Analytica Chimica Acta 822 (2014) 51–59

Contents lists available at ScienceDirect

Analytica Chimica Acta

journa l homepage: www.e lsevier .com/ locate /aca

Preparation of thin-sections of painting fragments: Classical andinnovative strategies

Emeline Pouyet a,*, Anna Lluveras-Tenorio b, Austin Nevin c, Daniela Saviello d,Francesco Sette a, Marine Cotte a,e

a European Synchrotron Radiation Facility, 6, rue Jules Horowitz, Grenoble F-38000, FrancebConsorzio Interuniversitario Nazionale per la Scienza e Tecnologia dei Materiali (INSTM), Via G. Giusti 9, Firenze 50121, ItalycConsiglio Nazionale delle Ricerche – Istituto di Fotonica e Nanotecnologie (CNR-IFN), Dipartimento di Fisica, Politecnico di Milano, Piazza Leonardo da Vinci32, Milano 20133, Italyd Politecnico di Milano – Dipartimento di Chimica Materiali e Ingegneria Chimica, Piazza Leonardo da Vinci 26, Milano 20133, Italye LAMS (Laboratoire d’Archéologie Moléculaire et Structurale) UMR-8220, 3 rue Galilée, Ivry-sur-Seine 94200, France

H I G H L I G H T S

* Corresponding author. Tel.:+ 33 476 882944.E-mail address: pouyet@esrf.fr (E. Pouyet).

http://dx.doi.org/10.1016/j.aca.2014.03.0250003-2670/ã 2014 Elsevier B.V. All rights reserved.

G R A P H I C A L A B S T R A C T

� New methods for the preparation ofthin sections frompainting fragment,in particular for FTIR microscopy.

� SES method: embedding-free ap-proach, suitable for the preparationof fresh or rich-binder paintings.

� ARE method: use of an AgCl barriercoating to prevent the penetration/contamination of embedding resin.

� Application to real cases: a deeperinsight into ancient manufacturingor degradationprocesses of historicalartefacts.

A R T I C L E I N F O

Article history:

Received 2 December 2013Received in revised form 13 February 2014Accepted 17 March 2014Available online 20 March 2014

Keywords:PaintingsSynchrotronSample preparationFourier transform infrared spectroscopyCultural heritage

A B S T R A C T

For more than a century, the analyses of painting fragments have been carried out mainly through thepreparation of thick resin-embedded cross-sections. Taking into account the development of innovativemicro-analytical imaging techniques, alternatives to this standard preparation method are considered.Consequently, dedicated efforts are required to develop preparation protocols limiting the risks ofchemical interferences (solubilisation, reduction/oxidation or other reactions) which modify the sampleduring its preparation, as well as the risks of analytical interferences (overlap of detected signals comingfrom the sample and from materials used in the preparation). This study focuses particularly on thepreparation of thin-sections (1–20mm) for single or combined fourier transform infrared (FTIR)spectroscopy and X-ray 2Dmicro-analysis. A few strategies specially developed for themFTIR analysis ofpainting cross-sections have already been reported and their potential extrapolation to the preparation ofthin-sections is discussed. In addition, we propose two new specific methods: (i) the first is based on afree-embedding approach, ensuring a complete chemical and analytical neutrality. It is illustratedthrough application on polymeric design objects corpus; (ii) the second is based on a barrier coatingapproach which strengthens the sample and avoids the penetration of the resin into the sample. Thebarrier coating investigated is a silver chloride salt, an infrared transparent material, which remainsmalleable and soft after pellet compression, enabling microtoming. This last method was successfully

[(Fig._1)TD$FIG]

F

52 E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59

applied to the preparation of a fragment froma gilded Chinese sculpture (15th C.) andwas used to unravela unique complex stratigraphy when combining mFTIR and mXRF.

ã 2014 Elsevier B.V. All rights reserved.

1. Introduction

1.1. Embedded cross-sections: a classical preparation method for themicro-analytical study of painting fragments

Within the large family of artworks, paintings have attractedsignificant scientific interest leading to an improved characteriza-tion of these complex artistic materials. Related scientific studiesgenerally aim at (i) obtaining information about paintingtechnology, and/or (ii) understanding degradation mechanismsaltering the works of art.

Over the last few decades, developments of non-invasiveinvestigation methods, avoiding or minimizing the need forsampling, have increased in the cultural heritage (CH) domainproviding valuable data at the painting scale. These techniques(X-ray fluorescence, X-ray radiography, infrared and ultravioletphotography, multispectral and hyperspectral imaging, etc.)should receive priority over more invasive methods since paintedcultural artefacts are unique, and the entire preservation ofpaintings is the ultimate goal of any conservation study.Moreover, their development is crucial since they providesupplementary information on larger areas of the artwork,corresponding to more statistically relevant information, andhelp in refining the area where sampling should be performed.These techniques are extensively applied for revealing over-painting, or preparatory drawings as well as painting compositionat the artwork scale which may provide valuable information forauthentication and technical art history.

