Talanta 80 (2009) 7183

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

Advanced combined application of -X-ray diffraction/-X-ray uorescence

with conventional techniques for the identication of pictorial materials from
Baroque Andalusia paintings
L.K. Herrera a, , S. Montalbani b , G. Chiavari b , M. Cotte c,d , V.A. Sol c , J. Bueno e,f , A. Duran a ,
A. Justo a , J.L. Perez-Rodriguez a
a
Materials Science Institute of Seville, CSIC-Seville University, Avda Americo Vespucio 49, 41092 Seville, Spain
b
Univ Bologna, CIRSA, Chem Lab, Via S Alberto 163, I-48100 Ravenna, Italy
c
European Synchrotron Radiation Facility, BP 220, 38043 Grenoble Cedex, France
d
Centre de Recherche et de Restauration des Muses de France (C2RMF), CNRS UMR 171, Palais du Louvre, Porte des Lions, 14, quai Francois Mitterrand, 75001 Paris, France
e
Faculty of Fine Arts, University of Seville, Spain
f
Pedrela Painting Restorers, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The process of investigating paintings includes the identication of materials to solve technical and his-
Received 15 October 2008 torical art questions, to aid in the deduction of the original appearance, and in the establishment of the
Received in revised form 5 June 2009 chemical and physical conditions for adequate restoration and conservation. In particular, we have focused
Accepted 9 June 2009
on the identication of several samples taken from six famous canvases painted by Pedro Atanasio Bocane-
Available online 21 June 2009
gra, who created a very special collection depicting the life of San Ignacio, which is located in the church
of San Justo y Pastor of Granada, Spain. The characterization of the inorganic and organic compounds of the
Keywords:
textiles, preparation layers, and pictorial layers have been carried out using an XRD diffractometer, SEM
Works of art
Pigments
observations, EDX spectrometry, FT-IR spectrometry (both in reection and transmission mode), pyroly-
Cultural heritage sis/gas chromatography/mass spectrometry and synchrotron-based -X-ray techniques. In this work, the
XRD advantages over conventional X-ray diffraction of using combined synchrotron-based -X-ray diffraction
SEM/EDX and -X-ray uorescence in the identication of multi-layer paintings is demonstrated.
FT-IR 2009 Elsevier B.V. All rights reserved.
PYGC/MS
Synchrotron radiation
-X-ray diffraction/-X-ray uorescence

1. Introduction
Analytical methods are an essential requirement in the cultural
heritage eld, as they provide the means to understand the objects
under investigation. By identifying materials (such as pigments,
binders, mixtures, and processes), we can reach back through
time and develop a deeper understanding of the craftsmanship
and technology used. Analytical methods allow us to contribute
to the blooming activity at the interface between art and sci-
ence. Contributing to the development of diagnostic techniques is
necessary for practical applied art conservation. The techniques
concerned include: scanning electron microscopy (SEM), energy
dispersive X-ray spectrometry (EDX), -X-ray analyses (includ-
ing synchrotron radiation -X-ray diffraction (XRD) and -X-ray
uorescence (XRF)), Fourier transform-infrared spectrometry (FT-
Corresponding author.
E-mail address: lkaren@icmse.csic.es (L.K. Herrera).
IR), and analytical pyrolysis in-line with gas chromatography and
mass spectrometry (PYGC/MS) [13]. The analytical methods used
in this eld of research are identical to those used at the cut-
ting edge of modern materials science. Methods developed for
advanced physics and chemistry can be applied to ancient and
modern materials, as the problems encountered in the areas of
advanced technology and cultural heritage are surprisingly similar.
Depending on the information required, one might use a combina-
tion of non-invasive techniques; however, one should aim at the
maximization of information and the minimization of the sam-
ple consumed volume [4,5]. The use of non-invasive techniques is
increasing due to the improved development of instruments (e.g.
able to determine the elemental and mineral composition of the
micrometric multilayers of micro-samples) [6].
Analytical studies of paintings performed at the microscopic
and at molecular level are the focus of this paper. Microscopic
and molecular studies are carried out on paint samples derived
from the painting itself. These samples are unique and in limited
supply. Assuming that such samples are representative of an area

