrite has suggest that the discontinuous precipitation of been developed by Dube (in Samuels 1980; see copper can be used to establish the authenticity of Appendix A fo r names of ferrite shapes in low car­ ancient silver. They extrapolate from experimental 3 bon steels and a glossary of terms). data to give a growth rate of about 10- microns per year for the rate of precipitation. This kind of Discontinuous Precipitation growth can lead to age-embrittlement of ancient Another type of phase separation of importance is metals. Thompson and Chatterjee (1954) also discontinuous precipitation. A good example is fo und that lead fo rmed at the grain boundaries of afforded by copper-silver alloys used in antiquity. impure silver and led to embrittlement. Figure 38. Discontinuous precipitation in Ag-Cu. TOC

100% Ag 100% Cu Tw o-phased Ma terials 22

Intermetallic Compound Formation peritectic reaction) at about 240°C (Rhines, When some metals are mixed together they can Bond, and Rummel 1955). CU3Au can fo rm fo rm phases that are essentially like ordinary berween 50% and 50.8% gold. CuAu can fo rm chemical substances in that they are effectively berween 70% to 85% gold. compounds of fixed composition. An example is Ordered phases such as these have to be dis­ the gold-copper alloys. These alloys were used in tinguished from those phases normally called antiquity, especially in the more base, copper-rich intermetallic compounds. Intermetallic com­ compositions in South America. This alloy was pounds are usually represented on the phase dia­ known as tumbaga and was used widely both fo r gram by a straight line that passes down vertically castings and fo r hammered sheerwork, often as the temperature falls. There may be rwo such being finished by a depletion gilding process lines close together that mark out a rectangular (Lechtman 1973, Scott 1983; see Fig. 39). block on the phase diagram, showing the areas The diagram shows that copper and gold are over which the intermetallic compound may completely soluble in each other with a eutectic­ fo rm. The ordered phases are rather different type low melting point, which occurs at a compo­ because they may be fo rmed over wider composi­ sition of 80. 1 % gold at 911°C. The rounded tional limits, they do not show vertical phase shapes at the bottom of the diagram show the boundaries in the fo rm of straight lines, and they regions where the ordered phases can fo rm. There may pass easily berween ordered regions and dis­ are essentially three different ordered composi­ ordered regions. tions: CU3Au, CuAu, and CuAu3' CuAu3 can However, these ordered phases are usually fo rm berween about 85% to 92% gold. It is a harder than the disordered alloy of the same com­ superlattice formed by a peritectoid (a solid state position, and they may make the process of Figure 39. Cu-Au phase diagram. TOC a

100% Cu CuAU3 100% Au Two-phased Materials 23

working and annealing to shape more difficult. usually as globules of one phase in grains of the For example, the quenched alloys in the gold­ higher melting point metal. An example is leaded copper system between about 85% gold and 50% copper, shown in Figure 40. gold are softer than the alloys that are allowed to The diagram shows that the microstructure cool slowly in air. This is the opposite of the situ­ consists of two distinct phases and that the copper ation that exists in alloys such as iron and carbon, grains that form will contain globules of lead. where the material is dramatically hardened by Practically all the copper will solidifybef ore the quenching because of the formation of new lead-copper eutectic fo rms. This lead-copper phase, martensite. The reason why gold-copper eutectic is, fo r all practical purposes, pure lead, as alloys are softer is that the quenching process sup­ it consists of 99.9% lead and 0.1 % copper. This presses the fo rmation of the ordered phases which means that the lead is segregated while the solidi­ need some time to fo rm, and it is these ordered fication process is taking place. Ordinarily, the phases that give rise to higher hardness values and separation of lead globules would be expected to to the difficulty sometimes experienced in the result in massive segregation and an unusable working of these gold-copper alloys. South Amer­ material would result. There is a monotectic reac­ ican Indians, in lowland Colombia fo r example, tion at 955 oc, which occurs when the liquid used water quenching after annealing in order to from which the copper is separating out reaches a make their alloys easier to work to shape and to composition of36% lead. At this point, a new liq­ avoid embrittlement. uid forms that contains about 87% lead. This There are many examples of intermetallic new liquid is heavier than the first liquid, and so compounds, such as cementite (Fe3C)' which it tends to sink under gravity. However, in prac­ contains 6.67% carbon and the delta phase in tice the gross segregation is limited by the fo rma­ bronzes, which is an intermetallic compound of tion of a dendritic structure upon casting in the the formula CU31SnS' copper-rich alloys and, with a high cooling rate, the lead is finely dispersed among the dendrites. Immiscible Structures With very high lead content alloys, the two liq­ In some cases metals may be completely insoluble uids that separate out form an emulsion when in each other. Examples of this type of micro­ they are cooled from about 1000 °C (Lord 1949). structure are shown by the alloys of copper and This emulsion results in a division into vety fine lead, zinc and lead, and iron and copper. As the droplets so that gross separation cannot occur. temperature falls from the melt of these mutually With leaded copper, brass, or bronze alloys insoluble metals, one ofthem will be precipitated, the lead usually occurs as small, finely dispersed Figure 40. Cu-Pb phase diagram. melting.l'� -_------point Cu I Cu + liquid ______-- two liquids ---.... 1084.5 °cI-- --.....;.-----.....;.::::::::=...::::;,.....___ ------� � , 36 95S"C 87 \ Cu + liquid melting 1-______---1 99.9% 326°C Pbpoint (327.5°C) Cu + Pb (eutectic of 99.9% Pb 0.1 % Cu) Cu Pb Two-phased Materials 24

spherical globules scattered at the grain bound­ diagram, the composition of the still liquid cop­ aries and within the grains themselves. Lead has per follows the liquidus until at 1095 DC the cop­ the effect of making the alloys of copper easier to per will contain something like 3% iron in cast; it can, fo r example, improve the fluidity of solution. At 1084.5 DC, a peritectic reaction will the alloy in the melt. Lead also makes copper occur between the liquid and the precipitated alloys easier to work since the lead acts as an area gamma phase to give a solid solution of96% cop­ ofweakness between the grains. This is of no per and 4% iron. This means that given a vety practical use if the alloy is used fo r the manufac­ slow cooling rate the alloy should consist of a ture of daggers or sword blades since they will be solid solution (eta [T]l phase) of 96% copper and weakened by the inclusion oflead, but it is advan­ 4% iron with residual gamma (y) iron particles. tageous for the production of cast objects. As the temperature falls, the copper(ex) is gradually It is not strictly true that iron is insoluble in precipitated out of the alpha iron and at the copper; the phase diagram is more complex than same time the eta (T]) phase loses iron. In ancient that, although the end result of admixtures of specimens, so far no evidence has come to light to copper with small amounts of iron is the presence suggest that the peritectic reaction had occurred. of small dendrites or globules of iron mixed with Because of the presence of alpha-phase iron (fer­ the copper grains. The phase diagram is given in rite), the copper alloys containing iron are usually Appendix G in this book. The cooling of a 6% ferromagnetic and can sometimes be picked up iron in copper alloy is examined by Cooke and with a magnet. Aschenbrenner (1975). As the 94% copper and Care must be taken when grinding and pol­ 6% iron alloy cools, it reaches the liquidus at ishing leaded alloys to ensure that lead globules about 1215 DC and solid gamma iron begins to do not drop out (without notice being taken of separate out. This gamma iron will contain about their existence) in the process. If they do fall out, 8.3% copper in solid solution. As the temperature they leave small spherical holes and it may then falls, more of the gamma iron-rich phase separates become vety difficult to distinguish between until at 1095 DC the precipitating iron contains porosity due to casting defects and lead inclusions about 8.5% copper. At the copper-rich side of the as an alloying constituent, since both appear as holes in the polished section. 4 THE MICROSTRUCTURE OF TIN BRONZES

Some of the features to be fo und in alloys of cop­ of interdendritic delta phase will be much per and tin, commonly referred to as bronze, or reduced or disappear entirely. However, at tin more correctly as tin bronze, have been discussed contents of about 10% it is very unusual in cast­ in previous chapters. The phase diagram fo r the ings from antiquity to get absorption of all the copper-tin system is rather complex and cannot delta phase and the dendrites will usually be sur­ be discussed fully here, and the one given in this rounded by a matrix of the alpha + delta eutec­ book ignores(ex) the low-temperature phase fieldof toid. the alpha + epsilon (£) phase region (Fig. 33 As the tin content increases the proportion of and Figs. 198, 199 in Appendix G). This is interdendritic eutectoid also increases. If a homo­ because the £ phase never appears in bronzes of up geneous copper-tin alloy is worked by hammering to abour 28% tin that have been manufactured by and annealing, then the typical fe atures seen in conventional means. Thousands of hours of face-centered cubic metals will be developed; annealing are necessary in order to make the £ namely, annealing twins, strain lines, progres­ phase appear and, of course, such conditions sively finer grains as a result of working, and flat­ never occur, even in modern bronzes. Despite tened grains if left in the worked condition, as this, common fo rms of equilibrium diagrams discussed earlier. The same fe atures will develop if contain no clue that equilibrium is practically the alloy is two-phased, although the eutectoid is never attained. Even more surprisingly, some rather brittle and may be broken up to some modern metallurgical textbooks appear to be extent. The usual microstructure shows the pres­ unknowledgeable on the subject and maintain ence of small islands of alpha + delta eutectoid that an alloy with 2% tin will decompose on cool­ between the recrystallized grains of the alpha solid ing from the usual alpha + delta region to give solution. If coring in the original cast ingot was alpha and eta without difficulty! pronounced, then this may be carried over in the Tin bronzes may conveniently be divided into worked alloy as a faint or "ghost" dendritic pat­ two regions: low-tin bronzes and high-tin tern superimposed upon the recrystallized grains. bronzes. Low-tin bronzes are those in which the When a bronze section is etched with ferric chlo­ tin content is less than 17%. This is the maxi­ ride, this difference in alloy composition due to mum theoretical limit of the solubility of tin in coring may only be apparent as vague differences the copper-rich solid solution. In practice, the in shading of the alloy, and a dendritic outline of usual limit of solid solution is nearer to 14%, the shading may be very difficult to see. Some although it is rare to find a bronze with this tin experience in the examination ofbronzes must be content in a homogeneous single phase. developed so that a worker can differentiate When a tin bronze is cast, the alloy is exten­ between uneven etching and uneven coloring of sively segregated, usually with cored dendritic the specimen surface due to coring. growth, and an infill of the alpha + delta eutectoid Apart from complications introduced by surrounds the dendritic arms. The center of the other alloying elements such as zinc, the struc­ dendrite arms are copper-rich, since copper has tural features seen in most low-tin bronzes are the the higher melting point, and the successive fo llowing: growth of the arms results in the deposition of J. Homogeneous bronzes, in which all the tin more tin. At low-tin contents, fo r example, has dissolved with the copper and which do between 2 and 5%, it may be possible fo r all the not display coring or residual cast fe atures. tin to be absorbed into the dendritic growth. This 2. Cored bronzes, in which there is an unequal varies considerably depending on the cooling rate distribution of copper and tin, but no eutec­ of the bronze and the kind of casting involved. If toid phase present. the cooling rate is very slow, there is a greater 3. Bronzes in which both the alpha phase and chance of reaching equilibrium, and the amount The Microstructure of Tin Bronzes 26

the eutecroid phase are present. phase and ro produce a martensitic structure). 4. Bronzes in which the alpha phase is exten­ Hammer marks and oxide scale could then be sively cored and where the eutecroid phase is removed by grinding with abrasives of various present. grades, often on a simple lathe, and then the object was polished. Surface decoration, if Most ancient alloys have less than 17% tin. At present, was cut into the surface with drills or an this level of tin content, bronzes can be cold­ abrasive wheel before finalpolishing . worked and annealed; however, if the tin content Although certain vessels made from this alloy is between 17% and 19% it has been fo und that possessed interesting musical properties, the prin­ the alloy is unworkable: it can neither be hot­ ciple reason fo r its use in regions where tin was worked nor cold-worked. A film of delta forms plentiful, was its color. The color of typical beta and this brittle phase then coats the grain bound­ bronze resembles gold. Beta bronzes were first aries with the result that the alloy breaks up inro fo und in India and Thailand from the early cen­ pieces. However, above 19% tin the bronze can turies B.C. and they spread slowly ro the Near be hot-worked. Bells and mirrors in antiquity East. The Islamic alloy, white bronze, safldruy, is were often made of ternary tin bronzes consisting an example of a high-tin alloy. It was also fo und ofabout 20-25% tin, 2-10% lead, the remainder in Java and Korea, but when brass became more being copper. Alloys of this type were almost widely known, high-tin bronze use became much invariably cast. Binary tin bronzes containing more limited. more than 17% tin often have about 23% tin, The alloy known as specuLum, which may con­ which corresponds closely ro the equilibrium tain up ro about 35% tin, is said by some ro have value of the beta phase of the bronze system, been used by the Romans fo r the manufacture of which has been mentioned in connection with mirrors. However, Roman mirrors were often peritectic transformations. Above 586°C, a made by tinning; the alloy itself was often a low­ bronze in the beta region can be readily worked, tin bronze. At high levels of tin, such as those whereas if allowed ro cool slowly ro room temper­ encountered in tinned surfaces, the fo llowing ature, the bronze would decompose into alpha intermetallic phases of the copper-tin system and delta and be impossible ro work. One advan­ must be considered carefully: (1) the delta (b) tage of beta bronzes is that the beta phase can be phase which has already been discussed, Cu 31 Sn 8' retained by quenching. A complete account of containing about 32.6% tin; (2) the epsilon (E) this process is quite complex, bur one ofthe most phase, CU3Sn, containing about 38.2% tin; (3) important points is that the beta phase is retained the eta (11) phase, containing 61.0% tin, CU6Sn5. by quenching as a structure of martensitic nee­ Here, in tinned surfaces, the epsilon phase does dles. This quenched beta bronze is very hard, but appear and is important in understanding the a lot less brittle than the same bronze slowly microstructure. When tin is applied ro bronze, cooled ro the alpha eutecroid room temperature layers of both the eta and the epsilon phase can fo rm. Apart from a few cast figurines, the major­ develop by interdiffusion between the bronze and ity of artifacts of beta bronze composition were the molten tin, which then can develop layered made by the fo llowing series of operations. The structures in the fo llowing sequence: surface tin, alloy was made up as accurately as the technology eta phase, epsilon phase, substrate bronze. of the time allowed, a blank was then cast in the Under the optical microscope, tin is light and approximate fo rm of the desired object, and the silvery in appearance, the eta compound is object was shaped by hot-working at a tempera­ slightly more grey-blue in color, the epsilon phase ture of about 650 °C. At the end of the working is the darkest grey-blue, and the delta is light blue. process, the alloy was uniformly reheated ro about The range of fe atures that may fo rm on tinned the same temperature and was then rapidly surfaces is complicated: CU6Sn5 is common and is quenched (ro preserve the high-temperature The Microstructure of Tin Bronzes 27

often misdescribed as tin (see Meeks J 986). gram shows that the presence of rwo liquids fo r Many tin bronzes are leaded. With low-tin many compositions above 734 °C may prevent bronzes, typically castings, the lead does not alloy quenching from temperatures high enough to with the copper or tin and occurs as small globules retain beta (see Appendix G, Fig. 2 J 2). throughout the srructure. Some gravity segregation Figure 4 J shows a section through a cast toggle may take the lead down to the base or bottom of pin from Iran, +etched 0 in FeCl3 at x450. The net­ the casting, but generally in cast structures the dis­ work of the a eutectoid can clearly be seen in rribution is fine and random. the interdendritic regions. The toggle pin has a With a higher percentage of tin, the structure composition of92.3% copper, 3.7% tin, 0.6% may become difficult to understand if lead is zinc, 0.4% iron, and 1.3% arsenic. Note that the a present as well. This is especially true ifthe bronzes phase (the copper-rich component) appears light are quenched. There are several Roman mirrors in color except near the eutectoid where some cor­ made in bronzes approximating to: 24.7% tin, ing occurs. The coring looks as though some+ of 0 the 3.2% lead, and 73.7% copper. When quenched tin is being depleted in the region of the a from intermediate temperatures, a very fine parti­ eutectoid as some tin from the a solidO. solution cle matrix develops that is difficult to resolve region migrates to join+ up0 with a + The darker under the optical microscope. This is due to the regions around the a phases are copper-rich very fine dispersion of the lead and the develop­ compared to the rest of the bronze. ment of a Widmanstatten srructure in the The bronze srructure of the incense burner in bronzes. If quenching of bronzes of beta composi­ Figure 42 is atypical, consisting of small polygo­ tion is not sufficient to retain beta, then decompo­ nal grains with patches of the eutectoid berween sition into a Widmanstatten srructure offine alpha them and interspersed with small globules oflead. and delta may occur. Some Roman mirrors have This leaded bronze must have been cast fo llowed this kind of srructure. The situation is even more by an annealing process during manufacture (see complicated, in fact, because in leaded-tin bronzes Figs. 109, 110 in Appendix F). There are very few the melting points of the alloys are much lower traces of rwinned grains present and these are not than in the binary system. The ternaty phase dia- as prominent as the coring that still remains from Figure 41. right. Cast toggle pin from1.3% arsenic.Iran. Bronze Etch: with FeC 3.7%I3: x 12tin0. and Figurebronze 42. incense far right. burner Chi ofnese the cast­ 19th lead.century. Etch: Bronze FeCI3: with x80. 8% tin and 4% The Micros/mellire o/Tin Bronzes 28

Figure 43. top right. A small mirror Suof matrabeta-quenched from Kota bronze Cina. froma 10th x15century0. A.D. trading complex. Figure 44. top far right. A high-tin thebronze classic mi rrordevelopment from Java of showing the thebeta-quenched microstructure. bronze x 13 structure0. in Figure 45. below right. Cast high­ 22%tin leaded tin; 6% bronze lead and of composition72% copper alcoholicshowing structure FeCI). x80. afte r etching in Figure 46. below far right. compositionLaboratory quenched 24% tin. 76%alloy copper of waterthat has at been650 °Cquenched showing in a coldfine innetwork this sample of � needles.occupies The the � phase fewstructure globules completely of copper with oxide only a present as an impurity. x 180. the cast metal. possesses a highly lustrous dark green patina. The bronze mirror in Figure 43 has slightly less Despite being a metastable phase, beta bronzes than the required amount of tin to make a com­ remain quite stable after thousands of years of plete beta phase throughou t, and small islands of a burial. phase can be seen that often fo llow the previous Figures 45 and 46 are examples oflaboratory­ grain boundaries of the high temperature p grains made bronze ingots, originally produced to study before quenching the alloy. The type of p needles the structural aspect of ancient alloys. For example, developed here is sometimes mistaken fo r strain Figure 45 was cast in a high-tin leaded bronze to lines in a copper alloy, but the context and type of replicate some of the structures that were fo und in lines to be seen is unmistakable. It is etched with a series of Roman mirrors. Figure 46 illustrates the FeCl3 at x150; about 22.5% tin, 77.5% copper. microstructure for a laboratory prepared alloy The a islands that are also present in many of made from an ingot of 24% tin, 76% copper the high-tin beta bronzes, such as the mirror in Fig­ quenched in cold water at 650 °C. The structure of ure 43, are sometimes rwinned as a result of hot­ this alloy consists of a fine interlocking nerwork of working the bronzes before quenching. The mirror p needles. Th e Microstructure of Tin Bronzes 29

Figure 47, top right. Japanese sword closeblade tofra gmentthe center of the of 18theth blade,century, showingstructure folds as a inresult the ferriteof the forginggrain operEtch:ations nital; xused 150. to make the blade. swordFigure 48,blade top fra fargment, right. close Japanese to the sitecutting which edge, has showing been tempered. lath marten­ tureNote between the great the dif fcentererence and in struc­ picral;cutting x300. tip of the sword. Etch: sectionFigure 49, of abelow Japanese right. sword Part ofblade the showingcutting edge transition where zone troosite near (very to canfine beirresolvable seen togeth pearlite)er with nodules Widmapicral; x220.nstatten pearlite. Etch: Figure 50, below far right. Japanese fromsword the blade cutting section, edge furtherand closer back to whitethe center phase than is pure Figure ferrite 48. Here grains the withpearlite an (fineinfill ofgrey the zones). eutectoid Note the two-phasedsection. Etch: slag picral; stringers x220. in the

5 NOTES ON THE STRUCTURE OF CARBON STEELS

The type ofsrructure seen in Figure 51, below, in low-carbona + steel, could arise from heat rreatment in the y region of the phase diagram. At room temperature, the structure will be fe rrite + pearl­ ite, and after heating this will become:

and after cooling the gamma phase will revert to: featuFigureres. 52. Grain boundary structure with subgrain

alpha + Fe3C. During cooling some of the fe rrite will take a The section in Figure 53 shows small grain Widmanstatten srructure. size in the cutting edge and a larger grain size in the body of a knife. There is some tempered mar­ ferrite (a) tensite on the edge.

slightly Widmanstatten (a) t Figure 51. Partially Widmanstatten steel. atsmall edge grains bodylarge ofgrains knif ein If the cooling rate is very fast then martensite Figure 53. Grain size of knife edge. can fo rm. Even if the cooling rate is too slow fo r this transformation to occur the proportions of This type of structure cannot originate from pearlite and fe rrite in the solid will not be present heating the tip of the blade fo llowed by quench­ in equilibrium amounts. The presence of more ing. If the tip was made by heat rreatment from pearlite than fe rrite will lead to an overestimation the same metal then the grain structure would be of the carbon content. larger. This indicates that the cutting edge must The structure in Figure 52 shows grain have been pur in at some point during manufac­ boundary cementite (Fe3C) with a subgrain ture; it could not have been made in one piece. structure of cementite + pearlite. How can this When etching quenched structures, some type of srructure arise? When cooling down an areas of the martensite may etch rather darker alloy from the gamma (y) region, the alloy con­ than others. This can be due to tempering of the tains more than 0.8% carbon. Upon decomposi­ martensite or, if fuzzy nodular shapes appear, it tion of the gamma phase, Fe3C will be fo rmed at could be due to the presence of troosite. Troosite the original austenitic grain boundaries after cool­ patches appear because the cooling rate is not ing from temperatures above 900 °C. If this alloy, quite fast enough to produce a martensitic struc­ typically with about 1.2% carbon content, is ture. There is often confusion between the terms cooled down and then subsequently reheated to troosite and sorbite. It is better to reserve the term abour 800 °C (in the gamma + [alpha + Fe3C] sorbite fo r srructures resulting from quenching region), then more Fe3C will be fo rmed on a finer fo llowed by heat rreatment (for example, marten­ scale, while the original austenitic grain bound­ site going to sorbite upon heating). Troosite aries will be preserved, so that on subsequent should be used fo r srructures resulting from the cooling, the gamma phase will decompose to rapid cooling of a carbon steel, but not fast Structure of Carbon Steels 32

enough fo r the production of martensite. In Figure 55, if the alloy is of composition A, In the banded structure of the quenched then the structure will consist ofalpha at the grain sword blade in Figure 54, the question arises as to boundaries with pearlite infill, but if the alloy is of why, if the whole structure has been quenched, composition B, then the structure will be cement­ are there some areas of martensite and some that ite at the grain boundaries with an infill of pearl­ are apparenrly only a mixture of fe rrite + pearlite. ite. Small changes in the direction ofB make litrle There are two possible reasons fo r the structural difference to the visible amount of cementite at differences. First, the difference may be due to the grain boundaries, but a small movement in carbon content. The amount of carbon present the direction ofA makes a large difference in the will affect the production of martensite when the amount of alpha at the grain boundaries. heated alloy is quenched. It is possible that the carbon content in the fine fe rrite + pearlite regions is not sufficientto produce martensite. Second, the shift may be due to other alloy ele­ ments in the regions concerned, which would be apparent with aHiJUfurther analysis. I quenchedFigure 54. swordBanded blade structure consisting of a of A +-1I -----.B verybands fine of martensite.pearlite and alternate 1I + � � . O.8%C msI ms ms martensite (ms) Figure 55. Part of the Fe-Fe3C phase diagram. Similarly, if in a homogeneous specimen the cooling rate is not sufficienrlyhigh to produce a totally martensitic structure, then it is possible to get a series of transitions between martensite, rroosite, fe rrite, and finepearlite, with incipienrly Widmanstatten fe rrite between the pearlite regIOns. 6 MARTENSITE IN LOW-CARBON STEELS

Martensite is a very hard needlelike constituent of appearance in section than the plate fo rm. steels that fo rms when steels are quenched in Mixtures oflath and plate martensite occur in water or other liquids at room temperature. A steels up to about 1 % carbon, while the plate steel artifact is usually heated up to the austenite fo rm predominates over 1 % carbon. Sometimes region of the phase diagram, abour 800-950 °c, darker etching networks in bisulfite or nital are then quickly plunged into cold water, thus fo rm­ composed of plate martensite. ing the extremely hard martensite constituent. The hardness of martensite is determined by Some of the hardness and brittleness could then its carbon content which can vary from about be removed to various degrees by tempering, usu­ 300 Hy at 0.1 % carbon up to about 900 Hy at ally at temperatures between 200 and 500 °C. 0.6% carbon. Many ancient artifacts, however, Untempered martensite is difficult to etch in were made under poorly controlled conditions, picral, but it can be etched in nital or sodium and the martensite may have tempered itself as bisulfite. There are two essential types of marten­ the heated steel cooled down to room tempera­ site: lath and plate. Most steels below a carbon ture; this is known as self-tempering. Tempered content of 0.6% give lath martensite. The indi­ martensite reacts much more quickly to etchants vidual laths cannot be resolved by optical micros­ than the untempered variety, and picral etchants copy even at x1000, and what can be seen is an can be used readily, as can bisulfite, although aggregation of small lath bundles. The number of according to Samuels (1980) there is only a slight possible orientations fo r the growth of lath mar­ change in the response to nita!' tensite within the parent austenite grain are less During the quenching process it is also possi­ numerous than those in plate martensite, which ble to retain some of the high-temperature auste­ becomes the dominant constituent as carbon con­ nite phase. Austenite can often be made apparent tent increases. Lath martensite has a more regular with Villela's etchant.

7 THE TEMPERING OF MARTENSITE

Althoughthe tempering process simply involves untempered martensite. Similarly the fo rmation heating the quenched artifact to temperatures of cementite at higher tempering temperatures that may lie between 100 and 600 °C, the pro­ (e.g., 500 °C) gives darker etching regions, and cesses involved are quite subtle and complex; fo r sometimes a mottled appearance can be seen at example, a particular fo rm of carbide, epsilon (E)­ high magnification at xlOOO-x2000. carbide, fo rms when steels containing more than When tempering is carried out at 400 °C or 0.2% carbon are tempered between 100 and higher, carbon is removed from the martensite 250 °C, while if the tempering is carried out and grains of fe rrite can start to fo rm. These can between 250 and 700 dc, cementite is fo rmed be platelike, but the morphology can barely be instead of E-carbide, in the fo rm of small plates of seen under the optical microscope. spheroidal particles. It is difficult to see any fine The hardness of tempered martensite decreases detail of these changes using optical microscopy. as the tempering temperature rises, and the influ­ Precipitation of E-carbides can only be detected ence that carbon content has on the hardness of indirectly since they etch more darkly than tempered structures decreases considerably.

WootzFigure 56. crucible A good-qual found inity the steel Deccan prill fromarea the of India.lid of The a medievalFigure 57. kn Marife tensiteblade from needles Ardingley, in the Sussex,cuttingedge England. of a probablyprill had become during agitation embedded to intest the cruciblethata molten lin ingproduct Etchedonly just in resolvablepicral at x380 at this magnification. magnification. The The needles overall are overhad been 0.8%. produced. The area Theshown carbon in thecontent photomicrograph, of the prill is effesteels;ct is, compare however, thistypical with formartensite martensite seen in inlow-carbon the cut­ austenticetched with grain nital boundaries at x65 shows with somecementite ferrite needles at prior and steel bead shown in Figure 59. This is lath martensite. thisfine magnificapearliteas ti on.the dark phase which is not resolved at The Tempering of Martensite 36

Figure 58. Photomicrograph of part of a small kris from Figure 59. A cross section through an 18th century India.Broken showing fragments the ofhigh cementite carbon contentneedles ofoccur the blade.as white (centerFrench )cut-steel and the beadoctagonal showing faceted part beadof the in su section.pport wire angularalong the fragm directionents throughout of working the of the section. blade. oriented The dark Lightly etched; x20. slagbackgr content.ound is Etch: finely nital; divided x 15 0pearlite with practically no

showingFigure 60. well French developed cut-steel pla tebead martensite of the 18 atth acentu ry contentmagnification of the of cut-steel x420. etched bead isin aboutpicral. 1.The2% carboncarbon. a veryEuropean high carbonproducts steel of thecompared time. with most other 8 STRUCTURE AND PROPERTIES OF CAST IRON

• It is remarkable that ancient Chinese metalsmiths They can be mixed with wrought iron and in the Han Wei period (206 B.C.-A.D. 534 ) and heated to produce a steel by lowering the car­ in the Western Han dynasty (206 B.C.-A.D. 24 ) • bon content. were already producing spheroidal graphite cast Grey cast irons may corrode badly and they iron. Just as remarkable are whiteheart malleable present difficult problems fo r the conservator. cast irons from the Province of Hubei, dating to There are three principal groups of cast iron: the Warring States period of 475-221 B.C. grey cast iron, white cast iron, and mottled cast (Hongye and J ueming 1983). Early apparatus fo r iron (mixed areas of white and grey) . providing efficient blasts of air enabled the attain­ In the grey cast irons there is free graphite in ment of very high temperatures. The commercial the structure-some of the carbon being rejected production of cast iron in the West did not com­ from solution and solidifYingas flakes ofgraphite. mence until the 13th century A.D., considerably In white cast irons all the carbon occurs as later. But what is cast iron and what are the dif­ cementite. In the mottled, mixed group, both fe rent microstructures associated with this mate­ graphite flakes and cementite can be fo und in dif­ rial? fe rent regions of the same sample, sometimes Cast irons have high percentages of carbon­ deliberately made by chilling part ofthe mold and greater than 2% but generally less than 5%-and increasing the cooling rate, which favors cement­ are castable solely because they can be kept in a ite fo rmation. liquid state at temperatures as low as about There are two important factors influencing 1148 °C (with a carbon content of 4-5%). Cast the fo rmation of graphite and cementite: solidifi­ irons are difficult to describe fully without some cation rate and composition. Since the iron-car­ analyses fo r additional alloying elements that are bon phase diagram (Fig. 30), incorporating fe rrite often present, besides carbon, and that can and graphite as stable products, is the nearest to change considerably both the physical properties equilibrium, slow cooling favors graphite produc­ and the microstructure. tion, while rapid cooling favors the metastable Present-day cast iron produced in a blast fur­ fe rrite and cementite system. nace is essentially pig iron and may contain sili­ The most important elements in cast irons are con, sulfur, phosphorus, manganese, nickel, and carbon and silicon, a high content of either favor­ chromium, in addition to carbon, and each of ing graphite fo rmation. Other alloying or impu­ these elements produces changes in microstruc­ rity elements that stabilize graphite are nickel, ture. The most important alloying additions or aluminium, copper, and titanium. impurities as fa r as ancient and historical period Manganese stabilizes cementite, although the cast irons are concerned are silicon, sulfur, and situation is more complicated if sulfur is present phosphorus. The special properties of cast irons as well. Sulfur stabilizes cementite as well, but sul­ may be summarized as fo llows: • fur also has a strong affinity fo r manganese and • They may be cast into shape in molds. fo rms manganese sulfide particles if both ele­ They have good fluidity and, when carbon is ments are present. This compound, as a result, precipitated, they expand on cooling and thus has little influence on carbon. • take a good mold impression. Primary additions, fo r example, of sulfur to Their melting point is low, on the order of an iron with a high manganese content, tend to • 1200 °C. stabilize graphite fo rmation. Phosphorus acts They have good wear properties since graphite chemically to promote carbide fo rmation, bur the • is a good lubricant. presence of phosphorus produces a ternary • They have a high damping capacity. phosphide eutectic (steadite) between fe rrite They have a reasonable strength in compres­ (usually with some phosphorus content), cement­ sion, although they are weak in tension. ite, and iron phosphide (a + Fe3P + Fe3C)' This Structure and Properties 0/ Cast fron 38

ternary eutectic melts at 960 °C and thus is the This CE value places the Han cauldron in the last constituent to solidify. This results in the y + hypereutectic region, although since the lede­ Fe3C solidifyingslowly and allows any silicon burite eutectic occurs at 4.3% carbon (at a tem­ present to promote graphite fo rmation. With low perature of 1140 DC), the alloy is quite close to amounts of phosphorus (about 0.1 %), graphite the eutectic region. Ledeburite is a eutectic com­ can actually be enhanced rather than destabilized. prising iron carbide and austenite. Nickel, like silicon, dissolves easily in fe rrite and also acts to stabilize graphite. Microstructural Constituents Because of the complexities introduced by The constituents, apart from graphite and trans­ alloying elements beside carbon, a carbon equiva­ fo rmed ledeburite, are similar to steels and in a lent value (CE) is often quoted instead: typical grey cast iron would include graphite = ---(Si + P) flakes, as well as varying amounts of fe rrite and CE total C% + 3 pearlite. The fe rrite may be harder than in ordi­ nary steels because of the silicon content. The This CE value may then be used to determine usual white cast iron constituents are pearlite and whether an iron is hypoeutectic or hypereutectic. cementite. The grey cast irons may have the fo l­ In general, the lower the CE value, the greater the lowing constituents: tendency fo r an iron to solidifyas a white cast iron or as a mottled iron. 1. Flake graphite, fo rmed during solidification, Irons that are hypereutectic (carbon equiva­ can determine the properties of the iron by its lent more than 4.2-4.3%) will precipitate kish shape, size, quantity, and distribution. ) graphite on solidification with normal cooling Graphite flakes are usually fo und in the fo l­ rates. Hypereutectic cast irons solidifyby direct lowing fo rms: randomly distributed, den­ fo rmation of graphite from the melt in the fo rm dritic, or rosette. ofkish, which has a low density and can rise to the 2. Pearlite, which may vary in composition be­ surface. When entrapped by metal it generally tween 0.5 and 0.8% because of alloy content. appears as long wavy or lumpy flakes. This phase 3. Steadite, the ternary phosphide eutectic previ­ is precipitated over a range of temperatures that ously discussed. will be examined with the aid of a phase diagram 4. Free fe rrite, which appears in appreciable (Fig. 61) later in this section, starting at the amounts only in low-strength cast irons. The liquidus surface and continuing until the remain­ fe rrite is usually rounded in appearance com­ ing melt is at the eutectic point when crystalliza­ pared with steadite or cementite because it tion of graphite and austenite occurs. This precipitates from the austenite solid solution eutectic graphite in hypereutectic cast irons is rather than from a liquid. It is often fo und in usually finer than primary kish graphite. association with graphite flakes. Free cement­ As an example of the application of the CE ite is unusual in grey cast irons. fo rmula, consider a white cast iran from Nang Yang City, Henan Province, China, dating from Preparation of grey cast irons by metallo­ (A.D. graphic polishing can give rise to problems the Eastern Han dynasty 24-220 ). The object is a cauldron with the fo llowing elemental because the graphite flakes may be smeared or analysis: C 4.19%, Si 0.14%, Mn 0.09%, removed preferentially upon polishing. Very cor­ S 0.06%, P 0.486% (Hongye and J ueming roded grey cast irons can also be very difficult to 1983). retain in a plastic mount because they may consist oflittle more than a loose mass of iran oxides in a = 0. 14 + 0. 486 graphite matrix. These have a tendency to lose CE 4. 19 + 4.39 3 part of the fr iable surface upon polishing wi th the consequent difficultyof obtaining a good polish. Structure and Properties of Cast Iron 39

