Ancient Man-Made Copper Silicate Pigments Studied by Raman Microscopy

ART AND CHEMICAL SCIENCES 942

CHIMIA 2001, 55, No. 11

Chimia 55 (2001) 942-951 © Schweizerische Chemische Gesellschaft ISSN 0009-4293

Ancient Man-made Copper Silicate Pigments Studied by Raman Microscopy

Soraya Bouheroura,b, Heinz Berke,a,* and Hans-Georg Wiedemannc

Abstract: This article describes the application of Raman microscopy (micro-Raman spectroscopy) to the characterization and identification of alkaline earth copper silicates, which were used in ancient times as blue and purple pigments and were man-made. Thus, Egyptian Blue (CaCuSi40lO), Chinese Blue (BaCuSi40lO) and Chinese Purple (BaCuSi206) give rise to well-defined Raman spectra (excitation 514 nm), which allowed the identification of Egyptian Blue in 11 original Egyptian samples from the Vlth Dynasty to Roman times and that of Chinese Purple in six ancient Chinese samples dated from the Warring States period to the Han Dynasty (479 BC to 220 AD). One greenish-blue Egyptian sample turned out not to contain Egyptian Blue but rather a mixture of cupro-wollastonite and libethenite. Some of the Egyptian Blue samples showed only minor admixtures of CaS04 or silica. Four of the Chinese samples were colour sticks. In one original Chinese stick sample Chinese Blue was recognized as an additional pigment component. One colour stick was revealed to

be highly contaminated with Sa-silicates, BaC03, CaC03, PbS04, BaS04, PbO, and Si02. The investigated two pigment layer samples of the Terracotta Army, Xi'an, China were analysed by Raman spectroscopy and were found to contain, in addition to Chinese Purple, azurite, vermilion, iron oxide red, PbC03, PbO, and BaS04' Chinese Blue was not found. Keywords: Archaeometry . Chinese Blue· Chinese Purple· Egyptian Blue· Raman spectroscopy

1. Introduction tion. One of them is Raman spectrosco- 2. Instrumental Conditions for the py, which has been utilized to an increas- Application of Raman Spectroscopy Various analytical methodologies effi- ing extent in the recent years and has as an Archaeometric Method ciently contribute to the quantification of proved to be a very powerful tool at the conclusions on materials of archaeologi- arts/science interface. Selected referenc- Raman spectroscopy involves the ir- cal and cultural value [1][2]. The use of es may witness the state of the arts [3- radiation of a sample with (laser) light analytical tools first of all provides phys- 34]. (usually in the visible or near-IR range) ical and chemical data, which may then The invention of Raman spectroscopy and the analysis of the components of the be transformed into valuable information by Sir Ch. V. Raman was acknowledged inelastically scattered light [36]. The concerning the history and other back- with the Nobel Prize in 1930 and this was sample excited by an incident beam emits grounds of such objects. In this regard the onset of a long-lasting, but eventually light at discrete wavelengths and with pigments of artefacts play a prominent very beneficial development toward an given intensities. However, the intensity role. Several analytical techniques are analytical technique that is nowadays ap- ratio between the excitation beam and the well-suited for their proper characteriza- plied in practically all matter-oriented emitted light is several orders of magni- sciences. For decades the utilization of tude, This caused detection problems in the Raman experiment suffered from in- the past, but more recently Raman instru- sufficient intensity of the scattered light mentation has become much more sensi- of the Raman effect. However, instru- tive [5] mainly due to the advent of very mental developments of the past ten effective holographic filters separating years have led to an enormous improve- the incident beam light out from the scat- ment in sensitivity, which allow that even tered light and due to enormously im- minute amounts of sample (~m and ~g proved sensitivity of the detectors for the 'Correspondence: Prof. Dr. H. Berkea range) are analysed in a non-destructive scattered light, such as photodiode arrays aAnorganisch-chemisches Institut manner [4]. Thus, Raman spectroscopy, or, even better, coupled cluster discharge Universitat Zurich Winterthurerstrasse 190 especially in conjunction with a micro- (CCD) detectors. CH-8057 Zurich scope (micro-Raman spectroscopy), has Laser beam energies of sometimes Tel.: +41 1 6354681 only a few milliwatts are sufficient to Fax.: +41 1 6356802 advanced to a microanalytical tool appro- E-Mail: [email protected] priate for analysis of micro-scale hetero- illuminate microscopic sample spots of a www.unizh.ch/aci/coord geneous mixtures such as those appear- few ~m2, and even these conditions bPresent address: Solvias AG, Klybeckstr. 191, CH-4002 Basel ing in forensic analyses [35] and archaeo- allow appropriate Raman spectra to be CMettler-Toledo GmbH, CH-8603 Schwerzenbach metric studies [3-34]. obtained. As mentioned above, a typical ART AND CHEMICAL SCIENCES 943