0039-9140/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2009.06.032
72 L.K. Herrera et al. / Talanta 80 (2009) 7183
under study, valuable general information about the painting can be
deduced. The information can be maximized by sequential analyses
using different complimentary techniques.
Synchrotron radiation has become an increasingly important
tool for research in the elds of art, archaeometry, and the con-
servation of objects of cultural heritage signicance. Synchrotron is
a source of intense, bright light of variable wavelength; the princi-
pal properties of synchrotron radiation are as follows: high ux and
brilliance, ability to produce a fast data collection and the possibil-
ity to use small sample size, small beam footprint: it enables area
mapping at millimeter to micron length-scale. The wavelength tun-
ability and the energy region can be selected to suit the problem at
hand [710]. Synchrotron techniques are particularly suited to non-
destructive or to the micro-characterization of different materials
including organic and inorganic materials, either amorphous or
crystalline. New developments in synchrotron radiation techniques
result in a signicant advance to study the samples at micrometer
length scale through the combination of different non-destructive
techniques. The most employed till now have been synchrotron
X-ray uorescence and diffraction and, to a lesser extent, small
angle scattering (on wood, bone and hair) and infrared microscopy.
The synchrotron beam can be focused in sub micrometric spots,
allowing the examination of very small samples. The list of artwork
studied by using SR is made up of a large variety of materials [7,11].
This article is intended to illustrate the chemical description of
the organic and inorganic components of the stratigraphy sequence
using, in parallel, the recent applications of microanalytical meth-
ods using synchrotron radiation and the information supplied by
laboratory techniques. This is exemplied by the identication of
pigments and binders in six famous canvases painted by Bocanegra
(XVIth century).
2. Experimental
2.1. Historical context
The rst reports of Pedro Atanasio Bocanegra as a painter are
from 1661 [12]. He included in his paintings many elements of his
master, Alonso Cano.
According to Gomez Moreno [12], the Bocanegra paintings of the
San Justo y Pastor of the church that represents the passages of the
S. Ignacio of Loyola life (14911540) were carried out in two steps. In
the rst, Bocanegra painted three canvases: San Ignacio herido a las
puertas de Pamplona, La Aparicin de San Pedro and San Ignacio
azotado por un armenio. In the second step, he created the other
three paintings between 1670 and 1671, where these are consid-
ered the most important works of oriental Andalusia painting [13].
These canvas paintings are La Visin de la Storta, La Aparicin de
la Trinidad a San Ignacio and S. Ignacio enviando a predicar a S. Fco.
Javier and S. Fco. de Borja. In these paintings, the author shows tech-
nical advances and a greater intensity of color that is reminiscent
of the art of Valdes Leal [14]. The painting details are described in
Fig. 1. The six paintings are a complex multi-layered composition
of organic and inorganic compounds.
2.2. Sampling
The samples were collected by taking small samples close to
areas that were already damaged. Twenty samples were careful
extracted from different zones of the paintings using a scalpel. The
obtained samples were a representative collection of white, red,
and blue colors.
Sample preparation is often a tricky point for the study of micro-
metric fragments of paintings [15]. It may be difcult to carry
out all the experiments of a unique sample; conversely, when
the quantity of painting fragment is enough, different sample
preparations may be followed to t the technical requirements of
each analytical technique. Three types of sample preparation were