Less corroded specimens polished with diamond Heat Treatment of Cast Iron abrasives usually do not present major difficulties. Castings produced in white cast iron can be made A good test to determine if graphite is pol­ malleable to some extent by a number of tech­ ished correctly is to examine the sample in niques. Two such methods are known as white­ 2 reflected polarized light. Graphite is anisotropic heart and blackheart. These names refer to the 3 and exhibits reflection pleochroism. When respective fractures of the diffe rent irons after examined under ordinary bright field illumina­ annealing, the whiteheart variery being white and tion with unpolarized light, graphite flakes appear crystalline due to the presence of pearlite and the to have a uniform brownish-grey color. If the blackheart being grey or black due to graphite. In same sample is then examined under plane polar­ whiteheart structures, annealing oxidizes some of ized light, some graphite flakes appear light, some the carbon producing a fe rrite structure that grad­ dark, and some have an intermediate color. If the ually changes to a steel-like mass of fe rrite and sample is rotated 90°, the flakes fo rmerly dark pearlite with interspersed nodules of graphite. In become light. The relationship between the plane blackheart structures, the heat treatment does not of polarization of the incident light and the posi­ result in much carbon oxidation and, depending tion of maximum or minimum brightness is such on other alloying constituents such as silicon and that when the plane of polarization is at right sulfur, the cementite may break down to graphite angles to the graphite flakes they appear dark, and aggregate nodules in a matrix offerrite and pearlite. when parallel they appear light (Nelson 1985). If the same sample is examined under polarized Temper Carbon Nodules light between crossed polars, upon rotation When cast iron is in the white condition, anneal­ through 360°, each graphite flake will lighten ing at temperatures in the range of 800-950 °C fo ur times; this is expected from the hexagonal may produce graphite nucleation. At this temper­ anistropy of graphite. ature, white cast iron usually consists of a eutectic Picral is probably the best etchant fo r pre­ matrix of cementite, austenite and, if many impu­ dominantly pearlitic grey, malleable, and ductile rities are present, inclusions. Nucleation ofgraph­ irons. Picral does not damage graphite which nital ite then occurs at the austenite/cementite can, but fo r fe rritic grey cast irons nital is prefer­ interfaces and at inclusions. Cementite gradually able. dissolves in the austenite and the carbon diffuses If the iron is sufficiently below the eurectic to the graphite nuclei to produce nodules. This value, has a very low silicon content, and contains kind of process of heat treatment appears to be appreciable carbide stabilizers, or if the cooling responsible fo r the spheroidal graphite cast irons rate is fast, then instead ofgraphite flakes solidifi­ made in ancient China. Ancient Chinese smiths cation occurs by fo rmation of austenite dendrites. could cast a white cast iron and then, by high Meanwhile the interdendritic regions solidifyas temperature heat treatment, produce spheroidal ledeburite, a eutectic mixture of iron carbide and graphite. The graphite is slightly polygonal in austenite. In appearance, ledeburite etched in shape, but in all other respects is the same as the nital would consist oflight etching cementite in a spheroidal graphite cast iron produced today by background of unresolved and transformed auste­ alloying a variery of elements such as cerium or nite, which usually appears as dark etching magnesium with cast iron. These alloying ele­ pearlite. ments do not occur in ancient Chinese examples. Ascast iron cools, the austenite transforms to Analysis by SEM indicates that spheroidal graph­ ferrite and cementite. The structure of a white ite nuclei are latent in white cast irons having car­ cast iron at room temperature therefore usually bon, low silicon, and the required ratio of consists of primary dendrites of pearlite with manganese to sulfur. interdendritic transformed ledeburite. Structure and Properties o/ Cast Iron 40

In the microstructure of a typical cast iron, both of the constituents present at t2. which cools down quickly, the nonequilibrium At tr tj Y + (y + Fej C) ----7 phase Fe3C tends to predominate: the equilib­ rium constituent is free graphite. Line AB has y (+Fej C) +[(y + Fej C) + Fej C] been drawn on the iron-iron carbide diagram that Finally, as cooling proceeds past t3 itself, all of the represents the cooling of a typical cast iron (Fig. austenite tellds to decompose to pearlite. This 61). At temperature t1, the liquid material starts ; means that the final change can be represented as: to solidifY and austenite grains, or rather den­ drites, start to fo rm, surrounded by liquid. As the At tj Y (+ Fej C) + [(y + Fej C) + Fej C] ---7 temperature falls, this liquid cools and reaches the (y + Fej C) + Fej C (ex + + Fej C) + Fej C + Fej C stage at which a simple eutectic decomposition occurs. This eutectic is called ledeburite and con­ This series of reactions leads to the fo rmation of sists of austenite + Fe3C. L2 typical white cast iron. There is a different series r--7+ (dendrites of At tJ L Y y in liquid) of phase changes that may occur if slow cooling is At t + + (y C) carried out or if there are alloying additions, such Y Y + Fe 2 L2 ----7 3 as silicon, made to stabilize the formation of As the temperature falls between t2 and t3 on the graphite. The series of changes that take place at graph, so the carbon content of the austenite t1, t2' and t3 are the same, except where Fe3C decreases and more Fe3C is precipitated from occurs in the phase relationships, it is replaced diaFiguregram 61. for Part cast of iron. Fe-Fe3C phase

B D Structure and Properties o/ Cast Iron 41

with graphite (G): in Figure 61 would then be represented by: Y + L Fe3 C + 2 At At tl LI ----7 tx LI -----7 L2 -----7Fe,C + (y + Fe,C) At + Y+ (Y + G) At ty Fe3C L t2 Y L2 ------7 + 2 + ( + G) G + f(y + G) tr t3 Y Y ----'7Y+ + G} At ty-tz Fe3 C + (y + Fe ,C) � Fe3 C (+ y) + f(y + Fe3C) + Fe3 C} At t3 y + G + f(y + G) + G} ------7 y+G+G+ y+G+G+ G White cast irons with a total carbon content of over 4% tend to be extremely hard with no In practice, the last stage never fu lly takes place. elongation at all and are not normally used in the More typically the first phase change is from aus­ fast cooled state. With slow cooling, very large tenite to austenite + graphite, but the austenite is graphite flakes can fo rm, and it is possible to get much more likely to precipitate Fe,C. The phases quite a large difference in the size of these flakes at t3 then become: depending on where they fo rm in relation to the At t3 Y + G + f(y + G) + G} ------0:> phase diagram. Graphite fo rmed with a carbon (a + Fe3 C) + G + (a + Fe3 C) + G + G content of 3% can be quite small, whereas with a carbon content of 5% the flakes can be enormous This should lead to dendrites of pearlite in a (Fig. 62). matrix of fe rrite and graphite which is most likely Although cast iron metallurgy was well devel­ in grey cast iron. oped in China, isolated examples of cast iron do Above a carbon content of 4% the primary occur in other Old World contexts, fo r example, reaction would be decomposition of the liquid to the 1st century A.D. find from the classic excava­ fo rm Fe,C and ledeburite (austenite + Fe,C). The tions of Bus he-Fox at Hengistbury Head and also phase changes upon fast cooling of an alloy shown from the Roman period cast iron from Wilders- Figure 62. Flake graphite in cast iron.1969(E). From x I 00.ISO Standard R 945- a.b. FlakeCrab-type graphite graphite d.c. QuasiAggregate flakeor graphite temper carbon f.e. IrNodularregular or spheroidalopen-type graphitenodules a b c df�(fiI'�� <.JfJ?;. .JJ�1� � �  •. " : .• ..- :. ••••. - . � 1)�""\. :J;"*.;z. -'I. tf.%,� (#�, � • '.• - • • � 1'(:1 'd-tl,\,e&> ir'l� t)>,. ��J � � � 1?:4'J ...... � .. -.. .. 4t"��-> J>,'4. ,,.1�:>-___ � �rr . r ff .•.•• .e. . ..• •..•. - .-• •. .e •.• �t-';i' .:.\ 'J �' !.J.l'Ot>? �� ir� \1i� ,.' ..- •..•. ..e· • .- :. d e StYlla"re and Properties o/ Cast Iron 42

Figure 63. Cast iron scales from the from18th century.England, Typicalshowing example the graphite flakesborders surrounded (ferrite). The by white grey infill is whpearlite,ile the which white has spotty variable phase spacing, is the ternarynital; x90. eutectic, steadite. Etched in Figure 64. Heavily corroded fragmentfrom the Towerof a cast of iron London, cannonball 18th century. Note severe corrosion of following analysis: C l.98, Si 0.16, Mn 0.04, the ferrite regions around the pool, Lancashire. Both of these pieces were of course graphite flakes and grey cast iron. The Hengistbury example had an S 0.048, P 0.297. corrosion of pearlite regions analysis of: C 3.4%, Mn tr, Si 0.38%, P 0.1 8%, phasesthroughout that thehave structure. survived Thebest twoare S 0.035%, while the specimen from Wilderspool Notes the steadite eutectic (the fine gave: C 3.23%, Mn 0.403%, Si l.05%, P 0.76%, 1. Kish graphite is the graphite that separates from spotty material) and the cementite S 0.49%. molten cast iron as soon as it cools to the solid. laths (the long white crystals). Many of the Chinese examples have lower sil­ It may sometimes float on top of the molten Unetched; x40. icon content than these, ranging from 0.08% to alloy. 0.28% in the examples published by Hongye and 2. Anisotropic substances do not have the same J ueming in 1983. The Chinese examples span a physical properties in all directions of the mate­ considerable range of different material as can be rial. . Reflection pleochroism results from different seen from the fo llowing: 3 • interactions in anisotropic solids when viewed An axe from the Warring States period (475- under polarized light. Generally, pleochroic 221 B.C.) excavated from the site ofTonglu­ substances will show some color variation on shan, Huangshie City, Hubei Province, was rotation when viewed under polarized light. fo und to be an example of a whiteheart mal­ leable cast iron with imperfect decarburiza­ tion and a total carbon content that varies from 0.7 to 2.5%. The analysis fo r this axe gave: C 0.7-2.5%, Si 0.13%, Mn 0.05%, • S 0.01 6%, P 0.108%. An example of a mottled cast iron is provided by a hammer from the same site with the fo l­ lowing analysis: C 4.05%; Si 0.19%; • Mn 0.05%; S 0.019%; P 0.1 52%. A white cast iron was fo und at Cheng Zhou City, Henan Province. This block dates to the ( ) Eastern Han dynasty A.D. 24-220 and gave the fo llowing analysis: C 3.97%, Si 0.28%, • Mn 0.30%, S 0.078%, P 0.264%. An example of cast iron with spheroidal graph­ ite was fo und in the iron workshop at Triesh­ enggou dating to the Western Han dynasty (206 B.C.-A.D. 24 ). This site is located in Gong County, Henan Province, and gave the 9 CORRODED MICROSTRUCTURES

Many ancient objects are, ofcourse, covered with also many crystalline and other morphological corrosion products that may have originated at details are revealed clearly. the time of manufacture (if the object had been It should be remembered that the interface of deliberately patinated, fo r example), or they may interest may not just be between the metal and have arisen from corrosion during burial or in the the primary (or currently existing) corrosion atmosphere. Many books on the subject of corro­ products. Important info rmation may also be sion do not discuss the types of structures that preserved in some other interfacial event between occur in ancient metals. Often they are subjected layers of corrosion products of different composi­ to several different corrosion processes, resulting tion or structure. in a nearly composite material consisting of Corrosion products may pseudomorphically metallic remnants and mineral alteration prod­ replace the metallic structure that previously ucts. Normally corrosion produces a buildup of existed. That is to say, the structural orientation insoluble products, both within and overlying the (for example, dendritic structure) of the metal is original metal volume. Corrosion products may preserved in the corrosion products that have be very informative; indeed, they may be all that replaced it. One of the ways in which this can is left of the original object. Therefore, corrosion happen is by epitaxial growth. This means, in products should not be cleaned from the surface general terms, the regular orientation of a partic­ of antiquities before metallographic examination. ular growth on a crystalline substrate. Most Loose material, soil, and soil and mineral concre­ ancient chemical corrosion processes (which are tions usually should be removed because contin­ best modeled by electrochemical reactions), if ual loss of them during polishing could create a they produce structural information preserved in very badly scratched surface unsuitable fo r micro­ the corrosion products, do so by chemoepitaxy. scopic examination. It is generally possible to pol­ This is a subdivision of the epitaxy structure. ish corroded samples in much the same way as Chemoepitaxy can be defined as a process leading more robust specimens, although ifthe material is to the growth of crystalline, regularly oriented very friable, vacuum impregnation with a low­ reaction layers on a material resulting from a viscosity epoxy resin should be considered before chemical reaction between this initial substance mounting, or the whole sample could be and any other substance. Examples of layers mounted in resin under vacuum. fo rmed from such a process are cuprite (CU20) It is most important to examine all features of growing on the surface or into the metal grains corrosion products before etching the specimen, themselves, and Ag20, the initial thin filmthat since many corrosion products are severely can fo rm on silver objects. attacked by the chemicals used to etch metal sur­ Valuable information can sometimes be faces: they could be dissolved completely. Under retrieved from corroded metallic fragments; parts normal bright-field reflected light microscopy of the structure may remain uncorroded, or there most corrosion crusts have a grey color. Examina­ may be pseudomorphic replacement of the phases tion under reflected polarized light yields a great by corrosion products. Appendix B illustrates deal of valuable information. The polarizer in some of the types of corrosion fo und in ancient reflected light microscopes is usually housed in specimens. Evidence of the authenticity of an the chamber of the light source and can be artifact can be obtained from metallographic rotated, while the analyzer is placed in a special examination of small chips of corroded metal. In holder in front ofthe eyepiece turret. By adjusting copper alloys, the presence of intergranular corro­ one or both of the polarizing elements, the "true" sion, intragranular corrosion, corroded slip lines, colors of corrosion products can often be twin lines, or cuprite next to the metal would be revealed. Not only does this aid considerably in very significant. It is extremely difficultto pro­ the interpretation of many microstructures but duce coherent cuprite layers by artificialcorrosion Corroded Microstructures 44

over short periods of time, and it is also practically remove smeared material, or ultrasonic cleaning impossible to fake the penetration of cuprite in baths of acetone or alcohol, may produce a bet­ along selective planes in the crystal, such as twin ter finish. One danger is that small specimens will bands or slip lines. If the corrosion process has not become rounded at the edges and part of the cor­ completely disrupted the original volume of the rosion crust or the edges of the metallic constitu­ metal but has preserved some interface between ents may then be out of fo cus microscopically internal and external corrosion, then this discon­ compared with the remaining polished surface. tinuity may be recognized as preserving informa­ The polishing hardness ofcorrosion products and tion relating to the original surface of the artifact. the metal from which they have fo rmed is usually When corroded samples are mounted fo r pol­ quite different. Surface relief effects are therefore ishing, the difficulty of preparing scratch-free sur­ common when examining corroded samples. It faces is evident. Abrasive particles fr equently can has been fo und that a much sharper preservation become embedded in the corrosion products. of detail in corrosion products is achieved by During polishing some of these particles may be using diamond polishing compounds rather than released, producing scratches on the metal. Pro­ alumina compounds (Figs. 65-68). longed polishing with intermittent etching to Figure 65. Typical variations in the earth minerals, quartz grains, etc. preservationresult of corrosion. of surf a.ace Preservation detail as a tionof shape of surf in acecorrosion. by corrosion. b. Disrup­ c. Sound metal with patina. _.,...���-:o:��77?::77�T"-;r-suroriginalface coin corrosionsecondary products sound metal corrosion products earth minerals, quartz grains, etc.

r.r:��,�_-;--- suoriginalrface coin

sound metal secondary corrosion products _��.����.... ����'"'_---- original surface "patina"lacquer, etc.)(oxide, sulfide, grease, sound metal 9 CORRODED MICROSTRUCTURES

Many ancient objects are, of course, covered with also many crystalline and other morphological corrosion products that may have originated at details are revealed clearly. the time of manufacture (if the object had been It should be remembered that the interface of deliberately patinated, fo r example), or they may interest may not just be between the metal and have arisen from corrosion during burial or in the the primary (or currently existing) corrosion atmosphere. Many books on the subject of corro­ products. Important information may also be sion do not discuss the types of structures that preserved in some other interfacial event between occur in ancient metals. Often they are subjected layers of corrosion products of different composi­ to several diffe rent corrosion processes, resulting tion or structure. in a nearly composite material consisting of Corrosion products may pseudomorphically metallic remnants and mineral alteration prod­ replace the metallic structure that previously ucts. Normally corrosion produces a buildup of existed. That is to say, the structural orientation insoluble products, both within and overlying the (for example, dendritic structure) of the metal is original metal volume. Corrosion products may preserved in the corrosion products that have be very informative; indeed, they may be all that replaced it. One of the ways in which this can is left of the original object. Therefore, corrosion happen is by epitaxial growth. This means, in products should not be cleaned from the surface general terms, the regular orientation of a partic­ ofantiquities before metallographic examination. ular growth on a crystalline substrate. Most Loose material, soil, and soil and mineral concre­ ancient chemical corrosion processes (which are tions usually should be removed because contin­ best modeled by electrochemical reactions), if ual loss of them during polishing could create a they produce structural information preserved in very badly scratched surface unsuitable fo r micro­ the corrosion products, do so by chemoepitaxy. scopic examination. It is generally possible to pol­ This is a subdivision of the epitaxy structure. ish corroded samples in much the same way as Chemoepitaxy can be defined as a process leading more robust specimens, although if the material is to the growth of crystalline, regularly oriented very friable, vacuum impregnation with a low­ reaction layers on a material resulting from a viscosity epoxy resin should be considered before chemical reaction between this initial substance mounting, or the whole sample could be and any other substance. Examples of layers mounted in resin under vacuum. fo rmed from such a process are cuprite (CuzO) It is most important to examine all features of growing on the surface or into the metal grains corrosion products before etching the specimen, themselves, and AgzO, the initial thin film that since many corrosion products are severely can fo rm on silver objects. attacked by the chemicals used to etch metal sur­ Valuable information can sometimes be faces: they could be dissolved completely. Under retrieved from corroded metallic fragments; parts normal bright-field reflected light microscopy of the structure may remain uncorroded, or there most corrosion crusts have a grey color. Examina­ may be pseudomorphic replacement of the phases tion under reflected polarized light yields a great by corrosion products. Appendix B illustrates deal of valuable information. The polarizer in some of the types of corrosion fo und in ancient reflectedlight microscopes is usually housed in specimens. Evidence of the authenticity of an the chamber of the light source and can be artifact can be obtained from metallographic rotated, while the analyzer is placed in a special examination of small chips of corroded metal. In holder in front of the eyepiece turret. By adjusting copper alloys, the presence of intergranular corro­ one or both of the polarizing elements, the "true" sion, intragranular corrosion, corroded slip lines, colors of corrosion products can often be twin lines, or cuprite next to the metal would be revealed. Not only does this aid considerably in very significant. It is extremely difficult to pro­ the interpretation of many microstructures but duce coherent cuprite layers by artificial corrosion Corroded Microstructures 45

corrodedFigure 66a, bronze top right. pin fromComp letely phasePalestine. are Theoriginal dark copper elongated sulfide andinclusions annealed from pin, the preserving worked their originalproducts. location From thineir the preser corrosionvation objectit is possible was not to cast,deduce but that worked, the evenx125. though no metal remains. corrodedFigure 66b, bronze top far rod right. from Part Iran. of a preservedThe original in sucorrosionrface of the products rod is isonly, clear. but x50. the circular cross section CompFigure letely66c, middlemineralized right. silver disk; andradiating patches crystals of silver of silver sulfide chloride (dark) withpreserve some the dis shapetortion. of x65.the object MuFigureltiple 66d, corrosion middle layersfar right. simil ar to fraLieseganggment from rings Iran.in a bronzeThe periodic precipitation of cuprite and outerFe, and areas Cu immediacontain telysome adjacent Sn, CI, Ca, malachitbanded corrosione result in structurethis finely in to the surface whichpreser vedsurf acein corrosion. outline is notx I 00. some regions of high Ca contentinclusions) (soil mineral

\....-----t------originalof the object surface Figure 67. Drawing of the cross sectionThe location of a bronze of the electron rod fragment. micro­ probe line scan is illustrated. A, B, compactprincipally layers cuprite that and are tin TheC, and line D scan indi catewas areasstarted of appr analysis.oxi­ oxide in segregated areas mately at position I and was only trace amounts of bronze metal grains terminated at position 2. tincorrosion in the outer crust containing 9.5% tin Corroded Microsmtctures 46

corrosionFigure 68a-d. of gold-copper Examples of alloys.

Top left, a. Sheet gold-copper alloy Top right, b. Fragment of a Bottom left, c. Cross section of a Bottom right, d. Enlarged view of fromMarta, the Colombia. Sierra Nevada A fragment de Sant ofa a Tairona34.8% copper, ear ornament: and 2.5% 39.6% silver. gold, In fromcorroded the Taironagold-copper area ofalloy Colom­ sheet Figureholic ferric 68c, chloride,x300. Etched revealing in alco­ a gildedlarge Tairona surfaces. sheet The withcomposition depletion­ is themagnification unetched ofcondition x 160, extensive at a 59.6%bia. The gold, composition 19.4% copper, of the andsheet is throughfine network the allo of y.cracks Some running cracks in 15.2% gold, 65.4% copper, and cracking can be seen. Cracking internal7.1 % silver, corrosion the low and total extensive reflecting the principally cuprite matrix are 0.6%largely silver. of dark The grey structure corrosion consi sts corrosion.results from Some volume cracks changes run just upon conversion to cuprite. The ture,related since to thea twin original can bemicr seenostruc­ in andproducts, poros mostlyity (seen cu asprite black with holes) gold underDepletion-gilding the surface hasof the created alloy. a polishedroded alloy section with showsa number the ofcor­ fine cracksone grain appear area to and be some following of the grain asmetal a result is white. of corrosion. Unetched; The x 16 sound0. gold-enrichedto embrittlement layer than more the resistant thatlamellae appear of sound bright. metal The sheetremn antshas boundaries.ing appears toSometimes be unrela theted crack­to the underlying layer. consibeen stentdeple gold-richtion-gilded sur andface the can be grainsome structure,pseudomorphic but here retention. there is section.seen clearly Unetched; on the xtop 16 0.of the Corroded Microstructures 47

Figure 69, top. Luristan ceremonial axe showing interdendritic poro­ The samples illustrated in Figures 69-70 (see sity (dark) with well-developed also Plate 10 and Figs. 112, 1 13), taken from the dendrites. Etch: FeCI3; x30. broken end of a hilt of a Luristan ceremonial blade show a well-developedex dendritic structure RedepositedFigure 70, middle copper (see resulting Fig. 69). from consistingex8 of an dendritic system with the usual in-depth corrosion. Etch: FeCI3; + eutectoid infill, showing that the hilt is in xISO. the as-cast condition. There is some porosity present in the section resulting from interden­ corrodedFigure 71, fra bottom.gment Sectionfrom an of a dritic holes, which could be due either to rapid Ecuadorian gilded copper solidification in the mold or to gas evolution in ceremonial axe, from Pindilig, the casting. The hilt section has distinct copper Canton Azogues, Ecuador. A gold globules, which originate from corrosion effects brightsurface line can at jus thet be top seen of the as section.a thin within the bronze structure and occur most prob­ This axe was gilded by the ably in the fo rmer eutectoid spaces that are electrochemical replacement attacked selectively in this particular bronze in a plating technique described by burial environment. The dissolution of the eutec­ Lechtman (1982). The photo­ toid sometimes leads to precipitation of copper extrmicraordinarograph illy ustrpreciatespitation the of (as here), somewhat akin to the process of copper as a result of the corrosion dezincification in brass. The copper globules of the axe. The redeposited copper often display twinned grain structures, barely vis­ outlines areas of remaining ible here at x 150. uncorrodedpure copper. al xloy, 120. which is nearly

10 REFLECTED POLARIZED LIGHT MICROSCOPY

Polarized illumination fo r the examination of electron microscopy, or transmission-electron ancient metallic samples has many applications. microscopy. The principal advantages gained by In bright-field unpolarized illumination, many using polarized illumination are the visual mani­ nonmetallic inclusions, corrosion products, festation of the morphology and the distribution burial concretions, soil minerals, core materials, of corrosion products by color and crystal size. etc., appear as various shades of grey. By using There are some color effects that can be seen if the polarized illumination and rotating the polarizer stage or specimen is rotated. Color change is due and analyzer until most or all normally incident to pleochroism-as the vibrational direction of light is excluded, fine color differentiation the polarized light is altered by rotation, so the becomes possible between many of the mineral resulting color of the observed anisotropic mate­ phases associated with ancient metals. rial will also change. Some of the metals themselves also give rise to Grain boundaries in metals, twin lines, and colored effects under polarized illumination if slip planes may react differently even in isotropic they are optically anisotropic, i.e., if they have materials such as copper or low-tin bronze. Den­ two or three principal refractive indices. Some dritic structures that have preferred orientations examples ofanisotropic metals are antimony, tin, of growth may also show chiaroscuro effects in and zinc. In addition, some alloys may also pro­ polished section. duce anisotropic effects such as prior austenitic Care must be taken to avoid inference from grain boundaries in martensitic steels, beta-grain the absence of an expected reaction to polarized structures in high-tin bronzes, different reflectiv­ illumination: the concomitant absence of the sub­ ities in duplex alloys, and strain or slip markings stance sought. For example, cuprite, which is usu­ within grains. Most of the common metals of ally a liver-red color in mineral fo rm under antiquity, such as iron, copper, silver, gold, lead, macroscopic conditions, is expected to appear and nickel are optically isotropic since they ctys­ scarlet-red under the microscope using polished tallized in the cubic system and are either body­ sections with polarized illumination. Often this is centered cubic (in the case of alpha-iron) or face­ how it looks; however, some small cuprite inclu­ centered cubic (as in copper, silver, gold, nickel, sions give no bright coloration because the orien­ and lead). Nevertheless, even optically isotropic tation of the crystal is not producing anisotropic metals have grain boundaries, and optical effects effects in that particular position. Although it is with, fo r example, cast metals, do occur, although not customary to use rotating stages with metal­ they rarely provide new information. lurgical microscopes unless the stage is an With completely clean and uncorroded sam­ inverted one, it may be useful to change the stage ples there may be little advantage in the use of if much polarized light work is to be carried out. polarized light microscopy. However, this is rare Massive corrosion, such as that preserving the among ancient metals which typically have both surface detail on a copper alloy artifact, will usu­ nonmetallic inclusions and corrosion products ally produce a wide range of colors. The color present, and often valuable information can be observed with copper corrosion products is very gained by examining them in polarized light. similar to the color usually associated with the Voids that appear dark in unpolarized illumina­ mineral concerned, as evidenced by cuprite. The tion suggest, fo r example, that they may be filled same limitations that apply to visual identifica­ with quartz or calcite, although the chemical tion apply to this type of microscopic examina­ identity of the material examined is not easy to tion. It would not be possible, fo r example, to establish. In fact, there is no straightfotward know that a green mineral seen under polarized guide to the identity of any material of this kind. light was malachite. It is also not possible to say, The best means of identification is by X-ray dif­ with a mineral such as cuprite, that the material is fraction, electron-microprobe analysis, scanning- solely composed of cuprite because a scarlet-red ReflectedPolarized Light Microscopy 50

color is visible under polarized light: there may be compositional variations or other minerals present in addition, such as stannic oxide, which alter the shade of the color. Thus the analyst can­ not detect the presence of minerals by this means alone, although polarized light microscopy used in conjunction with electron-microprobe analysis can give very useful information (Scott 1986). Slags and other materials, such as matte, speiss, cements, and plasters, can also be examined effec­ tively by reflected polarized light. Ancient plaster and cements are usually fine grained or contain fine-grainedaggregates which makes examination difficult in thin section (usually 30 microns thick). More information can be gained about the crystalline and glassy phases in many slags under polarized light, although mineral identification is not usually possible. I An idea of the possible range of color can be shown by looking at slag composed of copper sulfides (matte) in globular fo rm, surrounded by magnetite and fayalite crys­ tals in a dark glassy matrix. In un polarized light, the copper sulfides appear grey, the fayalite laths will be a different shade of grey, the magnetite is white, and the glassy matrix is dark grey. Under polarized illumination, the copper sulfides are light blue, the magnetite is black, the fayalite is grey, and the matrix can have various effects, but can appear translucent and glassy. Crystal size in cements and plasters can be revealed by polarized light examination, as can color variation, and lay­ ered mineral assemblages, as are commonly fo und in wall paintings, painted tesserae, or poly­ 2 chromed materials.

Notes 1. For more information on the structure of slags and other metal by-products see Bachmann (1982) where many slag structures are illus­ trated, both in color and black-and-white.

2. For fu rther information on polarized light microscopy (although little information has been published on ancient metals), refer to the works mentioned in the bibliography by McCrone et al. (1974) and Winchell (1964). 11 GRAIN SIZES OF ANCIENT METALS

A useful quantitative aspect of metallography is to an ASTM number that can be readily com­ the measurement of the grain size of the material pared by another investigator. The ASTM has or the dendritic arm spacing. Dendrite arm spac­ prepared two typical series fo r fe rrous and nonfer­ ing is best obtained by measuring the number of rous materials which are reproduced in Figures 73 arm intersections across a line traverse, the dis­ and 74. Strictly speaking, fe rrous alloys are esti­ tance being measured by means of a stage micro­ mated on a logarithmic scale based on the for­ meter or graticle, or on a photomicrograph of the mula: area concerned. Large, coarse dendrites generally mean that the cooling rate of the alloy in the mold was fa irly slow, usually implying that heated or where n is the ASTM number, and y represents well-insulated molds were employed. Finer den­ the number of grains per square inch. The most drite arm spacings suggest faster cooling rates. useful numbers in this series are from ASTM 1 to Grain size can be measured by a number of ASTM 8, with 8 being the smallest defined grain different methods. One of the simplest tech­ size. As well as ASTM numbers, it is possible to niques is to use a grain size comparator eyepiece. refer to a scale that gives average grain diameters, This eyepiece has inscribed around its circumfer­ usually from 0.010 to 0.200 mm. With twinned ence ASTM (American Society fo r Testing and nonferrous alloys, the structure of worked grains Materials) grain size scales. By direct visual com­ can sometimes be confusing because the grains parison at an objective lens magnification of xl 0, appear fragmentary. The intercept method can giving a magnification of xl 00 overall (the mag­ also be used to determine the average grain size of nification of microscopes is obtained by multiply­ the material. The equation relating the factors ing the objective lens by the eyepiece lens magni­ required is: fication), the ASTM number can be determined. = length of intercept (mm) Great accuracy in this type of measurement is sel­ N(D) magnification x no. of grain boundaries dom required: what is useful is the approximation Figure 72. Nomograph for grain 0 300 size. Estimation based on Hilliard's 250 60 method. 10 50 200 £L..Cl) � 2 150 V> ....Cl)L..V> 2015 4035 0.0:::J Cl)L.. .L..0c ECl) 30 <+=....V> ..0:::JE 3 � -.::J.� 4030 25 ....Cl) Cl)c 100 2-.: � 50 auE '';;CN 4 80 ....Cl)0.. 0c 70 20 c .�L.. uCl)L.. .'"u'" 100 15 V>0 :L0.0 5 60 .�....L.. <+='c0.0 150 .'"0c VI«l- SO cCl)'" E'" 200 L..V>Cl)u ....Cl)c 6 40 0.0Cl) 300 10 ....CCl) '".�:::J 30 >Cl)L..'" 400 87 0L.. Cl)r:r 7 25 '" 700500 6 ..0ECl) 1000 5 c:::J 8 20 4 9 15 Grain Sizes of Ancient Metals 52

Hilliard's circular intercept method can also estimation. The circle is used until at least 35 be used. In this technique, a photomicrograph of grain boundary intersections have been counted. known magnification is selected and a circle, One of the advantages of the circular intercept drawn on a sheet of plastic of diameter 10 cm or method is that it is easier to apply to deformed 20 cm, is placed over the photomicrograph. The structures. A nomograph fo r the graphical solu­ circle should intersect more than six grain bound­ tion of the Hilliard method is given in Figure 72. aries fo r the method to give a reliable grain size estimatingFigure 73. Typicalthe (austenitic) standard grain for samplessize of steel. carburized Photomicr at 17ographs00 of (930 of 0c) for 8 hours and slowly cooled toFrom develop Samans the (19 cementite63). Etch: network. nital; xIOO.

no. I grain size no. 2 grain size no. 3 grain size

no. 4 grain size no. 5 grain size no. 6 grain size

no. 7 grain size no. 8 grain size Grain Sizes of Ancient Metals 53

· estimatingFigure 74. Typicalgram. size . standard f annealed for • bnonferrousronze and materialsnickel silver.0 suc Actualh as brass x75.diameter of average g rain is noted.

