Raman Spectroscopy in Archaeology and Art History

3. Raman Spectroscopy in Archaeology and Art History

British Museum, 
London 20th November 2001
The Editor

This one-day meeting was held in the British Museum, an institution famous (or infamous, depending on your viewpoint) for housing the ‘Elgin’ Marbles removed many a century ago from the Parthanon in Athens and its Round Reading Room where Karl Marx carried out his research in writing Das Kapital. The Marbles are still there, but the books have gone to the brand new British Library nearby and the Reading Room has become redundant. Sir Norman Foster, the architect of the revamped Reichstag masterminded the construction of a magnificent Education Centre in the rotunda and the meeting was held in one of its superb new lecture theatres.

My first impression was one of surprise at the number of people attending. I expected to see perhaps 30 faces, many of whom would be old friends. The friends were there, but most of the attendees, almost 120 of them were from all over Europe and many indeed from the Art World. These were enthusiastic practitioners expert in Art History, Restoration, Conservation and Archaeology.

The lecture programme was extensive and the posters many and very varied. It is quite clear that Raman is now well on its way to becoming established as a primary tool in the analysis of artefacts. Again and again lecturers stressed that since no sampling was needed and the laser power so low that no damage was done to the specimen’s surface, the method was unequally attractive to conservators and curators. But what can Raman tell us? An amazing amount. There was so much on offer that I can only report a few examples.

Professor Couprey using minute amounts of laser power examined the precious Fecamp Scriptorium dated between 990 and 1140AD. Two blue dyes were extensively used in this period based on Indigotin – an organic product and Lapis Lazuli – a mineral: similarly Minium and Vermillion in the red. By careful scanning of several images it is possible to trace the work of individual artists since it is a reasonable presumption that monks favoured their own particular painting materials. However, another manuscript from Saint Germain incorporates two different dyestuffs in one image. This cannot be explained by simple progress because the Corbic Scriptorium (XIIth Century) uses only one blue and this is based on Ultramarine. The differences in technique were unsuspected and are now areas of investigation.

We heard of several cases where forgeries have been uncovered. By simple historical methods, it is easy to produce a table defining when particular pigments or dyes became available. If Raman can identify the dyes, the detection of forgeries produced by careless or ignorant forgers can easily be revealed. So, we were shown ‘Roman’ artefacts containing the modern copper phthalocyonine dyes or even Rutile (titania) whites [the Romans used CaCO3 , the Chinese CaSO4].

Dr Steve Bell is interested in Chinese documents. Vast numbers were found sealed in caves and hence in good condition by the Stein Expedition of 1907. Many of these documents found their way into private collections and museums in Europe. Included was the first ever printed book (dated in the 9th Century). Steven has used both resonance Raman and surface enhanced techniques to study the paper in these specimens. The paper is almost invariably dyed yellow and by Raman it is clear that the stain is Berberine. The trick is to adjust the laser wavelength until absorption occurs and then look for Raman bands. Steve Bell showed several examples but always against a huge and overwhelming fluorescence background, but undaunted we were shown how the problem can be eliminated – See Note 1 at the end of this report. Now Berberine was not the only dye used by the ancient Chinese. Some paper fragments show evidence of the presence of Palmatine. Where tree bark extracts have been used both Palmatine and Berberine are present together. It is quite clear that different artists used different dyes – the questions are who and why. Clearly, Raman is invaluable in this type of investigation.

Rob Withnall uses SERS, resonance SERS and resonance Raman to obtain even higher sensitivities [a very comprehensive, possibly the MOST up-to-date coverage of SERS and related techniques is available in the IJVS archive in Volume 4, Edition 2]. Dr Withnall is attempting to use his methods quantitatively – a very tricky prospect at such ultra high sensitivities but he showed us how considerable progress has been made detecting Raman signals in the 10-7–10-9M range. His work is not only developmental as he showed us spectra taken from the blue background (indigo) in a Tudor miniature portrait- see below.

SERRS spectra of Alcian Blue in the concentration range
1.8 x 10-9 M to 3.6 x 10-7 M.

Prof. Smith of the Muséum National d’Histoire Naturelle in Paris, is interested in much older artefacts – pre-Columbian axe heads. These are, of course, made of hard stone and the exact identity can tell archaeologists much about their origin, how far they have travelled and hence something of migration and trade patterns so long ago. Normally, Petromicro graphical methods and X-ray fluorescence would be used but these are very destructive.

The Pyroxene, garnet and titanite minerals all give good Raman spectra and all or some can be found together in the same sample axe head. The problem is to subtley analyse the spectra. Prof. Smith showed that within a mineral type, the untutored eye would deduce that all the spectra were essentially the same – hardly surprising – but it is possible to relate the subtle differences in frequency to the detailed composition of the mineral . [A data-base was referred to and I will attempt to identify this and include it in Spectroscopists’ Bookshelf. Editor’s Note].