CHIMIA2001. 55. No.11 set-up for a Raman spectrometer com- 4. Egyptian Blue and Chinese Blue Chinese Blue [43-45]. Chinese Blue and prises the spectrometer and a video- and Purple Purple appeared in Warring States period camera-controlled microscope (micro- (479-221 BC) and can later be traced in the Raman spectroscopy), which additionally Egyptian Blue and Chinese Blue Qin and Han Dynasty (221 BC-220 AD). enables the surface of a sample to be and Purple are defined chemical spe- It should be mentioned that an 'Egyptian scanned in the 11mrange. This is of enor- cies with the compositions CaCuSi40lo, Purple', which would correspond to mous advantage for archaeometric stud- BaCuSi40IO, and BaCuSi206• Their (his- CaCuSi206, has yet not been detected ies, since the heterogeneity of samples is torical) preparations and structures are and is probably non-existent. The prepa- quite often above this limit. The practical briefly discussed to the extent that this rations of the Chinese copper silicates spectral resolution of commercial spec- article needs as reference. A more de- demand more severe physical conditions trometers is normally between 1 and tailed background is given in earlier de- than Egyptian Blue and if the pure Chi- 5 cm-I, which is satisfactory with respect scriptions [39-45]. nese Blue and Purple compounds are of to the expected band widths of the Raman interest they also require strict control of emISSIOns. 4.1. Egyptian Blue stoichiometry of the starting materials. In current practice, rare cases oflimit- As denoted by its name Egyptian Blue The ancient synthesis started mostly from ed applicability of micro-Raman spec- was prepared by the Egyptians, and can be very stable barite, BaS04, sometimes troscopy still persist when the sample or a traced back to predynastic times. Earliest also from witherite, BaC03, as barium- distinct chemical component of a sample artefacts studied date back to 3600 Be. containing components, which, especial- mixture does not show the Raman effect Its use was perpetuated through all peri- ly for the chemical break-down of or when the samples absorbs most of the ods of Egyptian civilization and penetrat- BaS04, required the addition of lead ox- incident beam. Furthermore, the sample ed into the Greek and Roman cultures. ide or carbonate as catalyst. All these fac- may show fluorescence. The latter two The invention of Egyptian Blue has been tors indicate that the development of Chi- disadvantages can quite often be over- attributed to insufficient availability of nese Blue and Purple in ancient times come by changing the wavelength of the stable blue minerals. In fact, the only nat- have to be considered fine technical incident laser, i.e. changing the laser that ural blue mineral of stability and colour achievements (in modem terminology normally operates at fixed wavelength. properties comparable or even superior to they would be called 'High Technology'). the man-made Egyptian Blue and which The increased complexity of the resulting did not require chemical transformation pigment mixtures complicates the pigment 3. Pigment Applications of Raman prior to use was the rare mineral lapis la- analysis in comparison to Egyptian Blue. Spectroscopy zuli [46]. However, as we will see later Raman spec- The manufacturing processes of troscopy can efficiently contribute to un- As mentioned above, the use of Ram- Egyptian Blue in modem times were ravel the constituents of the mixtures and an spectroscopy in arts and archaeometry traced back starting in early 19th century. concomitantly contribute to elucidation for the identification of pigment and oth- These studies have been reviewed else- of the historical role of these pigments. er materials, such as binders, varnishes where [39][41] and are not the primary and matrices in general, became more focus of this article, however, implicitly 4.3. Structures of Egyptian Blue, common in the early nineties. In particu- selected aspects of them will appear Chinese Blue and Purple lar, micro-Raman spectroscopy has con- when original samples are analysed with Egyptian and Chinese Blue are iso- tributed recently to the pace of develop- respect to their chemical compositions. structural sheet silicates [47-50] consist- ment. Prominent applications with pig- Of relevance are therefore the starting ing of a puckered two-dimensional net- ments involve manuscripts, paintings and materials and remains of these and even- work of condensed silicate tetrahedrons. other artefacts and support the establish- tually side products; the latter became These are arranged in tetragonal arrays of ment of identity, manufacturing process- apparent from studies of independent four-membered silicate rings to addition- es, cultural and historical background, synthesis. The starting materials for ally generate eight-membered rings. The and even the nature of conservation prob- Egyptian Blue were normally calcite resulting network pattern is sketched in lems [4-34]. Raman spectroscopy identi- (CaC03), quartz (Si02) and copper-con- Fig. 1. fies chemical species and such in mix- taining components, such as malachite, The puckering in the final layer is ap- tures by IR emissions due to specific vi- azurite or even copper metal. Normally parently to the most part induced by the brations of the atomic framework of a the synthetic procedure required addition coordination of the cations and here component. In archaeometric and arts ap- of a flux, which, however, in most cases mainly by Cu2+, which demands a plications pigments are mostly solid inor- is difficult to trace due to its presence in square-planar environment. Cu2+ sits ganic or organic compounds and glassy amounts below (Raman) detection limits. above or below every other silicate phases, which in most cases give rise to square and pulls four adjacent squares to- characteristic Raman bands. In the recent 4.2. The Chinese Pigments gether to enable strong oxygen contacts. years libraries of spectra of ancient and Chinese Blue and Purple, sometimes In an alternating fashion with Cu2+, the contemporary pigments have been [37] also denoted Han Blue and Purple, Ca2+ or Ba2+ ions take the remaining po- or are about to be established [38], which are also copper silicates, but contain bari- sitions above and below the squares, greatly assist the identification processes. um instead of calcium. Chinese Blue, but moreover they interconnect the 2 This article will exclusively focus on the BaCuSi40IO, is the heavier isostructural CuSi40lO -layer by doubling their coor- Raman analysis of blue and purple cop- homologue of Egyptian Blue, and pos- dination sphere to establish eight-coordi- per silicate pigments, which were used in sesses very similar chemical and physical nation. As mentioned before the whole the ancient Egyptian and Chinese civili- properties. Chinese Purple constitutes framework attains tetragonal symmetry, zations and were produced by synthetic another unique phase of the composition however the loqal symmetry of an isolat- 8 procedures. BaCuSi206 containing less quartz than ed planar Si4012 - ring (D4/1) is lowered ART AND CHEMICAL SCIENCES 944

CHIMIA 2001,55, No. 11

to C2V in the condensed arrangement of Egyptian and Chinese Blue. The structure of Chinese Purple shows three significant features [49][51], condensation which are partly distinct from those of :- Egyptian and Chinese Blue: - There is no extended silicate framework, rather the structure is composed of planar four-membered silicate rings [5Ia] - Cu-Cu bonded units [51 b] hold the II puckering with four-member rings together to form 2 2 U Cu +/M + coordination layers of [Cu2Si40'24-]oo - Similar to Egyptian or Chinese Blue, the Ba2+ ions are located above and below the silicate rings and thus interconnect the layers. A top-view projection of a [Cu2Si40'24-]oo layer is schematically sketched in Fig. 2. Due to the presence of isolated four- member silicate rings, Chinese Purple may be viewed an intermediate in the condensation process to Chinese Blue. From experimental studies it was recognized that Chinese Purple is indeed a ki- netic product on the way to Chinese Blue [42-45]. Concerning the Raman properties of Chinese Purple, it is important to keep in mind the two above-mentioned Fig. 1. Schematic sketch of the condensation process of a four-membered silicate ring (Si40,2S-) structural features distinguishing it from 4 2 to build up a Si4010 - layer and its puckering induced by coordination of Cu + and M2+ (M = Ca, the Blues, because these two properties Ba). are principally well-suited for Raman spectroscopic detection.