0.010 mm 0.025 mm 0.035 mm

0.065 mm 0.090 mm 0.120 mm

0.150 mm 0.200 mm Grain Sizes of Ancient Metals 54

Figure 75, right. Mounting small specimensEmbedding in resin silicon used rubber here wascups. aboutBuehler 6-8 epoxidehours. resin which sets in mountedFigure 76, sample below. onGrinding wet silicon the carbide240--600 papers;grit are usually emplo papersyed. of Crain Sizes of Ancient Metals 55

Figure 77. right. Polishing the mountedMastertex sample. cloth is Here being Buehler used with andI-mi lappingcron diamond oil lubricant. spray abrasive Figure 78. below. Sample storage speciallyand selection purchased is best metallography carried out in storageobtained cabine in a varits. etyTrays of candiame beters formounts. storage Most of dioffferent the samplessize here epoxyare I 1/mounted4" diame samplester polyester that have or backlabels for embedded easy identification. in plastic on the Grain Sizes of Ancient Metals 56

Figure 79, right. Examination of metalspigments can or be corrosion best achie productsved by of micrpolarizedoscope light can micr alsooscopy. be used This for mountedreflected illumsamplesination can sobe thatexamined. Figure 80, below. Use of the micrinvertedoscope stage in whichmetallur thegical specimen stageis placed as illu onstrated top of thehere. micr Theoscope Epiphotmicroscope attached illustrated to a videois the Nikon camera,a 4" x 5" a color 35-mm or camera black-and-white back, and Polaroidreflected filmpolarized back. Darklight are field both and verymetallograph. useful features of this 12 METALLOGRAPHY AND ANCIENT METALS

Metallography offe rs one of the most useful 2. The thermal history ofthe object. Quenching means fo r the examination of ancient metals. It is and tempering processes may produce defi­ the study of polished sections of metallic materi­ nite changes in the microstructure that can be als using a special microscope that reflects light seen in section. passing through the objective lens onto the speci­ 3. The nature of the metal or alloy employed to men surface. The reflected light passes back make the object. For example, many debased through the objective to the eyepiece, which silver objects are made from silver-copper al­ enables the surface structure of the section ro be loys and both constituents are clearly visible srudied (see Fig. 80 fo r a typical example of a met­ in the polished and etched section; they show allurgical microscope). Reflectedlight micros­ up as a copper-rich phase and as a silver-rich copy is used fo r metallographic examination phase. Sometimes it may be very difficultto because metals cannot transmit light in thin sec­ obtain any idea of composition from looking tions in the same way as ceramic or mineral mate­ at the microstructure and here additional evi­ rials. Apart from gold fo il, which, if very thin, can dence must be obtained using a suitable ana­ transmit a greenish light through grain bound­ lytical method. One way of doing this if a aries, metals are opaque substances. sample has been taken and already mounted is It is usually necessaty to take a sample, which to trim the plastic mount and place the sam­ can be quite small, fr om the object being studied. ple in an X-ray fluorescenceanalyzer (XRF). I t is possible to polish a small area of a relatively 4. The nature of corrosion products. Much use­ flatsu rface on an object using a minidrill and fine ful and varied information can be obtained polishing discs, but this method is quite difficult. from corroded fragments or pieces of corro­ With many antiquities, useful information can be sion crusts. For example, there may be residu­ 3 obtained from samples as small as 1 mm , which al metallic structures within the corrosion can be removed from the object with a minimal layers, pseudomorphic replacement of metal amount of damage. Some methods of sampling grains by corrosion products, the existence of applicable to ancient material are listed in Chap­ gilded layers or other surface finisheswithin ter 13. The sample itself need not be metallic: cor­ the corrosion layers, pseudomorphic replace­ rosion products, paint layers, core material from ment of organic fibers or negative pseudo­ castings, niello, deteriorated enamels, patinations, morphic casts of fibers, unusual morphology and an array of other compositional materials are of the corrosion products themselves, and of considerable interest. Examination with a met­ changes brought about by methods of conser­ allurgical microscope is often a useful first step in vation. It should be noted that the scanning characterizing a particular component of an electron microscope is a very powerful tool fo r object. Some typical metallographic equipment is many of these investigations in addition to shown in Figures 75-80. optical microscopy. Metallography is, then, an important tool The principal difficultieswith metallography that may provide clues to the fabrication technol­ applied to ancient material lie in the problems ogy of the object or may assist in answering ques­ associated with sampling the object and the selec­ tions that arise during the treatment of an object tion of the sample. If any damage is to be caused, by a conservator. It may be possible to provide the owner should always be consulted firstso that info rmation on the fo llowing range of topics: necessary permission is obtained for the work. It 1. The manufacturing processes used to produce should be made clear to the owner whether the the object. For example, whether cast into a sample can be returned when the examination is mold or worked to shape by hammering and completed, and the kind of information that the annealing. sample is expected to provide. The owner should Metfillography findAnc ient Metflls 58

receive a copy of any written report that is pre­ mean that it is no longer necessary to remove as pared as a result of the examination, together with much sample bulk from the object as many earlier any photomicrographs of the structure, clearly investigations required. The extent of the damage identifiedwith the magnificationof the print, the is therefore greatly reduced; indeed, with much etching solution used, and the laboratory number corroded metalwork or fragmentary objects, the assigned to the sample. It may be necessary to loss may be insignificantand the resulting infor­ examine carefully a group of objects in the mation may be very important indeed. museum or laboratory before coming to a deci­ The only difficulty with taking small samples sion as to the number of samples to be taken or is the problem of the representative nature of the the objects from which it may be permissible to sample compared with the overall structure of the take a sample. If the damage is such that it can be object. It does not fo llow, fo r example, that repaired by fillingthe area concerned, the owner because the area sampled shows an undistorted should be given the option of deciding whether or dendritic structure (indicating that the piece was not to have the object gap-filled. Itis preferable to cast) that the whole of the object will reveal the fill small holes or missing corners with a synthetic same structure. Obviously there are cases in resin colored to match the surface color of the which this difficultywould be very unlikely to object. A fa irly viscous epoxy resin can be used fo r occur: worked, thin sheet-metal, pieces of wire, this purpose and can be colored with powder pig­ small items of jewelry, and so on, but with arti­ ments.1 facts such as axes, knives, swords, large castings, In many museum collections there are exam­ etc., it may present a very difficult problem. If a ples where large slices, pieces, or drillings have complete interpretation of the metallographic been taken from metallic antiquities causing gross structure of an artifact such as an axe is required, damage to the objects concerned. Missing parts then it is almost essential to take fWOsamples are sometimes filled with unsuitable materials from the axe (Fig. 81). such as Plasticine (normally a mixture based on Another example is that of a piece ofwire or putty, whiting, and linseed oil), which usually rod where both the longitudinal and transverse creates severe corrosion of exposed metal surfaces sections are of potential interest. Here the fWO over a period of years in the museum collection. sections can be obtained from a single sample by Sulfur-containing fillerssuch as Plasticine should cutting the wire into fWOpieces and mounting Figure 81, right. Drawing of an axe never be used fo r mounting, display, or gap-fill­ them in a mold (Fig. 82). showing the ideal location of two ing of metallic objects. Great improvements in The procedures fo r preparing metallographic samples removed for metallo­ metallographic techniques over the last 20 years samples are as fo llows: graphicone of the examination. cuts of the Note "Y" sectionthat principalshould be sides at right of the angles object to theto aid characteristicsinterpretation ofof anythe dimicrectionalro­ structure. �mold Figure 82, far right. Two samples of dirwireections or rod to mounted obtain structural in different JI cm wire: longitudinal/ section wire: \transverse section information relating to length and crossdifferent section. or di agnosticThere are featu oftenres visiblebe apparent in one in section the other. that may not Metallography andAncient Metals 59

1. Selecting the sample Note 2. Mounting the sample in a synthetic resin cast 1. For example, Araldite XD725 can be used with into a small mold good quality artists' powder pigments mixed 3. Preliminary grinding of the embedded sample into the resin and hardener mixture. Ifa missing 4. Polishing, using rotary discs impregnated piece is being gap-filled and it is required that with alumina or diamond powders the restoration can be removed at a later time 5 Examining the polished section with a metal- . then the metal surface can be coated in the area lurgical microscope to be filledwith a reversible synthetic resin such 6. Etching with a suitable etching solution as Paraloid B72 brushed on as a 10% solution in 7. Preparing a written report toluene applied in two coats. This should ensure 8. Doing photomicrography and drawings, if re­ that the fill can be removed, ifit is not keyed-in, quired, of the section to show microstructural by soaking in acetone fo r a few minutes. details 9. Employing an orderly system of sample stor­ age and documentation.

13 METALLOGRAPHIC SAMPLING OF METALS

There are often severe restrictions on the quantity provide X-ray fluorescence analysis fo r major of metal that can be removed from an artifact fo r constituents. metallographic examination. On the other hand, 2. A fine jeweller's saw may be employed. These 3 even a very small sample, smaller than 1 mm if are easily broken if they get wedged in the cur, necessary, can be mounted and polished fo r so a blade must be selected that is sufficiently examination, although great care has to be exer­ robust fo r the job at hand. It is then possible cised at all stages of preparation. It is much easier to remove very small specimens quite accu­ to work with larger samples, although by museum rately, often with minimal damage to the 3 standards samples of3 mm are already unusually object concerned. large, unless whole artifacts or substantial frag­ 3. A hollow-core drill bit can be used to remove ments are available fo r sectioning. a core that can then be mounted fo r examina­ There are. a number of criteria that samples tion. While only robust or thick-sectioned should meet: objects would survive this technique, it is sometimes the only way a metallographic 1. The object should be photographed or drawn sample can be taken. before the sample is taken. This is especially 4. A wafering blade can be used to cut a thin important if the dimensions of the object are slice or remove a small V-shaped section from fu ndamentally altered by the material the object. One suitable machine fo r this pur­ removed. pose is the Buehler "Isomet" diamond wafer­ 2. The microstructure of the samples should not ing machine with a cutting blade consisting of be altered in the process of removal. an oil-cooled copper disk impregnated with 3. The sample should be representative of the diamond powder along its circumference. object as a whole or of a selected feature or The principal problem with this machine is area of the object. the inherent difficultyof holding large or 4. The orientation of the sample in relationship irregular artifacts, although a variety of to the entire object should be carefully clamps are provided fo r this purpose. If an recorded and if it is not obvious where the object can be held securely and oriented in the sample was taken from, the position on the proper direction, then the machine is object concerned should be marked on a pho­ extremely useful. Not only can the length and tograph or drawing of the object. direction of the cut be precisely controlled, There are a number of possibilities fo r remov­ but the speed and weight applied to the cut­ ing a sample depending, inevitably, on the nature ting blade can be adjusted. The copper blade of the object and on the metal or alloy ofwhich it conducts generated heat to the oil bath so that is composed. the risk of any microstructural alteration is negligible. 1. A hacksaw with a fine-toothed blade can be Friable material may break along lines of used to cut large samples; however, high heat weakness, crystal planes, fractures, etc. and generated by friction during cutting can alter should not be cut using this machine, since the original microstructure. The blade should transverse pressure on the blade in motion be cooled periodically in water or ethanol. would result in cracking or chipping of the Removing a sample will usually entail consid­ diamond-impregnated cutting edge, which erable loss of solid as fine powder. This pow­ would be costly (approximately $300 each in der can be kept fo r analytical purposes 1989). although it may not be completely free from 5. With thin sheet or uneven edges it may be contamination. It can be used, fo r example, to Metallographic Sampling of Metals 62

possible to break off a small piece of metal by carefully gripping the area to be sampled with a pair oflong-nosed pliers, medical fo rceps, or fine tweezers. Since many ancient sheet met­ als are corroded or otherwise embrittled, the lack of plastic deformability of the material may assist in allowing the removal of a sam­ ple, since slight and carefully applied pressure may enable the removal of very small pieces. 6. A scalpel may be used under a binocular microscope. This technique may enable the removal of small pieces ofcorrosion products or metal fo r subsequent examination. It is occasionally possible to detach a specimen under the microscope by patient chiselling with a scalpel. Some deformation of the spec­ imen may result, but this zone can usually be eliminated in the grinding stage or, alterna­ tively, one of the undeformed sides polished. 7. A broken jeweller's saw blade held in a vibro­ tool can be used. There are a number of appli­ cations fo r this technique, which allows the removal of small pieces from hollow sheet work, lost-wax castings, cavities, etc., since the vibratory action of the blade in combina­ tion with a gentle sawing action is not as lim­ iting as using the blade rigidly held in its usual frame. A broken blade, about 3-5 cm in length, is satisfacrory. 14 MOUNTING AND PREPARING SPECIMENS

Once a sample is removed from an artifact, it 4. A disk of ash-toughened wax or a small quan­ must be mounted and prepared fo r examination tity of molten beeswax can be poured over the under a metallurgical microscope. Before the bottom of the mold. When set in position, sample is placed in a mold, it should be de greased the specimen can be partially embedded thoroughly in acetone or ethanol and dried. This before the resin is poured into the mold. is especially important if the sample has been cut When set and removed from the mold, the in a Buehler wafering machine. Very small sam­ specimens will stand in slight relief and will ples, pieces of thin wire, or thin sheet may require have to be carefully ground before polishing. physical support at the mounting stage, otherwise This technique is inadvisable fo r very brittle when they are embedded in resin and the block is materials. ground and polished, there will be no control 5. If the sample will be subjected to electrolytic over the metal surface exposed to view. There are or electrochemical polishing, it can be sup­ a number of ways in which this can be achieved: ported by a piece ofwire in the mold using the jig method detailed in No. 2. The wire can be 1. A small brass jig, specially made fo r the pur­ attached to the specimen using "aqua dag" or pose of b.olding specimens, can be used (Fig. some similar conducting paint applied in rwo 83a). The purpose of the jig is to allow the or more coats to effect good electrical contact. specimen to be attached to a toothpick with a The wire is cut to allow about 3-5 cm excess very small amount of acrylic or nitrocellulose above the top of the mold (Fig. 83b). adhesive. The attached specimen and stick 6. Larger specimens, especially if they have been can then be rigidly held in the jig and angled cut, can usually be mounted directly by placing or positioned as required in the mold. them in the required position in the mold. If 2. Similarly, the specimen can be mounted on a the embedding resin is then poured carefully, toothpick and then supported with more the specimen should not move in its position at toothpicks or matches laid across the mold as the bottom of the mold. shown in Figure 83b. It is, of course, possible to support more than one sample in the same Embedding the Sample mold using this technique, but the samples There are a number of proprietary synthetic resins are much more likely to move in the mold on the market that are designed as embedding when resin is poured in, and simultaneous materials fo r metallographic specimens. Epoxy, polishing of rwo or more very small samples is polyester, and acrylic resins are the most common. much more liable to result in the loss of the For routine use, "Scandiplast 9101" polyester embedded specimens once the block has been embedding resin can be used with peroxide catalyst ground and polished. "Scandiplast 9102." One drop per ml of catalyst is 3. Some texts suggest adhering the sample to the stirred into the resin, which begins to gel after bottom of the mold with adhesive or support­ about five minutes, and thus must be used ing the sample with a paper clip. These meth­ promptly. Fifteen ml of resin is sufficient to fill ods can be used but have disadvantages. For Metaservor Buehler 1-inch diameter or 1-118-inch example, the adhesive may result in slight diameter molds, which are made ofpolyethylene or unevenness in the resin layer surrounding the silicon rubber. Curing is best carried out at room edge of the sample with loss of good edge temperature; samples can be removed from the retention, or the paper clip may interfere with mold after two-four hours, but do not attain their the polishing of very soft metals if it is in con­ maximum hardness until about eight hours. Man­ tact with the resin surface. ufacturers state that the slightly sticky surface on Mounting and Preparing Sp ecimens 64

the set resin is due to oxidation reactions, which There are a number of ways in which this prob­ can be minimized by covering the mold while the lem can be tackled. One method is to use an resin sets. This, however, is rarely a serious problem epoxy mounting system that has low viscosity and the resin gives satisfactory edge retention and compared with polyesters and to add a filler of low porosity upon curing. specially graded black alumina granules. These With many archaeological materials, the are hollow alumina spheres whose hardness can major problem is to achieve good edge retention. be matched to that of the material to be mounted. Top,Figure a. 83. Use Holding of one smalltype ofsamples. __-- adjusting screw inmou a setnting position. jig to hold small samples Middle, b. Use of toothpick adjusting screw .. support over mold to attach to sample held with HMG or samples.Bottom, c. Use of wax embedding matchstick or toothpick .. other suitable cellulose or acrylic adhesive method;samples. only suitable for small silicon rubber mold� � brass mounting jig

wood bIOCk----.....� Usesmall of samples one type in ofa set mounting position. jig to hold � /'tooth picks n sample held with HMG adhesive .. J I or other suitable cellulose or acrylic adhesive toUse atta ofch tooth to samplespick supp ort over mold. � ::�:::�::::�"mpl" g OnlyUse of suitable wax embedding for small samples.method. Mounting and Preparing Sp ecimens 65

They also provide good edge retention, although laid on the resin surface next to one edge of the they are time-consuming to use. mold and the mold is then completely filled with If the metallographic sample is embedded at resin, embedding the sample without need of a an angle in the mold, some of the surface area of support stick. the sample will appear to be greater upon grind­ ing and polishing. This technique can be used to Grinding provide a magnifiedview of surface layers or other Once mounted and set, the resin block must be fe atures that might prove difficult to see in sec­ ground flat. The standard procedure at this stage tion. The sample, when embedded at an angle, is is to use wet silicon carbide papers with progres­ known as a taper section. sively finer grit sizes (120, 240, 400, 600). The Other ways of solving the problem include sample must be held so that it does not rock or electrolytic deposition of protective metallic coat­ move out of one grinding plane, otherwise it will ings, such as nickel, or the use of electroless dep­ be very difficultto obtain an optically flat surface. osition solutions. Care, however, should be Starting with the coarser grit paper, the sample is exercised when using these methods since a cor­ moved backwards and fo rwards over the paper roded sample is permeable to the penetration of until a uniform ground finish is obtained. It is applied layers and they will create a visually con­ then carefully washed under running water, fusing result. A label should be incorporated into examined, rotated 90°, and ground on the next the top of the resin giving at least the laboratory grade of paper. This process is then repeated with number so that samples can be properly identi­ the finer rwo grinding papers, rotating the speci­ fied. men 90° on each paper. It is very important to Another method of embedding very small eliminate completely the scratches from the pre­ samples, which is often used fo r paint cross sec­ vious grinding because they will not be removed tions, involves the use of a rectangular mold half­ by polishing. filled with resin (Fig. 84). The sample is carefully With very important specimens, a fresh strip samples.Figure 84. Embedding small

silicon rubber mold introduce samples and fill with more resin

half fill with resin and allow to set remove mounted blocks from mold Mounting and Preparing Sp ecimens 66

of grinding paper should be used, and successively paste, the sample should be washed in water, longer grinding times are necessary as the paper rinsed in ethanol or acetone, and dried. It can becomes finer. This ensures thorough removal of then be polished on one micron diamond fo r at deformed material at the specimen surface before least five minutes. For many routine purposes this polishing. The quality of 600 grit paper seems to is sufficient, and the sample should then be care­ vary, and if any large particles are present they will fully washed to remove all traces of polishing produce surface scratches difficult to remove compound and oil before it is ready fo r examina­ later. The care taken with metallographic prepa­ tion under the metallurgical microscope. For very ration of soft metals, such as gold alloys, zinc high-qualiry work, finish by final polishing on alloys, or some copper alloys, has to be much one-quarter micron diamond. more rigorous than is necessary with iron, steels, Electrolytic or electromechanical polishing or cast iron specimens in archaeological contexts. can be employed from the grinding stage and may give a more perfect finish than that obtainable Polishing from hand-polishing. Disadvantages occur with For most ancient metals, the best results are to be many archaeological samples because of the obtained by polishing on diamond-impregnated extensive corrosion that they may have under­ rotary polishing wheels lubricated with mineral gone; the presence of these corrosion products oil. Diamond powders are usually supplied as means that local anode/cathode reactions may tubes of paste or in aerosol cans in an oil-based occur with excessive sample dissolution as a suspension. The usual range of diamond powder result. If one phase of a two-phase alloy is slightly sizes are: six microns, ·one micron, and one-quar­ corroded it may be preferentially attacked at an ter micron. Some of the polishing can be carried accelerated rate. Mechanical methods are gener­ out automatically using a variery of machines or ally preferable unless there is a particular reason polishing attachments. Hand-finishing, however, fo r the use of other techniques. For example, is usually preferable fo r best results with one granulated gold spheres, difficultto etch, were micron or one-quarter micron diamond paste. polished and etched at the same time using an Polishing with diamond powders produces less electromechanical polishing technique on an rounding of surface details than is apparent when adaptation of the Struers "Autopol" machine. using a-alumina, y-alumina, or magnesium oxide A freshly polished section should be examined pastes. Alumina often can be used fo r less detailed metallographically as soon as possible: some sur­ work and is frequently satisfactory fo r the prepa­ faces tarnish rapidly and will have to be repol­ ration of iron and steel alloys. Magnesium oxide ished. Ancient metal samples should always be has to be used fresh as it undergoes atmospheric examined in the polished state to begin with since carbonation and the resulting magnesium car­ there may be many features already visible under bonate that is fo rmed may have a coarse crystal the microscope. fo rm. Polishing is carried out by holding the speci­ men against the rotating polishing cloth. It is dif­ ficult to quantifYhow much pressure must be used: too little pressure retards the rate of polish­ ing and may result in some pitting of the surface, too much pressure may distort the surface. The correct polishing pressure varies with different metals and can only be learnedthrough experi­ ence. After initial polishing on six micron diamond 15 RECORDING RESULTS

Visual evidence should be described, preferably 4. Grain sizes, or surface deformation features, with accompanying photographic records of the or heat-treated zones at cutting edges and microstructure at suitable magnifications. Inclu­ worked surfaces. sions or corrosion products thatare present when 5. The distribution of inclusions, weld lines, slag the section is examined in the polished condition particles, or porosity. should be noted because these will either be dis­ 6. The presence of any surface coating or gild­ solved or partially obliterated by etching. If nec­ ing. Sometimes careful examination at high essary, a photograph should be taken in the magnification is necessary to establish the unetched condition to show the range and type of presence of surface coatings, leaf gilding, inclusions present. They should also be examined amalgam gilding, etc. in reflected polarized light, which may assist in 7. The distribution of any corrosion products identifying the range and composition of non­ present and the existence within corrosion metallic material that may be evident. layers of pseudomorphic remnants of grain It is important to obtain an overall view of the structure or other microstructural features, specimen at a low magnification (about x30-x50) the presence of any remnant metallic grains, before proceeding to look at particular fe atures at and layering or unusual fe atures. higher magnifications. Some specimens will 8. Indications of grain boundary thickening or appear almost fe atureless before etching if the precipitation of another phase at the grain metal or alloy is uncorroded and relatively free boundaries. from slag particles, oxide inclusions, or other 9 . The presence of twin lines within the grains impurities. In these cases, it is permissible to pro­ and whether the twin lines are straight or ceed to an etched surface quite early on in the curved. examination. The details of the etching solution IO. The presence of strain lines within the grains. used should be recorded; it is not customary, II. Whether dendrites in cast alloys show indica­ however, to quote the time of immersion in the tions of coring and the approximate spacing etchant because the conditions of use and (in microns) ofthe dendritic arms, if these are strength of solutions vary from laboratory to lab­ clearly visible. oratory, making exact comparisons difficult.The 12. Do not fo rget that a polished section is a two­ magnification should, of course, be recorded with dimensional representation of a three-dimen­ any notes made about the structure, once detail sional object. If the structure is complex, as in can be observed. pattern-welded steel blades, fo r example, sup­ The range of fe atures that may be made visi­ plementary information, such as x-radiogra­ ble by etching is variable, depending on the type phy to reveal the internal pattern, will be of specimen examined. Details not apparent available. using one etchant alone may become visible after 13. The presence of intercrystalline or transcrys­ another reagent has been tried, so the assumption talline cracking in the specimen. that all microstructural detail is evident by using 14. Indications of grain-boundary thickening. one etchant should be resisted. 15. The presence ofsecond-phase precipitation at The fo llowing features should be noted: the grain boundaries (discontinuous precipi­ tation) or precipitation within the grains (as I. The range and type of grains present. Their in the case ofWidmanstatten precipitation). size can be compared with, fo r example, an 16. Evidence of martensitic transformation or eyepiece marked with grain sizes or with heat treatment used in the fabrication process. ASTM standard grain size numbers. 2. The presence of different phases. There may be, of course, other important 3. Gross heterogeneity or differences between structural details but this brief listcovers most of various areas of the sample. the fe atures that may be visible using the metallo- Recording Results 68

graphic microscope. Care should be taken not to confuse notes made on specimens under diffe rent conditions of polish or etching because it may be very difficult later to reproduce exactly the examined surface. Also, if careful note of the specimen number is not kept, it may become very difficult to establish which specimen had been examined. The same care should be applied to the photographic recording. 16 ETCHINGAND ETCHING SOLUTIONS

A satisfactory metallographic specimen fo r mac­ ide). Once the surface of the sample has been roscopic or microscopic investigation must exposed to the etching solution, it should not be present a representative plane area of the material. touched; it can be dried, after rinsing in ethanol To distinguish the structure clearly, this area or acetone, by placing it in front of a specimen must be free from defects caused by surface defor­ drier fo r a few seconds. Sometimes, at this stage, mation, flowed material (smears), plucking of the very obvious staining of the specimen occurs surface (pull-out), and surface scratches. In cer­ around the edges where it meets the resin mount. tain specimens, edges must be well-preserved in This is a consequence of the etchant seeping order to investigate gilding, tinning, etc. Even fo r down any fine crack that may be present and then routine examination, poor specimen preparation being drawn to the surface upon drying. Some can result in problems because the observation or specimens require considerable care if this prob­ conclusions may not be valid interpretations of lem is to be avoided. For example, it may be pos­ the visual evidence. sible to dry the sample more slowly in air, or to Polished metal surfaces do not reveal details of dry it using a stream of cold air. After the sample structure. In order to examine grain boundaries, has been etched, it is important to examine it other phases, and effects of alloying additions the immediately, as stored and etched specimens can polished metal surface must be attacked with rapidly become tarnished, and this obscures the selected chemical reagents that will reveal differ­ surface detail the etchant is trying to reveal. The ences in grain orientation and microstructure. sample, ifvery delicate, should be leveled in such There are three ways in which etching a polished a way that the impact of the leveler does not fall metal surface can be carried out: onto the etched surface of the metal. Damage to the surface structure can result if the leveler is 1. With chemical reagents either as a solution or depressed into soft alloys, such as gold and silver. as a gas. Sound metallographic principle dictates that 2. With electrolytic etching, by applying small all specimens should be examined in an unetched AC or DC currents to the specimen immersed condition before proceeding to the etched exam­ in an electrolyte. ination. This is partly because etching will often 3. With electromechanical etching, which is a dissolve out any inclusions that may be present combination of anodic dissolution in an (this would mean repolishing the entire speci­ electrolyte and mechanical polishing of the men), and partly because etching may not be nec­ speCImen. essary with some archaeological materials. If the specimen is two-phased or has surface enrichment This section deals with etching solutions that or other visible features, all this should be noted are appropriate fo r archaeological metals. The in the unetched condition before proceeding to polished metal surface must be free ofall traces of etching. oil from polishing, and of polishing compounds, grease, and particles of dirt. The usual procedure Etchants for iron, steel, is to pour a small quantity of etching solution into andcast iron a small petri or crystallizing dish so that the spec­ imen can be immersed fo r a few seconds or min­ Nital utes, as necessary. Some solutions need to be 100 ml ethanol, C2H50H freshly mixed before each use since they either 2 ml nitric acid, HN03 undergo very rapid chemical reaction (such as Picral solutions of potassium cyanide mixed with 100 ml ethanol, C2H50H ammonium persulfate) or they slowly deteriorate 2 g picric acid, C6H2{OH){N02)3 (such as solutions mixed with hydrogen perox- Etching and Etching Solutions 70

This is the most common etchant fo r wrought 24 g sodium thiosulfate, Na2S203 iron and iron carbon steels. Often useful fo r iron 3 g citric acid, CH2COHCH2(COOHh . and heat-treated steels, pearlite, and martensite. H20 Fe3C is stained a light yellow. Nital and pieral can 2 g cadmium chloride, CdCl2 . 2H20 be mixed together in 1:1 proportion. Use fo r 20-40 seconds. Chemicals must be Oberhoffer s reagent dissolved in the sequence given. Each constituent 500 ml distilled water, H20 must dissolve before adding the next. Store in the 500 ml ethanol, C .2 H ') OH dark under 20 °C. Filter before use. Solution 42 ml hydrochloric acid, HCI remains active fo r only fo ur hours. Used as a tint 30 g fe rric chloride, FeCl, etchant fo r iron and carbon steels. Ferrite is 0.5 g stannous chloride, SnCI2 (add HCl last) stained brown or violet. Carbides, phosphides, and nitrides may be only lightly stained. After etching, the section should be rinsed in With Beraha's reagent, fe rrite is colored but a 4: 1 mixture of ethanol and hydrochloric acid. cementite is not, so that both proeutectoid Useful fo r steels and fo r segregation studies in cementite and cementite in pearlite are in strong irons (e.g., arsenic segregation). contrast to the fe rrite, which can make even thin Hey n s reagent films of cementite easily visible. 20 ml distilled water, H20 Alkaline sodium picrate 20 g copper (II) ammonium chloride, 2 g picric acid, CGH2(OH)(N02)3 CuCI(NH4) 25 g sodium hydroxide, NaOH Copper precipitates must be wiped from the 100 ml distilled water, H20 surface with distilled water or washed off with dis­ This solution is useful fo r distinguishing tilled water fr om a wash bottle. Useful fo r phos­ between iron carbide and fe rrite in steels. It can phorus segregation in steel. be used as a boiling solution fo r ten minutes or Klemm s reagent longer if required. The iron carbide, Fe3C. is 50 ml saturated aqueous solution of sodium darkened by the reagent, while fe rrite is unaf­ thiosulfate, Na2S203 fected. Etching in this solution may provide a 1 g potassium metabisulfate, K2S205 good indication of pearlite lamellae spacing.

Phosphorus segregation in cast steels and cast Beaujards reagent Irons. 20 g sodium bisulfite, NaHS03 Baumann s print solution 100 ml distilled water, H20 100 ml distilled water, H20 To ensure that the specimen is evenly wetted, 5 ml sulfuric acid, H2S04 it is often a good idea to etch in nital fo r two to

Silver bromide paper is saturated with the fo ur seconds beforehand. Beaujard's reagent can solution and firmly pressed against the specimen then be used fo r 10-25 seconds; the surface should be very carefully washed and dried, other­ surface. After one to fiveminutes, rinse and fix wise the deposited surface filmwill be disturbed. with a solution of 6 g Na2S203 in 100 ml H20. Wash and dry. Useful fo r verification, arrange­ The reagent produces good contrast between fe r­ rite grains and between lightly tempered marten­ ment, and distribution of iron and manganese sulfide inclusions. site and fe rrite, as well as delineating cementite networks. The reagent works by depositing a Beraha s reagent complex oxide-sulfide-sulfate film on the metal 100 ml distilled water, H20 surface, in various shades of brown. Etching and Etching Solutions 71

Dimethylglyoxime nickel test Sometimes useful fo r variable carat gold jew­ Some ancient steels contain nickel as an elry alloys and Au-Cu-Ag alloys. important impurity. A simple test is to take a nickel print by pressing a blotting paper soaked in Etchant fo r tin dimethylglyoxime against the polished section Ammonium polysulfide when the nickel-rich areas are revealed by brown­ A saturated aqueous solution of ammonium staining on the blotting paper. polysulfide. Use fo r 20-30 minutes. Wipe off Vilella s reagent with cotton wool after etching. Can be used fo r 1 g picric acid, C6H2(OH)(N02)3 all types of tin alloys. Nital or picral can also be 100 ml ethanol, C2H50H used (see etchants for iron, steel and cast iron). 5 ml hydrochloric acid, HCI A reagent that can be used 5-40 seconds and Etchants for zinc reveals clearly the needles of plate martensite. It is Zinc alloys are difficult to prepare mechani­ useful fo r exposing the austenitic grain size of cally. Fake microstructures are common because quenched and tempered steels if this fe ature is at deformation is difficult to prevent. all discernible. Palmerton s reagent Wh iteley s method 100 ml distilled water, H20 5 g silver nitrate, AgN03 20 g chromic oxide, Cr03 100 ml distilled water, H20 1.5 g sodium sulfate, Na2S04 (anhydrous) or 3.5 g sodium sulfate, Na2S04 . 10 H20 A technique fo r revealing sulfide inclusions in steel and other alloys. Soak the polishing cloth in Can be used for seconds or minutes. the freshly prepared solution. The cloth is then 50% concentrated HCl in distilled water washed to remove all excess solution. The pol­ 50 ml conc. hydrochloric acid, HCI ished surface of the sample is then rubbed care­ 50 ml distilled water, H20 fully over the prepared surface of the cloth. Any This solution is sometimes useful fo r swab sulfide inclusions that are present should be application as well as immersion. stained a dark brown.