Almost all of the papers at this stage of the meeting had described applications based on visible laser Spectrograph CCD instruments. Howell Edwards then introduced work based on F-T instrumentation. In these experiments, much higher laser powers are used but the exciting laser wavelength is in the near infrared (1.064µ) and as a result fluorscence is almost eliminated as a problem. Howell described experiments on Mummies exhumed by the Petry expedition to Egypt of 1906. The skeletal remains and their clothing had been seriously corroded by Natron (Na2CO3/NaHCO3). Cellulose decomposes over the years through hexose ring opening but good diagnostic spectra were obtained. It was pointed out that Raman is far better than IR in this type of work because the vibrations of the numerous OH bands dominate much of the IR and also cause band broadening – not a problem in Raman scattering. Howell then showed results on a bead in the eye socket of a mummified cat. The bead proved to be cats claw, not as expected glass or amber. Why, no one has yet explained.

Mummification is not confined to Egypt. Much more recently ~1475 AD ice bound mummies were interred in Greenland. No chemicals are involved so degradation in these cases is much less severe than in Egypt. Howell obviously has a ‘thing’ about mummification because he then moved on to South American mummies of around 1000AD. Spectroscopy proved that rather crude mummification had been attempted. The range of materials studied is quite remarkable with spectra from various ivories (some fake) being described. Howell Edwards is an expert on Lichens [see report on the IRDG meeting in London to follow] and he has applied his methods to lichen infestations in frescoes in the Palazzo Fornese (16thCentury) and in an ancient Spanish Church. Amazingly, and very destructively some have penetrated up to 10mm into the surface of the artwork.

The next two papers described work on pottery. Prof. Colomban of the CNRS showed data on glazes, sub-glazes and the body of ceramics. The latter are described as hard paste – Kaolin + Sand + Feldspar or soft paste – Sand + Clays + Chalk + Frit. Hard pastes are Alumina rich, soft richer in silica. Both can readily be distinguished using Raman methods. Silicate structures are specific to particular glazes and again can be distinguished.

Examples of application came thick and fast – identification of components in glazes, characterisation of the paste type from well known European factories, proof of fluorescence etc. He was asked why he thought Raman was a useful tool in his work. He replied that XRF yields elemental data only, the detailed fingerprint from Raman tends to be far more specific.

Prof. A. Zoppi of the University of Florence, continued in a similar vein but his pottery subjects were very old e.g. from the Tell Beydar site and dated around 3000B.C. His analysis was aimed at identifying both the source material in the body of the ceramics and also the glazes. He is also interested in tracing the firing techniques – for example their temperature and whether reducing or oxidizing. The analysis is detailed and agrees with the results from electron diffraction and X-ray fluorescence but as Prof. Zoppi pointed out, Raman is non-sampling and fast.

(a) Hematite Fe2O3 is the most stable iron compound under oxidising conditions, responsible for the colour of typical reddish sherds.

(b)Black magnetite Fe3O4 forms under reducing conditions and is the most important mineralogical phase found in black sherds.

(c)Diopside (CaMgSi2O6) is a typical “high temperature” mineral that forms (irreversibly) in a-rich clays through the transformation of primary calcite, above ~850°C. Its presence is thus an indication that at least such temperature values were reached firing the firing process.

We then moved on to artists materials. Prof. Vandenabeele from Ghent discussed artists materials and the possibility of dating and/or proving authenticity based on analysis of pigments. Raman with its non-sampling ability is ideal, the ability too to analyse under excellent spatial resolution (just a few microns) and, a new development, the use of fibre optic probes allowing the art historian to have data from large objects and/or fixed ones. Too many examples were given to list all but two are described as typical. Two ‘medieval’ miniatures have been examined – one of Solomon, the other of St. Barbara. Pigment analyses revealed BaSO4 in the latter confirming that it is not medieval. René Magrithe produced pictures, many of which are being restored. Raman analyses is enabling art historians to follow the way his palette changed through his career. One unexpected find is that Magrithe used starch in his colours.

Prof. Robin Clark from University College in London, then spoke about his long standing work in the Raman field. Like other authors he demonstrated the value of the technique in identifying pigments and hence in some cases demonstrating that specimens were fake. One of the very unexpected results of his research were described e.g. he has found ultramarine in a Florentine ceramic glaze – showing that ultramarine can withstand 900ºC. He extended the range of pigments described by previous contributors by showing MnO2 in pottery and Pb2Sb7 as a yellow pigment in an Egyptian vase. On the other hand, some pigments suffer from the atmosphere and become difficult for Raman e.g. in a 13th Century lectionary PbCO3 white has become coated with a layer about 1µ thick of the black PbS. Similarly basic copper carbonate (Azurite – blue/green) goes to CuS over the years.

