AN

ALYST

FULL PA

PER

THE

www.rsc.org/analyst

Raman spectroscopic and SEM study of cinnabar fromHerod’s palace and its likely origin

Ray L. Frost,*a Howell G. M. Edwards,b Loc Duong,a J. Theo Kloproggea and Wayde N.Martensa

a Centre for Instrumental and Developmental Chemistry, Queensland University ofTechnology, 2 George Street, Brisbane, GPO Box 2434, Queensland 4001, Australia.E-mail: r.frost@qut.edu.au

b Department of Chemical and Forensic Sciences, University of Bradford, Bradford, UK BD71DP

Received 15th October 2001, Accepted 16th November 2001First published as an Advance Article on the web 2nd January 2002

Mineral samples of cinnabar were obtained from the ancient mining sites of Tarna and Almaden in Spain andfrom Hunan province in China. A comparison is made using a combination of scanning electron microscopy andRaman spectroscopy with the cinnabar coated on the wall of Herod’s palace. Scanning electron microscopy showsthat the cinnabar samples from Spain were impure, whereas the sample from the wall of Herod’s palace washighly pure. No conclusion can be drawn as to the likely origin of the Herod’s palace cinnabar and the mineral isjust as likely to have originated in Tuscany or China as Spain.

Introduction

Cinnabar (HgS) has been used through the ages as a colouringmaterial for ancient manuscripts.1,2 The term cinnabar comesfrom the Persian word zinjifrah meaning ‘dragon’s blood’. Thecolour of cinnabar ranges from scarlet through brownish red toblack and even grey. The variation in colour is simply anindication of the levels of impurity. Indeed the term ‘red letterday’ originated from the use of cinnabar as a pigment to makeletters in ancient and medieval manuscripts.3,4 The term miniumwas applied to both cinnabar (HgS) and red lead pigments inantiquity. In Roman times, minium was reserved for mercury(II)sulfide but was also applied to lead tetraoxide during theRenaissance. Confusion in the interpretation of ancient recipesfor pigment mixtures is inevitable and is compounded by thepractice of adulteration of mercury(II) sulfide with leadtetraoxide for economic or artistic reasons. Cinnabar was minedin Spain by the Romans at two sites, namely Almaden andTarna. Another Roman source was at Monte Amiata inTuscany, Italy. Ancient mining sites for cinnabar are alsoknown in China. Cinnabar was used to decorate King Herod’spalace in around 6 BC. The question arises: could the cinnabarused in King Herod’s tomb have originated in Spain? Not onlywas the mineral cinnabar used for decorative purposes inancient and medieval times but two other arsenic sulfides wereused, namely realgar and orpiment. The name ‘realgar’ isderived from the Arabic rahj al-ghar meaning ‘powder of themine’. The name ‘orpiment’ is actually derived from the Latinword auripigmentum meaning ‘golden paint’ alluding to its useas paint and because of its colour, it was supposed to containgold. These two minerals were also used as colorants inantiquity and mediaeval times.

Cinnabar is normally accepted to have a space group of D3h.This means that there are 2 A1A modes which are Raman active,3A2B + 5EA modes, which are both infrared and Raman active.Wilkinson has given the spectrum of cinnabar as the A2A modesat 36, 102, 336 cm21 and the EA modes at 83, 92, 100, 280, and343 cm21 as the infrared active modes.5 The Raman activemodes are given by A1B at 43 and 256 cm21 and the EA modesat 72, 88, 108, 282 and 343 cm21.5–7 Two unassigned bands

were observed at 290 and 351 cm21.3,5 Raman spectroscopy hasbeen widely used as the technique of choice for studyingsamples from antiquity or from mediaeval times.8,9

Realgar is monoclinic with a tetramolecular structure of C2v

symmetry. The structure consists of four sulfur and four arsenicatoms bound together by covalent bonds to form a square and atetrahedron with the sulfur square cutting through the tetra-hedron in the middle.10,11 Farmer reports strong Raman bands at341, 224 and 169 cm21 with less intense components at 374,367, 359, 210, 204, 193 and 183 cm21.11 Orpiment has atetramolecular monoclinic unit cell of C2h symmetry, withAs2S3 layers in which spiral chains of AsS are running parallelto the c-axis. The major Raman bands of orpiment occur at 354,346, 312, 300 and 140 cm21.10,11

One of the difficulties of determining the spectra of thesulfides of heavy metals such as Hg and As is that the bandsoccur in the region below 400 cm21. Compared with infraredspectroscopy, Raman spectra are more easily obtained, howeverdifficulties may arise through the scattered signal beingobliterated through filters, as is the case in this work. Then FT-Raman spectroscopy must be applied to determine bands below250 cm21. In this work, we report the scanning electronmicroscopy and Raman spectra of three cinnabar samples fromthe two Spanish sites, Hunan province (China) and a samplefrom the wall of Herod’s palace.