5. Raman spectra of Egyptian Blue, Chinese Blue and Purple

Exemplary spectra of pure CaCu

Si40IQ, BaCuSi40IQ, and BaCuSi206 samples are shown in Fig. 3. Overall the spectra of the Blues are very similar em- phasizing their close chemical relation- ship, while the spectrum of Chinese Pur- ple shows only little resemblance with the general band pattern of the other two pigments. In the following the main features of the Raman spectra are briefly discussed. The spectra have been analysed with respect to the principal types of structures by aid of lattice dynamics calculations for both types of compounds [52][53]. The MCuSi40IQ phyllo-silicates (M = Ca, Sr, Ba) possess P4/ncc [47][48] and BaCuSi206 l4/mmm symmetry [49][51]. Since both types of species thus belong to centrosymmetric space groups, total symmetric vibrations should give rise to intense Raman bands. For the Egyptian and Chinese Blues the bands at 1088 and 1102 em-I are respectively assigned to

Fig. 2. Schematic top view of a CU2Si40124- layer of Chinese Purple (BaCuSi20s). alg v(SiO) vibrations, while the other ART AND CHEMICAL SCIENCES 945

CHIMIA 2001,55, No. 11

Fig. 3. Raman spectra (514 nm) of Egyptian Blue (CaCuSi401o), Chinese Blue (BaCuSi40lO) and of Chinese Purple (BaCuSi206) in the range 1200- 200 cm-1. Major bands of Egyptian Blue: 1088(s), 992(w), 789(w), 575(w), 471(w), 435(s), 363(w), 201(w) cm-1; Chinese Blue: 1102(s), 997(w), 791(w), 562(w), 427(s), 383(w) cm-1 and Chinese Purple: 990(m), 588(s), 516(m), 459(w), 354(w), 276(w), 183(w) cm-1.

ported to Mentuhotep II (2069-2014 BC) as the commander of the troops. The next three samples studied came from the New Kingdom period. Blue from the crown of the famous bust of Nefertete (Aegyptisches Museum der Staatlichen Museen Berlin, Germany) and Blue of the Talatat Stones of the Temple of her husband Echnaton [41]. The cylindrical seal, which is dated also to the New Kingdom and falls in a very productive period for artisans, is the property of one Chinese Purple of the authors. It has been described in detail elsewhere [42]. The sample of Brick Nimrud (Fig. 4) originates from 1200 1000 400 200 Mesopotamia and is dated to the 13th- 7th century BC. Today it is property of the British Museum, London [58a]. The Egyptian Late Period is docu- mented by several other samples. The bands at 435 (Egyptian Blue) and 427 identify Egyptian Blue in ancient artwork blue Amulet Bes protected those who cm-I (Chinese Blue) are attributed to alg [3b][13] [29][45][57]. Our group has re- wore it by the powers of dwarfish God bridging oxygen breathing modes. The cently introduced Raman spectroscopy to Bes. It is dated to the 24th Dynasty and is relative constant position of these bands trace BaCuSi40lO and BaCuSi206 in blue property of one of the authors. The next points to almost exclusive motions of the and purple colour sticks dated to the War- sample of this period originates from a closely related silicate frameworks of ring States, Qin and Han Periods and also relief of the Tomb of Ibi, chief steward of

CaCuSi40lo and BaCuSi40IO. From the in pigment samples of the Terracotta the divine adoratrice under Psammet- bands of lower intensities it is worth Army of the Qin Dynasty from Lintong ichus I (Fig. 5). mentioning that those at 363 (CaCu near Xi' an, China [44][45]. A more I Si40IO) and 383 cm- (BaCuSi40lo) are detailed Raman analysis including that attributable to vibrations with strong of Egyptian Blue in ancient samples is Cu-O character [54]. The band of CaCu presented in this article, which nicely

Si40lO at 201 cm-] is presumably due to a demonstrates its straight-forward appli- Ca-O vibration, since it is missing in cability as an archaeometric method [5].

BaCuSi40IO· alg-Type bands of BaCuSi206 are those at 990,588,516,459 and 276 cm-I. 6. Raman Investigations of Original While the 990 cm-I band of medium in- Egyptian, Mesopotamian, and tensity belongs to a v(SiO) vibration, Chinese Samples those at 588 and 516 cm-I are attributed to a 8s(Si02) mode of the silicate groups The ancient samples of Egyptian Blue [49] and, similar to the MCuSi40lO com- and Chinese Blue and Purple investigated pounds, to an oxygen breathing vibration, by Raman spectroscopy are summarized respectively. Bands at 276 and 183 cm-I in the Table. The Egyptian samples cover are assigned to vibrations with Cu-O a span of almost 3000 years starting with character, however, based on earlier anal- Blue from the 6th Dynasty of the Old yses of complexes with Cu-Cu bonds, Kingdom to a sample of the Roman time. these vibrations are expected to bear ad- The oldest sample stems from the ditional Cu-Cu bond character [55][56]. Mastaba of Mereruka in Saqqara, Egypt, The mentioned lattice dynamics calcula- which possesses wall paintings with pre- tions predict the vibration with mainly vailing blue colour. Mereruka was vesir Cu-Cu character to appear at 63 cm-1 to the King Teti arriving in the Mastaba [52], unfortunately out of the analysis from his tomb through a ficti ve door [41]. range of standard Raman spectrometers. The second sample originates from the Fig. 4. Photograph of a fragment of Brick Nim- In several earlier cases Raman spec- tomb of General Antef, Middle King- rud from Mesopotamia dated to the 13th-7th troscopy has been successfully applied to dom, 11 th Dynasty. General Antef re- century Be (British Museum, London). ART AND CHEMICAL SCIENCES 946