However, painting presents usually very complexmulti-layeredstructures, with heterogeneities at the micrometer scale, conse-quently micro-sampling is sometimes necessary for probing theintrinsic stratigraphic structure ofmaterials (Fig.1). In this context,spatial information at sub-millimetre scale can be obtained bymanually separating the different layers composing the structureof the fragment, but this completely destroys the sample, andrequires a minimum thickness layer (ca. >20mm). Alternatively,obtaining a full section preserves the 2D structural information.Formore than one century, painting fragments have been preparedmainly as resin-mounted and polished thick sections (typically inthe order of 1mm3), called cross-sections. Before the well-knownpaper by Laurie on Egyptian blue [1], a pioneering article reportedthe preparation of the first painting cross-sections in 1910 [2]. In1934, Gettens extended the method to fine art in general [3]. Animportant and exhaustive paper by Plester, in 1956, discusses theadvantages and the necessity to prepare painting cross-sectionsand details the interest for using a very small fragment from

ig. 1. Systematic preparation stra

paintings which may yield significant information [4]. The samplepreparation described in this reference is almost the same as thepreparation used today and is based on four steps: (1) mounting ofthe sample, (2) embedding it in resin, (3) polishing the polymer-ized block to reveal the stratigraphy and (4) storing of the wholeresin block for future studies. Such cross-sections are relativelyeasy to achieve, to handle and to store, and are compatible withclassical investigations such as visible and electron microscopies.This preparation method slowly became a routine for micro-fragment examination in museums and related institutions, whichcontinues to give valuable information on painting layer struc-tures, pigments and painting techniques [5].

Nowadays, the set of micro-analytical techniques available forthe study of paintings is constantly expanding with theemergence and improvement of instrumentation. For exampletime-of-flight secondary ion mass spectrometry (TOF-SIMS), X-ray photoelectron spectroscopy (XPS), electron energy lossspectroscopy (EELS), X-ray computed tomography (CT), atomicforce microscopy (AFM), surface enhanced raman scattering(SERS) and scanning transmission ion microscopy (STIM) havebeen applied to analyse micro-samples. A careful and specificprotocol for sample preparation is a major determining factor inthe success of these new analytical approaches. In most cases,these suitable procedures are more time consuming than theclassical cross-section method mentioned above, but offer a gainin terms of data quality, reliability and consequently in terms ofdata processing. Accordingly, an increasing number of scientists,conservators, historians and archaeologists are interested in thedefinition of “best-practices” and protocols for painting fragmentpreparation. This is nicely illustrated by the work carried out onthis topic within the framework of the CHARISMA project [6].

The present work focuses on the analysis of painting fragmentsusing, individually or as a combination, a set of synchrotron basedmicro-analytical techniques: micro fourier-transform infraRedspectroscopy (mFTIR), micro X-ray fluorescence (mXRF), microX-ray absorption near edge spectroscopy (mXANES) and micro X-ray diffraction (mXRD). These techniques are increasingly appliedto reveal the organic and inorganic composition of paintingfragments, and more generally artistic materials, at a micrometricscale. These techniques are often applied on cross-sections andprobe different depths below the surface. Hard X-rays canpenetrate a few tens of microns on paint materials, while infraredlight penetration in attenuated total reflectance (ATR) mode istypically limited to the first 0.5 to 2mm, depending on thewavelength, the angle of incidence, the ATR crystal and the samplecomposition. By preparing and analysing thin sections, better

tegy of painting thin-sections.

E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59 53

control of the in-depth resolution is achieved and efficiency ofmicro-analysis is significantly improved. The preparation of thinsections also offers enhanced data quality (in particular for FTIRanalyses [7]), reduced acquisition time, new imaging capabilities(in particular for XANES data [8]), and correlated reduced dose [9].Finally, in case of potentially destructive analyses (staining, SIMS,Raman), the use of thin sections allows working on smallerquantities (few microns), without affecting the remaining part ofthe fragment. For the samples and the analytical techniquesconsidered here, the typical section thickness required is in therange of a few microns to a few tens of microns.

Therefore, in the following, after a generic discussion abouttechnical issues related to the classical preparation of sections(Section 1.2), a review of the alternative strategies developed forsample preparation for the analysis of painting fragments isproposed. Even if preliminary dedicated to cross-section prepara-tion their adaptability to thin-section preparationwill be discussed(Section 1.3).

1.2. Common issues with the preparation of sections

Cross- and thin-sections are usually obtained after embedding asample in a medium. This is generally a synthetic embedding resin(polyester, epoxies or acrylics) with a twofold function:

-

it consolidates the sample and allows the entire preservation ofits structure during polishing or microtoming,

-

it provides an easy support for manipulation, to position thesample for microscopic analysis and to store.

The long use of such organic embedding media demonstratesthe efficiency of this traditional approach. However, it also leads tocritical chemical and/or analytical interferences. In our terminolo-gy, chemical interference means that the sample has beenchemically modified (dissolved, oxidised or contaminated) duringthe sample preparation; similarly, analytical interference impliesthat exogenous materials used for sample preparation contributeto the detected signal, whatever the technique. This interferencemay hide or distort signal from the sample.

Regarding chemical interferences, the dissolution of someoriginal or degraded painting material in the liquid prepolymer isone of the most important issues [5]. The solubilisation of waxespresent in furniture finish samples, the solubilisation of organicdyes on inorganic carriers in modern pigments and the solubilisa-tion of natural resin samples of dammar, mastic and copal havebeen reported [10]. Moreover, solubilisation can also take placeafter embedding; for example, water sensitive compounds can bedissolved during wet polishing or wet cutting.