Experimental

Origin of samples

The sample from Herod’s palace was a piece of the palace wallaround 3 cm wide by 7 cm long. The sample was analysed insitu; no cinnabar was removed from the wall sample. Bothpowdered and crystalline cinnabar samples were used in thisstudy from the Tarna and Almaden sites (Spain) and from theHunan site (China). Other mineral samples including cinnabarsamples were obtained from The Queensland University ofTechnology School of Natural Resources and the University ofWestern Sydney through Peter Williams.

This journal is © The Royal Society of Chemistry 2002

DOI: 10.1039/b109368c Analyst, 2002, 127, 293–296 293

Publ

ishe

d on

02

Janu

ary

2002

. Dow

nloa

ded

by S

yrac

use

Uni

vers

ity o

n 25

/11/

2013

07:

48:4

5.

View Article Online / Journal Homepage / Table of Contents for this issue

Electron probe microanalysis

The electron probe uses the characteristics of X-rays generatedfrom the interaction between the electron beam and the sampleto identify elemental composition of the sample. Selectedsamples for electron probe were polished with diamond pasteand coated with carbon to provide a good conducting surface.The EPMA analyses were done on the Jeol 840 A ElectronProbe Microanalyser at 20 kV, 39 mm working distance. Threesamples were embedded in Araldite, polished with diamondpaste on Lamplan 450 polishing cloth with water as lubricant.The Moran EDS software was used to analyse the samples in thefully standard mode. The secondary electron images were takenfrom the Jeol 35CF Scanning Electron Microscope, at 15 kVusing Image Slave software to capture the image.

Raman microscopy

The samples were placed on a polished metal surface on thestage of an Olympus BHSM microscope, which is equippedwith 103, 203, and 503 objectives. The microscope is part ofa Renishaw 1000 Raman microscope system, which alsoincludes a monochromator, a filter system and a CCD detector(1024 pixels). The Raman spectra were excited by a Spectra-Physics model 127 He–Ne laser producing highly polarisedlight at 633 nm and collected at a resolution of 2 cm21 and aprecision of ±1 cm21 in the range between 200 and 4000 cm21.Repeated acquisition on the crystals using the highest magnifi-

cation (503) were accumulated to improve the signal to noiseratio in the spectra. Spectra were calibrated using the 520.5cm21 line of a silicon wafer. Raman scattering with 760 nmlaser excitation was also used for the low wavenumber part ofthe spectrum.

Spectral manipulation such as baseline correction/adjustmentand smoothing were performed using the Spectracalc softwarepackage GRAMS (Galactic Industries Corporation, NH, USA).Band component analysis was undertaken using the Jandel‘Peakfit’ software package that enabled the type of fittingfunction to be selected and allows specific parameters to befixed or varied accordingly. Band fitting was done using aLorentzian–Gaussian cross-product function with the minimumnumber of component bands used for the fitting process. TheGaussian–Lorentzian ratio was maintained at values greaterthan 0.7 and fitting was undertaken until reproducible resultswere obtained with squared correlations of r2 greater than0.995.

Results and discussion

Scanning electron microscopy

The SEM back-scattered image of cinnabar from King Herod’spalace is illustrated in Fig. 1a. This image was taken from afragment of the palace wall. The image shows the cinnabarparticles as grey, since this is a scattered image. The cinnabar

Fig. 1 a, SEM image of cinnabar from King Herod’s palace. b, SEM back scattered image of cinnabar from Tarna (Spain). c, SEM back scattered imageof cinnabar from Almaden (Spain). d, SEM back scattered image of impurities in the cinnabar from Almaden (Spain).

294 Analyst, 2002, 127, 293–296

Publ

ishe

d on

02

Janu

ary

2002

. Dow

nloa

ded

by S

yrac

use

Uni

vers

ity o

n 25

/11/

2013

07:

48:4

5.