CHIMIA2001,55,No.11

Table. Egyptian, Mesopotamian, and Chinese samples investigated by Raman spectroscopy

Period Name. origin, or present location Blue or Purple Other Impurity Pigments Pigments

2345-2181 BC Mastaba of Mereruka, Saqqara EB CaS04 2133-1991 BC Tomb Antef, Thebes EB CaSOd 1340 BC Bust of Nefertete. blue of crown, Amama EB CaS04·· 1353 BC Echnaton temple, blue of Talatat, Amama EB CaS04·· 1300 BC Cylinder Seal, Thebes EB CaS04 13th-7th Century BC Bnck Nimrud, Iraq EB SI02 712-332 BC Amulet 8es, Egypt EB 663-525 BC Tomb Ibi, Thebes W/L 500 BC Ba-bird, Soul, Memphis EB Si02 3rd Century BC Mummy cartonage EB 1st Century BC Ptah Sokarls OSlns Figure EB Roman time Mummy coffin, Cairo EB CaSOd Han Dynasty Stick Freer Gallery, Washington CP Si02, CaCOa, BaCOa, BaS04, PbSO.,. Ba-sillcates Han Dynasty Stick Royal Ontario Museum CP Han Dynasty Stick 4069. Oestas Museum CB/CP SI02'· Han Dynasty Stick 4070, Oestas. Museum CP BaCOa Qin Dynasty Sample I. Terracotta Army CP azurite, SiO:> vermilion, iron oxide red, PbCOa Qln Dynasty Sample II, Terracotta Army· CP BaSO.,. SiD

EB = Egyptian blue, CP = Chinese purple; W = cupro-wollastonite, L = libethenite; * the lower lying terracotta layer also contains FeTiOa and Caa(P04b as established by EDX [44]; ** uncertain due to low amounts.

The wooden falcon figure Ba-bird, BC stems originally from Memphis, Egyptian Blue from Ptolemaic and Ro- Kestner Museum, Hanover, Germany Egypt. It is now property of the Staatliche man times. Their present locations are contains blue areas, from which a sample Sammlung Aegyptischer Kunst, Munich, unknown. was studied. The Ba-bird from approxi- Germany (AS 19). It served as a contain- Four of the investigated six original mately 500 BC symbolized a protection er for corpses. A sample of the Blue was samples of Chinese Blue and Purple (Ta- god at burial ceremonies [41]. The figure taken [41]. The samples of the mummy ble) were taken from original Blue or Ptah-Sokaris-Osiris from the 1st century carton age and the mummy coffin contain Purple sticks of octagonal shapes. It is

Fig. 5. The photograph of a section of the relief of Theban Tomb 36 (Tomb Ibi),which shows craftsmen at work. Several areas have a ~ greenish-blue colour [58b]. ART AND CHEMICAL SCIENCES 947

CHI MIA 2001.55. No. 11 thought that these sticks were trade items. They were used as pigment bases in paints and applied by grinding. . • .•.. ,it The general phenomenology of sample Freer Gallery has been described ear- , lier by FitzHugh [59][60]. It is displayed in Fig. 6. Sample stick Royal Ontario Museum came from a blue octagonal stick said to have come from Jincun near Luoyang, China [60]. It is reported to be made of mainly blue particles with scattered purple ones. The two samples of sticks from the Museum of Far Eastern Antiquities in Fig. 6. Exemplary photograph of stick Freer Stockholm, K 4069 (Blue) and K 4070 Galery demonstrating its octagonal shape and (Purple) are described as worn and po- its true size (above). Microscopic view of a chip rous materials [60], from which powdery of stick Freer Galery revealing its porous and material was taken for investigation. In heterogeneous nature (below). Colour sticks of this kind are anticipated to have been trade addition the stick samples were discussed items. They were applied as pigment bases in recently [44]. paints by grinding. Distinct samples from different parts of the Terracotta Army were also studied by Raman spectroscopy (Fig. 7). A group As indicated in the Table, several of treatment of the samples during the prep- of Terracotta Soldiers is shown in Fig. 7a. the listed Egyptian Blue samples show aration, since in pure state at least a- From a fragment of the hat of a soldier CaS04 (gypsum or anhydrite recognized quartz can be converted to a-tridymite or (Fig. 7b) the pigment layer (sample I) by their Raman emission at 1012 em-I) a-christobalite only by heating them was investigated as an embedded cross [65] as a minor admixture, which pre- above 870°C and 1470 0c. Unfortunate- section (Fig. 7c). In Fig. 7d fragments of sumably stems from natural CaC03 re- ly the main Raman band of christobalite the trousers of a Terracotta soldier are sources used as one of the starting com- overlaps with that of Egyptian (or even displayed, from which the pigment layer ponents. It should be mentioned that Chinese) Blue at around 430 em-I. How- was studied (sample II). It is presented as Egyptian Blue also shows a very weak ever, the major Raman band of tridymite an embedded cross section in Fig. 7e. Raman band at this wavenumber, which at 516 em-I could be seen in the presence Earlier investigations of these samples obscures the analysis of very low CaS04 of Egyptian Blue. A mixture of both have clearly revealed the presence of contents. For that reason some of the should therefore be Raman detectable. In Chinese Purple and other pigment mate- CaS04 assignments in the Table are un- none of the Raman spectra of Egyptian rials [61 ][62]. Application of Raman certain as indicated. Blue in the Table could this band be spectroscopy were expected to confirm Two samples can be shown to contain traced. Therefore, it can be stated that these results, but perhaps also to deliver a-quartz, recognized by the strong Ram- these samples are tridymite-free. I further information on the nature of the an band at 468 cm- . Especially sample CaC03 impurities could in principle pigment content. Brick Nimrud shows considerable amounts be identified by the major Raman bands All the samples were compared to of quartz. Its admixture may be interpret- at 1084 and 280 cm-I. However, since the pure Egyptian Blue and Chinese Blue ed in terms of a tuning of the material strong 1084 band would overlap with the and Purple or mixtures of them, obtained properties with regard to construction 1088 cm-I of Egyptian Blue, the presence by contemporary independent synthesis purposes. In an exemplary way the Ram- of CaC03 could only be verified by the [63]. an spectrum of silica 'contaminated' 280 cm-I band. None of the Egyptian The Egyptian and Mesopotamian Brick Nimrud is displayed in Fig. 8 Blue samples in the Table showed this samples listed in the Table contain the showing besides the bands of Egyptian band. Major amounts of other impurities Blue the strong a-quartz emission. chemical species CaCuSi40lO (mineral could be detected only in the sample of name cuprorivaite) except for the blue- To a lesser extent a-quartz was seen Ptah-Sokaris-Osiris, which consists of green sample Tomb Ibi, which will be in sample Ba-bird, however, no obvious Zn compounds according to EDX studies considered separately. In most cases the function can be associated with this ad- [45], this presumably is indicated by a Raman spectra reveal an even high con- mixture. Other samples may also contain very strong Raman band at 745 cm-I. tent of this blue pigment. Earlier findings minute quantities of a-quartz, which can- This band could however not yet be at- of EI Goresy on other ancient Egyptian not be analysed due to overlap with the tributed to a particular chemical species. samples showed that these consisted of weak band of Egyptian Blue at around The unique greenish-blue sample of considerable amounts of wollastonite 470 cm-I. In this context the question Tomb Ibi does not contain Egyptian Blue (CaSi03) [64], which could not be detect- arises whether other metastable phases of according to the Raman analysis, rather it ed in our study, at least in amounts above silica could be detectable by Raman seems to be a mixture of cupro-wollas- the Raman detection limit. EI Goresy spectroscopy [65]. Of interest are a-tri- tonite and libethenite (Cu2(P04)(OH» even defined Egyptian Blue as a mixture dymite and a-christobalite. The former [66]. With the exception of a less struc- of wollastonite and cuprorivaite. Wollas- has been traced in preparations of Egyp- tured band area between 1100 and 1000 tonite could have definitely been traced tian Blue previously [41b]. The presence cm-I, the overlay of the two compounds by Raman spectroscopy, as a- or ~-wol- of these quartz modifications could prin- fits well with the spectrum of the sample lastonite or in its vitreous form [29]. cipally allow conclusions on the thermal Tomb Ibi, displayed in Fig. 9. ART AND CHEMICAL SCIENCES 948