Etchants for lead alloys Etchants fo r gold alloys See ammonium persulfatelpotassium cyanide Glycerol etchant (p. 72) used fo r silver alloys. A very useful solu­ 84 ml glycerol, HOCH2CHOHCH20H tion fo r a wide range of gold alloys. 8 ml glacial acetic acid, CH3COOH 8 ml nitric acid, HN03 Aqua regia Use fresh only. Gives grain boundary con­ 40 ml nitric acid, HN03 trast. Nital can also be used (see etchants fo r iron, 60 ml hydrochloric acid, HCI steel and cast iron). Used fo r a few seconds or up to one minute. Use fresh. Aqua regia is a strong oxidizing solu­ Glacial acetic acid tion and highly CORROSIVE. In most alloys it 15 ml glacial acetic acid, CH3COOH provides grain contrast. 20 ml nitric acid, HN03 Useful fo r lead solders and Pb-Sn alloys. If Hy drogen peroxide / iron (III) chloride difficulty is experienced in the preparation oflead 100 ml distilled water, H20 alloys, a good technique is to try finishing the pol­ 100 ml hydrogen peroxide, H202 (l ishing with fine alumina powder or 1/4 32 g fe rric chloride, FeCl3 Etching find Etching Solutions 12

micron) suspended in distilled water with a 1 % 5 g potassium ferricyanide, K2Fe(CN)6 solution of aqueous ammonium acetate. 100 ml distilled water, H20

This etchant can be used to darken some pre­ Etchants for copper alloys cipitates such as CU3P which may coexist with Aqueous ferric chloride alpha + delta eutectoid. CU3P can be seen as a 120 ml distilled water, H20 putple shade against the very pale blue of the delta 30 ml hydrochloric acid, HCl phase. After etching in FeCI3, the distinction is 10 9 fe rric chloride, FeCl3 lost. Some difficulties may be overcome by using Produces grain contrast. Very useful fo r all a relatively high amount of aqueous ammonia in copper containing alloys such as the arsenical the final polish (e.g., 5% NH40H). coppers, bronzes, brasses, etc. Etching time is given as a fe w minutes in some books but ancient Etchants fo r oxide layers on iron metals etch faster. Reduce time to 3-5 seconds at Solution 1 first. 10 ml distilled water, H20 5 ml 1 % nitric acid, HN03 solution Alcoholic ferric chloride 5 ml 5% citric acid, 120 ml ethanol, C2HsOH CH2COHCH2(COOH)3 solution 30 ml hydrochloric acid, HCI 5 ml 5% aqueous thioglycollic acid, 10 9 fe rric chloride, FeCl3 HOCH · COSH Same as aqueous fe rric chloride, except that Swab for 15-60 seconds. Etches Fe203; there may be some advantages in using an alco­ Fe304 is not attacked. holic solution (less staining, for example). Solution 2 Aqueous ammonium persu/fote 15 ml distilled water, H20 and 100 ml distilled water, H20 5 ml fo rmic acid, HCOOH solution (a) 10 g ammonium persulfate, (NH4)2S20S 15 ml H20 and Only a few seconds are necessary fo r all these 5 ml fluoboric acid, F2B03 solution (b) etchants unless specified to the contrary. Pro­ Swab fo r five seconds with solution (a) fo l­ duces grain contrasts. Must be used fresh; will not lowed by solution (b) for two seconds. Etches keep. Fe304 only. Saturated solution of chromium (VI) oxide

Chromium (VI) oxide solutions should be Etchants fo r silver alloys handled with great care. Mixtures of Cr03 and Acidifiedpot assium dichromate organics may be EXPLOSIVE. Grain boundary 10 ml sulfutic acid, H2S04 etchant. 100 ml potassium dichromate, satutated K2Cr207 in water Ammonia/ hydrogen peroxide 2 ml sodium chloride, saturated NCI solution 25 ml distilled water, H20 25 ml ammonium hydroxide, NH40H Dilute 1:9 with distilled water before use. Can 5-25 ml hydrogen peroxide, H202 also be used without the sulfuric acid addition. Useful fo r silver and copper-silver alloys. Make up and use fresh only. Adding larger amounts of H202 creates better grain contrast; Ammonium persu/fote / potassium cyanide adding less H202 creates better grain boundary 100 ml distilled water, H20 and etching. 10 g ammonium persulfate, (NH4hS20S (a) 100 ml H20 and 5% potassium ferricyanide Etching and Etching Solutions 73

10 g potassium cyanide, KCN (b) POISON Acidified thiosuLfateI acetate 12 g sodium thiosulfate, Na2S203 . 5H20 Solutions (a) and (b) must be mixed before 1.2 g lead acetate, Pb(CH3COOh . 3H20 used in the proportion 1:1. After use flush down l.5 g citric acid, CGH 07 . H20 the sink with plenty of water. Never mix with S acids. The solution of potassium cyanide is POI­ 50 ml distilled water, H20 SONOUS. Also useful fo r gold alloys, silver Use after first pre-etching with ammonium alloys, copper alloys. Make up the persulfate solu­ persulfate. A very useful etchant fo r copper alloys tion just before use. but must be made up fresh as the solution will not keep. Dissolve the thiosulfate before adding the Ammonia I hydrogen peroxide other ingredients. Best made by placing the pre­ 50 ml ammonium hydroxide, NH40H pared solution in a refrigerator fo r rwenty-four 50 ml hydrogen peroxide, H202 hours when a clear lead thiosulfate solution Must be used fresh. should fo rm with a sulfur precipitate. Acidified thiourea Acidified suLfateI chromic oxide 10% aqueous solution of thiourea, SC(NH2h 50 g chromium trioxide, Cr03 5-10 drops of either nitric acid, HN03 or 5 g sodium sulfate, Na2S04 hydrochloric acid, HCI 4.25 ml 35% hydrochloric acid, HCl 250 ml distilled water, H20 Interference film etchants for color effects Interference etchants fo r iron, steel, and cast iron It is sometimes useful to investigate the applica­ tion of color metallography to ancient metallic 6 g potassium metabisulfite, K2S20s object specimens. One of the potential advan­ 100 ml distilled water, H20 tages to interference film metallography is that This etchant is useful for both carbon and coring and segregation may be revealed when alloy steels. conventional metallographic techniques fail. 2 g sodium molybdate, Na2Mo04 . 2H20 Color metallography may also be used to enhance 200 ml distilled water, H20 the visual appreciation of microstructural fea­ Acidify to pH 2.5 to 3 using dilute nitric acid. tures. Further details concerning the techniques Useful fo r cast irons after first pre-etching with which may be employed can be fo und in Yakow­ nital. itz (1970), Petzow and Exner (1975), and Phillips 60 g sodium thiosulfate, Na2S203 . 5H 20 (1971). Some of the useful recipes are set out 7.5 g citric acid, CGHS07 . H20 below. 5 g cadmium chloride, CdCl2 . 2H20 250 ml distilled water, H20 Interference etchants fo r copper alloys Useful fo r cast iron and steels without pre­ Acidified seLenic acid etching. Must be freshly made. This solution will 2 ml 35% hydrochloric acid, HCI not keep longer than a week. 0.5 ml selenic acid, H2Se04 300 ml ethanol, C2HsOH Saturated thiosuLfatesoLution This solution can be used after first etching 100 ml saturated sodium thiosulfate, . the polished sample fo r a few seconds in a 10% Na2S203 5H20 in distilled water solution of ammonium persulfate. Dty after etch­ 2 g potassium metabisulfite, K2S20s ing in persulfate before using the selenic acid Can be made up fresh and used fo r cast iron etchant. and steel alloys. Etching ({ndEtching Solutions 74

Selenic acid solution 10 ml 35% hydrochloric acid, HCI 4 ml selenic acid, H2Se04 200 ml ethanol, C2HsOH

Can be used fo r fe rritic and marrensitic steels.

Interfe rence etchants fo r aluminum alloys

Molybdate / bifluoride 2 g sodium molybdate, Na2Mo04 . 2H20 5 ml 35% hydrochloric acid, HCI 1 g ammonium bifluoride, NH4FHF 100 ml distilled water, H20 Can be used fo r etching both aluminium and titanium alloys.

Notes on use of these reagents fo r color etching and tinting of copper alloys

Pre-etching, which is sometimes necessary, should be carried out with 10% ammonium per­ sulfate solution, after which selenic acid etchant or thiosulfate/acetate etchant can be employed. In either case it can be difficult to fo llow the change of colors in very small samples. The pre-etched sample should be immersed in the interference etchant and the color change observed. The sequence of colors should be from yellow, orange, red, violet blue and finally bright silver. Etching should not be allowed to continue beyond the blue or violet stage. Chromic acid/sulfate etchant should be used without pre-etching and is useful fo r improving grain contrast. The thiosulfate/acetate etchant, fo r example, when used on an a-� brass results in the a grains turning light green, while the � grains become orange-yellow. With a 5% cast tin bronze, the dendritic a regions are colored blue, +yellow, b or brown depending on orientation; the a eutectoid can also be colored brown in the thiosulfate/acetate etchant (after pre-etching in ammonium persulfate). 17 MOUNTING RESINS

Mounting resins come in different types and eight minutes; it can achieve a workable hardness some manufacturers offe r several grades fo r diffe r­ after 45 minutes. The polymerization of the resin ent purposes. The most suitable are epoxy or can be accelerated by heating to about 45 °C. In polyester resins that have been specially adapted fact, the manufacturers recommend heating fo r metallography or mounting of small speci­ because extended hardening time at room tem­ mens. perature may result in some surface oxidation, Two suitable resins used by the author are which can lead to incomplete polymerization and given as examples, one an epoxy resin and the a lowered chemical resistance of the cured resin, other a polyester. although this is rare. Similarly, to avoid a tacky surface on the set resin, it should remain in the Epoxide mold until it cools down to room temperature. Manufactured by Buehler Ltd., 41 Waukegan This prevents oxidation reactions at the surface of Road, P.O. Box 1, Lake Bluff, Illinois 60044, the hot resin. Because of excessive heat genera­ U.S.A. tion, the resin should not be cured in a drying oven or any similar closed container. Epoxide 20-81 30-032, with hardener 20- The sample should be thoroughly cleaned and 8132-008, is a cold-mounting epoxy resin system degreased; this will aid in the adhesion of the which is excellent fo r almost clear mounts fo r polyester resin to the sample. Scandiplast 9101 metallographic samples. The resin adheres well to will cure fully after eight hours and can beground samples and shows a very low shrinkage rate dur­ and polished to give a perfectly adequate mount­ ing curing. It sets in about eight hours at room ing material. The manufacturer maintains that temperature. Although the setting time can be there will be a complete absence of air bubbles in inconvenient, it is better to wait fo r room-tem­ the cured resin, but this is not always the case. In perature curing than to speed up the process by any event, care must be taken when pouring the curing in an oven at a higher temperature; shrink­ resin into the mold so that air bubbles do not age would be more likely to occur. Buehler Ltd. become trapped on the surface of the sample, also manufactures a complete range of good qual­ since they will interfere with the production later ity metallographic equipment and many other of a scratch-free surface by retaining abrasive resi n types. particles. To ensure particularly good edge retention of Polyester the sample, a small amount of the resin can first Scandiplast 9101 UK Distributor: be made up and poured into the bottom of the (Hardener 1 % Scandiplast 9102) mold to fo rm a layer about 1-2 mm deep. Once Manufactured by Polaron Equipment Ltd., this layer has cured, the mold is then filled up in 60-62 Greenhill Crescent, Holywell Industrial the usual way. (For further details concerning the Estate, Watford, Hertfordshire, U.K. molds and sample preparation see Chapter 9, This polyester resin, which is green, can be p. 43.) The best molds are made ofsilicon rubber; used as an embedding resin fo r sample examina­ Polaron Equipment supplies Scandiform embed­ tion in reflected light or as an adhesive fo r some ding molds which are excellent fo r this purpose as applications. Scandiplast 9101 is used with a per­ the Scandiplast resin is easily removed from the oxide catalyst, Scandiplast 9102. One drop of cat­ silicon rubber surface. alyst is added fo r each milliliter of resin. Stirring Scandiplast 9101 is not suitable fo r vacuum should be carried out carefully and not violently impregnation, and if a vacuum-potting technique is so that air bubbles are not incorporated into the necessary a low-viscosity epoxy resin is preferable. resin. The polyester must be used within six to

18 MICROHARDNESS TESTING

Metals are rather different from minerals in that on the Vickers scale and are available as attach­ Mohs' scale of hardness cannot be used accurately ments fo r inverted stage metallurgical micro­ to assess the hardness of a metal. Usually, meth­ scopes, such as the Vickers metallurgical ods must be used that deform the metal in a hard­ microscope, or as attachments fo r stage micro­ ness-testing machine under a certain load, em­ scopes that use reflectedlight. ploying a specially shaped indenter to press into One such instrument is the McCrone low­ the surface of the metal. level micro hardness tester which can be attached An early method, avoiding the need fo r defor­ to the Vickers metallurgical microscope. Four mation at the surface, was to drop a steel ball onto readings of the hardness of a sample are taken, the the surface of the object and measure the distance sample must be mounted in embedding resin in a that the ball rebounded after impact. Although 1" or 1-114" mold and the surface must be pol­ this device, called the scleroscope, is now a histor­ ished, otherwise the impression of the indenter ical curiosity, some early archaeological reports may not be visible. refer to this method of hardness testing, thus its The exact relationship berween the macro­ mention here. hardness and the microhardness of a particular Brinell hardness testing (HB) was the first sample may be difficultto compare because of a method to find universal acceptance. With this number of complications that may arise. One technique, a steel ball ofknown diameter is fo rced example is grain size: if the microhardness inden­ into a metal surface under a certain applied load. tation falls across only one or rwo grains it may The diameter ofthe impression left on the surface not compare with a macrohardness reading which can be measured accurately and a scale of hard­ averages the result of deformation over many ness calculated. Subsequent industrial develop­ grains and grain boundaries. ments have brought other scales and methods into operation, such as the Rockwell method (HR), the Knoop method (HK), and the Vickers method (Hy). The only suitable scales fo r archae­ ological metals are the Brinell and the Vickers scales. The Vickers method utilizes a fo ur-sided, 136°, diamond pyramid indenter and the results of the scale are sometimes quoted as DPN num­ bers (from Diamond Pyramid Number). This is the same as the Hy, merely being a different abbreviation fo r the result. The Brinell and Vick­ ers scales are roughly the same, although fo r accu­ rate work they diffe r and a table of equivalents must be resorted to fo r some comparisons (see The MetalsHa ndbook, 9th Edition fo r the most recent information). Macroscopic scale hardness tests leave impres­ sions in metal that can be seen with the naked eye-in most ancient metals, sound metal is not fo und at the immediate surface and hardness test­ ing must usually be carried out on a micro rather than a macro level on a polished and mounted section using a special microhardness testing machine. Microhardness testers usually operate

APPENDIX A COMMON MICROSTRUCTURAL SHAPES

carbonFigure 85.steels. Shapes of ferrite in low­ terms.Figure 86. Common descriptive grain boundary allotriomorphs equi-axed hexagonal grains /\1\/\Widmanstatte/\n side plates annealing twin Widmanstatten sawteeth strain lines (slip lines)

idiomorphs mechanical twinning

asintr transgranular)acrystalline (also crack known intergranular Widmanstatten plates

intercrystalline (crack) massed ferrite polygonalo acicular

dendritic Appendix A - - 8· 0

banded lenticular

fibrous fusiform c::JoGCJ nodular

(-triple point botryoidal odiscontinuous p ,,,ipirn,, ��� � rhomboidal ��� angu lar fragments •martensltlc (needle-like)

co I u mnar grains APPENDIX B MICROSTRUCTURE OF CORRODED METALS

intergranular corrosion pitting corrosion

intragranular corrosion uniform corrosion through metal

Cstress-cracking �J corrosion selective corrosion ./1{ \. ] c�selective corrosion (parting) originalcorrosio sun rfproducace ts over the cavitation corrosion

remnant metallic grains in a slip lines outlined by corrosion mass of corrosion

twin lines outlined by corrosion disrcorrosionuption productsof the su rfaceand warty corrosion J APPENDIX C MICROHARDNESS VA LUES FOR DIFFERENT ALLOYS AND METALS

The values given in the tables below fo r the vari­ Metal or Alloy ation in micro hardness of some of the copper­ pure copper, cast 40-50 based alloys illum'ate the advantages of using tin pure copper, worked and annealed 50-60 bronzes fo r obtaining hard materials, The pure copper, cold worked 100- 120 increase in the Brinell hardness fo r arsenical cop­ 70:30 brass, annealed 50-65 per in the range of 1-3% arsenic is of some inter­ 70:30 brass, cold worked 120- 160 est, since the proportion of arsenic included in the gunmetal, cast 65-70 alloy beyond the 3% level has relatively little 1.8% arsenic copper, cast 48 effect on the ability of the alloy to be hardened 1.8% arsenic copper, worked 65-70 any fu rther by working. 2.6% arsenic copper, cast 65-70 This is in contrast to the tin bronzes where 2.6% arsenic copper, worked 150- 160 increasing amounts of tin produce successively cast leaded bronze, 10%5n 5%Pb 70 harder alloys when they are cold-worked. An even cast leaded bronze, 10%5n 10% Pb 65 greater effect is noticed when the tin bronzes are cast leaded bronze, 9%5n 15%Pb 60 compared with low-zinc-content brass and pure lead 3-6 cupro-nickel alloys, both of which are not greatly pure tin 6- 10 hardened by cold-working at the 50% reduction pure aluminium 14-22 level. These figureshave a practical importance. pure silver, cast 15-30 The color of the product upon casting may be silver, work hardened 80 strongly affected by inverse segregation of arsenic 20% copper 80% silver, annealed 45-50 to the surface of the mold but as fa r as srrength or 40% copper 60% silver, annealed 45-50 hardness is concerned there is little or no advan­ 60% copper 40% silver, annealed 65-70 tage in having 7% arsenic in the alloy compared 80% copper 20% silver, annealed 65-70 with 3% arsenic. It should be borne in mind from 12% tin bronze, fu lly worked 220 a practical point of view, that it is very difficult to 0.45% C steel, water quenched 546 compare the results of hardness tests on ancient 0.45% C steel, annealed 400 °C 418 alloys where diffe rent hardness scales have been 0.45% C steel, annealed pearlite 184 used. 0.55% C steel, water quenched 876 The scale used should always be specified and 0.9% C steel, water quenched 965 the same scale employed whenever possible fo r all 0.93% C steel, normalized 323 comparisons with similar materials in different 0.93% C steel, water quenched 836 degrees of working or composition. ancient file, tempered martensite 535 The fo llowing list gives some values or ranges ancient arrowhead, tempered martensite 390 fo r the microhardness of materials of interest in knife blade, tempered martensite 720 the Vickers 5cale (Hv), which is one of the most ancient saw, fe rrite 185 generally useful scales. mortise chisel, fe rrite 129 axe, pearlite and fe rrite 269 sickle, pearlite and fe rrite 17l Appendix C 83

Hardnesses of some imponant alloys on the Brinell Scale (HB) in the 50% reduction cold- rolled (i.e., worked) condition and in annealed condition.

Alloy HB 50% Reduction Annealed

Bronze 1% Sn 110 50 2% Sn 115 52 3% Sn 130 54 4% Sn 140 59 5% Sn 160 62 6% Sn 178 72 7% Sn 190 80 8% Sn 195 82

Arsenical copper I%As 118 50 2% As 138 51 3% As 144 52 4% As 146 53 5% As 148 54 6% As 148 55 7% As 149 56 8% As 150 58

Cupro-nickel alloys I% Ni 102 52 2% Ni 104 52.5 3% Ni 108 53 4% Ni 110 54 5% Ni 112 55 6% Ni 114 56 7% Ni 119 57 8% Ni 121 58

Brass alloys with low zinc content 1% Zn 102 51 2% Zn 104 5l.5 3% Zn 106 52 4% Zn 107 52.8 5% Zn 110 53 6% Zn 112 54 7% Zn 114 55 8% Zn 118 56 APPENDIX D ALLOYS USED IN ANTIQUITY

Copper Iron

copper-tin (bronze); also high-tin bronzes iron-nickel (meteoric/unusual ores) such as saffdruyand speculum) iron plus slag (wrought iron) copper-arsenic (arsenical copper or arsenic Iron-Carbon Alloys bronze) grey cast iron (free C) copper-antimony white cast iron (no free C) copper-zinc (brass) Steels copper-silver (shibuchi) hypereutectic (>0.8) copper-gold (tumbaga, shakudo) hypoeutectic « 0.8) copper-zinc-lead (bidri) iron-nitrogen (nitriding) copper-zinc-tin (latten, sometimes Pb) iron-hydrogen (embrittlement in steels) copper-cuprous oxide

copper-tin-arsenlc Lead

copper-nickel (cupronickel) lead-tin (soft solders, ancient pewters)

leaded bronzes, brasses, etc. Gold lead-gold (for gilding) gold-mercury (amalgams)

gold-copper (tumbaga) Tin

gold-silver (electrum) tin-lead (ancient pewters)

gold-platinum (sintered alloys) tin-antimony (modern pewters)

gold-iron (Egyptian rose golds) tin-copper (e.g., coatings, such as CuGSnS)

gold-lead (sometimes used fo r gilding)

Silver

silver-copper

silver-gold

silver-lead (sometimes from cupellation)

silver-silver oxide

silver on iron (example of gama hada) APPENDIX E TERMS AND TECHNIQUES IN ANCIENT METALWORKING

Joining Techniques Casting and Working Techniques

Mechanical Joins Casting

CrImpIng casting on overlapping seam lost-wax overlapping sheets solid riveting over a core tab-and-slot from a master matrix other refinements (e.g., mold carving) Metallurgical Joins centrifugal casting brazing open mold hard soldering piece mold fusing welding (2 parts locally heated) two-piece fu sion weld multi-piece filler of same composition in solid state slush casting fillerof same composition in liquid state Wo rking pressure welding (solid state) reaction soldering (as in granulation) annealing soft soldering chasing cold-working Surface Treatment Techniques drawing engraVIng Surface Coloration filing chemical solution treatments hot-working gold powder in binding medium machining pigments and dyestuffs turnIng drilling Surface Coatings boring amalgam gilding milling arsenic coatings pIercIng fusion gilding punching gold-iron alloys quenching gold or silver electrochemical deposition raISIng leaf gilding repousse platinum coatings spInnIng silver coatings over copper stamping tin coatings tempering

Surface Enrichment

depletion gilding of Au-Cu alloys depletion silvering of Ag-Cu alloys gravity segregation inverse segregation natural corrosive processes selective surface etching selective surface removal APPENDIX F METALLOGRAPHIC STUDIES

This section contains a miscellany of microstruc­ iron, and steel, and some examples of silver and tures with a limited amount of information con­ gold alloys. cerning the composition of the objects studied. TheMeta ls Handbook, Volume 9, published Most of the microstructures have been chosen by the American Society fo r Metals, includes because they are of good quality and illustrate a many microstructures relevant tothe study of wide range of fe atures seen in ancient metallic ancient metals and the 9th edition has been alloys, many of which have no modern equi­ expanded to cover precious metals, such as silver, valent. It would be ideal if the section included so that the text is more useful toa broader audi­ representations of the full range of microstruc­ ence. In order to make this appendix accessible. tures to be fo und in ancient and historic metal­ examples are indexed with regard totype of work, but this is velY difficult to achieve. The microstructure. alloy, object. and period or prov­ selection that fo llows is therefore somewhat arbi­ enance. These appear in italic type in the index at trary, but includes a good range of copper alloys, the back of this volume.

Silver-copper alloy coin This is a late and degraded fo rm of the drachm, whose origins may be traced back over 600 years to the Indo-Greek kings. This coin bears the king's head on the obverse side and a fire altar on the reverse. It is small, thick, and struck in a much-debased silver-copper alloy that was almost certainly surface-enriched by pickling, resulting in copper depletion at the outer surfaces. In the un etched condition, the rwo-phase nature of the coin can be seen, although it becomes much more evident upon etching in either potassium dichromate or fe rric chloride. The dendrites, which are silver-rich, are not con­ tinuous, indicating a rather low silver content, about 18%, analogous to other struck silver­ copper alloy coins. The silver-rich phase is elon­ Figure 87. Base silver-copper alloy gated and flattened along the length of the coin. coin of Skandagupta. Western Also, the contours fo llow the depressions in the India. c. A.D. 470. I I mm. surface suggesting the coin was struck while hot. The corrosion is quite limited and does not Figure 88, top. Low magnification showing part of the extend much below the surface into the interior of struck surface. Note the curvature in the structure as a the alloy. result of deformation. Etch: FeCI3; x32. Figure 89, bottom. The copper-rich grains can now be phaseseen with passing the alongflattened the remnantslength of theof the coin. silver-rich Note the discrdendriticete elo remnanngatedts plates. are not From continuous similar cast but and occur worked as aboutsections, 15 -20%.the silver Etch: content FeCI3; can x 12 be3. roughly estimated at Appendix F 87

Islamic inlaid inkwell cast in a copper­ tin-zinc-Iead alloy The inlay on this fine inlaid inkwell has been car­ ried out in silver, copper, rock asphalt, and prob­ ably, ground quartz. It was made in the eastern Iranian province of Khurasan, probably in Herat during the period from the late 12th through early 13th century A.D. The inkwell measures 103 mm high (with lid) and has a diameter of approximately 80 mm. The composition of the metal used in the manufacture was fo und to be: copper 65%, tin 5%, zinc 20%, lead 10%, with Figure 90, above. Islamic inlaid smaller amounts of iron and silver. inkwell from the eastern Iranian A sample taken from the rim of the body of province of Khurasan. the inkwell was removed using a fine jeweller's Figure 91, top right. Coring is saw. Although there are fo ur major constituents, evident, as are some globules of the alloy can be considered a ternary mix of cop­ lead. The larger dark holes are per, zinc, and tin because lead is immiscible in isporosity difficult into the interpret alloy. The fully structure because copper. There is a random scatter oflead globules of coring. Some strain lines are visible in the polished section. The section etches visible (near bottom right). Etch: well with alcoholic fe rric chloride and reveals a FeCI3; x63. cored structure to the grains showing that the ink­ Figure 92, middle right. Overall well has been cast to shape. The inner areas of view of the section at x21 etched each grain are richer in cop�er and the outer areas in alcoholic FeCl3 showing coring, richer in zinc and tin. Also evident in this section lead globules, and complex are a number of holes that are due to some poros­ structure. ity present in the casting. The prominent coring Figure 93, bottom right. Some of shows that the inkwell was not extensively worked the second phase areas may be or annealed fo llowing fabrication. However, gamma phase from the copper-tin­ some strain lines can be seen near the surface sug­ zinc system which appears as light­ gesting some final working in the as-cast condi­ coloredlead globules regions appear in the dark picture. grey. The tion, probably due to chasing of the surface to Etch: FeCI3; x41. receive the inlaid metal strips (see Plate 1 1). Appendix F 88

Cast bronze arrowhead This arrowhead is from Palestine and was exca­ .." ...'!O. vated by Sir Flinders Petrie between 1930 and : .. . �: .....: ...... 1938 from the site at Tell Ajjul. During his exca­ vations Petrie fo und many artifacts associated Figure 94. Drawing of cast bronze with buildings and graves dating to the early arrowhead from Palestine. XVIIIth Dynasty. This arrowhead, as well as the Palestinian bronze sword (below), were fo und during the excavations. The composition of the arrowhead is a ternary copper-tin-arsenic alloy with some lead present (Ghandour 1981): Sn 9.86%, Pb l.49%, As 3.87%, Zn 0.85%, the remainder copper. The arrowhead is 30 mm in length and triangular in shape with three fins and a hole fo r the socket on the tang. The microstructure of the arrowhead can eas­ ily be observed after etching in fe rric chloride fo r 3-5 seconds. The central region of the arrowhead as well as the three fins, which are seen here in sec­ tion, all have the same undistorted, cast, dendritic Figure 95, top. Fine, very long, and undistorted copper­ pattern. This is clear evidence that the arrowhead rich dendrites with eutectoid infill of a. + I)phases was cast in a mold directly to shape and that the between the arms. Note the occasional interdendritic fins were not beaten out or work-hardened after pores. Etch: FeCI): x32. manufacture. At low magnification, very long and Figure 96, bottom. The dendritic structure is clear and finely developed dendrites can be seen together the occasional a. + I)eutectoid areas are more resolved with some of the interdendritic eutectoid infill. at this magnification than at x32. Etch: FeCI): x65. The eutectoid itself is difficult to resolve, some structure becoming apparent at x600.

Figure 97. Drawing of Late Bronze Age sword. Etch: FeCI). Palestinian bronze sword This sword is dated to the Late Bronze Age. It has .. an intact tang rectangular in cross section and a ===:r:� blade about 285 mm long and 31 mm wide at the maximum width. There is a pronounced midrib � (see "Cast bronze arrowhead," above, fo r a brief background reference). in alcoholic fe rric chloride reveals the grain struc­ The sword is composed of Sn 8.20%, Pb ture to be rather variable. Toward the cutting 0.27%, As 0.66%, Zn 1.72%, the remainder cop­ edge of the sword the grains appear quite large, per. In the unetched state, some grain boundaries possibly from over-annealing during some stage are evident, as well as quite large but irregularly in its working. Toward the midrib, the grain size distributed pores. There are no visible sulfide or is smaller but still rather variable in size. Some of oxide inclusions present in this section. Etching the grains are clearly twinned with straight twin Appendix F 89

showingFigure 98, in tercrystright. Overallalline cracks,view thelarge section. grains, Unetched;and some porosityx30. in grainFigure structu 99, far re right. with Recrystallized twinned crbands.ystals Etch: and corrosionFeCI3; x62. along slip

lines, showing that the sword was finished by a lining many slip bands in some pans of the srruc­ finalannealing operation. The dark, thick lines ture. The extenr of these slip bands suggests that that oudine most of the grain boundaries are due the finalannealing operation was not sufficienr ro ro corrosion advancing through the alloy in an remove the srresses within all the grains as a resulr inrergranular fo rm of attack. The copper has also of working the sword ro shape (see Plate 17). corroded along slip planes within the grains, out-

Roman wrought iron This sample is a section fro m a large Roman billet (bar) of iron about 25 x 7 x 7 cm and weighing about 6 kg. It represenrs the primary fo rging of the bloom. During this stage of production most of the incorporated slag was squeezed out and the iron was consolidated in preparation fo r rrade or manufacrure inro fu nctional anifacrs. As with the bloom, this iron can have varying degrees of carburization. Forging can remove or inrroduce carbon depending on the area of the fo rge ro which the iron is exposed. Near the tuy­ Figure 100. View showing variable grain size of the ere (blast of air), conditions are more conducive wrought iron with extensive slag particles and some ro oxidization and carbon is lost from the struc­ largermaterial slag and fra gments.typically Noteconsis thatts of the wustite slag is (FeO)a two- phasedas ture. Variable carbon conrenr can be advanra­ rounded dendritic shapes in a glassy matrix. Etch: nital; geous when iron is fo rged, since fe rrite regions x83. (low-carbon areas) are malleable and easily worked, while peariitic regions are rougher. Appendix F 90

Gold-copper alloy sheet This fragmenr is a circular disc approximately analyzer revealed a composition of about 80.6% 110 mm in diameter. These gold-copper alloys copper, 17.1 % gold, and 2.6% silver. are usually referred to as tumbaga alloys. They The polished section shows a typical worked were usually finished by a depletion-gilding pro­ sheet with small grains and a lot of inrergranular cess. This particular fragment has deep-yellow corrosion. The grains can be etched, with diffi­ colored surfaces with some patches of purple­ culty, in an ammonium persulfate and potassium brown staining. A circular hole has been pierced cyanide mixture that shows the nature of the through the sheet which shows some fine surface gold-enriched surface as well. The grains are Figure 10I. Depletion-gilded disc of scratches in the metal surrounding the hole. twinned with most of the twin lines being tumbaga alloy showing matte gold These scratches probably arose as a result of fin­ straight, but a few are nor. There are some cuprite and burnished gold surface design ishing the pierced hole when metal burrs were inclusions as well as cuprite lamellae running made by surface-etching the removed with an abrasive. Analysis of sound along the length of the section which also appears enriched gold after depletion­ metal grains with an electron microprobe to be very even in thickness (see Scott 1983). gildinPupialesg. From in the the Department Municipio of probablyNariiio, southern dates from Colombia, the Piartal and AD.period, about I 0-12th century

suFigurerface 10of2. the The worked depletion-gilded sheet is clearly structure evident. on theThe andFigure the 10 small3. Note globules the very of cuprite even thickness that can beof theseen sheet whileetchant etching has left the the grains gold-rich which sur areface now largely visible un asattacked recrys­ elonfollowinggated somelamellar of theregions grain are boundaries. caused by The some dark, tallizedphased. withEtch: straight KCN/(NH4 twin hS20S;lines. The x200. alloy is single­ corrosion to cuprite during burial. Unetched; x64.