Analytical interferences will strongly depend on the problem-atic, the sample permeability and on the analytical methods used.Such interferences are particularly critical for the analysis of (i) thesurface of the fragments and (ii) the bulk of porous and permeablesamples (Fig. 1). In the latter case, when the binder content is low[11],more precisely containing less than 20% of bindingmedia [12],infiltrated embedding material fills the empty network around thepigment and coat the particles. It consolidates the structure forfurther sectioning, however the presence of such an embeddingmediumwithin the sample may affect the direct observation withvisible microscopy by hiding some aspects of the stratigraphy,thereby preventing the identification of the different paint layersand their interfaces [10]. The presence of an organic embeddingmediumwithin the sectionmay also affect the identification of theconstitutive original organic components of the painting. Thisissue concerns a large range of analytical techniques such as FTIR,Raman microscopy, TOF-SIMS, STIM and visible microscopy withor without staining [5]. For vibrational spectroscopies and more

precisely for FTIR, the strong absorption features related to theresin C��H and CQO molecular groups, may compete or hidetypical bands of the paint materials. Many authors refer to thedifficulties of identifying characteristic absorption bands of paintcomponents due to bands overlap issues [5,13]. The identificationof organic binders and varnishes, in particular those made ofpolyester compounds such as oil, is complex. Moreover, overlaps inthe 1400–900 cm�1 region can compromise the identification andlocalisation of inorganic compounds such as sulphates orphosphates. Data treatment which may involve computationalsubtraction of the embedding media contribution is inherentlylimited, and in extreme cases can lead to distortions of the samplebands and limit the detection of other components, or even lead toerroneous absorption bands [14].

For X-ray based techniques, an organic embedding mediumusually presents less risk of analytical interference. Some additivesor fillers may contain elements such as Al, Si, Fe, which, in the caseof element-sensitive techniques (e.g. XRF, XAS) can prevent theaccurate localization of the endogenous contribution of theseelements.

In addition to the embedding step itself, grinding and/orpolishing of any embedded sample can also affect the compositionof the section, by contaminating the sample surface with theformation of a thin resin film, by smearing components of differentlayers throughout the entire sample, and by depositing exogenouspolishing materials such as SiC [15].

Consequently, depending on the sample and the analyticaltechniques, risk of chemical or analytical interferences must beassessed and a preparation strategy must be adapted (Fig. 1). Thediscussion here focuses on the preparation of sections, howeversimilar attention should be paid to the choice of materials used forhandling the sample before preparation, as well as during and afterthe analyses.

1.3. Strategies for section preparation: state of the art

Regarding analytical interferences, mFTIR, and other moleculargroup imaging techniques (e.g. Raman, ToF-SIMS, fluorochromestain tests) are the most demanding in terms of adapted samplepreparation. For mFTIR, various strategies have been proposed asalternatives to classical embedding in resin. The followingdiscussion reports the major innovations [5,12,16]. Fig. 2 summa-rizes the main strategies explored in previous works, as well as thenew protocols specifically developed in the present study.

In an optimal approach, the sample is prepared without anyembedding process, either by compression or with a microtome.The diamond anvil micro-compression cell has long beenidentified as an efficient device avoiding any use of embedding/coating material, first reported for artistic materials in 1975 [17].Even if this process avoids analytical interferences, the maindrawback of this approach is the compression step, which maydistort the stratigraphy, and mix the different layers [18].Moreover, in the case of highly heterogeneous samples, with astratigraphy of successive layers presenting different hardness, it isdifficult to achieve homogeneous compression.

Alternatively, another preparation strategy relies on the fullsubstitution of the organic resin by an IR-transparent salt, such asKBr, BaF2 or AgCl. These salts are commonly used for FTIR bulkanalyses, since they have no absorption bands between 4000 cm�1

and 400 cm�1. The sample is embedded in these salts, usuallypressed into a pellet, subsequently polished or cut to producecross-sections allowing stratigraphic micro-FTIR analysis withoutanalytical interference from the embedding material [14,18–20].Some problems appear during pellet preparation, such as thedifficulties related to the positioning of the sample or thedisaggregation of paint fragments during pressing [14]. From an

[(Fig._2)TD$FIG]

Fig. 2. Schematic view of the main alternatives for the preparation of painting thin-sections for mFTIR analysis in transmission mode.

54 E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59

analytical point of view, this fulfils the requirements regardinganalytical neutrality for mFTIR analysis. However, it was initiallyadapted for the preparation of cross-sections rather than to thin-sections. The pellet is usually brittle and sectioning with amicrotome ends with the crumbling of the salt, which no longermaintains the sample. Alternatively, thin-sections can be obtainedbya double-side polishing approach, but this consumesmost of thesample, is time consuming, and challenging for the preparation ofonly fewmicrometer-thick sections. A significant risk is the loss ofthe sample during the last steps.

In order to find an optimal balance between the advantageousanalytical neutrality of these salts, and the difficulties related to themechanical properties of the pellets, an intermediate solutionmerging the use of IR-transparent salts with the classicalembedding resins has been proposed [21]. This solution belongsto a more general strategy where the sample is first protected witha barrier coating, before being embedded in a standard resin. Thebarrier coating acts as a protection towards the infiltration of theembedding resin, and strengthens the painting fragment. Differentbarriers have been tested: (i) two IR-transparent salts: KBr andNaCl [21]; (ii) two metals: aluminium, used as a thin foil [5,7], andgold, spread with a sputter coater [5]; and (iii) organic mediaincluding cyclododecane (CDD, C12H24) which sublimates in roomconditions [12].

The previous strategies were chiefly developed for ATR-mFTIRanalysis carried out on samples prepared as cross-sections, but arenot always fully applicable to the preparation of thin-sections.Therefore, new investigations were conducted focussed ondedicated strategies for the preparation of thin-sections and fromthe previous established protocols, two novel approaches areproposed. They are illustrated through applications on both modeland historic polymeric-based samples (Section 3). In both cases,considering the drawbacks listed above regarding the double-polishing approach, the two protocols were based on the use of amicrotome.