CHIMIA 2001,55, No, 11

Fig, 7. Samples of the Terracotta Army Xi'an, China. a) A group of terracotta soldiers; b) shows the fragment of a painted hat of a terracotta soldier (fragment 009-98) and below this in c) a microscopic view of a cross section (horizontal extension 1 mm). Bottom left: Pur- ple particles of this sample (sample I) correspond to Chinese Purple and scattered blue grains to azurite. Other pigment content see text. Fragments of the purple trousers of a soldier (fragments 003-92) are shown in d). A microscopic view (enlargement 500 times) of a cross section of its pigment layer corresponding to sample II is presented in e), The displayed part has a horizontal extension of 0.22 mm. The pigment layer contains grains of Chinese Purple and other pigment particles (see text). Below the pigment layer one can find a layer of lacquer and next to this terracotta.

It should be mentioned that cupro- wollastonite has been detected previously in other Egyptian artefact samples [29]; it has been defined as a-wollastonite with some Ca2+ replacement by Cu2+. By Raman spectroscopy, however, cupro- wollastonite cannot be distinguished from wollastonite. Our Raman studies on the various ancient Chinese barium-copper-silicate samples (Table) and reference samples also brought about well-defined conclusions. From the Chinese stick samples, stick 4069 contains Chinese Blue in mixture with Chinese Purple. All other stick samples show no bands for Chinese Blue and consist mainly of Chinese Purple as Sample 1 Sample II the colouring matter. In all Chinese samples the following potential impurities were checked, which might originate from the minerals used, sumably be identified, however spots In addition it became apparent that the added flux or chemical transforma- containing BaS04 or BaC03 expected BaS04 was used as a starting material tions occurring during synthesis: CuO, from the powder X-ray results [45] could which partly remained unreacted. Pre- CU20, BaC03, BaS04, PbC03, PbC03, 2 not be traced. Stick 4070 unambiguously sumably the bands at around 940 and I PbO, PbS04, CdO, and Si02. The Terra- showed the presence of BaC03. 420 cm- account for glassy phases with 2 cotta Army samples I and II are pigment A slice of Stick Freer Gallery was in- polysilicate (Si03 -) units and the bands application cases. They were studied in vestigated by Raman spectroscopy in eventually present at 914 cm-I for disili- 6 order to reveal Chinese Blue or Purple, greater detail. The Raman studies at vari- cate (Si207 -) [67]. Due to a great stoichi- but also other pigments. Several Chinese ous spots made clear that this stick is a ometric excess of Si02, glass formation samples presumably show a Raman very heterogeneous mixture, since other seems to be very plausible and indeed the emission for silica, which varies in rela- components found in significant amounts quite hard general appearance of the disc tive intensity from sample to sample. have been detected besides Chinese speaks for a relatively high amount of However, identification of the Raman Purple. Several bands were assigned to glassy phases. It is maybe quite amusing band of minor amounts of Si02 in mix- crystalline or glassy silicate phases. Oth- to see that an early German patent from tures with Chinese Purple is obscured by er attributable phases include CaC03, 1900 describes the preparation of glassy the fact that Chinese Purple also possess- BaC03, BaS04, PbS04, PbO and perhaps highly coloured barium-copper-silicates es an emission of medium to weak inten- also Si02. An exemplary Raman spec- for the use as pigments [68]. This modern sity in this region. The Raman spectrum trum is displayed in Fig. 10 demonstrat- times discovery was certainly made with- of the Chinese sample, stick Royal On- ing the heterogeneity of the sample and out knowledge of the existence of ancient tario Museum, allowed no definite con- some of the impurities. Chinese Blue and Purple and may simply clusions on impurities due to the low From this it became clear that the stoi- reflect man's timeless need for artificial quality of the spectrum caused by an in- chiometry of the starting materials did colouring matter. sufficient amount of sample. For stick not fully match those theoretical amounts The Terracotta Army samples I and II 4069 the major a-quartz band could pre- required to obtain pure Chinese Purple. consist of Chinese Purple as indicated by ART AND CHEMICAL SCIENCES 949