Ecuadorian copper-alloy nose-ring The object is one of several metal finds from the of a two-phased silver-copper alloy. secondary chimney urn burial at La Compania, There is a very clear demarcation between the Los Rios province, Ecuador. It is of intermediate applied silver coating and the copper body of the size in the collection of nose ornaments from this nose-ring. The polished section was etched in burial of the late Milagro phase. The section alcoholic ferric chloride, resulting in a well­ through the nose ornamenr shows that the copper fo rmed twinned crystal structure with perfecrly incorporates a considerable amounr of cuprous straight twin lines in the copper grains. The oxide inclusions that run along the length of the grains near the cenrer of the nose-ring are quite object in a striated fo rm, typical of worked cop­ large with prominenr twinning (ASTM grain size per. The nose-ring has on its outside surface a number 2-3), while on the outer surfaces of the very thick (about 0. 1 mm; 1000 microns) coating nose-ring, which have received slighrly greater Appendix F 91

hammering and recrystallization during manufac­ enriched, or depletion silvered, as is apparent ture, the grain size drops to about ASTM 4-5. from a metallographic examination of the intact The silver coating is a duplex structure con­ surface. This is not uniformly apparent on the sisting of copper-rich islands of alpha phase with section taken because some has broken away dur­ an infilling of the copper-silver alpha + beta ing burial as a result of corrosion or during prep­ eutectic in which the eutectic phase fo rms a con­ aration fo r metallography during the cutting tinuous network throughout the alloy. The junc­ stage. The observation that the silver alloy coating tion between the applied silver coating and the has itself received only slight working and has, copper base is quite abrupt, and the silver coating essentially, a slightly modified cast structure, has not received anything like the degree of defor­ means that it must have been applied hot, i.e., in mation and recrystallization to which the copper a molten condition, probably by dipping the has been subjected. I t is quite apparent from this cleaned nose ornament into a molten silver-cop­ fact, and from the very narrow diffu sion zone per alloy. The manufacturing process can there­ between the silver alloy coating and the copper, fo re be almost entirely reconstructed from the that the silver alloy must have been applied at a metallographic evidence. velY late stage in the manufacturing process, i.e., I. A raw, cast copper blank, which contained when the nose-ring was practically completed. The copper-rich alpha phase in the silver alloy extensive cuprite inclusions as large globules, coating betrays occasional slip lines and a few was worked and annealed in cycles to shape bent twins are also evident. This suggests some the nose-ring. 2. Oxide scale was removed fr om the surface to relatively light working and annealing fo llowed by some cold-working of the coating after appli­ present a clean surface fo r coating. 3. An alloy of silver and copper was made up cation to the shaped copper base. The silver coating has itself been superficially containing about 30-40% silver, which melts at approximately 200°C below the melting Figurenature 10of 4,the right. silver-copper View showing alloy the point of pure copper. The alloy had to be held coating over worked and in a suitable container whether dipping or recrystallized grains. A few bent surface brush coating was employed. twins in the alpha region of the 4. The copper nose-ring was dipped or coated silver-copper alloy coating suggest with the molten silver alloy. some light working following the 5. The nose-ring received some final working to application of the coating to the shaped nose-ring. Etch: FeCI); smooth out the silver coating and the surface x120. was cleaned by an acid plant or mineral prep­ aration to remove oxidized copper and to cre­ Figure 105, far right. Lower ate a good silvery appearance. twinnedmagnification grains view of thein whichbase thecopper aresilver-copper clearly revealed.alloy su rfTheace thickcoating chlorideis well-contrasted etchant. x78.by the ferric Appflldix F 92

Cast arsenical copper axe Simple axes of this kind were common, not only clearly evident upon etching. The fact that the axe in many parts of the New World but also has a recrystallized grain structure with twin lines, throughout the Old Wo rld, and the methods of as well as coring, and virtually undisturbed poros­ production and manufacture are essentially the ity, indicates that it was finished by a working and Figure 106. above. Small copper same. Many of these axes were made by casting annealing process, probably to improve the shape axe from Guayas. Ecuador. 70 mm. into simple, open molds. The molds themselves of the axe rather than to work it extensively. The Figure 107. top right. The structure could be made of stone, clay, or metal. It is some­ grains have straight twin lines indicating that the is surprisingly clean. There is the times possible to obtain evidence for the position final process was either hot-working or an anneal­ expected porosity of a cast metal of casting by metallographic examination. For ing operation. but no oxide or sulfide inclusions. example, slag and dross may be segregated to the The achievement of casting the copper alloy The subtle shading passing across top surface of an open-mold casting with the into a mold and making an axe like this with no orthe the grains casting is due that to persists residual despite coring result that one side viewed in section looks quite apparent oxide or sulfide inclusions present dem­ the twinned grains which are clean while the other may contain more oxide onstrates considerable metallurgical skill. clearly visible. Etch: FeCI3: x32. inclusions and gross impurities in the fo rm of debris or slag. This is not the case with this axe, theFigure twin 10 lines8. bottom are straight right. Note showing that although we know from other examples that axes that an annealing process was were cast horizontally into one-piece molds. probably the final stage in manu­ Bushnell (1958) believes that the axe is from facture. The relatively large grains. the Manteno period. Spectrographic analyses the absence of strain lines. and showed that the axe was made of arsenical copper theundi amountstorted ofporosity working indica waste not that with about 1 % arsenic content. The other ele­ extensive. Etch: FeCI3: x65. ments detected were present only in very small amounts (all less than 0.05%). These were anti­ mony, bismuth, iron, lead, manganese, nickel, and tin. It is difficult to cast copper into open molds wi thou t extensive oxidation occurring and yet the microstructure of this axe is remarkably clean: there are no cuprous oxide inclusions to be seen. The axe has been cast-that, is evident from the pattern of distribution of porosity as well as from the coring resulting from dendritic segregation. Coring occurs as subtle patches of darker etching material that ignores the grain structure and is 93 Appendix F

Chinese bronze incense burner The fragment mounted fo r study shows a cross section through the lid of the incense burneLThe sample shows an undistorted cast structure with a fa ir amount of delta tin-rich eutectoid.

Figurebe clearly I 10 seen.. The The outline a + of8 eutectoidthe equi-axed phase structure occupies can the graininclusions boundaries scattered for through the most the part. section. with Thissmall structure cuprite annealing.is unusual forEtched a normal in alcoholic cast bronze FeCI3; and x98. ind icates ancestralFigure 10 9.incense Chinese burner. 19th centuryusually placedHeight on19 0the mm. ancestral table.

Thai bronze cast bell A bronze bell from Ban Don Ta Phet (see Rajpi­ tak 1983) with a smooth surface and dark green patina. cast by the lost wax process. The compo­ sition is Cu 85.6%, Sn 3.5%, Pb 15.5%, As 0.1 %, Ag 0.3%, Ni 0.1 %, Co 0.01 %. The micro­ structure shows the bell was cast in one piece. It has a coarse dendritic structure with extensive porosity which is evident in the section.

extensiveFigure I I I.corrosion. Note the especially large. coarse toward dend therites. bottom. the and areas.the segregation Etch: FeCI3; of thex35. lead and tin to the interdendritic Appmdix F 94

Luristan dagger handle The materials used to make the handle and blade were well-chosen. The tin content of the handle is about 20%, while the hilt of the blade has a tin content of about 13.5%. The casting-on of the Figure I 12. Fragment of a Luristan blade from Iran: Cu handle over the finishedblade involved the use of 76.6%. Sn 20.8%, Pb 0.9%, Ni 0.5%. The handle was cast­ a clay mold built up around the blade into which on in a tin-rich bronze; analysis of the dagger hilt yielded: molten bronze was poured. Cu 83.6%, Sn 13.5%, Fe 0.3%, Pb 0.8%, As 0.4%, Ni 0.6%. It is important that the alloy used in casting­ Length 89 mm. on has a lower melting point than the metal to be incorporated in the cast-on product. This was fa cilitated in this case by the careful selection of materials. The blade has a worked microstructure, but the structure of the hilt incorporated into the handle is an annealed casting. The hilt is equi­ axed compared with the heavily dendritic struc­ ture of the handle. The redeposited copper that occurs in the crevice between the rwo components may have been caused by low oxygen availability in corro­ sive burial environments. That sort of situation Figure I 13. The left side of the photomicrograph is the would be ideal fo r this fo rm of corrosion-redepo­ hilt of the blade, while the region on the right is the cast­ sition phenomena. A slight potential gradient on handle of the dagger. The irregular strip between the could also be set up berween the rwo components two regions is redeposited copper resulting from themselves, that is, berween the hilt and the blade. corrosion of the bronze. This is the result of the Redeposited copper also replaces delta, tin-rich corrosion of the ex +8 eutectoid phase in the higher tin regions in the hilt (see Plate 10 and Figs. 69-70). content alloy of the handle. x90.

Fragment of a Thai bronze container This fragment is part of a circular container with Sn 22.19%, Pb 0.3%, As 0.03%, Sb 0.01%, a polished outer surface and a rough inner sur­ Fe 0.03%, Ag 1.2%. face. Also, the alloy is corroded on the surface, Examination of the container reveals a well rim, and part of the body. The diameter is developed dendritic structute with an infilling of approximately 3.4 cm and the thickness of the a matrix that appears superficially to be scratched. metal about 1.5 mm. However, this structure is due to the development The object is one of a group of metal finds of a martensitic structure in the bronze as a result from Ban Don Ta Phet, a village in the Arnphoe of quenching the bronze alloy fo llowing casting. district of Phanom Thuan, Kanchanaburi Prov­ The quenching procedure prevents the bronze ince, southwest Thailand. The date of the site is alloy (which is high-tin) from decomposing into thought to be about 300-200 B.C. The composi­ the usual alpha + delta eutectoid structure. This tion of the alloy was determined by inductively­ was carried out as a deliberate process in the man­ coupled plasma spectroscopy as eu 70%, ufacture of the container with the result that the Appendix F 95

Figure 114, top right. Cast dendritic structure of cored alpha grains with delra-phase formarion is suppressed and, insread, an infill of beta-phase needles. a bera-phase marrensire grows due ro rhe reren­ There seem to be two types of rion of rhe high remperarure bera phase. The needles present: fine martensitic microsrrucrure makes ir possible ro reconsrrucr needles and thicker lenticular rhe manufacruring process of rhe container. The needles. Etch: FeCl3: x62. objecr was firsr casr in a mold. Afrer casring, ir was Figure I 15, bottom right. Enlarged briefly annealed and rhen quenched. The remper­ view of the same area in which the arure ar which rhe quenching was carried our is beta martensitic structure is clearly esrimared ro have been 586-798 0c. visible. The needlelike martensitic phase is designated � I and the The needlelike marrensiric phase is designared dendritic copper-rich phase is the as bera, and rhe dendriric copper-rich phase is rhe alpha. Microhardness measure­ alpha. Microhardness measuremenrs were made ments were made and gave results and gave resulrs of 188 Hv for rhe dendriric phase of 188 Hv for the dendritic phase and 227 Hv fo r rhe bera phase. andEtch: 227 FeCl3: Hv forx250. the beta phase.

Colombian cast Simi ear ornament This rype of fine filligree ear ornamenr was origi­ nally rhoughr ro be made by rhe soldering rogerher of a number of wires. However, merallo­ graphic examinarion has proven rhar rhey were made by rhe losr-wax process. The secrion cur rransversely rhrough rhe ear Figure I 16. Part of an ear ornament ornamenr shows rhe casr narure of rhe gold-cop­ from the SinG zone in north­ per alloy. western Colombia. 42 mm. The loop fo r suspension is slighdy uneven ar rhe rop, as is common in many orher ear orna­ Figure I 17. The orientation of the corroded copper­ menrs from rhis area, showing rhar rhe probable rich region of the cored dendrites reveals the original casring posirion was wirh rhe fu nnel and channels castthis arestructure difficult of to the etch gold-copper satisfactorily alloy. with Structures either KCN/ like siruared above rhe suspension loop, arrached ro ir (NH4hS20S or HN03/HCI. FeCI3 alone will only etch ar rhe uppermosr poinr. copper-rich gold alloys with up to about 30% gold by The polished secrion shows an undisrorred weight. Unetched; x36. casr dendriric srrucrure in which considerable corrosion has raken place. An esrimarion of rhe dendriric arm spacing gave a value of 33 microns averaged over nine arms (SCO(( 1986). Analysis yielded: Au 16.9%, eu 71.2%, Ag 1.7%,= Pb 0.07%, Fe nd, Ni nd, Sb nd, As nd (nd nor derecred). Appendix F 96

Figure I 18, top right. Overall view showing the large graphite flakes Cast iron cannonball from withwhite, infilling unetched of pearlit ternarye and phosphide the Sandal Castle eutectic, which cannot be readily This polished fragmenr is a section from a can­ resolved at this magnification. Etch: nonball used in the siege of Sandal Castle in 1645 nital; x70. between July and September ofthat year. One of the reasons fo r examining these cannonballs is to pearliteFigure I with19, middle variations right. in Fine the size see if there are any differences between the 64- of the graphite plates. Etch: nital; pound balls brought from Hull specially fo r the xISO. purpose, and the 32-pound cannonballs used ear­ lier in the campaign. The structure of both types, Figure 120, bottom right. The in fa ct, proved to be the same. (Further details ...... "'. pearlite is clearly visible. Some ,." ...... pearlite regions have very fine concerning the history and the cannonballs them­ spacing and appear grey. The white selves can be fo und in Mayes and Buder 19 77.) phase on the left is the phosphide Examination of the sample shows the cannon­ eutectic, steadite, while carbon, in ball to be made of a fa irly typical grey cast iron. .J -""'.\ theas dark form lin ofes.graphite Etch: nital; flakes, x390. appears The structure consists of rather variable-sized but \S' , quite large graphite flakes set in a matrix that is \. �� principally pearlite. Some of the pearlite eutec­ :ft,.'. toid is very finely spaced and is barely resolvable at xl 000 magnification. Much pearlite can clearly be seen, however, and there are a fe w areas that look fe rritic. In some places, a white phase can be seen that has a patchy appearance with small globular holes. This is a ternary phosphide eutec­ tic between fe rrite (usually with a little phosphide conrenr), cemenrite, and iron phosphide, Fe3P' This ternary eutectic, called steadite, has a melt­ ing poinr ofabout 960 °C (Rollason 1973) and so it is the last constituenr to fo rm as the cast iron cools down. It is a very brittle phase, but is usually only present in small amounts scattered as iso­ lated islands (see Plate 13). Appendix F 97

Bronze Age copper ingot The small plano-convex ingot illustrated here was the copper produced would need refining or sectioned. It is typical of the convenient products remelting and alloying before being used to pro­ used fo r trading cast copper. The cut face shows a duce small artifacts because of the unusually high columnar appearance, which suggests directional cuprite content, otherwise the grain boundaries solidification. XRF analysis detected a small of the material would be very weak. Figure 121, above. Copper ingot amount of nickel as an impurity in otherwise pure found in association with Bronze copper. Age material in Hampshire, The microstructure shows a remarkable scat­ England. 27 mm. ter of cuprous oxide inclusions. Grains are large (ASTM 1-2), with no annealing rwin or slip lines Figureshowing 12 the2, right. large Overallgrain boundaries view present, indicating that the copper was cast. The and the scatter of cuprite globules purity of the copper explains the absence of cor­ at the grain boundaries and within ing and visible dendrites. The large and numerous the grains. Note also the presence globular oxide inclusions occur both within the theof some absence porosity of any in twin the inlinesgot orand grains and as slighdy lenticular shaped discs that strain lines in the copper grains. run along the grain boundaries. There is some Etch: FeCI3; x I 05. porosity present. With this particular cast ingot,

Roman brass coin Many Roman coins were made from brass rather than bronze. The microstrucrure of this coin, etched with fe rric chloride, shows a typical recrys­ tallized grain structure with extensive evidence of deformation produced in the process of striking the coin. Both rwin lines and slip lines can be seen in the sections. Many fine sets of parallel slip (or strain) lines can be seen, both at x66 and x132 magnification. Some of the twin lines are more difficult to observe, and there is some variation in their appearance. Many of the twin lines seem quite straight, while others seem curved, indicat­ ing either hot-working or some cold-working after the final stage of recrystallization. Since the usual manufacruring technology involved strik­ ing heated coin blanks, there may be a dual char­ acter to these type of structures. isFigure from 12the3. Thisperiod Roman of Augustus, brass coin whocentury reigned A.D. dur48 ingmm. the first Figure 124, top. Roman brass. Etch: FeCI3; x66. x1Figure32. 125, bottom. Same as Figure 120. Etch: FeCI3; Appendix F 98

Thai bronze container fragment This is a fragment from the rim of a Thai bronze container from the site of Ban Don Ta Phet with an associated radiocarbon date of 1810 ± 210 B.P. The alloy is 77.22% copper, 22.7% tin, with traces of iron and lead. Examination of the section reveals alpha­ phase copper-rich islands, which are sometimes jagged and occur in areas with specific orienta­ tions as well as random scatter. The background matrix of these alpha-phase copper islands is dif­ ficult to establish. It could be either gamma phase or a homogeneous beta phase. Banded martensite is present and this modification is sometimes called beta'. The acicular martensite is often labeled as betal' 1 udging by the microstructure and composi­ tion, one can say that the object was probably cast initially and then allowed to cool. It was then heated to between 520 and 586 °C, which corre­ sponds to the alpha + gamma region of the phase diagram. After annealing, the container was Figure 126, top. a-phase islands, some clearly outlining quenched, preserving some of the high-tempera­ grain boundary regions. Etch: FeCI3; x40. ture beta phases. Figure 127, bottom. The very fine � needles appear to Microhardness readings from the two princi­ be stain lines, but are not. Note that the orientation pal phases gave the fo llowing results: alpha phase, of the a precipitates is not random but occurs as 257 Hv; gamma or beta grains, 257 Hv. grain boundary and as intragranular forms. This formFeCI3; of x280. martensite is called banded martensite. Etch: ------<)

Gold necklace bead The bead consists of eight small gold spheres joined together. It comes from a necklace in the collection of the Museo del Oro, Bogota, ' Colombia. . fromFigure Colombia. 128. Gold Calima necklace cu lturalbead This bead weighs 0.245 g and is 2.41 x 2.39 x t ... area. 2.38 mm. It is fr om the area of Calima-style met­ alwork in the zone of the western Cordillera of • Colombia. The necklace bead was fa ceted on two , opposing sides after joining. In the polished state the beads appear to be made of a homogeneous gold-rich alloy. There is some porosi ty present, mostly as spherical holes, but there are no inclusions present. Etching is best carried out with a potassium cyanide and ammonium persulfate mix, when the slightly cored dendritic structure can be clearly seen. There are also a fe w recrystallized grains in the vicinity of the joins between the spheres, showing that the bead was made fr om eight indi­ vidual small gold spheres welded together with­ out the use of copper salts to create a granulation Figure 129, top. Unetched view of join region between mixture. The heating of the positioned spheres spheres showing porosity. x55. had to be carefully controlled to avoid melting the Figure 130, bottom. Etched in cyanide/persulfate bead entirely. showing dendritic structure. x55. Analysis by atomic absorption spectropho­ tometry showed the bead to be made of a gold alloy containing 84.3% gold, 13.3% silver, and 0.8% copper. It is a typical native gold composi­ tion fo r this area of Colombia.

Gold alloy nail and gold ear spool The ear spool is made up of sheets of beaten gold gold sheet. The nail did not appear to have a large that were held to some internal support by means head but instead tapered slowly to a point rather of fine gold nails that pass through the sheets. like a wedge. This internal support, which was probably The polished section showed no visible inclu­ wooden, has now disappeared, leaving the small sions, little porosity, and no sign of any depletion Figure 131. Fragment of a Calima nails projecting out on the inside of the sheet. gilding. The section was etched in a diluted aqua ear spoolmade in a gold-rich alloy. This fo rm of construction is characteristic of the regia which revealed the grain structure. From pre-Hispanic Colombia. 45 Calima style of working. Large equi-axed grains with prominent twin­ mm. A section was cut through the sheet so that a ning can be seen. The twinned grains were nail could be examined together with the sur­ slightly deformed by working after the final rounding sheet. The polished section examined annealing as the twin lines are slightly curved. showed that the object consisted of three layers of The nail appears to be made of the same gold Appendix F 100

alloy as the sheet itself. It has been worked rather more heavily, as the grain size is smaller and there are some flaw lines resulting from lamination of the structure during working.

M icrohardness Weighr Diagonal average Hy

1 50g 15.5 56 2 50g 16.5 49 3 50g 15.0 60 4 50g 19.0 37 5 50g 16.5 49

Results 1, 2, and 3 were obtained from the nail, with the highest reading of 60 obtained from the nail tip. Results 4 and 5 were taken from rhe gold sheet and show that the hardness values are quite low. However, it seems unlikely that the nail would be able to penetrate the sheet effec­ tively and the evidence suggests that the sheets Figure 132. above. Section through a nail showing the were pierced after positioning them and before folded gold sheet and the overall shape of the nail. x60. the nails were applied. Analysis: Au 83.7%, eu 1.5%, Ag 12.8%, Fe nd, Pt nd, Sb nd, Ni nd, Figure 133. bottom. Gold sheet etched in aqua regia As nd, Pb nd, Si 50 ppm. withdeformation large twinned after final grains anneali and bentng. x twins 160. showing some

Figure 134. top right. Tang and blade from a late medieval site at Blade section Ardingley. Sussex. The section through the blade reveals that a steel Figure 135. bottom right. This is strip was welded to a principally fe rritic strip. the transition zone between Therefore, the variations in structure are due to martensite at the right side. the change in the rate of cooling and in carbon through a Widmanstatten content between fe rrite grains and tempered mar­ structure to the beginning of some tensite. The angular holes evident in the micro­ ferrite grains. Etch: nital; x 140. structure correspond to the position of slag inclusions, many ofwhich have fallen out during grinding and polishing. A blade fo rged from a strip ofsteel and a strip of wrought iron was a fa irly common method of manufacture. Vickers DPN hardness measurements on dif­ fe rent constituent areas of the blade gave the fo l­ lowing results: fe rritic area in the center, 89-93 Hy; tempered martensite towards the edge, 185- 294 Hy. Appendix F 101

Iron knife The polished and etched section shows that the entire blade was made from carbon steel quenched to produce martensite throughout. Martensite etched in nital is a dark straw color toward the edges. Slag content is moderately low with glassy fr agments flattened out along the length ofthe section. The martensite near the two surfaces of the blade appears tempered with occa­ sional suggestions of troosite. The fact that the martensite etches darker could be due to a change in the carbon content near the surface or to tem­ pering after quenching. No weld lines are evident in this section. The carbon content near the mid­ Figure 136, top. Blade section from a late medieval site dle region of the blade shows some banding, oth­ at Ardingley, Sussex. erwise it appears to be very even. Figure 137, bottom. Shows transition between light etchinga cutting and edge dark of etchingthe blade. tempered Etch: picral; martensite x280. towards Figure 138. Coin of the Emperor Domitian which dates from A.D. Roman copper alloy coin 81. It is from the Mint of Rome, The structure shows a well-formed, recrystallized CoinaCOS.YIIge 237).DES.VI TheII (Roman inscription Imperial on grain matrix with straight twin lines and very lit­ the obverse is IMP. CAES. DIVI tle porosiry present. The inclusions present are VESP. F. DOMITIAN AUG. P. M. not attacked by ferric chloride during etching, with a head laureate. The reverse is suggesting that they may be sulfide inclusions TR. P. COS.YII DES.YIII P.P.S.c., rather than cuprite inclusions, which are exten­ javMinerelinva and advancing, holding abrandishing round shield. a sively dissolved by fe rric chloride solutions. The View of the etched, twinned, and inclusions are flattened out along the length of recrystallized grains with part of the coin and are small-smaller than the grain the midsection crack through the size, which is about ASTM 6. There is no evi­ A deep crack penetrates the midsection of the coin showing as a black line. Note dence of any second phase. X-ray fluorescence that the structure appears single­ coin. This crack does not appear to have been phased. Etch: FeCI3; x39. analysis carried out on the coin itself showed that fo rmed as a result of corrosion of the coin, but only minor alloying constituents occur, with cop­ rather as a manufacturing fa ult. Midsection per being the only major constituent; tin, zinc, cracking is often associated with over-working at and antimony were detected in small amounts. some stage during manufacture. This can pro­ This evidence fits the metallographic evidence duce cracks in the worked alloy which, although that the alloy is single-phased. the grains are recrystallized upon annealing, are not removed by subsequent working and annealing. Appendix F 102

Figure 139. Roman iron nail from the site of Inchtuthil, Perthshire, Roman iron nail Scotland. 140 mm. At the Roman Agricolan fo rtress built at Inch­ •••7 Figure 140. Part of the nail shank tuthil in Perthshire, an enormous mound of iron showing small ferrite grains with nails was uncovered. The outer crust of the nails elongated slag stringers passing had corroded and protected the interior of the along the length of the shank. The mound from extensive corrosion. The cache of slag stringers in this nail are two­ nails, numbering hundreds of thousands, dated FeO).phased This (glassy is a typicalmatrix wroughtand wustite, iron the abandonment of the settlement by the Roman structure. Etch: nital; x90. army to A.D. 100 ± 80. The section taken fro m the nail is a longitudi­ nal cut through the head of the nail shank. The shank and head appear to have been made in one piece by fo rging from bloomery iron. A fe w slag stringers can be fo llowed from the section through the head as they curve down along the length of the nail. These slag inclusions do not can be seen under polarized light to be red-brown recrystallize when the iron bloom is heated by the and are typical features of ligh t1y corroded iron­ blacksmith to fo rge the nail and, as a result, they work. become elongated into stringers upon heavy There is a slightly corroded area near the head working. This can be seen in several areas in the of the nail where iron oxides outline some parts of polished section. Toward the center of the nail the grain boundaries of the wrought iron. No fu r­ shank there is a large, elongated slag inclusion ther fe atures are visible in the unetched condition. which at a magnification of x 1 00 can be seen to The nail was etched in nital fo r 5-10 seconds be composed of two phases. There is a glassy and this revealed a primarily fe rritic grain struc­ dark-grey phase with a fine dispersion of globular ture with no indication of carbon content. The dendritic wilstite precipitates (FeO). This kind of raw material varies somewhat in carbon content, two-phase slag inclusion is a common fe ature of because another nail fr om the same site showed ancient bloomery iron, although high-quality some pearlitic regions. The grain size in the shank products may show very litrie evidence of slag of the nail is slightly larger (ASTM 5-6) com­ content. The veined, grey areas intruding into the pared to the heavier worked area of the nail head nail surface are scale and corrosion products that (ASTM 6-7).

Native copper This sample is cut fro m a dendritic mass of native In the etched condition (etch in alcoholic copper. In its polished condition, there are fe w FeCI3), large grains can be seen with some very inclusions present in the copper mass which long twin lines that are quite clearly curved. In appears to be broken up into a number oflarge some parts of the specimen, strain lines are evi­ segments. The cracks between the different seg­ dent, but there are other fe atures tobe seen as ments are filled with cuprite and green-colored well. Long-zoned lines appear to mark the posi­ Figure 141. Native copper from the copper corrosion products. The outer surface of tion of impurities or growth spirals within the Great Lakes region in North the cut sample also shows the ingress of corrosion cross section of the dendritic arms. America. 42 mm. with patchy conversion to cuprite. Appendix F 103

Figure 142, top right. Cross section throughFeCI), x20, the showing dendrite large arm grainetched size in withinand unusual the grains. structural features viewFigure of 14Figure3, top 13 far8 at right. x 16 5A showing closer andunusual well-de featuresfined withingrain boundaries. the grains Figure 144, lower right. A high x500magnification showing view a fine of train Figure of small138 at bandedprecipitates regions occurring of the withindendrite feaarmstures that occurring may be depositional as a result of growthcopper. phenomena in the native

Head of a toggle pin This fragment is the head of a toggle pin, proba­ bly dating from the Late Bronze Age to the Early Iron Age period. The microstructure of the toggle pin head can be clearly seen after etching in ferric chloride fo r 3-5 seconds. The micrograph at x50 magnification (Fig. 146) shows that the structure is of cored dendrites with considerable interden­ Figure 145, above. Toggle pin head dritic porosity, which appears in the photomicro­ from Iran. 20 mm. graph as black zones between the dendrite arms. Figure 146, top right. Etch: FeCI); The structure is heavily cored, as can be seen from x50. the photomicrograph at xl 00. The eutectoid infill that also occupies the Figure 147, bottom right. Etch: spaces between the dendrite arms has a diffe rent FeCI); x I 00. character from that which can be seen, fo r exam­ ple, in Figures 95-96. Because of the range of composition and of the different rates of cooling, the extent of coring and the appearance of the eutectoid in tin bronzes can be quite variable. Appendix F 104

Javanese iron blade This blade fragment fr om Java is thought to date grains that show subgrain strucrure. The grains from about 1 0-12th century A.D. It is potentially themselves are quite large (ASTM 4) and they of interest since very fe w Javanese objects of this etch a light straw color in nital. The subgrains date have been examined. It is thought that nickel appear as occasionally equi-axed small areas out­ occurs in some of these products as an additional lined by finedark lines that may be, in some cases, alloying constiruent in addition to carbon. fine precipitates (not resolved at xl000). At high A section taken through the complete length magnification some precipitates can be seen at the of the blade fr agment reveals quite a complex grain boundaries in this region. The carbon con­ structure. There are a number of different fe a­ tent is greater in one of the piled strips and the rures that should be described. Overall the frag­ amount of pearlite is greater and the grain size menr has a piled structure made from a number smaller (ASTM 8). Towards the butt end of the of differenr components fo rged together. A draw­ blade the iron strip has clearly been fo lded over. ing of some of the fe arures is illustrated in Figure There is a middle piece also, which occurs clearly 149. in the midsection region of this sample, and joins Parts of the structure show flattened fe rri te with one of the hammered over pieces toward the grains, which is rather uncommon. This may cutting edge. have arisen from cold-working of the iron at some The area that contains a large scatter ofangular stage during manufacture. There is some pearlite slag inclusions passes along the length of the sec­ present at the fe rrite grain boundaries as well as a tion. The slag is a glassy-phase material which here fe w small slag inclusions. This region also displays appears homogeneous, and it has been broken up a banded structure. The bands are free from pearl­ into small angular remnants upon working. ite and have only occasional grain boundary pre­ There is no indication that this blade was cipitates crossing the bands perpendicular to the quenched. The carbon content is well below the length. This type of banding is generally associ­ eutectoid composition, and in some areas would ated with other alloying constiruents besides car­ not have been sufficienrly high to produce much bon, such as nickel, arsenic, or phosphorus. change in properties upon quenching. Toward the other side of the blade there are some "exThe etching has revealed in some grains the veining" substructure that arises when the Figure 148. Fragment of Javanese nital attacks subgrain boundaries as well as grain iron blade. I 10 mm. boundaries (Samuels 1980). It is thought that this develops as a result of etch pitting at the walls of Figure 149. Illustration of cross the dislocations at the subgrain boundaries. Both section of i ron blade. the development of subgrain structure and the ease with which the etch pits develop is depen­ some loss through dent on the physical and thermal history of the flattened ferrite grains corrosion sword. It can arise from hot-working or fr om quench-ageing effects due to the emergence of carbide growth and etch pitting in fe rrites due to (i�corroded- �E�tip heating at low temperatures (e.g., 50-150 °C will strip suffice). carburizedareas near surface slagangular inclusions Appendix F 105

structureFigure 150, with right. carburized Part of the sur face, grains.angular Etch: slag region,nital; x70. and ferrite inclusionsFigure 151, with far right.some Angularpearlite and assubgrain veined structure lines passing that through can be seen the ferriteare angular grains. slag Very fragments dark regions which areinclusions. mostly Etch: glassynital; single-phasedx280. slag

Bronze axe fr agment This fr agment is thought to be a part of a Luristan-sryle ceremonial bronze axe dating to about 8-1 0th century B.C., from Iran, composed of87. 1 % copper and 7.2% tin, with no other ele­ ments detected by X-ray fluorescence analysis. Figure 152. Bronze axe fragment The sample is taken from the cut end of the trian­ from Iran. 90 mm. gular fin which was probably integral with the blade. The triangular plate is carefully made and has smooth surfaces with a flange around the cir­ cumference on both sides of the plate. The sample Figure 153, above. Slight' coring visible together with removed fo r examination is a transverse section twporosityinned crystals.and recryst Noteallized that grainsthe twin with lines well-developed are straight. through part of the broken end of one of the pro­ This evidence suggests that the axe was roughly cast to jections of the plate. shape and lightlyfinished by working and annealing, with annealingFeCI3; x62 being (see thePlate final 19 ).fabrication process.Etch: Figure 154. Well-developed 60:40 hexagonal grain structure, usually Laboratory cast brass outlined with a precipitation of u­ This 60% copper, 40% zinc alloy is etched with phase (the copper-rich whiter FeCl3 in which case the � phase region appears material) showing variation in color as a result of different grain darker or black. Note the well-formed Widman­ statten precipitation of a within the � grains pro­ orientations in the brass: 60 Cu, 40 Zn, u + P brass. Etch: FeCI3; x31. ducing a precipitates only on certain crystal planes. See page 20 fo r a brief discussion of the Widmanstatten precipitation. Appendix F 106

Figure 155. 60 Cu 40 Zn brass. Two-phased a + � brass showing Note that the fu ndamental difference detail of the Widmanstatten between the Widmanstatten and martensitic formation of a plates within the � transformation is that the Widmanstatten crystals. Etch: FeCI3; x I 00. involves the decomposition of one solid phase into two phases while the marrensitic transforma­ tion involves the rapid cooling of one phase pro­ ducing one metastable phase on cooling.