2. Experimental

2.1. Samples

The complexity and the diversity of paintings as well as thediversity of analytical techniques make the establishment of aunique and universal sample preparation protocol rather difficult,if not impossible. The best approach to prepare thin-sectionsshould be defined, case by case, in order to fit most of the technicaland analytical requirements. The two newapproaches described in

this paper are increasingly used at the ID21 beamline, ESRF. About50 samples have been prepared following one of these twoprotocols. A few examples are chosen here, to highlight pros andcons of these two methods. They illustrate two completelydifferent cases, in terms of composition and specific analyticalquestions. The first example refers to the study of in-depthoxidation phenomenon occurring on a set of plastic ABS(acrylonitrile-butadiene-styrene)-based Italian design objectsfrom the 1960s. The full study encompasses fragments from bothhistorical objects, and artificially aged polymer blocks. Theexample presented here is an artificially aged sample exposed toradiation for 1000h.While the samples are not painting fragments,they show similar hybrid inorganic/organic composition andsimilar analytical challenges to those found in paint analysis.The second example is a painting fragment from a gilded Asiansculpture (15th century) which shows a characteristic complexmultilayered composition. Its preparation and analysis present agreater degree of complexity in comparison to models.

2.2. Materials and methods

Analysis was carried out at the ID21 beamline, ESRF. Thisbeamline hosts two independent end-stations, one in the tender X-ray domain (2–9keV), and the other in the mid-infrared domain(4000–700 cm�1). The scanning X-ray microscope [22], operatesunder vacuum and is used to acquire 2D mXRF and mXANES datawith submicrometric beam (typically 0.3�0.7mm2). The FTIRspectro-microscope is based on a commercial instrument,composed of a Thermo Nicolet Nexus FTIR bench associated withan Thermo Continumm microscope [23]. The infrared beam isemitted at the edges of a bending magnet. In the microscope, two�32 Schwarzschild objectives are used in a confocal mode and anaperture defines the spot size of the beam, which is diffractionlimited and in the range of 3–10mm. The signal is detected using a50mm MCT detector. Measurements can be carried out intransmission, reflection, trans-reflectance or ATR mode as well.In this work, spectra are acquired in transmission; samples aremounted horizontally, deposited on a 10�10� 0.2mm3 BaF2window.

mXRF and mFTIR maps are analysed using the PyMca softwarepackage [24]. For FTIR analyses, the OMNIC package is used as wellfor spectral identification based on comparison with the librarydata.

For sample preparation, the resin used as an embeddingmedium is a glycol methacrylate resin: Historesin (Leica). Micro-toming was performed using a motorized rotary microtome

[(Fig._3)TD$FIG]

Fig. 3. Description of the SES protocol: the sample is maintained between 2 peek foils before microtoming.

[(Fig._4)TD$FIG]

Fig. 4. SRmFTIR study of the oxidation of an ABSmodel film (sample exposed toUV-irradiation for 1000h): (a) visible light microscope image of a 2mm thin-sectionprepared following the SES method; (b) UV light microscope images of 1000h agedABS sample on the left, and of pure resin thin-section (2mm) on the right and (c) FT-IR spectra acquired at different positions corresponding to three different depthsfrom the sample surface and one spectrum characteristic of the absorption of a resin2mm thin-section.

E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59 55

(RM2265, from Leica) including binoculars for examination undermagnification of the cutting procedure, in dried conditions. Withthis instrument, the section thickness setting range goes from 0.25to 100mm. Carbide tungsten blades were sufficient during the firsttrimming step. However, in order to obtain homogeneous slicesthinner than 3mm, the use of a diamond knife was required.Sectioning at decreasing thicknesses (here from 20mm to less than1mm) improved the quality of sections. Moreover, particularattention was paid concerning the sample size as well as excess ofembedding material; which were both reduced to the minimalamount.

3. Novel strategies

3.1. Sample enclosing system (SES) approach

The procedure described hereafter relies on the preparation ofembedding-free thin-sections, using a microtome. In order toovercome the drawbacks related to diamond compression, wedeveloped an innovative procedure named “sample enclosingsystem” (SES) based on the use of polycarbonate foils (Fig. 3). Thepainting fragment, smaller than 1mm3, is placed on a first foil ofpolycarbonate (3�6mm2), before being covered with an identicalfoil (Fig. 3a and b). A sealing tape may be used to stabilize thesandwiched sample (Fig. 3c). A commercial sample holder clampsthe system for microtoming (Fig. 3d). Thin-sections of differentthicknesses are obtained using a microtome, in regular grades atroom temperature.

The SES technique has been largely and successfully applied tothe analysis of models and reconstructed paintings, as well as insome cases of historical samples to obtain thin-sections from 10 to1mm. Sections with high surface quality and without visiblestratigraphic deformation defaults have been achieved. Avoidingthe use of embedding media/polishing/grinding processes, SESclearly overcomes the risk of chemical and analytical interferencesand offers the opportunity of keeping a part of the sample intact forfurther future experiments (e.g. natural or artificial aging,conservation treatment, staining etc.).

As an example, SES was successfully applied for the study ofphoto-ageing in polymers. The set of samples contained bothartificially photo-aged ABS blocks and fragments from historicalItalian design objects from the 1960s. The objective of this studywas to assess the depth of oxidation in terms of chemicalmodifications [25]. In-depth analysis of such phenomena hasalready been reported but based on the progressive removal of thin(50mm) layers parallel to the exposed area of the plastic blocks,using a microtome. However, considering the fragility of thedegraded surface area, the in-depth resolution with this approachin the first �100mm is critical [26]. In the present study, themicrotome was used as well, but thin-sections (2mm) wereobtained perpendicularly to the block surface. Accordingly, in asingle map, both the internal pristine polymer and the superficialproducts formed during photoageing can be identified andlocalized. The visible image (Fig. 4a) and UV fluorescence images(Fig. 4b) highlight a luminescent upper layer of about 45mmon thesurface of the block. In Fig. 4c, three points are shown as references

at three different depths: 40,194 and 268mm. The spectra from theexposed area of the sample show a clear increase of the signalrelative to CQO at 1735cm�1 and the appearance of an intenseand broad signal relative to O��H stretching at 3436 cm�1. Thesebands are attributed to ketones, aldehydes, esters,a,b-unsaturated carbonyl, carboxylic acids and alcohols. Thedepth of oxidation in samples was evaluated, for samples aged fordifferent amounts of time. mFTIR highlighted a selective andprogressive gradient of degradation, greatest at the surface of thesample and which decreases in-depth. The oxidation compoundstend to form a passivation layer, limiting the further oxidation ofthe bulk ABS. Results will be detailed in a forthcoming publication[27].