CHI MIA 2001,55. No. 11

468

435 457

1090

970

1043

1200 1000 800 600 400 200 1200 1000 800 600 400 200 [el11"] [em"]

Fig. 8. Raman spectrum of sample Brick Nimrud in the range 1200- Fig. 9. Raman spectrum of sample Tomb Ibi in the range of 1200- 200 cm-1 (514 nm). The spectrum shows the presence of Egyptian 200 cm-1 (514 nm). The spectrum shows the presence of cupro- Blue (see Fig. 1) and a-quartz (468 cm-1). wollastonite (1043,970,633 cm-1) and of libethenite (457,197 cm-1).

the appearances of the major Raman amounts of crystallites of azurite (major 7. Conclusions Drawn from the bands for this compound (Table ). A typ- Raman band at 402 em-I) [65], which ap- Investigations of Original Samples ical Raman spectrum of Chinese Purple parently provided the blue colouring tone of Egyptian Blue, Chinese Blue obtained from sample I is displayed in in the pigment layer. The Raman spectra and Purple Fig. I I. of the additional pigments of vermilion, In contrast to the sticks the Terracotta azurite and iron oxide red are exemplari- The described investigations have Army samples represent mixtures of ac- ly displayed in Fig. 12 as they were ob- demonstrated that Raman microscopy is tual paints, which are naturally more het- tained from different spots of sample I. a very powerful archaeometric tool to erogeneous. Their a priori heterogeneity, The Raman spectrum of sample II shows study samples of the ancient man-made however, makes it sometimes difficult to besides the mentioned components the copper silicate pigments Egyptian Blue, decide whether a specific component has presence of BaS04 as a whitening agent. Chinese Blue and Purple. Even if they are to be attributed to a mineral or to a chem- Furthermore a Raman band at 1035 cm-I embedded in matrices or are constituents ical origin or whether it was related to a is detected at several spots of the pigment of heterogeneous mixtures, like in paint colouring function. A significant PbC03 layer, which however could not yet be at- applications, Raman microscopy will en- (major Raman band at 1050 em-I see tributed to a pigment species with certainty. able these pigments to be traced. Specific Fig. 11) or PbO (major Raman band at 143 em-I) [37] content was noted. This could be related to their function as synthetic additives, but could also point to their role as colouring components. For 988 sample I and II vermilion (cinnabar, HgS) [37] could be traced, The other red pigment definitely detected in sample I via Raman analysis is iron oxide red 591 516 (haematite, Fe203) [15][37]. I and II also 969 showed substantial quantities of a-quartz. This is assumed to be a natural circum- stance associated with their use in paints.

It should however be noted that neither 941 sample showed the presence of Chinese 464 Blue. Sample I possessed considerable

1069

Fig. 10. Raman spectrum of sample stick Free Gallery inthe range 1200- 200 cm-1 (514 nm).The major bands are attributed to BaC03 (1069 cm-1), 1200 1000 400 200 BaS04 (988 cm-1) PbS04 (969 cm-1), Basilicates (941 cm-1) and Chinese Purple (591, 516, 464 cm-1). ART AND CHEMICAL SCIENCES 950 CHIMIA 2001, 55, No. 11

585

985

514

1050 466

Vermilion

400 200 800 400 200

Fig. 11. Raman spectrum (range 1200-200 cm-l, 514 nm) of a spot Fig. 12. Raman spectra (range 900-200 cm-1, 514 nm) of selected of sample I of the Terracotta Army, Xi'an, China displaying bands for pigment spots as obtained from sample I of the Terracotta Army. Chinese Purple and a band attributed to PbC03 (1050 cm-1).