View Article Online

appears to be finely milled and free of impurity. Particle sizevaries from ~ 0.1 mm to 10 mm and the particles appear rounded.Such roundness indicates the mineral was finely ground. Thesize and shape of the particles suggest that the samples had beenmechanochemically treated through grinding. It should beunderstood that there are two types of impurity that might befound in minerals of archaeological and medieval significance.Firstly there are mineral impurities; i.e. impurities of otherminerals such as calcite, quartz, pyrites as is found in thisresearch. Scanning electron microscopy lends itself to thedetermination of these phase impurities. There is a second typeof impurity, which is an elemental impurity or even a molecularimpurity within the crystals. Such impurities result fromisomorphic substitution or by solid solution formation. Such animpurity must be determined through a method of elementalanalysis such as X-ray fluorescence or ICP-AES. These types ofimpurities are at the sub-crystalline level.

The back-scattered image of the cinnabar from Tarna (Spain)(Fig. 1b) shows much larger particle sizes with a significantimpurity of calcite. The particles appear as white in this imageas the image is a back-scattered image. The particles are > 10mm and are irregular in shape. The back-scattered image of thecinnabar sample from Almaden (Spain) shows impurities ofpyrites and quartz (Fig. 1c and d). The areas marked on theback-scattered image indicate where the energy dispersive X-ray analyses were undertaken. The conclusions reached fromelectron microscopy suggest that the samples from the Spanishmine sites when compared to that of Herod’s palace are not thesame. This leads to the concept that the cinnabar may haveoriginated from some other source. A likely site is the ancientRoman cinnabar mine at Monte Amiata in Tuscany, Italy.Another possible site is in Hunan province in China wherecinnabar has been mined since antiquity.

The energy dispersive X-ray analysis of the samples fromHerod’s palace, Tarna and Almaden are shown in Fig. 2. Energydispersive X-ray analysis shows the sample from Herod’spalace wall to be of high purity. No impurities could be found.The EDX analysis for the Almaden sample shows impurities ofSi and Fe originating from quartz and pyrite impurities (Fig.2b). In contrast, the cinnabar sample from Tarna containedimpurities of both quartz and calcite. This is illustrated by thehigh Si and Ca readings in the electron probe analyses (Fig. 2c).The EDX analysis shown in Fig. 2d shows some of the cinnabarfrom Tarna to be of high purity. If this cinnabar was used inRoman times, then some purification process must have beenused.

The EDX analyses raise a number of questions as there is anapparent disparity between the phase impurities of the Spanishcinnabar samples and that of the sample from Herod’s palace:(a) Was some purification process before the application of thecinnabar used in Herod’s palace? (b) Are the samples we areanalysing the same as those obtained from the ancient minesites? Or are the ancient mine sites thoroughly worked out? (c)How representative are the cinnabar samples of a particularmine site? The fact that the cinnabar from the wall sample ofHerod’s palace is of high purity with no apparent additionalphase impurities suggests that the cinnabar was purified throughsome process. Two possible techniques could have beenemployed: (i) sublimation of the mercury sulfide could havebeen used. Such a technique would have certainly made small-particle cinnabar of high purity. (ii) The ore may have beenprocessed by crushing and washing. The less dense fractions inthe ore such as calcite could be fractioned from the cinnabar. Itis possible that the cinnabars we are examining are the same asthose mined from the ancient sites, since the cinnabars wereobtained from the same sites. The colour of the cinnabars in the

Fig. 2 a, Electron probe analysis of Cinnabar coated wall sample from Herod’s palace. b, Electron probe analysis of ancient cinnabar sample from Almaden.c, Electron probe analysis of first area ancient cinnabar sample from Tarna. d, Electron probe analysis of second area of an ancient cinnabar sample fromTarna.

Analyst, 2002, 127, 293–296 295

Publ

ishe

d on

02

Janu

ary

2002

. Dow

nloa

ded

by S

yrac

use

Uni

vers

ity o

n 25

/11/

2013

07:

48:4

5.

View Article Online

samples analysed suggests that the samples contained noelemental impurities.