The spectrum of a standard embedding resin (Historesin, Leica)presents strong overlaps with all the FTIR bands characteristic ofthe chemical bonds formed during photo-ageing (Fig. 4). Therefore,the identification and precise localisation of these degradationcompounds would have been strongly compromised with astandard resin embedding protocol. Moreover, samples preparedin this way could be examined successfully with other microscopictechniques (UV and visible microscopy).

The SES technique is very well suited for soft materials such aspure polymeric samples, replica or historical paintings withstrongly bound and cohesive layers. For leanly bound samples,loss of material can occur during microtoming. Consequently analternativemethodwas developed, for more brittle samples, basedon an appropriate medium able to tightly maintain the sampleduring handling and sectionion.

[(Fig._5)TD$FIG]

Fig. 5. Description of the ARE protocol: the sample is first embedded in AgCl. The obtained pellet is embedded in resin and microtomed.

56 E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59

3.2. AgCl resin embedding (ARE) approach

While salt-barrier strategies focus mainly on the preparation ofcross-sections for ATR analyses, this approach has, to ourknowledge, never been mentioned for the preparation of thin-sections and reliesmainly on the use of soft materials, which couldbe sliced with a microtome. Accordingly, considering its highsoftness and lowhygroscopic properties, AgCl was identified as thebest salt to pre-embed the sample. In Fig. 5, the proposed strategy,hereafter named ARE for “AgCl resin embedding”, is detailed: first,a 3mm-diameter soft pellet-bed of AgCl is pressed; then themicrofragment (typically 1mm3) is placed on it, with the stratigraphyparallel to the pellet surface (Fig. 5a). Some AgCl powder is addedon top of it and the whole set is pressed a second time (Fig. 5b). Astandard embedding procedure is then followed using a glycolmethacrylate resin, in our case HistoResin (Leica), which isrelatively soft and therefore well adapted to microtomingoperations (Fig. 5c–e).

The method was tested and applied on an historical samplecollected from the Shuilu’an temple in Lantian (China) dating fromthe Ming dynasty (1563–1568). In this site, there are more than1372 clay sculptures ranging from several centimetres to 5m. Thefragment analysed corresponds to an architecture found in the firstniche of the western wall. The fragment shows unburnishedgilding with metal foil (probably gold) on the surface. The samefragment named SL02 (�3�3�1mm3) was prepared followingthree different protocols:

-

[(Fig._6)TD$FIG]

Figdif(3of

with the SES approach, a section of 10mmwas obtained (Fig. 6a).It was impossible to achieve a thinner section, and unfortunate-ly, this thickness was too thick for acceptable mFTIR

. 6. SRmFTIR study of a fragment fromMing dynasty painting sculpture at Shuilu’an temferent protocols: (a) sliced without embedding media (SES approach, 10mm), (b) sectiomm), (d) sliced after embedding in a resin/AgCl barrier coating (ARE approach, 3mm), (e) sthe section (d) corresponding to layers III–V.

transmission measurements. Accordingly, after a first mFTIRanalysis, this sectionwas spread over diamond cells and presseduntil achieving FTIR maximum transmittance (Fig. 6b).

-

with the classical resin embedding followed by microtoming, asection of 3mm was obtained (Fig. 6c). This thickness wasappropriate for mFTIR in transmission.

-

with the new ARE approach, it was possible to reach a thicknessof 3mm, thanks to the consolidating action of the AgCl salt andsurrounding resin (Fig. 6d).

From the visible observations (Fig. 6a–d) the sample iscomposed of five layers: a white priming layer (layer V), a red/pink layer (layer IV), amordant layer (layer III), a metallic foil (layerII) and a superficial layer, not systematically observed, probablyresulting from old restoration treatment (layer I). The set of layersII–IV is characteristic of a mordant technique. The stratigraphy iswell preserved on Fig. 6a, c and d, but partially destroyed on b. Thesample has been analysed by gas Chromatography-mass spec-trometry (GC/MS) [35] showing the presence of milk in the whitepriming layer (layer V) and of a drying oil, animal glue and anotherproteinaceous material in layers IV and III. Distribution of theorganicmaterials in those layers is missing as far as layers III and IVcould not be mechanically separated and were analysed together.