performed.
Conventional cross-section. The samples were prepared as cylin-
der moulds and were embedded in epoxy resin as follows: The paint
piece was placed on top of the hardened block and carefully covered
with resin, and then cast and left to cure for 24 h at room tem-
perature. Next, the sample was cut to expose a face showing the
cross-section of the paint layer. Finally, the surface was carefully
polished on a rotating disc covered with abrasive SiC paper which
spanned a grain size from 800 to 1200 mesh, and then nished with
a cloth [16,17]. This method is useful for observation with an optical
microscope, SEM/EDX, or FT-IR spectrometer that requires a more
careful preparation in terms of surface polish.
Thin transversal cuts. Samples were embedded in a resin (Leica
Historesin) to prepare thin sections with a microtome. Several
thicknesses were tested, and 2030 m was chosen as the most
appropriate thickness [18] for SR XRF/XRD analyses.
Raw fragments of the samples. Small fragments of the samples
were ground to a powder for examination by conventional XRD and
for PYGC/MS [18,19].
Single bres (a few micrometers in diameter) were carefully
extracted from the canvas for textile identication by XRD.
2.3. Analytical techniques: instruments and methods
Observations of the cross-sections were performed by opti-
cal microscopy using a stereo microscope (Nikon HOPITHOT) with
objectives of 20, 50, 100, 200 and equipped for microphotog-
raphy using a digital camera (Nikon COOLPIX 4500), which was also
used to identify the stratigraphy sequences of the different paint
cross-sections. The inorganic pigments have been characterized as
cross-sections on a Jeol model JSM 5400 SEM. Analyses of elemental
compositions of the different painting layers were obtained using
EDX on a Link ISIS X-ray energy dispersive spectrometer equipped
with an ultra-thin beryllium window, using an accelerating volt-
age of 20 keV. The cross-sections were coated with gold lm before
the SEM/EDX investigations. X-rays analyzed with an energy disper-
sive detector (EDX) result in a characterization of different chemical
elements in the paint cross-section.
The crystalline phases were characterized by X-ray diffraction
(XRD), using a Siemens diffractometer, model Kristalloex D-5000
using Cu K1 radiation (40 kV, 30 mA), a diffracted-beam graphite
monochromator and a scintillation counter. Measurements were
carried out on powders of the sample in the range of 3 < 2 < 65
with steps of 0.05 and a 2 s counting time. Subsequently, the EVA
data bank of standard X-ray powder diffraction patterns was used
for phase identication. For textile identication by XRD, the bres
were introduced into a 0.3 mm glass capillary and these were then
used to collect X-ray diffraction patterns in transmission mode on a
Siemens D5000 dual diffractometer, equipped with a Ge monochro-
mator in the incident beam and Cu K1 radiation.
Fourier transformed-infrared spectrometry (FT-IR) is relevant for
the identication of organic compounds, but the surface of the
cross-section must be totally planar in order to obtain good reec-
tions [20]. Unfortunately, when organic compounds are mixed
with large amounts of inorganic matter, such as in an argilla-
ceous earth layer or lead white, the peaks representative of the
inorganic composition predominate in the FT-IR spectrum. The
organic constituents are often not visualized with FT-IR, as their
reectance and relative concentrations are low [18], and the signal
is reduced in a heterogeneous layer due to scattering of the infrared
light within the cross-section. The KramerKrnig transformation
is applied to the specular reectance spectrum to transform it
into an absorbance-like spectrum, which is more workable and
easier to interpret. The FT-IR spectra were registered on a Nicolet
 L.K. Herrera et al. / Talanta 80 (2009) 7183 73