Gilded silver earring The complete earring, drawn in Figure 156 is one tron probe yielded 92.9% silver, 2.9% chlorine, of a pair from the site ofTell Farah in Jordan and 1.47% copper, and 0.7% gold. is dated to the MiddleB.C.). Bronze Age (first half ofthe second millennium They were both made from gilded silver, the second example being in a fragmentary condition from which a sample could be taken across the broken end by careful Figure 156, above. Drawing of gilded sawing with a fine jeweller's saw. Both objects silver earring from Jordan. 22 mm. were decorated with granulated gold spheres Figure 157, top right. Part of the attached to the gilded surface. The terminals of gold foil covering with overlapped the earrings end in loops, which are secured with join in the outer gold foil. The foil twisted gold wires. Before conservation, both silverhas been beneath. diffusion-bonded Note that the to silver the objects were covered in dark grey corrosion of a rod is heavily corroded with few waxy consistency. X-ray diffraction analysis .,[ . remaining sound metal regions. showed that the main constituent of this corro­ ... •. ', ,r/• It'fl\;;,,· . Unetched; x21. sion was silver chloride. Conservation work was ". Figure 158, middle right. Hexagonal carried out both chemically and mechanically grain boundaries in the silver rod. using 15% ammonium thiosulfate and hand r The dark areas are corrosion, and tools. The corrosion products had erupted the grain boundaries that can be through the gilding in several places and had seen are not straight lines. This is a fo rced the gilding outwards. Characteristic fe a­ silver.consequence Over the of thepast age 3,500 of theyears, tures of leaf gilding were observed, in particular copper has precipitated from the the overlapping and corrugating of the thick silver rod and the grain boundaries gold fo il. � ",-'''''IIIIII_a •••,...,. '' have moved as a result. Etch: The metallographic sample, revealed a par­ acidified potassium dichromate; tially corroded silver substrate covered with a x200. thick layer of gold. The gilding had become Figure 159, bottom right. "Wiggly" detached from several corroded areas, but in some grain boundaries as a result of areas a good interface between the gilding and the discontinuous precipitation of silver could be seen. This interface is several acidifiedcopper from potassium the silver dichr alloomate;y. Etch: microns thick and suggests that a deliberate diffu­ x500. sion bond was created between the applied leaf gilding and the silver core. Analysis with the elec- Appmdix F 107

Figurebrass medallion 160, top right. from Fragmentthe La of a Fragment of a brass medallion Perouse, which sank off the coast This brass has major constituents of copper and of Australia in 1726. Unetched; zinc, with minor amounts of tin, lead, iron, and x20. chromium. The first micrograph, unetched at Figure 161, bottom right. Etch: x20 magnification shows the remarkable depth of FeCl3; x75. corrosive penetration into the brass. Corrosion has outlined the dendritic fo rms of the cast brass medallion which has been lightly etched with fe r­ ric chloride. Nthough loss of both copper and zinc has occurred here, it should be noted that the dendritic arms have been preferentially corroded while the interdendritic areas are in some regions preferentially preserved. This is evident in the sec­ ond photomicrograph at x75 magnification and suggests that the dendritic arms are rich in zinc compared with the interdendritic fill, since nor­ mally zinc is lost preferen tially in sea-water burial such as this shipwreck site (see Plate 4).

Fragment of a small Simi ear ornament These ear ornaments were cast as pairs by the lost­ .wax process and represent very skillful "false-fili­ gree" wire work. The whole ornament is modeled in wax in one operation and then cast. fragment.Figure 16 2.Pre-Hispanic Sinu ear ornament Colombia.

Figure 164. Enlarged view at x I 00 showing the light showingFigure 16 extensive3. Unetched, corrosion low magnification to the gold-copper view at x20allo y enrichment in gold at the surfaces as well as the wire section. The evidence shows that the ornament selectiveresult of coringretention of theof some gold-copper areas of alloy the dend on casting.rites as a dendritichas been caststructure by lost-wax of the alloywithout is clearly any working. evident. The The grey matrix is principally cuprite. Appendix F 108

Figure 165. Section of a Roman mirror fragment at x35 magnific­ Roman mirror fr agment ation. etched in alcoholic ferric The early examples of Roman mirrors made in ofchloride large lead showing globules the at precip the bottomitation Europe, in a specialized production center in Ger­ of the cast mirror (on the left-hand many, were often made in a leaded high-tin side of the picture). The grain bronze+ which0 is partially quenched to fr agment boundaries can just be seen as a the IXeu tectoid phase. The mirrors date to the suresultrface of with the differricferent chloride. etching Itof is the a early centuries A.D. and this example is fr om very fine �tructure for a cast Roman Canterbury and dates to the 4th century bronze showing that despite the A.D. It is made in an alloy of8% lead, 19% tin gross segregation of the lead glob­ and 73% copper (see Plate 5). faules.st enough the cooling to prevent of the thealloy usual was developidendriticng. structure from

Figure 166. next page. Roman bronze figurine of Roma AD. Roman bronze figurine 40-68. This small figurine, about 33 em in height, is Figure 167. right. Roma bronze from the J ulio-Claudian period of A.D. 40-68 microstructure. The proper sur­ and she is shown wearing a short Amazonian face of the Roma casting is to the tunic. The torso and head were hollow cast in one left hand side of this photomicro­ piece with the left leg just above the knee. The castgraph structure which shows of the the bronze. annealed The right arm was attached by casting-on or solder. grain size is large with small inclu­ The legs and fe et were then cast-on. The Roma sions of lead and copper oxides or was cast in a leaded bronze of 89.8% copper, sulfides. Etch: FeCI3; x90. 7.1 % tin, 2.7% lead, with traces of zinc, iron, nickel, and silver. Appendix F 109

There is an unusual fe ature on the back of this Roma: a series of hexagonal networklike structures appear in the metal. A core sample was taken in this area on the back to try to resolve the question concerning thenature of the grain struc­ ture in this area. The microstructure is shown here in Figure 167. The structure is etched in alcoholic fe rric chloride with the cast surface of the Roma at the left of the photomicrograph. The structure of the bronze is essentially that of an annealed casting. The exgrains appear hexagonal 8in shape in many areas of the section with the ex + eutectoid preferentially occurring at the grain boundaries and outlining some of the hexagonal exgrains. Small oxide or sulfide inclusions can be seen scattered through the section along with some small lead inclusions and porosity. The absence of a typical dendritic structure suggests that there may be a microstructural explanation fo r the network of hexagons observable on the surface of the bronze, since the hexagonal annealed appearance of the section is unusual fo r a cast bronze. The microstructure shows that the grains are not mammiform but are equi-axed. The coalescence of these hexagonal grains during solidification could be responsible fo r the surface appearance of the bronze.

Figure 168. Renaissance silver basin from Genoa dated to about A.D. Renaissance silver basin 1625. 755 mm. This basin is an elaborate repousse work with many cast additions and soldered components. It has been fi nished by chasing, engraving, and the addition of many components to the worked sheet of the basin itself. Before the late 16th cen­ tury, fo rks were uncommon in Europe and eating was carried out using the fi ngers. This bowl was probably only fo r show, and many were probably melted down when they went out of fashion and turned into silver knives and fo rks. The plate is made of an alloy of 94.9% silver, 3.2% copper; small amounts oflead, arsenic, zinc, tin, and antimony are also present (Scott 1991). Appendix F lIO

The samples fo r study were removed with a diamond-tipped core drill, 1.5 mm internal dia­ meter, after removing some of the components of the silver basin, so that no damage was visible from the front of the object when the parts are reassembled. It was customary in the Renaissance to start important projects by preparing a detailed wax or gesso model, usually full-scale. The silversmith then carefully copied the model in silver, first working the parts that needed to be raised. This repousse working was carried out to enable the metalsmith to model the location of the cast addi­ tions. The cast components were modeled in wax, Figure 169. Part of a solder blob most probably on the basin itself and then from a repair to the outer radial removed and prepared fo r casting. When cast and panel. The solder shows a cast soldered in place. all the surfaces were finishedby morphology of the copper-rich and chasing and engraving exactly as if the entire silver-rich phases. which are not basin had been made in one piece. copperequilibrium binary phases system. of the but si havelver­ individual phase compositions sectionFigure 17through0. top right.the basin. Phot omicretchedograph with acidiof parrfied of the values.removed This from structure their predic of alternted ate potassium dichromate under partially polarized light to dendritic morphology is typical of revealsilver cr theystals continuation across the of boundar the grainy intostructure a surface­ of the allomanyys. castThe silver-copper overall composition solder of altered zone. This micrograph illustrates the continuity the solder used here is about 50% veryof the di overfferentall structuralmicrostructure regions within of the two silver apparently basin. x30.silver. 50% copper. Etch: FeCI); x155. plugFigure from 171. the middle silver right. basin. Overall The top view sur faceof a coreof the-dril basinled andis uppermost nature of inthe this sur photographface-altered and zones shows of the the silver extent correspondingbasin. An unusual surrounding feature is zonethe su ofrfoxidized ace crack metal. and The Figure 172. bottom above. View of part of the silver surface alteration zones here are due to the grainsheet. structure etched in of acid theified silver pota canss iumbe seen. dichr Theomate. grain The toagglomeration cuprite. since of silvercopper is veryparti clessusceptible and their to oxidationoxygen boundaries are not straight and have a "jigsaw" absorption during annealing. This oxygen can then appearanceseen here occurs in places. as conti but nuousthe preci zonespitation at the of graincopper acidifiedcombine potwithas siumthe copper dichroma to makete; x50. cupr ite. Etch: boundaries. suggesting that the basin had been heated forsubsequent the casting to itsof manufthe smallacture. additions. probably x40. fo r repairs and Appendix F III

Figure 173, top right. Overall, the structure is typical of a cast alloy; Circular bracelet fr om Thailand the bracelet has been cast integrally This is a transverse section cut across a circular Largeand joined areas near of extensive the center porosit region.y bracelet from Thailand. It is from the site of Ban are clearly evident together with Chiang and is dated to about 500 B.C. In compo­ smaller zones, sometimes filled with sition, the bracelet is a tin bronze with 89.2% corrosion and, interestingly, a patch copper, 5.4% tin, and 0.6% zinc. The inclusions ofhollow casting between core material the two in regions the of present in the metal are copper sulfides and when the bracelet. The casting core has analyzed one was fo und to contain 76.5% copper, been prepared as a fine mix and 0.5% tin, and 19.1 % sulfur, with silver, lead, and appears in reflected light as a grey zinc undetected. The second inclusion studied coremineral appears assembla to bege. principallyThe casting min­ contained 71.8% copper, 5.7% lead, and 18.3% erai in composition with no organic sulfur with tin, zinc and silver undetected. debris apparent. One or two small Cuprite inclusions also occur and the tin content areas that are of darker color may can vary up to about 10.7%. be remnants of the ex + () eutectoid, Examination in the unetched condition anow thick rath cruster corroded. of corrosion There is also reveals a number of different fe atures. Toward surrounding the surfaces of the some of the shaped edges of the bracelet fine bracelet section. cross-hatching outlined in corrosion can be seen, The dendritic structure of the which indicates that there has been some working bracelet has probably been modified of the bracelet fo llowing casting, probably to fin­ by heating during manufacture since ish the surface rather than to produce a great it is not very pronounced, and the coring that remains is subtle. change in shape. Unetched; x40. Figure 174, middle right. Structure after etching in alcoholicstructure ferric together chloride. with Someevidence coring for canslip beplanes seen (some in the cupevidentrite corrosionas pseudomorphic crust adjacent remnants to the in theremaining principally etchingmetal). Somein this wayof the and inclusions may be copper are not sulfide affected inclusions; by metotherallic rather regions, dark which grey againregions sug appeargests remnants to be corroded of the sampleeutectoid is the phase. many Another copper-color interestinged areas feature present of this in the copper,eutectoid unallo regionsyed withor as tin, small may scattered be present blobs. as a Thisresult of morphicinsufficient replacement melting in theof the original original cast ex or + ()aseutectoid a pseudo­ corrosion-redepositionregions. It is more likely mechanisms to be present that as havea result resulted of x70.in the corrosion of the tin-rich phase preferentially. Figure 175, bottom right. High-magnification view coppershowing in a asmall corroded globule ex (l ight+ () region, in color) together of redeposited with showingcopper sulfide strain inclusionslines from (lworking.ight grey) Etch: and FeCI3;bronze x metal 170. Appendix F 112

Roman bronze bowl The sample was examined carefully in the unetched condition. Extensive inrergranular cor­ rosion has occurred oudining most of the grain boundaries with cuprite very clearly. Some srraight lines of cuprite corrosion can be seen passing occasionally inco a grain inrerior suggest­ ing that the metal must have been stressed at some stage during manufacrure co produce some slip, alrhough no twin lines are apparenr in the section before etching. Some of the inclusions bowl.Figure The176. bowl Section has froma composition the base of of a 87.0% Roman copper. bronze presenr in the metal are of cuprite, often conrinu­ 11.5% tin. and 0.9% arsenic. with silver. lead. zinc. sulfur ous now with the network of corrosion through and iron undetected. x I 00. the grain boundaries. There are other regions where the rypical variegated surface appearance of lead inclusions occur and these are more concen­ The section has co be etched fo r about 10-l2 rrated in the thicker region of the section where seconds in alcoholic fe rric chloride co bring out they appear CO have segregated. Note also that the grain conrrast. The grains are not twinned there is considerable porosiry as well as some areas which shows that this sample is essentially an where lead globules may have fallen out as a result annealed casting. Some of the grains show coring of grinding and polishing. That the alloy in ques­ where the disrribution of copper and tin in solid tion here is a leaded tin bronze, can almost cer­ solution is still unequilibrated. Note that the thin tainly be deduced from the occasional presence of srrip that appears co be benr back co produce a ex + 8 uniform edge has multiple srrain lines apparent in some regions of the euteccoid. Where the euteccoid phase occurs, it is usually surrounded the crystals. A small island of metal grains in this by islands of corrosion where the copper-rich region oudinedex by+ 8corrosion is also heavily grains have been attacked. In this example then, srrained. The euteccoid is much easier co see the primary corrosion process is the attack of the in this section in the etched condition, and the ex ex + 8 grains have color contrast as a result of differing grains surrounding the euteccoid region: ex orienration in the polished section. the within the euteccoid is also attacked. If the corrosion crust is carefully examined, The overall structure is a cast one that has some areas of remaining euteccoid can be clearly been annealed and then partly worked in some seen within the corrosion surrounded by copper areas co produce the srrain lines visible in some oxides. This is an inreresting example of the pref­ regions. The casting has resulred in some segrega­ ex tion of the lead componenr, while annealing has erenrial retenrion of the copper-rich phase. Careful examination ofthe corrosion layers under not been extensive enough co eliminate all of the polarized light is also useful in the study of this euteccoid. material since the crust can be resolved inro dif­ ferenr layers. Appendix F 113

Figure 177, top right. Brass sheet, etched in alcoholic ferric chloride Early Medieval brass sheet and then potassium dichromate. This fragment of a broken bronze sheet has three coredHeavily dendr etcheditic section structure reveals of the the studs pushed through the plate. The sheet is brass. Small lead globules can be English from the Early Medieval Period and has seen in interdendritic regions. xSO. 83.6% copper, 15.7% zinc. Before etching, the sample was examined in tallizedFigure 17and8, bottomheavily workedright. Recrys­ grain the polished state which reveals the fe atures of structure of the stud probably corrosion to be seen in the section. Corrosion of hammered at this point to attach to the bronze stud head can be clearly seen outlining the brass sheet. The grains are the worked and recrystallized grain structure in heavily strained and deformed, this part of the artifact. Corrosion has selectively afprobablyter working by additional and annealing hammering cycles removed part of the grains of the metal, advanc­ to shape the stud. Corrosion has ing along twin boundaries or slip planes. outlined some of the strain lines by After etching in alcoholic fe rric chloride fo r a dark corroded regions passing few seconds fo llowed by etching in potassium through the grains. Etch: potassium dichromate, the microstructure can be examined. dichromate; x 140. It is immediately clear that the sheet brass has been cast to shape, as is apparent from the undis­ torted dendritic structure, which is cored. There are some depressed areas on the surface of the cast sheet that show the presence of strain lines from some surface working of the casting. The stud, on hammering to attach the cast plate to a backing. the other hand, has been extensively worked to The stud appears slightly bent when exam­ shape and has a completely recrystallized grain ined by eye and this may have been caused by structure in which most of the twinned crystals hammering into place, although the cast plate have straight twin lines indicating that a final would probably have had the holes fo r the studs annealing operation overall was the last stage in cast in situ. Hammering through such a thickness manufacture. However, just past the area of the of metal could not have been accomplished with cast metal sheet, there is a zone of the stud that this stud, and there is no indication of deforma­ shows a concentration of slip lines that could only tion to the cast structure in the vicinity of the hole have resulted from severe strain of the bronze stud which would have occurred had the hole been in this particular region. This is perhaps due to drilled fo r the insertion of the stud. Appendix F lf4

Figure 179. Part of the arrowhead's microstructure showing a typical Romano-Greek iron arrowhead wrought iron with slag stringers This small iron arrowhead is square in cross sec­ rises, almost certainly due to deliberare carburiza­ passing(dark in alongthis photomicr the lengthograph) of the tion and tapers to a point. It is Romano-Greek of tion here, since there is no carbon conrenr to be arrow. The grain size of the ferrite about the 2nd century A.D. and is made of seen elsewhere in the seerion. The fe rrire begins to is variable and it is typical for a wrought iron. assume Widmanstarren characreristics and pearl­ piece of bloomery iron to find this In the polished state the wrought iron shows ire makes up the infill showing thar a sreel edge kind of variation in apparently a scatter of slag inclusions, some of which are has been produced here. The depth of pencrra­ homogeneous material. Etch: nital; elongated across the width of this section while rion of carbon is quire limired, as might be x75. others are rounded and globular. All look like expecred fo r a case-carburized sreel. wiistite in a glassy matrix. The ingress of some iron oxides as corrosion products can also be seen on the surfaces of the section. After etching in nital fo r about 10-15 sec­ onds, the fe rritic grain structure can be seen. The fe rrite would technically be described as massed fe rrite grains and, if properly etched, the subgrain structure can be seen within some of the fe rrite. The grain size in many areas of the arrowhead approximates to about ASTM 8. Toward one surface of the arrowhead, the carbon content Figure 180, top right. Unusual corrosion pattern through the Byzantine leaf-shaped blade bronze blade resulting in the The secrion from rhis small Byzanrine leaf-shaped gradualremnants isolation in a matrix of small of cupr grainite. blade is raken across parr of rhe edge of rhe blade. The ingress of corrosion is also Ir is made in a low-rin bronze wirh abour 92% severe and both surfaces of the copper, 7% tin. blade are retained in a thick There is very lirrle metal remaining in this corrosion crust. Unetched; x70. sample, which has mosrly been convened to Figure 181, bottom right. Heavily cuprite. The meral thar does remain shows rarher etched view, in which some fine disseminarion of cortosion ar rhe surface, surface scratches are visible as well suggesring rhar rhe grain strucrure is quire fine. as the bronze grains and corrosion After erching wirh alcoholic ferric chloride, somecrust, strainshowing lines twinned just evident. grains withThe rhe grain strucrure can be seen, in faer, to be very shading in the metallic area is due small suggesring that rhe alloy used to made the to coring from the original cast blade has been well-worked and annealed in a ingot, still not entirely removed by number of cycles to shape the blade. Many grainsworking conti andnue annealing, across areasand the of rwinned crystals are visible as well as exrensive slightly different composition. strain lines throughour rhe grains. There are some Etch: FeCI); x 160. blue-grey inclusions in rhe metal that may be small sulfide inclusions hammered out along rhe lengrh of the secrion. Appendix F 115

Figure 182, top right. Overall view of the corroded blade showing Medieval iron knife darker etching region toward This iron knife blade is from Medieval Britain contentcutting tip is greater.where the Outer carbon areas are and is mostly wrought iron rather than steel. The corrosion products, mostly iron sample taken is across the cutting edge. oxides. Etch: nital; x30. In the polished state the heavily elongated slag Figure 183, middle right. View of a stringers can be seen passing along the length of weld where different pieces of iron the section. Corrosion of the surfaces is also have been joined together. The apparent: the structure of the corrosion here does carbon content in the upper and not reveal any pseudomorphic preservation of the lower zones is greater while the structure of the iron. wroughtcenter triangular iron with region some is slag just After etching in nital, the comparatively large stringers. Etch: nital; x70. size of the original austenitic grains can be seen toward both long surfaces of the section, where carbonFigure 18steel4, bottom area of right.the blade Low with decomposition of the austenite has occurred partial Widmanstatten structure of upon cooling resulting in grain boundary fe rrite ferrite and pearlite. Etch: nital; with an infill offerrite and pearlite. In some areas x120. the beginnings of a Widmanstatten structure can be seen in the fe rrite, but this is not very pro­ nounced which means that the blade has been cooled fa irly slowly from the austenitic region and has not been quenched, at least not in this part of the structure. The low-carbon steel edges gradu­ ally grade into a fe rritic grain structure. Another steel region with low carbon content appears about midway through this section which suggests that some welding of diffe rent carbon­ content pieces may have been used to fa bricate the blade.

Set of panpipes This is a fragment of tumbaga alloy from a set of passing along the length of the curved section panpipes from the Department ofNarifio, which are areas of loss due to some segregation Colombia, dating to about the 8-10th century passing over from the cast ingot into the worked A.D. Found in the municipio ofPupiales. The sheet of the panpipes. This segregation from cast­ analyses of this sample gave 25.9% gold, 62.4% ing has resulted in unequal distribution of the copper, 4.0% silver, and 0.01 % platinum. gold and copper of the alloy, with the conse­ The structure has become quite corroded as a quence that the copper-rich segments of the result of burial and the fact that it is a tumbaga worked sheet are preferentially corroded. Apart alloy (a South American gold-copper alloy) that from this gross removal, there is also corrosion is quite debased. There are some dark striations along twin bands, slip lines, grain boundaries, Appendix F 116

Figure 185. This section is etched in cyanide/persulfate. The and through some of the grains themselves, isolat­ depletion gilded surface appears at ing small patches of residual metal. Note that in the top of the photomicrograph. some areas there is very good pseudomorphic Thecopper lamellar alloy structureinterior is increated the gold­ by replacement of the original grain structure with differential corrosion of the cuprite, especially along some twin lines. Corro­ worked and annealed sheet. Some sion has already outlined most of the features of twin crystals can be seen in the this gold-copper alloy, and there is little point in gold alloy, showing that annealing attempting to etch the sample. The grain struc­ was the final stage in manufacture. ture can be seen to be worked and annealed with x65. recrystallized grains and predominantly straight twin lines. Careful examination of the surface cuprite layers reveals a thin line embedded in the corrosion which is more golden in color than the alloy itself and which is, in fact, the remnants of the depletion gilded layer on this tumbaga alloy. The depletion gilded layer here is deceptively thin and could easily be mistaken fo r some other fo rm of gilding if the alloy itself did not contain gold, and the technology of the region was not understood. Figure 186. Photomicrograph of part of the blade edge, showing a Byzantine dagger blade low-carbon steel with Widman­ This is a fragment of a Byzantine iron dagger The edge of the blade, the detail of which is statten side plates and sawteeth in a blade, with the section taken across the cutting ferrite and pearlite steel of about still preserved in corrosion, can be seen to have 0.4% carbon. Etch: Villela's reagent; edge of the blade. the same structure as the carbon-rich zone in xIOO. In the unetched condition, the sample section which the Widmanstatten fe rrite and pearlite is quite corroded, wi th thick crusts of iron oxides appears to continue down into the edge, showing evident. There is a fine scatter of glassy slag inclu­ that the blade had not been quenched, but had sions, slightly elongated, toward the center of the been made from a low-carbon steel. blade section and some finer slag stringers more heavily elongated along part of the length of the blade. After etching in nital for about 10 seconds, very little relief could be seen and instead, Ville­ la's reagent was used fo r about eight seconds. This revealed that the blade has been made from a steel of variable carbon content. One side is predomi­ nantly rich in carbon with Widmanstatten side plates and sawteeth visible. The carbon content in this region must be of the order ofO.4%. In most other regions the structure is that of blocky fe r­ ritic grains with some fairly uniform pearlite infill. Appendix F 111

Figure 187, bottom left. Overall Roman coin of Victoren us view of the coin, showing a cast structure with superimposed This is a debased silver coin of the Roman Here there are some recrystallized and twinned recrystallized grain structure Emperor Victorenus, from A.D. 268-270. In grains superimposed on the as-cast structure suitableindicating size that and the then flan wasstruck. cast Et toch: a composition the coin is about 96. 1 % copper, where some limited working of the coin has FeCI3; x30. 2.5% silver. occurred. The corrosion products were examined under This type ofpartially recrystallized structure is Figure 188, bottom right. The grain reflected polarized light. Below the immediate very suggestive of hot-working of the cast coin shownstructure in thisof the photomic coin is rograph.clearly surface, a very crystalline mass of green corrosion flan. There is some silver content to the alloy, but The crystals are lightly twinned products are to be seen and there are some blue the amount of the silver-rich phase, which can be showing that some working and crystals also, and adjacent to that, some bright red seen elongated along the length of the coin, is not annealing (in this case hot-working) cuprite layers together with corrosion products very great. has been carried out. The light passing into corroded metal. Passing along the section one sees the heavily particles present between the worked grain structure; again the fo rm of the def­ grains are silver-rich particles. There are deep blue azurite crystals on the Variation in shading is due to coring surface of this coin, while in the microstructure inition of these grains against the slightly cored in the original cast structure. Etch: there are thin zones of grey corrosion which background suggests hot-working to shape. Note FeCI3; x 145. appear to be fo rmed around some of the grain the flow in the grain structure ofthe coin towards boundaries. parts of the design which are less heavily worked After etching in alcoholic fe rric chloride fo r by the die. Strain lines also occur in this region. about 10 seconds, the grain structure can be seen. Passing along to the ocher end of the section, Parts of the structure of the coin betray an almost there is a very uniform degree of grain size appar­ cast structure towards one end of the coin flan ent, with twinned crystals in which the twin lines where it has not been effectively deformed on are straight. working: the punch has not touched this region. Appendix F 118

Figure 189, top right. Overall view Grey cast iron of the cast iron microstructure, showing distribution of graphite This grey cast iron of the early 20th century con­ flakes. Light areas are mostly tains about 4% carbon and a little phosphorus. ferrite. Etch: nital; x3S. Before etching, the outlines of some of the Figure 190, bottom right. View of thick graphite flakes can be seen at low power, the graphite flakes and pearlitic while at higher power the coarse nature of the infill. The white areas associated structure is apparent. After etching in nital fo r withpearlite the is graphite mostly aredark ferrite. grey here The about 8 seconds, the high carbon content of this due to the fine spacing of the cast iron can be seen, since the whole of the infill ferrite and cementite. Etch: nital; surrounding the graphite flakes is made up of x130. pearlite of different spacings. Some white, unetched regions in this cast iron are of the ter­ nary eutectic, steadite. This is a common constit­ uent of many cast irons because of the phosphorus content.

Figure 191, top right. The cast structure of this iron-carbon alloy Grey cast iron can be seen in some dendritic areas This is a grey cast iron, circa 1920, with about 3% revealed by etching in nital and carbon and 0.2% phosphorus. picral. Grey areas are fine pearlite. In the polished state, the fine graphite flakes x40. can be seen in this fairly typical grey cast iron, Figure 192, bottom right. Graphite together with small inclusions that are probably flakes in a ferrite matrix with an sulfides. In some areas, clear rosettelike regions of steaditeinfill of pearlite patches. cont Etch:aining nital some and growth ofthe graphite flakes can be seen, together picral; x 130. with some zones in which thicker flakes have been deposited. After etching in nital for about 8-10 seconds, the structure can be seen to consist of small graphite flakes in a fe rritic surround with fine pearlite accounting fo r the remainder of the structure. These are the typical constituents of many modern and historical cast irons. Appendix F 119

Figure 193, top right. Overall Wh structure of white cast iron ite cast iron showing long cementite laths and This white cast iron dates to the early 20th century small globular regions of pearlite. and has about 4% carbon and 2% phosphorus. Etch: picral; x40. Before etching, some structural detail can be Figure 194, bottom right. Cast iron seen in the fo rm of slightly whiter and harder etched in Murakami's reagent and needlelike growths in a more cream-colored picral. The solid white areas are matrix. Note that there are several intergranular cemenpearlitetite, with dark occasional grey zones non­ are cracks that affect this sample, as well as a region of metallic inclusions and the finely porosity on one side of the ingot. structured white areas with dark After etching in picral, the structure can be spots are ferrite in a white matrix seen to consist of cementite laths with a fine infill of iron carbide and iron phosphide. of pearlite some of which can be resolved under x160. the microscope. The remaining material between the cementite laths and the pearlite is the ternaty eutectic, steadite. Here it tends to be present as fine particles of fe rrite, sometimes showing a her­ ring-bone effect set in a white matrix of iron car­ bide and iron phosphide. Using alkaline sodium picrate etch or picral wilI darken the carbide, although it should be used hot, while Murakami's reagent can be used fo r examination of the iron phosphide in these cast irons.

Tin ingot fr om Cornwall Many tin ingots have been recovered from marine excavations off the coast of Cornwall, England. This smalI tin ingot is heavily corroded and frag­ mentary with little metallic structure remaining. The components are now mostly the tin oxides cassiterite, romarchite, and hydroromarchite. The first ofthese is Sn02 while the other minerals CornwalFigure 19l.5, 85 above. mm. Tin ingot from are SnO and hydrated SnO respectively. Figure 196, right. Heavily veined tinand in corrodedgot in which fra gmentno metal from remains, the morphicnor is there retention any clear of thepseudo­ original shapecast structure of the in ofgot the is pringot.eser vedThe entirelyUnetched; in corrosionx35. products.

APPENDIX G PHASE DIAGRAMS

The fo llowing phase diagrams have been selected ternary mixtures that may occur in ancient as being the most useful fo r investigations into metals and are given here because of their use­ ancient metals. Many of the figures drawn earlier fulness as reference material. For further details in the text are schematic representations of the concerning phase diagrams the reader is complete phase diagrams included here. In most referred to the books by Cottrell (1975), Lord cases the different regions of each diagram have (1949), West (1982), and Bailey (1960). Since been labeled with the appropriate Greek letter by the £ phase in ancient bronzes is not present which the phase region concerned is usually below about 28% in tin, some different ver­ known. Some ternary diagrams are given sions of the copper-tin diagram are included to although discussion of their application has not represent the variable limits of a solid solution been dealt with in the text. They have a direct in practical terms, such as under normal cast­ relationship to some of the binary phases and ing conditions or if well annealed.