The mFTIR analysis carried out on the different samples with abeam size of 6� 6mm2 allowed the identification of some majorcomponents, from bulk to surface (Fig. 6f): thewhite priming layer(layer V) is composed of mica type muscovite (3624, 1029 cm�1),clays, identified as montmorillonite (3627, 1050 cm�1), and oforganic material in low quantities. Characteristic CQO ester andCH fatty chain bands could indicate the presence of a lipidicmaterial, possible result of the penetration of the oil identified in

ple, China. Visible light microscope images of fragment SL02 prepared followingn (a) spread over the diamond cell, (c) sliced after embedding in a resin mediacheme representing themain layers of the gilded sample and (f) SR-mFTIR spectra

E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59 57

the above layers. In layer IV a higher concentration of this oilybinder as well as silicate-based compounds have been identifiedtogether withmuscovite. This red layer prepared the application ofa large mordant layer (layer III) composed of oil (1740–1710 cm�1)and lead white (1410 cm�1), more precisely cerussite. Hydro-cerussite which present an OH characteristic band at 3450 cm�1

was not detected. In addition, the peak at 1550 cm�1 is ascribed tothe presence of carboxylates, possibly lead carboxylates [28],which could indicate that the oil was pre-saponified with a leadsiccative [29]. Similar compositions made of lead white andpartially saponified oil was reported in mural Asian paintings [30].Layer II does not show any absorption bands in the mid-IR domain.Finally, layer I was barely detected, and contains mostly clays suchas kaolinite and montmorillonite, gypsum as well as oxalates inhigh proportion. Calcium oxalate, more precisely weddellite type,was identified in all the different layers (1314 cm�1), with anenrichment in the upper layer.

To evaluate the potential of the three preparationmethods citedabove, FTIR spectra from layer III are compared in Fig. 7d. Besides,chemical maps of muscovite, esters and acids n(CO) signals areplotted in Fig. 7a for the ARE protocol, in Fig. 7b for the classicalembedding in a resin medium, and in Fig. 7c for the SES protocolsubsequently spread over diamond cells. As expected, thecomparison of the different sample preparations highlights severalissues, in particular over-absorption and contamination.

For the paint fragments studied, the appropriate thickness wasfound to be �3mm, however this value highly depends on thesample composition and density. Even after compression withdiamond cells, the SES pre-sliced fragment (initially 10mm) wasstill too thick; the low transmitted signal distorted dramaticallythe absorption bands of the resulting spectra (point 5).

With the ARE procedure, it was possible to obtain the properthickness of 3mm and unique information were acquired, inparticular regarding the fine structure of the mordant layer (layerIII). Both esters and acids were found, however, not co-localizedsince esters are more concentrated in the upper part of the layer(point 1 and 2 in Fig. 7d and a). Two hypotheses can be proposed toexplain the multi-layered distribution of esters and acids in layerIII. Either, this structure results from the successive applications oftwo different organic binders, or it derives from the migration orformation of compounds as a result of ageing [31]. The combinedoccurrence of these two phenomena can also be considered.

The observation of this micrometric complex sub-structurewasnot achievable with the other preparation protocols. The compres-sion procedure led to a clear spatial mixture of components as

[(Fig._7)TD$FIG]

Fig. 7. SR mFTIR study of a fragment fromMing dynasty painting sculpture at Shuilu’an(blue) in the samples: (a) sliced after embedding in a resin/AgCl barrier coating (3mm), (bembedding (SES, 10mm) and spread over diamond cells (d) SR mFTIR spectra from the

illustrated in Fig. 7c, where esters and acids cannot be separated. Inthe standard embedding method, the resin (point 3 in Fig. 7d) haspenetrated over a fewmicrons and clearly interferes with the C��Hand CQO bands of the sample (point 4) making the identificationof the organic materials far less easy and reliable. With ARE, AgCldoes prevent any diffusion of the embedding resin inside thesample as seen in Fig. 7a (no resin signal detected around thesection).

This example demonstrates the advantages of ARE method forFTIR analyses: in addition to the analytical neutrality, it strength-ens the sample allowing microtome sections at extreme thinnesscombined with the complete conservation of the stratigraphicstructure. The same section was analysed with mXRF to obtaincomplementary elemental distributions (Fig. 8). In the priminglayer (layer V), Mg, Al, Si, K and Ca were found, in agreement withthe identification of muscovite. The same elements were alsopresent in the layer IV, colocalized with Fe and Ti grains outliningthe use of iron and titanium oxides to colour this layer. Thesepigments were difficult or impossible to assess in the mid-IRdomain due to strong overlapwith other components or absence ofabsorption bands.

A complex stratigraphy was also revealed for the layer III, withtwo sub-layers. With the improved lateral resolution (0.5mm onthe zoom map), mXRF map shows the presence of lead-basedgrains of about 2–3mm in the bottom of the layer III (named 3b inFig. 8) mixed with Ca-based compounds. They are correlated bothwith the presence of cerussite and with the higher acid vs. esterratio observedwithmFTIR. In the upper part of the layer III (named3a in Fig. 8) no specific XRF signal was detected hypothesising thepresence of a pure organic layer. In this sub-layer, the esters aremore concentrated (cf. Fig. 7a and Fig. 8b).

Moreover, mXRF allowed the identification of IR-transparentmaterial such as a very thin gold layer (ca. 2mm thick) applied togild the sculpture. While Ag and Cl were mostly detected aroundthe sample, the presence of both elements in the entire samplereveals smearing of the barrier coating (cf. Fig 1 Supplementaryinformation).

Based on the combined results (Fig. 8) obtained frommFTIR andmXRF, hypotheses can be further formulated regarding the gildingprocess. The fact that lead is detected only in the above layer (IIIb)supports the assumption of a bi-layer application. The deeperpriming layer is made of oil mixed with lead compounds (cerusiteand carboxylates). The lead carboxylates could result from a longageing reaction of oil with a lead pigment, but an intentional andfast saponification should be considered as well. Indeed, curing oil

temple, China. Chemical maps of esters (green), acids (red) and OH from muscovite) sliced after classical embedding in a resinmedia (3mm), (c) section sliced withoutorganic mordant (Layer II) and the embedding on the 5 points indicated in (a)–(c).