Fig. 1. Bocanegra paintings. (a) San Ignacio herido en las puertas de Pamplona, (b) Aparicin de San Pedro a San Ignacio para curarle, (c) San Ignacio azotado por un
armenio a su vuelta del Monte de los Olivos, consolado por Cristo, (d) San Ignacio enva a predicar a San Francisco Javier a las Indias y a San Francisco de Borja a Occidente,
(e) Visin de La Storta, camino de Roma, en que se aparece la Trinidad a San Ignacio, and (f) Aparicin de la Trinidad a San Ignacio.

510 spectrometer equipped with a nitrogen-cooled MCT detector
and an Olympus optical microscope. The spectra were collected in
reectance mode, in the 4000700 cm1 range, with a resolution
of 4 cm1 . The nal format of the data was in absorbance, where
the background was collected through a gold lm substrate. For
each spectrum, at least 100 scans were accumulated. Peak posi-
tions were determined using the Nicolet Omnic software (based on
a polynomial least squares method). The advantage of infrared spec-
troscopy is to simultaneously obtain information on both organic
and mineral compounds of the painting layers.
74 L.K. Herrera et al. / Talanta 80 (2009) 7183

Fig. 2. Images of the conventional cross-section of paint samples (100) (a) originating from Trinidads dress in the Visin de La Storta, (b) originating from San Ignacios
dress in the Aparicin de San Pedro a San Ignacio para curarle, (c) originating from San Ignacios dress in the San Ignacio azotado por un Armenio, and (d) originating from
the dress in the San Ignacio Herido en las Puertas de Pamplona.

2.3.1. Synchrotron-based -X-ray diffraction and -X-ray multi-technique X-ray microscope. The double-crystal monochro-
uorescence techniques mator, equipped with both Si[1 1 1] and Si[3 1 1] crystals, offers
-X-ray diffraction and -X-ray uorescence were performed a wide energy range (670 keV) for optimal excitation conditions
at the ID18F beamline, at the ESRF. This beamline is a high-energy of a broad range of elements [21]. In this experiment, the excita-

Fig. 3. Optical micrographs (a) without any clearing treatment, (b) the organic compounds were removed by water with continuous heating for 24 h, (c) SEM image of the
bres of Bocanegra paintings, and (d) X-ray diffractogram of the bres. Linen pattern and Bocanegra bre pattern.
 L.K. Herrera et al. / Talanta 80 (2009) 7183
Fig. 4. FT-IR spectra corresponding to the preparation layer of the Bocanegra painting sample.
tion energy was xed at 28 keV, corresponding to a wavelength
of 0.44281 A. The acquisition time was 20 s per pattern and the
beam was focussed down to 15 m (horizontal) 1 m (vertical).
A two-dimensional CCD based X-ray detector (rened detector dis-
tance 134.1967 mm) was used to collect X-ray diffraction patterns
in transmission mode. The X-ray uorescence spectra were simul-
taneously detected with a Si(Li) detector. For these measurements,
thin transversal cuts were mounted vertically, between two foils
of ultralene, and the whole set was raster-scanned horizontally
and vertically to obtain two-dimensional images. Thin transver-
sal cuts were oriented horizontally such that the best resolution
of the beam (1 m, vertical) was aligned with the axis of maximum
sample heterogeneity, whereas the average in the horizontal direc-
tion (15 m) was obtained over the rather homogeneous colored
layer. Three vertical line scans were acquired in each painting cross-
section. X-ray uorescence spectra were treated with PyMCA code
[22] and used to identify the elemental composition of the differ-
ent layers of the samples. In this program, a LevenbergMarquardt
algorithm is implemented to t the spectra with constraints on
the tting parameters. A complete emission line series is tted
by taking into account theoretical line emission energies and the-
oretical energy ratios corrected by the experimental conditions
(detector characteristics, detection geometry, matrix composition,
excitation energy, etc.). A more detailed description of this code
is given in [22]. The tting conguration used to t the average
spectrum of a map is applied to each pixel of the 2D map to cal-
culate the different elemental maps through a batch treatment.
For each 2D X-ray diffraction pattern, the DebyeScherrer diffrac-
tion rings were unwrapped and integrated versus the azimuthal
Table 1
Analyses of the preparation layer of the six samples of the six paintings.
Painting Sample EDX
1 3 C, O, Mg, Al, S, Si, K, Ca, Fe
2 6 C, O, Mg, Al, Si, S, K,Ca, Fe
3 8 C, O, Mg, Al, Si, S, K, Ca, Fe
4 1 C, O, Mg, Al, Si, S, K, Ca,Fe
5 11 C, O, Mg, Al, Si, S, K, Ca, Fe
6 4 C, O, Mg, Al, Si, S, K,Ca, Fe
75
angle to produce a 1D diffraction pattern using the ESRF package
Fit2D [23]. The set of data thus generated has a format simi-
lar to uorescence data, with a 1D diffraction pattern for each
pixel.
Analytical pyrolysis was performed in the Chemical Department
of Bologna University. Analytical pyrolysis, on-line with GC and
mass spectrometry (PYGC/MS), was used to generate a qualitative
characterization of the organic components in a complex mixture
such as the painting layers [24]. PYGC/MS was carried out at
600 C for 10 s at the maximum heating rate using a CDS pyroprobe-
heated lament pyrolyzer (CDS, Oxford, USA) directly connected
to the injection port of a Varian 3400 gas chromatograph cou-
pled to a Saturn II ion-trap mass spectrometer (Varian Analytical
Instruments, Walnut Creek, USA). A Supelco SPB-5 capillary col-
umn (30 m, 0.32 mm i.d., 0.25 m lm thickness) was used with a
temperature programmed from 50 (held for 10 min) to 300 C at
5 C/min step intervals, with helium as the carrier gas. The tempera-
ture of the split/splitless injector and the PYGC/MS interface were
kept at 250 C. Mass spectra were recorded at 1 scan/s under elec-
tron impact at 70 eV in the mass range of 45650 m/z. Structural
assignment of the pyrolytical fragments was based on matching
with the NIST mass spectra library. To improve the gas chromato-
graphic behaviour of polar pyrolytic fragments, such as carboxylic
or dicarboxylic acids, 10 l of tetramethylammonium hydroxide
solution (TMAH 25% (w/w) in water) was added to the solid sample
(few micrograms) before the pyrolysis. The study of the pyrogram,
a gas chromatogram of the fragments of the volatile compounds of
the sample, gives an indication about the organic chemical nature
of the paint layer samples.
Phase identication (XRD) Phase identication (SR XRD)
Gypsum Calcite
Anhydrite Quartz
Goethite
Gypsum
76 L.K. Herrera et al. / Talanta 80 (2009) 7183