Figure 197. Tensile properties, 120 ,------,------,------,------,------�300 wroughtimpact value, copper-tin and hardness alloys. of

100 f------+------+------f------+------,f----1250

80 200 NEu �s·..,OJZ .:LboE 60 150 �c-c:3

40 100

20 ��--���-----+------f--�+_--�------�50 ______o � � � � !L___=r��_� 0 o 5 10 15 20 25 Tin Percent Appendix G 122

Figure Copper-tin system. °C ,-----,-----,-----,------,-----,-----,-----,-----,,-----,-----, 198. 1-1� \\---�.------+------+------I__----+_----+_----+_----_+_----+----_1 1000 \ a+ L 900 1__��,, ���---- �-----1------r-----T-----+-----+------+------r----�

I � 79 � 800 �--���.5 �� Sh"'-�13��2 �s...... --�--� -- _r--��--_r--�----_r--� a+� 700 �--_+-1�1_+4r��+���--��--�d_----r_--_+----�----r_--� 6 � (a) _ 0 'L1I� 4�� 600 �---i--��,,�t_V��II� �tjl Ht-- --!_----t-��- ....��-- t_----I_--_I ....a+--rr Y V...&..,1-I ,! ...... I .8 '20' E+ L � 500 �----+_�s r-+_--'Y�+�.--+�-----+----�-----1------r-��-+----� a+ , Il- ___ )__ 8__ 9. -t-t+- 8 4 92 400 _'��---,�====�===s1/ l�===t==l s='=t====�����.4 I 350' E + / 11.0 T] T]+L \ 300 ! 2 1.968� a + E E" 189' T] .,., 1022,7 '' nn ' 200 �---r----t_--_t--�������==== � == �� I � =+�Sn += 60.3' T]+ (Sn) E + T]' T]'..... __ Cu Sn 10 20 30 40 Weight 5Percent0 Tin6 0 70 80 90 Appendix G 123

Figure 199. Part of the copper­ tinconditions. diagram under different II ,--- -- � I ,----,------,------,------, 1100 00 ,------,---- ,----- 100 � I r-;\-�--tT- --�--�------1 IOOO � 1000 \ a� L \ .'\ L 000 � a,'\ L 900 \ 900 �4_-�-��--�-� 1\ 7981\0 �� r--\-.-\\ -\-4- ;..;.79.;;.8.....' !\__-'- -ffi+ +----1 800 I 800 , �� y+L I � �y+L f3r� --y ��II-+____'_a�+--.!f3=--Hr+f3 hf�R-L"'FV-�� 700 ....y+£,L 700 I I-- Y/ oJ" , 0 y IT : en 0 If 600 a a li�� 600 al a+y " � 5 £ Y520V4 1;,+�£ I �-"""";�-'&�2!)' V" 1;,T + £ sOO ��I-�--�-�y+8�� �-+ 500 y+8 0.01� J a+8 I� �--��-�--HH-�a+8 .�_:::J .iI .�c:::J a+8 400 8+£ 400 400 . / / 350' �--��-�--�-�f---5+--IjI f---� f-- -- f----}�f-++___ -I-I 300 I a+£ £ � 300 8+ � 300 I 200 � 200 �+_! 1-�--�--�+-� __ L-__ L-__ llJ_� L-__ -'--__-'-- __L-.L-L- ----1l 100 Sn% 1 00 L- Sn% 100 Sn% 10 Full Equilibrium20 30 10 Annealed20 30 10Usual Casting20 Conditions30 Appendix G 124

Figure Copper-arsenic system. °C 200. \\:5. 1000 \ \1\ L 900 \ n L /82 ,8° "' 800 "'\ Il n� 700 6.77\ 21.0 11+ -- AsS �blimes a 6� - �- -- at 03°C 600 a+� 0 .... �y 50.1 500 11+X

39 5° 4.00 + or _ 7.8 325° 25° 300 �£ 11XI (As) __ 200 £+0

Cu As 10 20 30 40 50 60 70 80 90 Weight Percent Arsenic Appendix G 125

Figure 20 I. Copper-lead binary °C system. � l 1000 Cu + l � ...."'-- 1 701 r-- 36. (l) a .,.+ l 955" 87.' 900 , \ 800 Cu + a(l) \ 700

600

500

400 27.502� 326" j 300 99.9 _(Cu) Cu + Pb (Pb) __

Cu 10 20 30 40 50 60 70 80 90 Pb Weight Percent Arsenic Appendix G 126

Figure 202. Copper-iron binary D system. C L 148 1500 V'f..:.:l..t''� --- -- •I �� 1400 I 1394' IV , 1300 I liquid + s lid Y '\\ Y -Fe 1200 I ' � 1084. 1095' I 1100 �. ,..4.0(E) \\ 1000 solid E + olid Y \ 912'

900 \ 836' \L' )'/,, 800 -Pf solid E + olid a -Fe ,,, Cu 10 20 30 40 so 60 70 80 90 Fe Weight Percent Iron Appendix G 127

203. D Figure Copper-gold binary C 4S systems. � 106 -� t--. L 41 1000 �r-t-== a+L F==::::� � 911D� a+L 900 [tlv.\ �

800

700 a 600

500 0 C/'I 4 0 400 85O �/ a' -,�I �" "" a,\'I" � 300 /1 � n \t,...... 200 / ! � TT0"''' ' ,-....

Cu 10 20 30 40 50 60 70 80 90 Au Weight Percent Gold Appendix G 128

Figure 204. Copper-antimony °C binary systems. I�\ 084S 1000 , " \ \ L \ \ 900 \ 1\ 800 \ 700 , 6450 � 6 ex 11.0 31.0l4\' 31.5 � 3/ 600 �0 \ 1">0 ...... 62.0 ...... f3 � 0 4RRO \ I, '// 526 500 ,..,\ .I 71-."- YJ--l�""...... " IIII 0 ��\ -�/ 400 I � I 400 / {,--K jl t" t� ,tr1 y (Sb) __ 300 £ _--II!I / '-_ IIII 200 I ilIiIl Cu 10 20 30 40 50 60 70 80 90 Sb Weight Percent Antimony Appendix G 129

Figure 205. Copper-silver binary °C system. � 1000 � \ �� ...... L 961.93° 900 ...... , a+L � ./� �/ A SOO 'ov � ...... - a f--- '2 /so 71.9 � 700 7 V 600 a+� \ 500 \

400

Cu 10 20 30 40 50 60 70 80 90 Ag Weight Percent Silver Appendix G 130

Figure 206. Copper-nickel binary °C system. 145 L - ---� 1400 � -�- - �I---- a+L _,..l..------1--- 1200 �------I�IOS4S � - 1000 a SOO

600 400 ./'35S'/ 200 / ,/Ma �n. / /" Tr, ns. Cu 10 20 30 40 50 60 70 80 90 Ni Weight Percent Nickel Appendix G 131

Figure 207. Copper-zinc binary °C 84S system. 1000 I�� ...... ex L � 9 r � + L 900 � 3 1� �� I �835' 800 '\ ex , � I � 700 \ \\ / y � 1.:\ 600 �+y "' +� 9. 8' 558'V ,\ \ 500 468' I 456' 1\E\ 42 , �48.9 / 88 �,fr""" 400 kslI -,I �' + y I I 97.3 ex��' I I I I I (Zn) ..... 300 I I /., / �' 250' I II I II II Cu 10 20 30 40 50 60 70 80 90 Zn Weight Percent Zinc Appendix G 132

Figure 208. Iron-carbon system. 1535 �------�1550 �------'------'T"""------, °C liquid 1400

cementite & liquid 1450 solidus uo austenite �------�------�1130 ...,::J'­., (y) ______D-.,r: 1400 o_ -'-0.2 ...... 0.4 __ -1.....0.6 �E 910 austenite & cementite Percentage of Carbon The 8-Region of the Iron-Carbon Diagram 695 �--�------�695 eutectoid horizontal

pearlite & cementite

ferrite & I pearlite i ------�--�------�------�6 % carbon 2.0 4.3 The Complete Iron-Carbon Diagram 1000 Y

a +y austenite 0u 0.04% 7100 ou'­., ...,'"::J'-.,600 ...,r:::J D-.,'- 400 ED-., austenite & ferrite austenite & cementite I-.,E I-., a + Fe3C ferrite 200 pearlite & ferrite pearlite & cementite 0.006% 0 0 0.05 0.1 0.15 Percentage of Carbon % carbon 0.8 1.4 Constitutional Diagram Indicating The Steel Portion of the Iron-Carbon Diagram Solubility of Carbon in a-Iron Appendix G 133

Figure 209a. The lead-tin system 350 327" (pewters). 300 �\� Liquid 250 f\ �jo...., -...... 252· a ft L � 200 \ �...... --'� , IW � ---- �+L J a 9.2 61.9 97.5 \ ISO /' 10so0 I n . R o I I o 10 20 30 40 SO 60 70 80 90 100 Wt % Sn

Figure 209b. The gold-silver system. 1100

1050 �---+-----r----+-----r---�-----+----�--��-1

1000 �----+-----�-----4�-��-----4------�----+-----�

950 �+-----�-----4------�----+------+-----4------�----+-----� ______900 L- � _L � L_ � _L � L_ � � o 10 20 30 40 SO 60 70 80 90 100 Wt %Au Appendix G 134

Figure 210. Copper-silver-gold ternary liquidus.

Au

Cu 10 20 30 40 Wt %Ag 60 70 80 90 Ag Appendix G 135

Figure 21 I. Copper-silver-gold ternary solidus.

Au

Cu 10 20 30 40 Wt %Ag 60 70 80 90 Ag Appendix G 136

Figure 212. Copper-tin-Iead ternary system.

Sn

Schematic of Pb Pb Corner

40 80 90 GLOSSARY

Acicular Amalgam fu rther described by such terms Bee Possessing an elongated or An intermetallic compound or as stress-relief anneal, solid Body-centered cubic. A unit needle-shaped structure. mixture of mercury with other solution anneal, normalizing, cell in which atoms are situated metals. Mercury may fo rm an etc. at each corner of a cube with Admiralty brass amalgam with gold, silver, tin, one atom in the center of the An outdated term fo r an alpha zinc, lead, copper, and other Annealing twin cube. Each atom at the corners brass in which some zinc is metals. The microstructure of In FCC metals, a process of is shared by each neighbor. replaced by tin; usually only these amalgams may be com­ recrystallization (often of Because of the close packing, about 1-2% tin is added. plex. worked and annealed metals) BCC metals such as barium, in which a mirror plane in the chromium, iron, molybdenum, Age-hardening Amalgam gilding crystal growth results in two tantalum, vanadium, and tung­ The process of hardening spon­ The process used fo r the gild­ parallel straight lines appearing sten cannot be heavily worked taneously over time at ambient ing of many copper alloys in across the grain when the metal but they have a good combina­ conditions. Some steels age­ ancient and historic times. is etched. tion of ductility and strength. harden as do other alloys. The Gold becomes pasry when process was first studied in mixed with mercury and may Anode Bidri aluminium-copper where be applied as a paste over a sur­ Component of a system that is Name for an Indian copper­ coherent regions of different face. This can be fo llowed by usually corroded. In an electro­ zinc-lead alioy, finished by sur­ structure form as the first stage heating to drive off most of the chemical reaction, the electrode face treatment to color it black of the process. mercury, or the mercury can be at which oxidation occurs. and often inlaid with silver. applied to the clean surface of Alloy the object to be gilded, fo l­ Arsenic Bloom A mixture of two or more met­ lowed by the attachment of Element with atomic number A roughly finished metallic als. The term alloying suggests gold leaf or fo il. 33, atomic weight 74.9214, product; specifically, the the mixture was deliberate. specific graviry (grey form) spongy mass of iron produced Antimony 5.73. The usual variety is grey in a bloomery furnace in which Alpha brass Element with atomic number arsenic which sublimates at the iron is reduced in situ and is An alloy of copper and zinc 51, atomic weight 12l.75, mp 610 °C. Arsenic is a steel-grey not molten during reduction. with no more than 38% zinc. 630 °C, specific graviry 6.62. A color with a metallic luster and The crude iron bloom must be In antiquiry, the cementation lustrous metal with a bluish sil­ was the first alloying element of extensively worked at red heat process fo r the manufacture of very-white appearance. The importance. Arsenical copper to consolidate the iron and brass meant that only up to metal does not tarnish readily alloys precede the use of tin remove excess slag and char­ 28% zinc might be absorbed in on exposure to air and can be bronzes in most areas of both coal. the copper when the zinc ore used as a decorative coating. the Old and New Worlds. was reduced in situ. Most Antimony is fo und in some Arsenic contents usually vary Brass ancient brasses do not contain copper alloys of antiquiry and from about 1 to 8%. An alloy of copper and zinc, over 28% zinc. in the region of 3% has a con­ usually with copper as the siderable hardening effect. Bainite major alloying element and Aluminum A term more commonly zinc up to 40% by weight. Atomic weight 26.98, atomic Annealing applied in the past to the Early brasses were binary alloys number 13, mp 660.37 °C, A process of heat-treatment decomposition of martensite in containing 90-70% copper specific graviry 2.69. It is the carried out on a metal or alloy, steels by tempering at about and 10-30% zinc. The color of most abundant of the metallic usually to soften the material to 200-350 °C. brass changes with increasing elements but is very difficult to allow further deformation. zinc content from a rich cop­ extract and was not properly Strictly speaking, if an alloy is per-red through pale yellow to known until 1827. involved, annealing should be white as the zinc increases. Glossary 138

Gilding metal containing 10- method is becoming less com- object that has been made by and as pearlite. These irons are 15% zinc is suitable fo r cold mon, but does allow some casting the metal into a shape. usually hard and brittle. (3) working. It is used fo r orna- comparison with results from The simplest forms are open Malleable cast irons are usually mental work and jewelry. Red the Vickers Scale, or other molds that are uncovered at the obtained by heat-treatment of brass contains 30% zinc and scales used more fo r industrial time of casting. This form was white cast irons by converting 70% copper and has good purposes, such as the Rockwell often used fo r simple early the combined carbon into free working properties. The com- Scale. Bronze Age axes. Piece molds carbon or temper carbon. In mon fo rm of brass is 60% cop- are made oftwo or more fitting the whiteheart process, fo r per, 40% zinc and is known as Carat pieces in stone, bronze, or example, a certain amount of yellow brass or Muntz metal A term used to express the refractory clay. Hollow-cast carbon is removed from the (see also alpha brass). degree of purity or fineness of objects are usually piece molds surface by oxidation. gold. Pure gold is 24 carat or with false cores. A figure was Brazing 1000 fine. The fineness of modeled in clay and a piece Cathode Joining of metals together with alloyed gold can be expressed in mold was built up around the In an electrochemical cell, the an alloy of copper and zinc. the number of partS of gold model. The model was component on which reduc- Modern brazing alloys may that are contained in 24 parts removed and could be shaved tion takes place. In many corro- contain copper, zinc, and silver of the alloy. For example, 18 down in size to provide the core sion processes, the cathodic and are often called silver sol- carat gold contains 18/24 partS around which the mold pieces regions are protected during ders. In ancient times, silver- of gold and is 75% gold or 750 and mother molds would be corrosion, while attack takes gold, copper-silver, and silver- fine. assembled. Many variations are place at anodic regions. gold-copper alloys were used possible (see also lost-wax cast- fo r brazing (or soldering) oper- Carburization ing). Cementation ations on precious metals in The process of increasing the The term has several meanings. particular. carbon content of the surface Casting-on Cementation of gold alloys layers of a metal (often wrought The process of making a cast with salt in a cupel may remove Bronze iron) by heating the metal part attached to an already silver leaving pure gold behind. In antiquity and historical below its melting point with existing object or component. Carbonaceous material may be usage, an alloy of copper and carbonaceous matter such as In antiquity, a lost-wax addi- used to cover the wrought iron, tin. Usually with up to 14 % wood charcoal. tion, made by creating a small which is then heated. Carbon tin, but many examples of mold around part of an object can diffuse into the structure ancient alloys are known with Case-hardening and casting on metal directly to creating a low-carbon steel sur- higher tin contents. 14% tin is A process consisting of one or it. Often used fo r dagger han- face by cementation. the limit ofsolid solution of tin more heat-treatments for pro- dies or repair or construction of in well annealed fJ. bronzes. In ducing a hard surface layer on large bronze figures. Cementite modern usage, the term bronze metals as in carburization. In The hard, bri ttle component of is associated with a number of the case ofsteel the carbon con- Cast iron iron-carbon alloys, containing copper alloys that may contain tent of the surface can be An iron-carbon alloy that usu- about 6.6% carbon, corre- no tin at all and the composi- increased by heating in a ally contains 2-4% carbon. sponding to the phase Fe3C tion of the alloy must be speci- medium containing carbon fo l- Generally divided into three and crystallizing in the ortho- fied. lowed by heat treatment. groups: (1) Grey cast iron in rhombic system. It is soluble in which free carbon occurs as molten iron, but the solubility Brinell Casting flakes of graphite. It has excel- decreases in austenite to about The Brinell Hardness Scale in The operation ofpouring metal lent casting properties and can 0.9% at the eutectoid tempera- which the hardness is measured into sand, plaster, or other be machined. (2) White cast ture. Pearlite, the eutectoid of by the resistance to indentation molds and allowing it to solid- iron in which all of the carbon fe rrite and cementite, is the of a small steel ball. This ifY. More generally, a metallic is taken up as cementite, Fe3C most common component 139

containing cementite in ment usually consists of rolling, changes in composition occurs Cupellation ancient and historic steels. hammering, or drawing at as the dendrite arm is fo rmed. Often applied to the removal of room temperatures when the Especially common in ancient basic metals from silver or gold Centrifugal casting hardness and tensile strength cast bronzes and cast silver­ by use of a cupel and either oxi­ Casting by the lost-wax process are increased with the amount copper alloys. Coring is accen­ dation of base metallic constit­ (usually) followed by rapid of cold-work, but the ductility tuated in alloys with a wide sep­ uents or chemical combination rotation of the casting mold to and impact strength are aration between liquidus and with salt. fo rce the molten metal into the reduced. solidus curves. casting spaces. Often used in Cupro-nickel modern dental casting tech­ Columnar CPH Alloys containing copper and niques and more associated Long columnlike grains that Close-packed hexagonal. A nickel, usually from 15% to with modern castings than with can fo rm when a pure metal is hexagonal net in which the 70% nickel, but in ancient ancient practice. cast into a mold. atoms are arranged in a repeat­ alloys often with less nickel ing sequence ABABABA...... than this. Alloys with about Chaplets Continuous precipitation One unit lattice has a hexago­ 25% nickel are now used fo r Small pegs, wires, or other The fo rmation of a precipitate nal prism with one atom at coinage metals. Early examples materials used to hold a hollow or inclusion distribution uni­ each corner, one in the center of copper-nickel alloys are also lost-wax casting in position in fo rmly through the grains of the bottom and top faces, known, the most fa mous being the mold by securing the themselves. and three in the center of the the Bactrian coinage. investment or casting core to prism. CPH metals tend to be the mold so that it will not Copper brittle, e.g., cadmium, cobalt, Damascening move when the wax is molten An element with atomic num­ titanium, and zinc. An ancient process of orna­ out. ber 29, atomic weight 63.54, menting a metal surface with a mp 1083 0c, specific gravity Crimping pattern. In the early Middle Chasing 8.96. Pure copper is reddish in Mechanical join between two Ages swords with this pattern Displacement of metal by use color and malleable and duc­ pieces of metal in which they were said to be from Damascus, of a chasing tool, often of brass tile. It occurs in native copper are deformed to shape an over­ made by repeatedly welding, or bronze or wrought iton. in dendritic masses and has lap or attachment. drawing out, and doubling up a Unlike engraving, metal is dis­ been known fo r thousands of bar composed of a mixture of torted around the chased years. Many copper minerals Cupel iron and steel. The surface was design and is not removed. were used to extract copper A porous ceramic, often made later treated with acid to metal in primitive fu rnaces, in from bone-ash or other refrac­ darken the steel areas. Ferrite Cire perdue the fo rm of copper prills and tory components. The cupel is remains bright. In the East the Term meaning lost-wax later as cakes and ingots of cop­ used to melt small amounts of process of inlaying metal on casting. per. There are about 240 metal, usually silver fo r the metal is common, particularly copper-bearing minerals and extraction of lead, or the assay in parts of Iraq and India, Coherent precipitation both copper oxides and sulfides of gold. In the extraction of where it is known as Kuft work. An atomic rearrangement of were smelted to obtain the lead from silver, the oxidized structure that is imperceptible metal. lead is absorbed into the cupel Dendritic by optical microscopy. leaving a button of silver. The Shaped like the branches of a Coring cupel can then be broken and tree. Dendrites are common in Cold-working The segregation of an alloy on smelted to recover the lead. cast alloys and may look like an The plastic deformation of a successive freezing to the solid. intersecting snowflake pattern. metal at a temperature low Zones are fo rmed, especially in enough to cause permanent dendritic castings, in which a strain hardening. The treat- continuous series of small Glossary 140

Depletion gilding Dislocation thin gold surface. Used in Eutectoid Gilding by removal of one or Defects in the crystal structure ancient Peru as a gilding tech­ Decomposition from a solid more baser components. Com­ ofa metal that allow movement nIque. phase into two finely dispersed monly used in ancient South of planes or atoms within the solid phases creates a structure America fo r the gilding of tum­ lattice. Edge and screw disloca­ Electrum called a eutectoid. The eutec­ baga or gold-silver-copper tions are two common types. Naturally occurring alloys of toid point is fixed in binary alloys by removing copper gold and silver, usually white or alloys and in morphology may from the outer surfaces by pick­ Dislocation silvery in color and containing resemble the eutectic. ling. entanglement up to about 40% silver. The density of dislocations Fayalite Engraving Depletion silvering increases on working the metal The most common component Silver-copper alloys usually until dislocation entanglement Use of an engraver, usually of of ancient slags, often occur­ develop a scale when worked is reached. At this point the hardened steel to remove metal ring as slag stringers in wrough t and annealed. Removing oxide metal is brittle since no further from a surface. iron. Fayalite is an iron silicate, scale enriches the surface in sil­ movement can occur. Anneal­ 2FeO. Si02 which melts at ver, creating a depleted copper ing will restore working prop­ EPNS about 989 °C, crystallizes in the zone and making the alloy sil­ erties. Electroplated nickel silver. An orthorhombic system, and usu­ ver in color. alloy of copper, nickel, and zinc ally takes the fo rm of broken, Drawing typically with 60% copper, elongated grey laths in silicate Diffusion The act of pulling a wire, usu­ 22% nickel, 18% zinc, electro­ based slags. The migration of one alloy or ally of silver or gold through a plated with silver. metal into another. Interdiffu­ drawplate of hard material. Filing sion also occurs with the sec­ Drawing is not thought to have Equilibrium diagram Removal of metal with a file. ondary metal migrating into occurred before the 6th century A synonym fo r a phase dia­ Filing is uncommon in ancient the first. Usually heat is A.D. gram. metalwork since hard steel files required fo r this process to were not available. occur. Ductility Equilibrium structures The ability of a metal to be Microstructures that represent FCC Diffusion bonding drawn or deformed. Ductile full equilibrium phases pre­ Face-centered cubic. In this lat­ Bonding or joining of two met­ metals are usually FCC types dicted from phase diagrams. In tice system there is an atom at als by heating them together. such as silver or gold. ancient metals the structure each corner of a cube and an Each will diffuse into the other, may be far from equilibrium. atom at each center of the cube at different rates, creating a Electrochemical faces. Face-centered cubic met­ Equi-axed strong and permanent metal­ corrOSIOn als tend to be soft and easily Of equal dimensions or proper­ lurgical bond. Corrosion of a metal in which worked, such as silver, alumin­ ties in all directions. Equi-axed anodic and cathodic reactions ium, gold, copper, lead, and Discontinuous grains are hexagonal in fo rm. result in metallic dissolution. platinum. precipitation The most common fo rm of Eutectic Fusion gilding Precipitates laid down at grain corrosion of buried or marine In binary alloys the composi­ boundaries, often by a process metals. A process used in ancient South of aging or of exsolution of tion with the lowest melting America, especially Ecuador, point. The eutectic is a fixed metastable phases. An example Electrochemical fo r the gilding of copper alloys composition in binary alloys by dipping or fu sion of molten is the precipitation of copper replacement plating and is often a fine intermixture gold alloys to the surface, from silver-copper alloys. Cleaned copper will exchange of two phases, typically (J. resulting in thick gold alloy with gold solutions to form a and �. coatings. May also be used to Glossary 141

create silver alloy coatings over Gold fo il Gunmetal Intaglio copper. Any gold sheet greater than 1 May have different composi­ The process of engraving or micron thick. tions, but usually an alloy of removing metal to create a Gama-hada copper, tin and zinc. design. The depression so Japanese decorative technique Grain fo rmed may be filled with making use of immiscible met­ In crystalline metals, the grain Hard soldering niello or enamel. als, such as silver or silver­ is an area or zone of crystal An alternative term fo r the use copper alloy droplets on iron. growth in a uniform and of a brazing alloy or a copper­ Interstitial homogeneous fo rm. Most met­ silver alloy fo r joining, as A small element that may German silver als consist of grains and the opposed to the use of lead-tin occupy lattice spaces without Alloys of copper, nickel, and grain boundaries are the inter­ alloys. causing too great a distortion of zinc usually comprising about face between a succession of the original lattice structure. 52-80% copper, 5-35% grains in the solid mass of Hot-working An example is carbon in iron. nickel, and 10-35% zinc. This crystals. Defo rmation of the metal or Carbon is a small element and alloy was fo rmerly used fo r alloy above the temperature can insert into the cubic iron many decorative purposes as a Grain boundary necessary fo r plastic deforma­ lattice. cheap substitute fo r silver, since segregation tion of the metal. it does not readily tarnish and is The precipitation of a phase or Iron of silvery hue. inclusion at the grain bound­ Hy An element with atomic num­ aries of a polycrystalline solid. Hardness on the Vickers or ber 26, atomic weight 55.85, Gilded Diamond Pyramid Number mp 1528 0c, specific gravity Covered with gold. Granulation (DPN) scale. 7.87. A heavy whitish metal Usually refers to small gold and one of the most abundant Gold grains attached to an object Hypereutectoid and widely distributed. Iron Element with atomic number with solder, either made in situ Containing a greater amount was known early as meteoric 79, atomic weight 196.96, mp by reduction of a copper salt (often of carbon steels) than iron, which usually contains 1063 dc, specific gravity 18.88. some nickel. Extraction by the with glue or by use of diffusion that required to fo rm a com­ Native gold usually contains bonding of the gold grains. pletely eutectoid structure. In bloomery process was common some copper and silver. Typical steels this would require more by the early centuries B.C., and Common in Etruscan gold­ gold concentrations are 85- work. than 0.8% carbon, the amount this continued until the inven­ 95% with the remainder being needed to create a completely tion of the blast furnace. Later mostly silver. Gold is bright Graphite pearli tic structure. methods of extraction in the yellow, but with increasing sil­ An allotropic fo rm of carbon, Western world were able to ver content the color is white, occurring in the trigonal sys­ Hypoeutectoid melt iron and, by the 13th cen­ while copper provides red tints Containing a lesser amount tury A.D., make cast iron. In tem as grey, soft, lustrous to the color. Platinum from 20 plates. It is the fo rm of carbon (often of carbon steels) than China, iron (with carbon) to 25% and nickel make the fo und in steels and cast irons that required to fo rm the eutec­ could already be molten during alloy with gold white. and usually occurs in grey cast toid sttucture of 0.8% carbon. the time of Christ and cast iron irons as thin flakesor nodules. Most ancient steels are hypoeu­ was produced much earlier Gold leaf tectoid, except for Wootz steels than in the West. Leaf in ancient technology is Grey cast iron made in a crucible, or later his­ rare. The term is reserved fo r Cast iron with a grey fracture. torical products, such as cut Latten gold less than 1 micron thick. The fracture color is indicative steel beads from France. Copper-zinc-tin alloy (some­ that the cast iron contains free times with lead as well) used in carbon as graphite. the medieval period fo r cheap Glossary 142

decorative and functional met- liquidus is a surface, not a line. uid at room temperature. The Peritectoid alwork. earliest extractions were carried An isothermal reaction in Lost-wax casting out by roasting cinnabar, mer- which two solid phases in a Lead Casting from a wax model. The cury (II) sulfide, in an oxidizing binary system react to fo rm a Atomic number 82, atomic object to be made is shaped in atmosphere and collecting the new solid phase. A peritectoid weight 207.19, mp 327.4 oc, wax (either solid or hollow) and mercury by distillation. reaction occurs in the bronze specific graviry 11.35. Pure is covered in a clay mold. system, for example, in which lead recrystallizes at room tem- When the wax is molten out, Meteoric iron at 65% copper a reaction perature when deformed. The the space can be filled with Iron from outer space. Usually occurs between CU3Sn, and the metal can readily be extruded molten metal, usually bronze an alloy of iron and nickel. solid solution gamma, produc- into pipes or rod but lacks the or brass. Small amounts of cobalt and ing a new phase, CU4Sn at tenaciry to be drawn into wire. manganese are rypical. Some about 580 0c. Lead was commonly extracted Martensite early iron exploitation made from galena, lead sulfide, and Often used only for the hard, use of meteoric iron. Pewter was often a by-product of the needlelike component of Ancient pewter is an alloy of extraction ofsilver from galena, quenched steels, but more gen- Nitriding lead and tin, much used in since many of these lead ores erally, any needlelike, hard In steels, the hardening due to Roman times. The poisonous are argentiferous. Lead is a use- transformation product of a nitrogen content that may nature of lead has resulted in fu l addition to bronzes and quenched alloy. The most result in nitrides being formed the replacement of lead with brasses, especially fo r making common in ancient materials is in the alloy. antimony, although antimony castings and is used as an alloy martensite in low-carbon steels, is also inadvisable in high with tin as a soft solder. or martensite in beta-quenched Open mold amounts fo r utensils. Common bronzes. A primitive fo rm ofcasting into pewter in antiquity may consist Leaf gilding an open-shaped depression in of 60-80% tin, 40-20% lead, Covering with gold by the Martensitic stone, sometimes covered par- while modern pewter may be application of gold leaf (or transformation tially to prevent excessive oxi- 15-30% copper, 5-10% anti- fo il). Sometimes held mechani- A product formed by rapid dation. mony, and 87-94% tin. cally by roughening the surface cooling of an alloy. Some alloys or by a diffusion bond to the may be specially fo rmulated to Pearlite Phase substrate metal. allow martensitic events to The fine mixture of fe rrite and A homogeneous chemical com- occur on cooling. cementite fo und in steels. The position and uniform material, Ledeburite eutectoid, pearlite, will be com- describing one component in a Name applied to the Mechanical twinning plete when the carbon content metallic system. cementite-austentite eutectic at Twinned crystals produced by reaches 0.8%. In most ancient 4.3% °carbon which freezes at mechanical strain alone, as in steels, a mixture of fe rrite and Phase diagram 1130 C. During cooling the zinc. pearlite is common. A diagram with axes of temper- austenite in the eutectic may ature and composition describ- transform into a mixture of Mercury Peritectic ing the different phases that cementite and austenite. Atomic number 80, atomic Reaction of a phase that has may occur in an alloy with weight 200.59, mp -38.84 0c, fo rmed with a liquid of a diffe r- change in either composition Liquidus specific graviry 13.55. Mercury ent composition to fo rm a new or temperature. A binary phase The line on a binary phase dia- has been fo und in Egyptian solid phase. The new phase diagram consists of two metal- gram that shows the tempera- tombs of 1500 B.C. and was may consume all of the liquid lic components. A ternary sys- ture at which solidification widely known in the centuties to fo rm a totally new solid, ryp- tern, which is usually more begins upon cooling from the B.C. in China and India. A sil- ically beta phase in the bronze complex, consists of three melt. In a ternary diagram the very white metal which is liq- system. metals. Glossary 143

Piece-mold particles are called prills and create raised designs on the One of the noble metals that is A mold taken from a model were often extracted by break- front. not oxidized by heating in air. that may be assembled in a ing up the smelted product and It is white in color and very number of pieces before being sorting the metal. In crucible Riveting ductile and malleable. Silver used fo r slushing wax over the processes, prills are small drop- Joining of metal sheet by small was usually obtained by cupel- mold interior fo r lost-wax cop- lets of metal adhering to the metal pegs passing through and lation of lead ores, although it ies of a master model. Such crucible lining. hammered down. may also be extracted directly techniques were common in from silver sulfide deposits. the Renaissance. Pseudomorphic SaHdruy Pure silver is often stated to be The replacement in the corro- Islamic term fo r high tin 1000 fine and alloys are based Platinum sion process of a material with bronzes, often white in color. on this nomenclature. For Metallic element atomic num- another that mimics the fo rm example, sterling silver (qv) is ber 78, atomic weight 195.08, of the replaced product. Segregation 925 fine. mp 1772 0c, specific gravity Pseudomorphic replacemen t of In alloys usually of three fo rms: (l) Sinking 21.45. First discovered in organic materials is common normal segregation, (2) South America by Ulloa in on iron artifacts and can occur dendritic segregation, and (3) A technique with which a vessel 1735, but used by the Indians on copper alloys and silver- inverse segregation. In normal can be produced by hammer- of Ecuador and Colombia who copper alloys as well. segregation, the lower melting ing from the inside. The sheet sintered the metal with gold. point metal is concentrated metal is hammered either on Finds are known particularly Quenching towards the inner region of the the flat surface of an anvil or from La Tolita dating to the The act of quickly cooling a cast. In dendritic segregation, more commonly hammered early centuries B.C. It can be metal or alloy by plunging into fe rnlike growth occurs from into a shallow concave depres- present in the native state, usu- cold water or oil. local compositional gradients. sion in the anvil. Also called ally alloyed with some iron. In inverse segregation, the blocking or hollowing. The metal is malleable and Quenched structures lower melting point constitu- ductile. It is used in early scien- Usually nonequilibrium struc- ents, such as tin or arsenic in Slag tiflc instruments since its coef- cures or phases that have been bronzes is concentrated toward A glassy phase or mixture of flcient of expansion is very made metastable in an alloy by the outer cast surfaces. phases usually to be fo und in similar to soda-lime glass. quenching in water or oil. The ancient or historic wrought most common quenched prod- Shakudo iron or steel. The slag is an Po lycrystalline ucts are martensite in steels and Japanese term for the deliberate important by-product of the Consisting of many individual martensite in high-tin bronzes. use and manufacture of gold- smelting of metals. It may be crystalline grains. Most metals Quenching may also be used to copper alloys. incorporated in copper or iron are polycrystalline solids. suppress ordering reactions, alloys as a result of incomplete especially in gold alloys and Shibuichi separation or incomplete melt- Polygonal some ancient texts refer to this Japanese term fo r decorative ing during extraction, as in the Many-sided. Some grain shapes practice to avoid embrittle- silver-copper alloys often bloomery process. may be polygonal. ment. worked and annealed to create decorative surfaces when col- Slag stringers Prill Repousse ored by chemical etching or Small pieces of slag that have In the extraction of copper Working from the back of a staining. become incorporated into the from primitive smelts, the metal, often on the slight relief metal and are then strung out metal is produced as small of a chased design on the front. Silver as small elongated ribbons as a The metal is then displaced by An element of atomic number result of working the metal to droplets or particles in a slaggy matrix. These small metallic hammering, often on a soft 47, atomic weight 107.87, mp shape it. support such as pitch, so as to 960 °C, specific gravity 10.50. Glossary 144