[(Fig._8)TD$FIG]

Fig. 8. Combination of SRmFTIR andmXRF for the study of a fragment fromMing dynasty painting sculpture at Shuilu’an temple, China. (a) Visible lightmicroscope images offragment SL02 prepared following the ARE protocol; (b) FTIR mapping of esters (1770–1720 cm�1), acids (1720–1693 cm�1), O��H from muscovite (3654–3600 cm�1) andPbCO3 (1427–1400 cm�1) in the sample, step size of 6� 6mm2; (c) XRFmaps at 7.2 keV of Ag, Cl, P, Pb, Ca, Fe, K and Au acquired for the entire samplewith 1�1mm2 pixel sizeand 0.5�0.5 mm2 in the zoomed region.

58 E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59

with a siccative, e.g. lead oxide, as described in recipes of leadplasters, turns the liquid oil into a soft white paste with uniquemechanical properties, particularly optimal for priming [29].Conversely, the upper sub-layer (IIIa) contains only organicmaterials, with variable proportions of esters vs. acids, in particulara higher concentration of esters just below the gold foil. This couldbe related to the application of a raw oil size, before the deposit ofthe final gold layer. These results show some similarities withthose reported byKatsibiri and Boon, on gilding paintings from twopost-Byzantine wall paintings [34].

In this example, the ARE protocol was fundamental since itallowed keeping the complex stratigraphy without distortion ofthe structure and merging of layers, reaching the appropriatethickness of 3mm needed for mFTIR analysis in transmission andavoiding the penetration of the embedding resin inside the sample.Localization of compounds was fundamental to highlight differ-ences in composition within the layers, revealing a more complexsample build-up that allowed establishing the gilding techniqueand ageing mechanisms.

4. Conclusion

This work presents new methods to overcome difficultiesrelated to the preparation of sections, and in particular thin-sections for the analysis of painting fragments with mFTIRspectroscopy. Based on the review of solutions proposed toprepare cross-sections, two new dedicated protocols were devel-oped and evaluated, one based on embedding-free technique (SES)and the second on the use of an IR-transparent salts barrier coating(ARE). SES technique gave promising results on photo-aged ABSsamples, as well as on fresh and soft model paintings containingorganic binders in high concentration. Even if this technique is notreliable for leanly bound samples, it is well suited for the simpleand fast preparation of thin-sections of rich polymeric materialand is the best protocol in terms of chemical and analyticalneutrality.

AgCl saltwas applied for thefirst timeas a barrier coating for thinsections, and acted satisfactorily against the penetration of the resinfor mFTIR measurements. Thanks to its mechanical characteristics,

AgCl presents unique properties for microtoming procedures. Incomparison with other classical protocols, mFTIR data quality andreliability were improved and, compared to the SES method, AREoffered a better handling of the paint fragment during and afterslicing.While the present discussion is focused onmeasurements intransmission mode, the ARE protocol is fully adaptable to cross-sectionspreparation. PerformingmFTIR inATRmodeonthin-sectionshould be considered in the future. This could avoid the difficultly inreaching samples to extreme thinness (a few microns) and allowsworking with enhanced spatial resolution on thicker section (a fewtens microns) optimized for X-ray analyses. In the future, otherbarrier coatings should be tested based on the promising resultsobtained with the ARE method.

Acknowledgments

The European Synchrotron Radiation Facility is thanked forproviding beamtime (EC-977 and in-house research), assistanceand economical support. The Shuilu’an sample was kindlyprovided by Mr. Ma Tao and Mrs. Yang Qiuying (Research Institutefor the Preservation of Cultural Heritage of Shaanxi Province, Xi’an)and Catharina Blaensdorf (Technische Universität Munich, Chair ofRestoration, Art Technology and Conservation Science).

Appendix A. Supplementary data

Supplementary data associatedwith this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.aca.2014.03.025.

References

[1] A. Laurie, W. McLintock, F. Miles, Egyptian Blue, Series A, Containing Papers ofa Mathematical and Physical Character (1905–1934), Proceedings of the RoyalSociety of London, 891914, pp. 418–429.

[2] E. Raehlmann, Uber die Maltechnik der Alten, George Reiner, Berlin, 1910.[3] R.J. Gettens, The cross-sectioning of paint films, Technical Studies in the Field

of the Fine Arts 5 (1936) 18–22.[4] J. Plesters, Cross-sections and chemical analysis of paint samples, Studies in

Conservation 2 (1956) 110–157.[5] M. Derrick, L. Souza, T. Kielich, H. Florsheim, D. Stulik, Embedding paint cross-

section samples in polyester resins: problems and solutions, Journal of theAmerican Institute for Conservation 33 (1994) 227–245.

E. Pouyet et al. / Analytica Chimica Acta 822 (2014) 51–59 59

[6] M. Spring, R. Morrison, Evaluation report on protocols for representativesampling (2011) pp. 1–39.

[7] M. Cotte, E. Checroun, V. Mazel, V.A. Solé, P. Richardin, Y. Taniguchi, P. Walter, J.Susini, Combination of FTIR and X-rays synchrotron-based micro-imagingtechniques for the study of ancient paintings, A Practical Point of View e-PRESERVATIONScience 6 (2009) 1–9.

[8] F. Meirer, Y. Liu, E. Pouyet, B. Fayard, M. Cotte, C. Sanchez, J.C. Andrews, A.Mehta, P. Sciau, Full-field, XANES analysis of Roman ceramics to estimatefiringconditions – A novel probe to study hierarchical heterogeneous materials,Journal of Analytical Atomic Spectrometry 28 (2013) 1870–1883.

[9] E. Pouyet, B. Fayard, C. Gervais, Y. Taniguchi, M. Salome, F. Sette, M. Cotte, (inpreparation).

[10] M. Derrick, D. Stulik, J.M. Landry, Infrared Spectroscopy in ConservationScience, The Guetty conservation institute, Los Angeles, 1999.