results were obtained from 12 original objective, a spectrometer with a 1200 Canada for their loan of samples. Also we would Egyptian and Mesopotamian samples. 11 grooves/mm grating and a NIR en- like to thank E. Emmerling and C. Blansdorf of of them revealed the prevailing presence hanced, Peltier-cooled CCO camera. For the Bayerisches Landesamt fUr Denkmalpflege, MUnchen, Germany, providing embedded cross of Egyptian Blue (CaCuSi 0 ). Only 4 IO the blue pigments an air-cooled argon ion sections of original samples I and II from the one sample turned out to be a mixture of laser served as the excitation source with Terracotta Army in Xi' an (fragments 009/98 wollastonite and libethenite. Traceable a wavelength of 514 nm and an output of and 003/92), China and the corresponding pho- impurities of some samples were CaS04 10mW. When necessary, the laser power tographs. and Si02. on the sample was reduced. The back- From the six investigated ancient Chi- scattered laser radiation is prevented Received: October 3, 200 I nese samples four were colour sticks and from entering the spectrograph by two two paint applications from the Terracot- holographic supernotch filters. Spectra [1] P. Mirti, Ann. Chim. 1989, 79,455. ta Army. According to our Raman inves- were usually recorded in the frequency [2] C. Lahanier, F.O. Preusser, L. Van Zelst, tigations the colour sticks consisted of range 2000-200 cm-I with the excitation Nucl. Instr. and Method. Phys. Res. 1986, B14, I. Chinese Purple (BaCuSi206) and in one at 514 nm. All spectra are baseline-cor- case a mixture of Chinese Purple and rected. [3] Reviews: a) S.P. Best, R.J.H. Clark, R. Withnall, Endeavour 1992, /6, 66-73; b) Chinese Blue (BaCuSi 0 ) was found. 4 IO C. Coupry, Analllsis 2000, 28, 39. Obviously due to difficulties in the Acknowledgments [4] R.J.H. Clark, J. Mol. Strllct. 1999, 480- preparation, the sticks contained impuri- We are indebted to the following institutions 481,15. ties, which were quite prominent for stick for their donations of original samples of Egyp- [5] R.J.H. Clark, J. Mol. Struct. 1995, 347, Freer Gallery. Major admixtures were re- tian Blue: Crown of Nefertete: Aegyptisches 417; L. Burgio, R.J.H. Clark, H. Toftlund, maining barium starting materials and Museum del' Staatlichen Museen Berlin, Ger- Acta Chem. ScalU!. 1999,53, 181; R. J. H. Clark, Chem. Soc. Rev. 1995, 24, 187; B. many, Preussischer Kulturbesitz. Ptah-Sokaris- lead compounds, which functioned as a Wehling, P. Vandenabeele, L. Moens, R. Osiris: Staatliche Sammlung Aegyptischer synthetic additive. The Terracotta Army Klockenkamper, A.V. Bohlen, G. Van Kunst, MUnchen, Germany. Ba-bird: Kestner samples revealed besides Chinese Purple Hooydonk, M. de Ren, Mikro Chim. Acta, Museum, Hannover, Germany. Brick Nimrud: 1999, 130, 253. the presence of azurite, vermilion, iron Laboratory of the British Museum of London, [6] S.E.J.Bell, E.S.O. Bourguignon, A.C. oxide red, PbCO}, PbO, BaS04 as pig- London. Great Britain. ments. Chinese Blue was not detected. Dennis, J.A. Fields, J.J. McGarvey, K.R. Furthermore, we would like to acknowledge Seddon, Anal. Chem. 2000, 72,234. G. Bayer, Nichtmetallische Werkstoffe, ETH [7] C.J. Brooke, H.G.M. Edwards, J.K.F. ZUrich, for the synthesis of reference samples. Tait, J. Raman Spectr. 1999, 30, 429. 8. Experimental The authors would then like to thank Elisabeth [8] L. Burgio, R.J.H. Clark, J. Raman Spectr. West FitzHugh, Freer Gallery of Art, Smithso- 2000,3/, 395. nian Institution, Washington DC, for providing A Renishaw System 1000 Raman mi- [9] L. Burgio, O.A. Ciomartan, R.J.H. Clark, the slice of original Chinese purple. Finally we J. Mol. Struct. 1997, 405, I. croscope was used for the Raman meas- are indebted to the Museum of Far-eastern An- [10] L.F. Cappa de Oliveira, J.d.C.R.P. urements. This system comprises a Leika tiquities, Stockholm, Sweden (sticks 4069 and Boscan, P.S. Santos, M.L.A. Temperini, OM LM microscope equipped with a SOx 4070) and the Royal Ontario Museum, Toronto, Qu{mica Nova 1998, 21, 172. ART AND CHEMICAL SCIENCES 951