Slip planes Speculum working due to slip of planes of fo rm is stable and below this FCC metals may show slip A name sometimes applied to atoms past each other. the grey cubic fo rm may exist. planes, a fine series of lines in Roman or bronze mirrors, con- Above 170°C, tin is rhombic rwo intersecting directions taining a high percentage of Striking in crystal structure. The metal upon heavy deformation of the tin. Historic speculum may A method of making coins and has low tensile strength and metal. contain about 67% copper, medals. The impression is cut hardness but good ductility. 33% tin. Ancient mirrors made in negative in a very hard mate- Slush casting use of similar alloys, usually of rial and this die is then placed Tinning A method of casting in which beta bronze composition and over the coin blank and given a The operation of coating a base metal is often spun or agitated up to about 24% tin. single heavy blow thus com- metal (usually) with tin. The in the mold so that a thin shell pressing the metal of the blank coating may be obtained by hot is fo rmed. More common in Spinning into the recesses of the die. dipping into molten tin or by ancient and historic metalwork Turning on a lathe fo llowed by Before the introduction of flowing the molten tin over the is slush wax work in which wax depression of metal while in steel, bronze coins could only surface of the object. is slush cast over a piece mold motion. have been struck using stone or interior before investment. bronze dies. Striking may cause Troosite Stamping stress-related fe atures in the A very fine mixture of pearlite, Soft solders Displacement of metal by a struck metal such as surface not resolvable by optical A term applied to lead-tin hard die often fo r assay pur- cracking or internal defects. microscopy. Nodular troosite is alloys used in soldering, usually poses. fo und in steels not cooled not of precious metals. The Tempering quickly enough to fo rm mar- upper limit of the melting Steel Usually applied to steels in tensite. On etching in low- range is about 300 °C, and A malleable alloy of iron and which some of the hardness is carbon steels, troosite appears many alloys melt at about 130- carbon that contains about removed by heating at moder- as a blue etching component in 180°C. 0.1-1.9% carbon. The carbon ate temperatures, from 450 to nital. is present as cementite, usually 650 °C, depending on the type Solidus as a component of pearlite. of tempering required. In Tumbaga The line in the phase diagram Low-carbon steels contain ancient metals, tempering, or The name given in ancient that separates the pasty stage of from 0.09% carbon to 0.2% self-tempering of martensite South America to the alloys of the alloy, usually a mixture of and are soft. Medium carbon would have been carried out to copper and gold, of wide range solid and liquid, from the com- steels contain 0.2-0.4% carbon reduce the brittleness of the of composition and color. pletely solid alloy below the and high carbon steels more fu lly quenched steel cutting temperature of the solidus line. than 0.4%. edges of swords. Tutenag The solidus temperature may Copper-zinc-nickel alloy of be different at diffe rent alloy Sterling silver Tin silvery color imported from compositions, depending on A common coinage binary An element of atomic number China to Europe in 18-19th 50, atomic weight 118.69, mp century A.D. the type of phase diagram. alloy of 92.5% silver, 7.5% 231.8 dc, specificgravity (grey) copper. Sorbite 5.75, (white) 7.31. A soft white TTT diagrams A decomposition product of Strain hardening lustrous metal obtained almost Time-temperature-transforma- martensite fo und in steels. It A synonym for work entirely from the mineral cas- tion diagrams. Most useful in consists of fine particles of hardening. siterite, Sn02' Tin is not considering the nature of the cementite in a matrix of fe rrite. affected on exposure to air at quenched microstructure to be Sorbitic structures may be Strain lines ordinary temperatures. At tem- fo und in steels and high tin rounded cementite not neces- Same as slip planes. Often seen peratures above 13.2 °C the bronzes. The quenching rate sarily derived from martensite. in FCC metals after heavy white tetragonal allotropic and composition results in a Glossary 145

differing series ofTTT dia­ feature of wrought low-carbon was not known in Europe until grams that are commonly used steels. rediscovered in 1746. Zinc is a to investigate transformation bluish-white, lustrous metal, effects on components fo und Wootz brittle at ordinary temperatures during quenching of alloys. Wootz is a kind of steel, made but malleable at 100-150 0c. in small crucibles in ancient Vickers India and often of hypereutec­ A hardness scale that uses a dia­ toid steel with a very low slag mond having an angle of 136 content. This cast steel was degrees between the faces. The widely used for the manufac­ loading can be used easily fo r ture of sword blades and other microhardness measurements. quality products. It is one of the most useful hardness testing scales. Wrought The process of hammering or Weld deforming a metal or alloy, as A term used to describe a joint opposed to casting. made between two metals made by the heating and join­ Wrought iron ing the separate parts with no Iron that has been produced solder applied. Ancient welds from the bloomery process and were often made in precious has been consolidated by ham­ metals, such as gold and silver mering and annealing into a and in the joining of iron com­ wrought product. Wrought ponents, especially in the fabri­ iron usually contains slag cation of wrought iron. stringers that have been elon­ Wh gated and flattened in the pro­ ite cast iron cess of working from the Cast iron with a white fracture bloom. due to the presence of the car­ bon in the cast iron as cement­ Zinc ite rather than free graphite. Element of atomic number 30, atomic weight 65.38, mp Widmanstatten 419.58 °C, specific gravity A type of structure that fo rms 7.133. Zinc ores were used fo r when a new solid phase is pro­ making brass by cementation duced from a parent solid phase long before the metal was used as plates or laths along certain in its pure fo rm. The limit of crystallographic planes of the zinc that can enter into solid original crystals. The structure solution in copper by this pro­ is associated with many mete­ cess is 28%. The Romans made oric irons, and with changes extensive use of brass and, in upon cooling in worked iron India, zinc was being made by and copper alloys. Most com­ distillation in retorts during the monly fo und as an incidental 13th century A.D. The metal

BIBLIOGRAPHY

American Society fo r Metals Annotated Metallographic Charles, ]. A. and ]. A. Leake B. Floyd Brown, Harry C Bur­ 1972-1973 Specimens. 1972 nett, W. Thomas Chase, Mar­ Atlas of Microstructures of Problems in the Fluorescence tha Goodway, Jerome Kruger, Industrial Alloys. In Metals 1972 Analysis of CuiAg Byzantine and Marcel Pourbaix (eds.). Handbook. Vol. 7, 8th ed. The role of microstructure in Trachea and Metallurgical NBS Special Publication 479. Metals Park, Ohio: American metals. Betchworth, Surrey: Information from Sections. In Washington, D.C: U.S. Society fo r Metals. Annotated Metallographic Methods of Chemical and Metal­ Department of Commerce. Specimens. lurgical Investigation of Ancient Allen, I. M., D. Britton, and Coinage. E. T. Hall and D. M. Cottrell, A. H. H. Coghlan Beraha, E. Metcalf (eds.). Royal Numis­ 1975 1970 1977 matic Society Special Publica­ An Introduction to Metallurgy. Metallurgical Reports on British Color Metallography.Metals tion No. 8. London: Royal London: Edward Arnold. and Irish BronzeAge Implements Park, Ohio: American Society Numismatic Society. Fink, C G. and A. H. Kopp and Weap ons in the Pitt Rivers fo r Metals. Museum. Pitt Rivers Museum Coghlan, H. H. 1933 Occasional Papers on T echnol­ Brandon, D. G. 1975 Ancient Egyptian Antimony ogy No. 10. Oxford: Oxford 1966 Notes on the Prehistoric Metal­ Plating on Copper Objects. Universiry Press. Modern Techniques in Metallog­ lurgy of Copper and Bronze in Metropolitan Museums Studies raphy. London: Butterworths. the Old Wo rld. 2nd ed. Pitt 4:163-167. Anderson, E. A. Rivers Museum Occasional Fink, C G. and E. P. 1959 Brown, B. F., H. C. Burnett, Papers on Technology No. 4. Zinc: the science and technol­ W. T. Chase, M. Goodway, ]. Oxford: Oxford University Polushkin ogy of rhe metal, its alloys and Kruger, and M. Pourbaix Press. 1936 compounds. In Th e Physical (eds.) Microscopic Study of Ancient Metallurgy of Zinc. C H. 1977 Coghlan, H. H. and R. F. Bronze and Copper. Transac­ Mathewson (ed.). New York: Co rrosion and metal artifocts: a Tylecote tions of the American Institute of Reinhold Publishing; London: dialogue between conservators 1978 Mining and Metallurgical Engi­ Chapman and Hall. and archaeologists and corrosion Mediaeval Iron Artefacts from neers. Metals Division. scientists. NBS Special Publica­ the Newbury Area of Berk­ 122:90-1 17. Bachmann, Hans-Gert tion 479. Washington, D.C: shire: Metallurgical Examina­ 1982 U.S. Department of Com- tion. Jo urnal of the Historical Ghandour, N. The Identification of Slags from merce. Metallurgy Society12 :23-29. 1981 Archaeological Sites. Institute of Elemental and Metallographic Archaeology: Occasional Pub­ Bushnell, G. H. S. Cooke, R. B. S. and S. Aschen­ Analysis of Objects from Tell lication No. 6. London: Insti­ 1951 brenner Ajjul, Palestine. M.A. Disserta­ tute of Archaeology. The A rchaeologyof the Santa 1975 tion (unpublished). University Elena Peninsula in South- West The occurrence of metallic iron of London, Institute of Archae­ Bailey, A. R. Ecuador. Cambridge: Cam­ in ancient copper. Jo urnal of ology. 1960 bridge University Press. Field Archaeology2: 251-266. Goodway, M. and H. C A Textbook of Metallurgy.Lon­ Charles, ]. A. Corrosion and metal art cts: a Conklin don: Macmillan. ifo 1968 dialogue between conservators 1987 The First Sheffield Plate. and archaeologists and corro­ Quenched high-tin bronzes 1967 Antiquity42:2 78-285. sion scientists from the Philippines. Archaeo­ Th e structure and strength of metals. Betchworth, Surrey: 1979 materials 2: 1-27. Bibliography 148

Hall, E. T. and D. M. Metcalf Lechtman, H. N., A. Erlij, and some experimental and archae­ Petzow, G. and H. E. Exner (eds.) E. J. Barry ological considerations. Archae­ 1975 1972 1982 ometry28: 133-162. Recent Developments in Met­ Methods of Chemical and Metal­ New perspectives on Moche allographic Preparation Tech­ lu rgical Investigation of Ancient metallurgy: techniques of gild­ Needham, Sir J. niques. Microstructural Science. Co inage. Royal Numismatic ing copper at Lorna Negra 1954 3:296-306. Society Special Publication No. Northern Peru. American Science and Civilisation in 8. London: Royal Numismatic Antiquity47:3 -30. China. Vol. 5, Part II. Cam­ Phillips, V. A. Society. bridge: Cambridge University 1971 Lord, J. O. Press. Modern Metallographic Tech­ Hendrickson, R. C. 1949 niques and their Applications. 1967 Alloy Sy stems. New York: Pit- Nelson, J. A. New York: ]. Wiley & Sons. Applications of Metallurgical man. 1985 Microscopy. New York: Bausch Metallography and Micro­ Poluschkin, E. P. & Lomb. Maryon, H. structures. "Cast irons." In 1964 1971 Metals Handbook. Vol. 9, 9th Structural Characteristics of Hongye, G. and H. Jueming Metalwork and Enamelling. ed. Metals Park, Ohio: Ameri­ Metals. Amsterdam: Elsevier. 1983 New York: Dover Publications. can Sociery for Metals. Research on Han Wei spheroi­ Republished from 1959, 4th Rajpitak, W. and N. J. Seeley dal-graphite cast iron. Sundry edition. London: Chapman Oddy, Andrew W. 1979 Trade Jo urnal International and Hall. 1977 The Bronze Bowls from Ban 5:89-94. The production of gold wire in Don T a Phet, Thailand: an Mathewson, C. H. antiquiry: hand making meth­ enigma of prehistoric metal­ Kiessling, R. and N. Lange 1915 ods before the introduction of lurgy. Wo rld Archaeology 1963-1964 A Metallographic Description the draw plate. Gold Bulletin 11:26 -31 . Non-metallic inclusions in of Some Ancient Peruvian 10:79-87. steel. Iron and Steel Institute: Bronzes from Machu Picchu. Rajpitak, W. Sp ecial Report No. 90. American Jo urnal of Science 1984 1983 (190)240:525-6 16. Ancient jewellery as a source of Th e develop ment of copper alloy Lechtman, H. N. technological information: a metallurgy in Thailand in the 1971 Mayes, P. and L. Butler study of techniques for making pre-Buddhist period, with special Ancient Methods ofGilding 1977 wire. In Fo urth International reference to high-tin bronzes. Silver: Examples from the Old Excavations at Sandal Castle Restorer Seminar: Veszprem PhD thesis (unpublished). 4 and New Worlds. In Science 1964- 1973. Wakefield Histor­ 241-252. Budapest: National vols. University of London, and Archaeology. R. H. Brill ical Society. Centre of Museums. Institute of Archaeology. (ed.). Cambridge, Mass: MIT Press. 2-30. McCrone, W. C. and O'Neill, H . Rhines, F. N., W. E. Bond, J. G. Deily 1934 and R. A. Rummel 1973 1974 Th e Ha rdness of Metals and its 1955 The Gilding of Metals in Pre­ Th e Particle Atlas. 2nd ed. Vol­ Measurement. London: Chap­ Constitution ofordering alloys Colombian Peru. In Applica­ umes I-IV. Michigan: Ann man and Hall. of the system gold-copper. tion of Science in the Exa mina­ Arbor Science Publication. Transactions of the American tion of Works of Art. W. ]. Petzow, G. Society of Metals 47:578-597. Young (ed.). Boston: Museum Meeks, N. D. 1976 of Fine ArtS. 38-52. 1986 Metallographic Etching. Metals Tin-rich surfaces on bronze- Park, Ohio: American Sociery fo r Metals. Bibliography 149

Rollason, E. C. 1991 1986 1973 Technological, analytical and Th e Prehistoryof Metallurgy in Metallurgyfor Engineers. Lon­ microstructural studies of a the British Isles. London: The don: Edward Arnold. Renaissance silver basin. Institute of Metals. Archeomaterials 5:21-45. Samans, Carl H. 1987 1963 Slunder, C. J. and W. K. Boyd The Early History of Metallurgy Metallic Materials in Engineer­ 1971 in Europe. London: Longman. ing. New York: Macmillan. Zinc: its corrosion resistance. London: Zinc Development Tylecote, R. E. and B. J. J. Samuels, L. E. Association. Gilmour 1980 1986 Op tical Microscopy of Ca rbon Smith, C. S. Th e Metallography of Early Fer­ Steels. Metals Park, Ohio: 1960 rous Edge Tools and Edged American Society fo r Metals. A Historyof Metallography. Weapons. BAR British Series Chicago: University of Chi­ number 155. Oxford: British Schweizer, F. and P. Meyers cago Press. Archaeological Report. 1978 Authenticity of ancient silver 1971 Untracht, Oppi objects: a new approach. Masca The Technique of the Luristan 1975 Journal :9-10. Smitho In Science and Archaeol­ Metal Techniques for Crafts­ ogy.R. H. Brill (ed.). Cam­ men. N ew York: Doubleday. Scott, D. A. bridge, Mass: MIT Press. 1980 West, D. R. E. 32-54. The conservation and analysis 1982 of some ancient copper alloy 1981 Temary Equilibrium Diagrams. beads from Colombia. Studies A Search for Structure: Selected 2nd ed. London: Chapman in Conservation 25:157-164. Essays on Science, Art and His­ and Hall. tory.Cambridge, Mass: MIT 1983 Winchell, A. N. and Press. Depletion gilding and surface H. Winchell treatment of gold alloys from Thompson, S. C. and A. K. 1964 the Narifio area of ancient Chatterjee Th e Microscopical Characters of Colombia. Journal of the His­ 1954 Artificial Inorganic Solid Sub­ torical Metallurgy Society The age-embrittlement of sil­ stances: Optical Properties of 17:99-115. ver coins. Studies in Co nserva­ Artificial Minerals. 3rd ed. tion 1:11 5-126. New York: Academic Press. 1986 Yakowitz, H. Gold and silver alloy coatings Tylecote, R. F. over copper: an examination of 1968 1970 some artefacts from Colombia Th e Solid Phase Welding of Met­ Some uses of color in metallog­ and Ecuador. Archaeometry als. London: Edward Arnold. raphy. Applications of Modern 28:33-55. Metallographic Techniques. 1976 ASTM STP 480:50-65. Met­ A History of Metallurgy.Lon­ als Park, Ohio: American Soci­ don: The Metals Society. ety fo r Testing and Materials.

INDEX

A solid solution 16 Brinell hardness testing 2, 77 microstructure 40 acidified potassium dichromate austeni tic grain boundaries 31, bronze 5, 25 scales xvii, 42 72 35 alloy 23 spheroidal graphite 37 acidified selenic acid 73 austenitic grains 21, 115 annealed 26 whiteheart 37, 42 acidified sulfate/chromic oxide austenitic region 17 bell 93 cast metals 6 73 Australia 106 beta-quenched 28 cast, quenched, worked 94, 98 acidified thiosulfate/acetate 73 axe 92 cold-worked 26 casting 5 acidified thiourea 73 fragment 105 dagger hilt xvi defects in 24 aggregate carbon 41 azurite crystals 11 7 examination of25 cathodic graphite flakes xviii alcoholic fe rric chloride xvi, 72 grain size 53 cementation 19 alkaline sodium picrate 70 B high-tin 25 cementite xvi, xix, 23, 32, 37 alloys Bachmann 50 ingots 28 free 38 annealing 7 banded structure 32 low-tin 25 grain boundary 31 two-phase 14 Baumann's print solution 70 mirror 28 laths xviii, 42 alpha 14, 15 bead sculpture xvi needles xix, 13,35, 36 alpha + beta brass 19 French cut-steel 36 system 26 networks 70 alpha + beta eutectic 12, 14 gold necklace 99 Bronze Age 88, 97 cements 50 alpha + delta eutectoid 25 Beaujard's reagent 70 sword xx ceremonial axe alpha + epsilon phase region 25 bell 26, 93 Buehler copper 47 alpha brass 19,20 bending 10 epoxide resin 54 Charles and Leake 7 alpha dendri tes 13 Beraha's reagent 70 Mastertex cloth 55 chasing 109 alpha solid solution 12 beta 14, 18 wafering machine 63 chemoepitaxy 43 alumina 64 brass 19 Bushnell 92 chill castings 6 in polishing 66 btonze 26, 28 Byzantine 114, 116 China 19, 27, 37, 38, 42, 93 ammonia/hydrogen peroxide needles 28 cast iron 39 72, 73 beta-phase brass 20 C Chinese cast iron lion xvi ammonium persulfate/ beta-phase martensite 95 cannonball 42, 96 chromic acid/sulfate etchant 74 potassium cyanide 72 beta-quenched bronze 28 carbide 37 chromium (VI) oxide 72 billet 89 carbon 23 close-packed hexagonal 1 ancient steel 16 anisotropic 39, 49 binary tin bronzes 26 alloy 23 coin 86, 97, 101, 117 annealed casting 93 blackheart 39 equivalent value 38 zmc xx annealing 3, 6, 14 blade fragment 114 steel 15 cold-work temperatures 7 blast furnace 37 carbon content 32 degree of9 twins 25 bloom 89 of ancient objects 16 Colombia 23, 46, 90, 95, 99, aqua dag 63 bloomery iron 102, 114 of cast iron 37 107, 115 aqua regia 71, 99, 100 body-centered cubic 1, 49 cast 87, 88, 92, 93, 95, 96, 97, color etching aqueous ammonium persulfate bowl 94, 98, 112 99, 103,10 5,10 7, 108, 117, copper alloys 74 72 Roman 112 118, 119 color metallography 73 aqueous ferric chloride 72 bracelet 111 cast and worked 111, 11 2 columnar growth 6 Araldite XD725 59 brass 19, 26 cast bronze xx compounds 22 archaeological materials 11, 14, alloy 23 cast bronze incense burner 27 intermetallic 22 15 grain size 53 cast iron 16 concentrated HCI 71 arrowhead 88, 114 jig 63 cannonball xviii, xix, 42, Cooke and Aschenbrenner 24 ASTM number 51 medallion xvi 96 cooling 115 atomic packing factor 1 bright-field unpolarized etchants 69 cooling rate 14, 17, 23, 24, 25, austenite 38, 39, 41 illumination 49 interference etchant 73 31, 32, 39, 51 Index 152

copper 9, 12,22, 47, 102 D E eutectoid copper alloy 23, 90, 97, 101 daggers 24 ear ornament 46, 95, 107 interdendritic 25 etch ants 72 blade 116 ear spool 99 pearlite 13 interference etchants 73 handle 94 early 20th century 118, 119 phase 15 copper and tin alloys 25 damage Early Medieval 113 point 17 copper corrosion products xvi due to sampling 58 earring 106 structures 15 copper globules 47 damping capacity 37 Eastern Han dynasty 38 transformation 16 copper oxide 28 debased silver 21 Ecuador 47, 90, 92 -type transformations 15 copper sulfide inclusions xvi alloys 7 edge dislocation 2, 3 copper sulfides 50 coin 117 edge retention 64, 75 F copper-arsenic alloy 92 deformation edged tools 16 face-centered cubic 1, 7,49 copper-gold alloy 95, 115 percentage of 9 electrochemical replacement false-filigree 107 copper-lead alloy 23 deformed grains 8 plating technique 47 fayalite 50 copper-silver alloy 21, 117 degree ofcold-work 9 electrolytic polishing 66 fe rric chloride xvi, 25 etchants 72 delta 15 electromechanical polishing 66 ferrite xvii, xix, 21, 24, 31, 32, copper-tin alloy 25, 88, 93, 94, dendrite arms 13,25 elongation 41 41, 42 98, 103, 105, 111, 112 native copper 103 embedding grain-boundary 21 copper-tin phase diagram 18 dendrites 5, 40 small samples 65 grain structure 29 copper-tin system 16,26 alpha 13 emulsion 23 grains 35, 104 copper-tin-arsenic alloy 88 ghost 25 England 35, 42 massed 21 copper-tin-lead alloy 93, 108 growth 5, 25 medieval 100-102, figurine 108 copper-tin-zinc-lead alloy 87 noncontinuous 86 113, 115 filligree 95 copper-tin-zinc system xviii segregation 5 engraving 109 Fink and Kopp 6 copper-zinc alloy 97, 105, 107, silver-rich 86 epitaxial growth 43 flake graphite 38, 41 113 structure 6, 23, 49 Epoxide 75 France 36 copper-zinc system 19 dendritic arm spacing 5 epoxy 63 free cementite 38 cored bronzes 25 measurment 51 epoxy resin 75 free fe rrite 38 coring xvi, xx, 5,12 ,25, 27, 87 depletion epsilon phase 19, 25, 26 free graphite 37, 40 Cornwall 120 of eutectic phase 14 epsilon-carbide 35 French cut-steel bead 36 corrosion 107, 113 of the eutectic phase 14 equi-axed grains xvi colors 49 depletion-gilding 46 equilibrium 5, 31 G of gold-copper alloys 46 diamond polishing 44 defined 5 gamma iron 24 corrosion crusts 57 Diamond Pyramid Number diagram 11, 25 gamma phase 16 corrosion processes 43 2, 77 structure 6 gamma solid solution 16 metallic fragments 43 dimethylglyoxime nickel test eta and tin 19 Ghandour 88 corrosion products 43, 44, 57 71 eta phase 19, 24, 26 gilded silver 106 distribution 49 disc etch pitting 104 gilding 116 examination 56 decorative 90 etching 67 glycerol etchant 71 morphology of 49 discontinuous precipitation 21 polished metal 69 gold 22, 107 crab-type graphite 41 dislocation entanglement 3 etching solutions alloys 11 crystal fault 2 dislocations 2 fo r archaeological gold alloy cuprite 43, 44, 46, 107 dispersion materials 69 etchants 71 cuprite lamellae 90 oflead 27 eutectic 14 gold fo il 57 curing 63 drawing 6 eutectic phase diagram 12 gold-copper alloy 22, 23, 46, Dube 21 eutectic point 12, 14 90, 95, 107 ductility 3 eutectic structures 12 gold-copper system 23 Index 153

gold-silver alloy 99 hot-working 7 iron oxide 38 leaf gilding 106 gold-silver system 11 Hv 2 crust XVI Lechtman 22, 47 grain boundaries 20, 21, 26, Hydrogen peroxide/iron (III) iron phosphide 37 ledeburite 40, 41 28, 32, 46, 49 chloride 71 iron-carbon alloy 96, 104, 116, eutectic 38 austentitic 31 hypereutectic 38 118 of cast iron 38 grain boundary cementite 31 hypereutectoid steel 21 iron-carbon phase diagram 16, transformed 38 gram size hypoeutectic 38 17 liquidus 24 measurement 51 hypoeu tectoid steel 21 iron-carbon-phosphorus alloy liquidus line 11 grain structure 119 lost wax 93, 95, 107 equi-axial 6 I iron-iron carbide diagram 40 low-carbon steel 31 hexagonal 6 immiscible structures 23 Islamic 26, 81 low-tin bronzes 25, 27 grain boundary fe rrite 21 incense burner xvii, 93 Islamic inkwell xviii, 81 Luristan 94 , cast bronze 27 isotropic 42, 49 Luristan ceremonial axe xx 47 grams austenitic 21 inclusions 101 Luristan dagger handle xvi graphite xvii, 37, 40 cuprite 90 J flakes 38, 42 cuprous oxide 90 Japanese sword 29 M free 40 in ancient metals 7 Japanese swordblade xix magnesium oxide matrix 38 nonmetallic 49 Java xix, 26, 28, 104 in polishg 66 nucleation 39 sulfide 71 jeweller's saw 61 magnetite 50 polygonal 39 India xix, 19, 26, 35, 36 Jordan 106 manganese 37 spheroidal 39 Indian Wootz ingot xix manufacturing processes 57 graviry segregation 27 Indian zinc coin 9 K martensite 20, 23, 32, 35, 98, Greek Herm xviii, xx Indo-Greek 86 kish graphite 38, 42 101 grey cast iron xviii, 37, 38, 96, ingot 97, 105, 119 Klemm's reagent 70 martensite needles 35 118 inkwell 81 knife 101, 115 martensitic 26 grinding 24, 26, 54, 65 intercept method 51 grain size 31 transformation 20, 106 growth spirals 102 interdendritic channels 5 Knoop 77 Maryon 6 interdendritic porosiry 47 Korea 26 massed fe rri te 21 H interdiffusion 26 kris 36 matte 50 hacksaw 61 interference film McCrone et al. 50 hammering 6 metallography 73 L McCrone low-level Han Wei period 37 intermetallic compounds 22 Late Bronze Age 88, 103 micro hardness tester 77 hardness 2 internal oxidation 7 lath martensite 29, 35 medallion 101 heat treatment 39 interstitial materials 2 lattice structure 1 medieval 115 Hengistbury Head 41 inverse segregation 5 lead 9, 24 Indian zinc coin xx Heyn's reagent 70 inverted stage metallurgical lead alloy knife blade 35 high-tin alloy 26 microscope, 56 etch ants 71 Meeks 27 high-tin bronze 25, 26 Iran xviii, xx 27, 87, 94, 105 lead globules 23, 24, 87, 112 melts 11 mirror xix, 28 Iranian Iron Age dagger hilt lead inclusions 24 metallic bonds 1 high-tin leaded bronze xvii, 19, XVlll lead-copper eutectic 23 metallic zinc 19 28 iron 9, 23, 89, 115 leaded copper 23 metallographic examination 61 Hilliard's circular intercept etch ants 69 leaded high-tin bronze 108 metallography 57 method 52 interference etchant 73 leaded zinc brass xvi metals homogeneous bronzes 25 iron alloy 23, 100, 101, 102, leaded tin bronze 27 deformation 1 homogeneous solid solution 15 114 lead-iron alloy 23 metastable phase 28 Hongye and Jueming 37, 38, iron carbide 38, 39 lead-tin alloys 15 metastable state 21 42 Index 154

mirror xvii, 18, 19, 26, 108 peritectic Q Scott 22, 50, 90 high-tin bronze 28 reaction 18, 22, 24 quasi-flake graphite 41 screw dislocations 2 Javanese 28 structures 17 quench ageing 104 segregation 9, 12 Roman 19, 26, 27 transformation 19, 26 quenching 20, 26, 31 inverse 5 modern 105 peritectoid 22 with water 23 normal 5 molybdate/bifluoride 74 Petzow and Exner 73 selenic acid solution 74 monotectic 23 phase 5 R sheet fragment 113 mottled cast iron 37, 42 defined 5 raising 6 silicon 37, 38, 40 mottled iron 38 diagram 11 rapid cooling 1, 37 silicon carbide 65 mounting resins 75 Phillips 73 recrystallization silver 12, 21, 86, 109 phosphorus 37 temperatures 9 alloys 11 N pickling 86 recrystallized grains 8 impure 21 nail 102 picral 39, 69, 71 redeposited copper 47 silver alloy native copper 102 pig iron 37 reflected light microscopy 57 etchants 72 Near East 26 pigments reflected polarized light 39 silver and gold alloy 106 necklace bead 99 examination 56 reflection pleochroism 39, 42 silver objects 57 Nelson 39 plasters 50 Renaissance 109, 110 silver-copper alloy 12, 13, 15, nickel silver plastic deformation 1 Rhines, Bond, and Rummel 22 57, 86, 109 grain size 53 Plasticine 58 Rockwell 77 eutectic phase diagram 12 nital 39, 69, 71 plate 110 Rollason 96 silver-copper phase diagram 21 nodular graphite 41 plate martensite 36, 71 Roman xvii, 26, 89, 97, 101, Sinu 107 nomograph 51 pleochroism 102, 108, 112, 117 Skandagupta 86 normal segregation 5 color change 49 period 19 slag 7, 36 North America polarized illumination 49 coin 97, 117 globules 7 Great Lakes 102 polarized light 43 figurine 108 inclusions 116 nose ornament 90 polishing 24, 38, 44, 66 mirror xvii, 27, 28, 108 stringers 7, 14, 29, 102, nose-ring 91 electrochemical 63 wrought iron 89 115 electrolytic 63 Romano-Greek 114 slip 3 o mounted sample 55 bands 9 Oberhoffe r's reagent 70 polycrystalline 1 S planes 3 optical microscopy 35 polyester 63, 75 saffdruy 26 slow cooling 37 ordered regions 22 polygonal grains 27 sample of cast iron 41 oxide layers on iron porosity 6, 24, 46, 47, 87, 112 criteria to be met 61 Smith 7 etchants 72 of cast metals 6 embedding 63 soft solders 15 oxide scale 26 potassium fe rricyanide 72 mounting 63 solid solubility 11 Pratt Hamilton xvi preparation 58, 63 solid solution anneal 7 p pre-etching 74 removal 61 solidus 11 Palestine xx,88 process anneal 7 small 64 sorbite 31 Palmerton's reagent 71 proeutectoid cementite 21 storage 55 speculum 18, 26 panpipes 115 proeutectoid fe rrite 21 Sandal Castle xix, 96 speiss 50 Paraloid Bn 59 pseudomorphic 43 saturated thiosulfate solution spheroidal graphite 39, 41, 42 pearlite xvii, xix, 17, 21,29, 31, pseudomorphic replacement 73 cast iron 37 32, 35, 36, 38, 96 by corrosion products 57 scalpel 62 splat cooling 1 lamellae 70 pseudomorphic retention 46 Scandiplast 63, 75 stannic oxide 50 percentage of deformation 9 Schweizer and Meyers 7, 21 steadite xvi, 37, 38, 42, 96 scleroscope 77 eutectic 42 Index 155

steel 16, 20 tinting Winchell 50 ancient 16 copper alloys 74 wire blade 100 toggle pin 103 longitudinal sample 58 etchants 69 cast 27 sampling 58 interference etchant 73 Tower of London 42 transverse sample 58 prill 35 troosite 29, 31, 32, 101 Wootz crucible 35 strain lines 9, 25 tumbaga 22, 90 Wootz steel 13 stress factor 2 turning 6 worked 86, 88, 89, 90, 91, 97, stress-relief anneal 7 twin lines 8, 102 99, 100, 101, 102, 104, 105, stress-strain diagram 1 twin planes 10 106, 109, 113, 114, 115 structure in zinc 10 worked and cast 94 banded 32 twinned work-hardening 3, 6 Struers "Autopol" machine 66 crystals xvi working 5, 6, 14 subgrain structure 31 , grains 8, 27, 90 wrought iron 7, 14, 114 sulfide inclusions xvii, xx 71 nonferrous alloys 51 wlistite 89, 102, 114 sulfur 37 twinning Sumatra 28 in CPH metals 9 X superlattice 22 two-phase alloys 14 X-ray fluorescence analyzer 57 surface detail two-phased solid 12 of ancient metallic y artifacts 44 U Yakowitz 73 sword 88 ultrasonic cleaning 44 Young's Modulus 1 sword blade 13 , 24, 104 untempered martensite 35 Untracht 6 Z T zinc 9, 19 taper section 65 V alloys temper carbon 41 Vickers 77 ancient 19 nodules 39 Vickers test 2 corrosion products xvi tempered martensite 31 Villela's reagent 6, 71, 116 crystals 9 tempering 29, 35, 101 etchants 71 ternary eutectic 42 W zinc-lead alloy 23 steadi te xviii wafering blade 61 ternary tin bronzes 26 Warring States period 37 textured effect 9 water quenching 23 Thai high-tin bronze vessel xvii Western Han 37 Thailand 26, 93, 94, 98, 111 white bronze 26 thermal history 57 white cast iron 37, 38, 39, 40, thiosulfate/acetate etchant 74 42, 119 thiosulfate-acetate xvi, xvii whiteheart 39 Thompson, and Chatterjee 21 cast iron 37, 42 tin xx 9, 26 Whiteley's method 71 alloy 119 Widmanstatten 20, 106 bronze 15, 25 pearlite 29 etchants 71 side plates 116 tinned surfaces 26 structure xvii, 27, 31, 115 transformation 20 "

ISBN 0-89236-195-6

Printed in Singapore