Robin described a run in with a slippery looking gent who showed him an ‘Egyptian’ papyrus he planned to offer at auction and hoped it would raise two million pounds. Robin found anatase and oddly the owners never came back to collect their loot! To finish Prof. Clark described the case of the Vinland map in the library of Yale University. If genuine the map clearly demonstrates that America was known to Europeans well before Columbus ‘discovered’ the continent. He wants to apply his wizardry to the map.

To finish this part of the one day meeting, Prof. Gilbert of the University of Liege concentrated on green copper pigments in illuminated manuscripts. Again we heard a favourable comparison with X-ray fluorescence – the value of the Raman fingerprint was stressed. As an example Prof. Gilbert’s group have found four different types of copper sulphate. It was also shown how pigments can vary from country to country and can be identified by Raman methods and used for proving provenance.

Click here to view pdf file [Please note the file size is huge so may take a while].

Raman methods can be used in depth profiling if the sample is transparent – we were shown an example on a yellow specimen – the red laser will pass through such a material. Dr. Robinet of the British Museum then explored another area – corrosion deposits on iron and bronze artefacts. Again, a favourable comparison as made with X-ray methods. Dr. Robinet has also used infrared but finds Raman preferable for much of his work on iron objects and also on Egyptian copper bronzes. He showed spectra of surface corrosion and in cracks. Peculiarly, he sees Acetates possibility from out gassing from adjacent wood. We saw a wide range of spectra many of very high quality indeed demonstrating the formation of soaps on surfaces but one spectrum fascinated we spectroscopists – a Raman band of high intensity near Δν=2200 cm-1. What CAN it be due to, we all asked, but no-one came up with a suggestion.

Dr. Bouchard of the Muséum National d’ Histoire Naturelle in Paris, had the unenviable task of giving the last lecture. To my amazement, no-one had left, so Dr. Bouchard had an excellent and still enthusiastic audience. His interest in several different areas including mediaeval stained glass and rock art and corroded artefacts including coins. Mediaeval stained glass lasts well but it does deteriorate, reactions occur over very long periods between the pigments and the glasses themselves. By running Raman spectra on real samples and comparing with especially produced model mixtures the deterioration can be followed in detail. A submarine wreck contained iron ingots (as ballast to stabilise the craft?). The corroded surface contained basic iron III chlorides. Hardly surprising, but nevertheless useful. Although fluorescence proves to be a persistent problem, Prof. Bouchard emphasised the unique value of Raman Spectroscopy stressing that the information available is so detailed that the palette for stained glass artists can be developed.

In addition to the lectures, we were also able to inspect some excellent posters. I picked up, or was sent extracts from these and we have included them as pdf files. 

G.Di Lonardo, F.Ospitali and F. Tullini, Università di Bologna, Italy
Danilo Bersani, Physics Department, University of Parma, Italy
[Please note the file sizes are quite big so may take a while]

To summarise – I don’t think I have attended a better meeting for years. I was bowled over by the enthusiasm and dynamism of everyone involved – the attendees, the lecturers and the organisers. From my own point of view I found it fascinating that attendees and lecturers fell into two groups: established spectroscopists who had been contacted by conservators or museum staff and Scientific conservators who had “got into” Raman spectroscopy. Both have their value but in the end the latter are the real driving force in the development of new techniques.

Note 1: Steve Bells’ method of coping with fluorescence

In many Raman spectra, the bands appear against an overwhelming background fluorescence. If a visible laser is being used, the fluorescence can be reduced by exposure to the laser, but even so fluorescence can still be a severe problem in that the bands tend to be obscured – if the fluorescence background can be reduced perhaps the Raman bands can be revealed.

  1. Steve records a conventional spectrum → memory.
  2. He then moves the frequency scan by about ½ width of the Raman bands [This can be done on a CCD/spectrograph by tweaking the grating or in an F-T Raman instrument by changing the laser line frequency]
  3. Run another spectrum.

Subtract 1 from 3. The background will essentially cancel out but the Raman lines will appear something like a first differential spectrum against a fairly flat background. Steve Bell uses ‘curve fitting’ to resurrect the spectrum, but a first differential (dI/du) will do the job. Although background noise will be apparent, the Raman spectrum now looks very adequate.

As a result of this wonderful meeting, I am convinced that Raman will provide a really significant input into this fascinating field, an input much more significant than sampling currently used.

After the meeting, I declined a glass of wine and hurried to the nearest Underground Station. As I reached the train level, we were informed that the station was CLOSED due to a “security event”. After walking to the next station I duly missed by train home – perhaps I should have stayed for that drink!!

REF: P.J.Hendra. Int.J.Vibr.Spec., [www.irdg.org/ijvs] 5, 6, 3  (2001)