Raman microscopy

One of the difficulties of determining the spectra of the sulfidesof heavy metals such as Hg and As is that the bands occur in theregion below 400 cm21, which makes the measurement byinfrared spectroscopy not easy. On the other hand, Ramanspectra are more easily obtained, however difficulties may arisethrough the scattered signal being obliterated through filters asis the case in this work. Then FT-Raman spectroscopy is moreappropriately applied to determine bands below 250 cm21.Forneris suggested that the spectra of these sulfides of Hg andAs may be conveniently divided into three sections namely: (a)the region centred upon 355 cm21 where the stretchingvibrations are observed (b) the region centred upon 250 cm21

ascribed to the bending vibrations and (c) the region below 100cm21 assigned to the lattice modes.10 The complexity of thespectra and the large number of bands previously reported arenot observed in this work. The bands between 250 and 400cm21 are readily observed using the Raman dispersiveinstrument; however the filters cut in at around 220 cm21, andas a consequence bands below this value must be determined byFT-Raman spectroscopy. The Raman spectra of cinnabar,realgar and orpiment are illustrated in Fig. 3. Clearly Ramanspectra may be used to distinguish between the three minerals.The Raman spectra of the cinnabars from Herod’s palace andfrom the three ancient mine sites of Tarna, Almaden and Hunanare shown in Fig. 4. Clearly the spectra all appear similar. Thusno distinction can be made between the Raman of cinnabar fromthe three mine sites and that of Herod’s palace wall.

Conclusions

It is of scientific interest to attempt to identify the origin of thecinnabar used for the decoration of Herod’s palace. Scanningelectron microscopy shows the cinnabar coating of the wall ofHerod’s palace is of high purity and has been finely ground. Thesamples of cinnabars from Spain all show impurities which are

not observed in the Herod’s palace cinnabar. The Middle East isas close to China as to Spain and the cinnabar from Chinahaving been used for the decoration of Herod’s palace seemsjust as likely as the cinnabars from Spain. Further, the ancientRomans had a cinnabar mine in Tuscany, and this site is highlyprobable. At this stage, no definitive conclusions can bemade.

Acknowledgement

The Centre for Instrumental and Developmental Chemistry ofthe Queensland University of Technology is gratefully ac-knowledged for financial and infra-structural support for thisproject. Martin Crane from the University of Western Sydney isthanked for the supply of the minerals realgar and orpiment. DrSilvia Rosenberg is thanked for the sample of King HerodAspalace wall-painting and Professor Fernando Rull for thesample of cinnabar from the Tarna mine.

References

1 H. G. M. Edwards, D. W. Farwell, E. M. Newton and F. R. Perez,Analyst, 1999, 124, 1323.

2 H. G. M. Edwards, D. W. Farwell, E. M. Newton, F. R. Perez and S.J. Villar, Appl. Spectrosc., 1999, 53, 1436.

3 G. Van Hooydonk, M. De Reu, L. Moens, J. Van Aelst and L. Milis,Eur. J. Inorg. Chem., 1998, 5, 639.

4 H. G. M. Edwards, D. W. Farwell, F. R. Perez and G. J. Medina,Analyst, 2001, 126, 383.

5 G. R. Wilkinson, in Molecular Spectroscopy (Conference Proceed-ings) 1968, Institute of Petroleum, London, 1968, p. 43–81.

6 S. D. Ross, Inorganic Infrared and Raman Spectra, McGraw-Hill,London, 1972.

7 W. Scheuermann and G. J. Ritter, Z. Naturforsch., 1969, 24, 409.8 L. Burgio, D. R. Ciomartan and R. J. H. Clark, J. Raman Spectrosc.,

1997, 28, 79.9 C. J. Brooke, H. G. M. Edwards and J. K. F. Tait, J. Raman

Spectrosc., 1999, 30, 429.10 R. Forneris, Am. Mineral., 1969, 54, 1062.11 V. C. Farmer, in Infrared Spectra of Minerals, Mineralogical Society,

London, 1974, p. 117.

Fig. 3 FT-Raman spectra of the three mineral colorants of archaeologicaland mediaeval significance. (a) Cinnabar. (b) Realgar. (c) Orpiment. Fig. 4 Raman spectra of four cinnabars of archaeological and mediaeval

significance.

296 Analyst, 2002, 127, 293–296

Publ

ishe

d on

02

Janu

ary

2002

. Dow

nloa

ded

by S

yrac

use

Uni

vers

ity o

n 25

/11/

2013

07:

48:4

5.

View Article Online