CHIMIA2001, 55, No.l1

[I I] M. Castillejo, M. Martin, D. Silva, T. Stra- [41] a) G. Bayer, H.G. Wiedemann, Sandoz ings, Florence, Italy July 1994, Faenza, toudaki, D. Anglos, L. Burgio, R.J.H. Bulletin 1976, 40, 20; b) G. Bayer, H.G. 1995,675. Clark, J. Mol. Struct. 2000,550-551,191. Wiedemann, Naturwissenschaften 1975, [62] C. Thieme, E. Emmerling, C. Herm, Y.Q. [12J R.J.H. Clark, M.L Curri, C. Laganara, 62, 181; c) H.G. Wiedemann, G. Bayer, Wu, T. Zhou, Zh. Zhang, Techna - Mono- Spectrochim. Acta, Part A 1997,53,597. Anal. Chern. 1982,54, 619A. graphs in Materials and Society 2, Ed. P. [13] R.J.H. Clark, P.J. Gibbs, J. Raman Spectr. [42] H.G. Wiedemann, G. Bayer, A. Reller, Vincenzini, CIMTEC Conference Pro- 1997, 28, 99; R.J.H. Clark, P.J. Gibbs, J. Actes de Table Ronde Ravello. - Edipul- ceedings, Florence, Italy July 1994, Faen- Chern. Soc., Chern. Comm. 1997, 1003. gia, Bari, Italy, 1997, 195. za, 1995, 591. [14] H.G. Wiedemann, C. Marcolli, J. Therm. [43] H.G. Wiedemann, G. Bayer, Proceedings [63] G. Bayer, unpublished results. Anal. Calorim. 2001,64,987. of an International Conference on the [64] H. Jaksch. W. Seipel, K.L. Weiner, A. EI [15] R.J.H. Clark, M.L Curri, J. Mol. Struct. Conservation of Grotto Sites, Mogao Goresy, Naturwissenschqften 1983, 70, 525. 1998, 440, 105. Grottoes, Dunhuang, The People's Re- [65] Inorganic Materials Database, Renishaw [16] P. Colomban, G. Sagon, X. Faurel, J. Ra- public of China, Ed. Neville Agnew, Los pic Spectroscopy Products Division, Old man Spectr. 2001,32,351. Angeles: Getty Conservation Institute, Town, UK, 1998. [17] P. Colomban, F. Treppoz, J. Raman Spec- 1997,379. [66] S.N. Reddy, R.V.S.S.N. Ravikumar, B.J. tr. 200], 32, 93. [44] H. Berke, H.G. Wiedemann, East Asian Reddy, Ferroelectrics, 1995, 166, 55. [18] C. Coupry, A. Lautie, M. Revault, J. Du- Science, Technology and Medicine 2000, [67] J.D. Frantz, B.O. Mysen, Chem. Geol. filho, J. Raman Spectr. 1994,25,89. 17,94. 1995,121, 155. [19] R. Davey, D.J. Gardiner, B.W. Singer, M. [45] H.G. Wiedemann, H. Berke, 'Chemical [68] A.F. Le Chatelier in 'Verfahren zur Her- Spokes, J. Raman Spectr. 1994,25,53. and Physical Investigations of Egyptian stellung von gri.inen, blauen und violetten [20] Derbyshire, R. Withnall, J. Raman Spectr. and Chinese Blue and Purple', Proceed- Mineralfarben aus Kieselsaure und Bary- 1999, 30, 185. ings of the Conf. Polychromy of the Terra- um- und Kupferverbindungen', Patent- [21] H.G.M. Edwards, E.M. Newton, J. Russ, cotta Army of the First Chinese Emperor schrift, Kaiserliches Patentamt, Nr. J. Mol. Struct. 2000, 550-551, 245. Qin Shihuangdi, Xian, 1999, p.13 112761, Klasse 22f, July 16,1900. [22] H.G.M. Edwards, D.W. Farwell, S. Ro- [46] D. Reinen, G.G. Lindner, Chem. Soc. Rev. zenherg, J. Raman Spectr. 1999,30,361. 1999,75. [23] H.G.M. Edwards, D.W. Farwell, E.M. [47] A. Pabst, Acta Cryst. 1959, 12,733. Newton, F. Rull Perez, S. Jorge Villar J. [48] W. Bensch, M .Schur, Z. Krist. 1995, 210, Raman Spectr. 2000, 31, 407. 530. [24J H.G.M. Edwards, DW. Farwell, E.M. [49] L.W. Finger, R.M. Hazen, R.I. Hemley, Newton, F. Rull Perez, Analyst 1999, 124, Am. Min. 1989, 74, 952; H.C. Lin, F.L 1323. Liao, S.L. Wang, Acta Cryst. Sect. C [25] H.G.M. Edwards, L. Drummond, J. Russ, Cryst. Struct. Commun. 1992, C48, 1297. J. Raman Spectr. 1999,30,421. [50] M. Henry, Coordination Chemistry Re- [26J H.G.M. Edwards, L. Drummond, J. Russ, views 1998, 178-180, 1109-1163. Spectrochim. Acta, Part A 1998,54, 1849. [51] a) J. Janczak, R. Kubiak, Acta Crystallo- [27] Z. Jian, W. Changsui, X. Cunyi, Q. Ping, graphica 1992, C48, 1299; b) V. Asago, Spectr. Lett. 1999,32, 841. K. Uchinokura, A. Zheludev, G. Shirane, [28] Z. Jian, W. Changsui, X. Cunyi, Spectr. Phys. Rev. 8,1997,55,8357. Lett. ]998, 31, 1431. [52] D.A. Mc Keown, M.l. Bell, Phys. Chern. [29] S. Pages-Camagna, S. Colinart, C. Cou- Minerals 1998, 25, 273. pry, J. Raman Spectr. 1999,30,313. [53] D.A. Mc Keown, M.I. Bell, Phys. Rev. B [30] L. Sabbatini, M.G. Tarantino, P.G. Zam- 1997, 56, 3114. bonin, G.E. De Benedetto, Fres. J. Anal. [54] L Botto, E.J. Baran, G. Minelli, An. Asoc. Chern. 2000, 366, 116. Quim. Arg. 1987, 75,429. [31] D.C. Smith, A. Barbet, J. Raman Spectr. [55] V.M. Miskowski, S. Franzen, A.P. 1999, 30, 319. Shreve, M.R. Ondrias, S.E. Wallace-Wil- [32] D.C. Smith, M. Bouchard, M. Lorblan- liams, M.E. Barr, W.H. Woodrouft, Inorg. chet, J. Raman Spectr. 1999,30,347. Chern. 1999, 38, 2546. [33] P. Vandenabeele, B. Wehling, L Moens, [56] A. AI-Obaidi, G. Boranovic, J. Coyle, B. Dekeyzer, B. Cardon, A. von Bohlen, e.G. Coates, J.J. Mc Garvey, V. Mc Kee, R. KJockenkamper, Analyst 1999, 124, J. Nelson, Inorg. Chern. 1998,37,3567. 169. [57] a) C. Coupry, in 'La Couleur dans la Pein- [34] P. Vandenabeele, F. Verpoort, L Moens, ture et I'Emailiage de I'Egypte Ancienne', J. Raman Spectr. 2001,32,263. Eds. S. Collinart, M. Menu, p. 77, Edi- [35] A.H. Kuptsov, J. Forens. Sci. 1994, 39, puglia, Bari, 1998; b) S. Bruni; F. Cariati; 305. F. Casadio, L. Toniolo, Vibr. Spectr. [36] Raman Microscopy, Eds. G. Turrell, J. 1999,20, 15. Corset, Academic Press, London 1996. [58] a) H.G. Wiedemann, E. Doehne, D. Stu- [37J L. Burgio, R.J.H. Clark, Spectrochim. lik, F.D. Preusser, Proc. 28th Intern. Sym- Acta, Part A 2001, 57, 1491; I. M. Bell, posium on Archaeometry, 23-27 March R.J.H. Clark, P.. Gibbs, Spectrochim. 1992, Los Angeles, California, 1992, 143; Acta, Part A 1997,53,2159. b) 1. Baines, J. Malek, 'Atlas of Ancient [38J P. Vandenabeele, A. Hardy, H.G.M. Ed- Egypt', Phaidon Press Ltd, Oxford, UK, wards, L. Moens, Appl. Spectr. 200],55, 1986. 525. [59] E.W. FitzHugh, L.A. Zycherman, Stud. [39] J. Riederer in 'Artists Pigments. A Hand- Conserv. 1983,28, IS. book of their History and Characteristics', [60] E.W. FitzHugh, LA. Zycherman, Stud. Ed. E.W. Fitzhugh, Vol. 3, 23; Oxford Conserv. 1992,37, 145. University Press, Oxford 1997. [61] C. Herm, C. Thieme, E. Emmerling, Y.Q. [40] W.T. Chase in 'Science and Archeology', Wu, T. Zhou, Zh. Zhang, Ed. P. Vincenzi- Ed. R.H. Brill, MIT Press, Cambridge, ni Techna - Monographs in Materials and Mass., 1971, 80. Society 2, CIMTEC Conference Proceed-