[11] E.F. Hansen, R. Lowinger, E. Sadoff, Consolidation of porous paint in a vapor-saturated atmosphere, Journal of the American Institute for Conservation 32(1993) 1–14.

[12] C. Martin de Fonjaudran, A. Nevin, F. Piqué, S. Cather, Stratigraphic analysis oforganic materials in wall painting samples using micro-FTIR attenuated totalreflectance and a novel sample preparation technique, Analytical andBioanalytical Chemistry 392 (2008) 77–86.

[13] A. Lluveras, S. Boularand, J. Roqué, M. Cotte, P. Giráldez, M. Vendrell-Saz,Weathering of gilding decorations investigated by SR: development anddistribution of calcium oxalates in the case of Sant Benet de Bages (Barcelona,Spain), Applied Physics A 90 (2008) 23–33.

[14] J. Pilc, R. White, The application of FTIR-microscopy to the analysis of paintbinders in easel paintings, National Gallery Technical Bulletin 16 (1995) 73–84.

[15] S. Prati, F. Rosi, G. Sciutto, R. Mazzeo, D. Magrini, S. Sotiropoulou, M. Van Bos,Evaluation of the effect of six different paint cross section preparationmethods on the performances of fourier transformed infrared microscopy inattenuated total reflection mode, Microchemical Journal 103 (2012) 79–89.

[16] L.A.C. Souza,M. Derrick, The use of Ft-Ir spectrometry for the identification andcharacterization of gesso-glue grounds in wooden polychromed sculpturesand panel paintings, MRS Proceedings (1995) 573.

[17] M.E. Laver, R.S. Williams, the use of a diamond cell microsampling device forinfrared spectrophotometric analyses of art and archaeological materials,Journal of the International Institute for Conservation-Canadian Group 3(1978) 34–39.

[18] J. van der Weerd, Microspectroscopic analysis of traditional oil paint, AMOLFThesis, Universityy of Amsterdam, Amsterdam, 2002, pp. 93–95.

[19] S. Prati, E. Joseph, G. Sciutto, R. Mazzeo, New advances in the application ofFTIR microscopy and spectroscopy for the characterization of artisticmaterials, Accounts of Chemical Research 43 (2010) 792–801.

[20] R. Mazzeo, E. Joseph, S. Prati, A. Millemaggi, Attenuated total reflection-fouriertransform infrared microspectroscopic mapping for the characterisation ofpaint cross-sections, Analytica Chimica Acta 599 (2007) 107–117.

[21] S. Prati, G. Sciutto, E. Catelli, A. Ashashina, R. Mazzeo, Development ofinnovative embedding procedures for the analyses of paint cross sections inATR FITR microscopy, Analytical and Bioanalytical Chemistry 405 (2013)895–905.

[22] M. Salomé, M. Cotte, R. Baker, R. Barrett, N. Benseny-Cases, G. Berruyer, D.Bugnazet, H. Castillo-Michel, C. Cornu, B. Fayard, E. Gagliardini, R. Hino, J.Morse, E. Papillon, E. Pouyet, C. Rivard, V.A. Solé, J. Susini, G. Veronesi, The ID21scanning X-ray microscope at ESRF, Journal of Physics: Conference Series 425(2013) 182004.

[23] J. Susini, K. Scheidt, M. Cotte, P. Dumas, F. Polack, O. Chubar, Why infraredspectromicroscopy at the ESRF? ESRF Newsletter 41 (2005) 24–25.

[24] V.A. Sole, E. Papillon, M. Cotte, P. Walter, J. Susini, A multiplatform code for theanalysis of energy-dispersive X-ray fluorescence spectra, Spectrochimica Acta– Part B Atomic Spectroscopy 62 (2007) 63–68.

[25] F. Toja, D. Saviello, A. Nevin, D. Comelli, M. Lazzari, M. Levi, L. Toniolo, Thedegradation of poly(vinyl acetate) as a material for design objects: a multi-analytical study of the effect of dibutyl phthalate plasticizer, Part 1, PolymerDegradation and Stability 97 (2012) 2441–2448.

[26] F. Toja, A. Nevin, D. Comelli, M. Levi, R. Cubeddu, L. Toniolo, Fluorescence andfourier-transform infrared spectroscopy for the analysis of iconic Italian designlamps made of polymeric materials, Analytical and Bioanalytical Chemistry399 (2011) 2977–2986.

[27] D. Saviello, E. Pouyet, M. Cotte, L. Toniolo, A. Nevin, Synchrotron FTIRmicrospectroscopy for themapping of photo-oxidation and additives in ABS inmodel samples and historical objects (2014) in preparation.

[28] Z. Zeng, S. Wang, S. Yang, Synthesis and characterization of PbS nano-crystallites in random copolymer ionomers, Chemistry of Materials 11 (1999)3365–3369.

[29] M. Cotte, E. Checroun, J. Susini, P. Dumas, P. Tchoreloff, M. Besnard, P. Walter,Kinetics of oil saponification by lead salts in ancient preparations ofpharmaceutical lead plasters and painting lead mediums, Talanta 70 (2006)1136–1142.

[30] M. Cotte, J. Susini, V.A. Solé, Y. Taniguchi, J. Chillida, E. Checroum, P. Walter,Applications of synchrotron-based micro-imaging techniques to the chemicalanalysis of ancient paintings, Journal of Analytical Atomic Spectrometry 23(2008) 820–828.

[31] M.J. Plater, B. De Silva, T. Gelbrich, M.B. Hursthouse, C.L. Higgitt, D.R. Saunders,The characterisation of lead fatty acid soaps in protrusions’ in aged traditionaloil paint, Polyhedron 22 (2003) 3171–3179.