Fig. 5. (a) The XRD pattern of the powder total sample whose peaks can be attributed the presence of gypsum (Gyp), calcite (Cal) and anhydrite (Anh). (b) XRF elemental
mapping of the rst layer: calcium and iron; elemental mapping of second layer: lead. (c) SR XRD of the rst layer: goethite, calcite and quartz. SR XRD of the second layer:
cerussite and hydrocerussite.

3. Results and discussion from textile fragments are shown in Fig. 3a and b. The bre has
a diameter of 16.816.9 m, which, together with its morphology,
Fig. 2 depicts the light-optical micrograph of the different sam- is a very typical for linen (Fig. 3c) [25]. The XRD diffractograms
ples taken from (a) San Ignacio azotado por un armenio, (b) La of the bre obtained from the painting and linen bre used as a
Visin de la Storta, (c) La Aparicin de San Pedro and (d) San Igna- standard are included in Fig. 3d. The XRD pattern of the bre sam-
cio Herido en las Puertas de Pamplona and the information gained ple recorded for 2 in the range of 350 revealed pronounced
concerns mainly the sequence and the thickness of the paint layers Bragg peaks at 2 values of around 15.216.6 and 22.8 , indi-
and the distribution of the pigments. cating the presence of a cellulose crystalline phase. The linen
used in these canvases contains a greater proportion of ligni-
3.1. Textile identication ed material that has survived the retting process. This factor
imparts strength and color, but makes the linen more prone to
The morphological studies of bres were realized using opti- deterioration. Factors that contribute to the deterioration of can-
cal and scanning electron microscopy. SEM images of single bres vas can be summarized as follows: light, external pollution, and,
to a lesser extent, the volatile acids trapped within a frame enclo-
Table 2 sure.
Analyses of the third white layer of the six samples of the six paintings. These results show that Bocanegra used a traditional canvas sup-
port that normally consisted of stretched linen fabric, onto which
Painting Sample EDX Pigment identication by SR XRD
was brushed an aqueous glue, which will be discussed later. Stud-
1 3 C, O, Pb, S, Ca Lead white (cerussite, hydrocerussite), ies of paper and linen show that the oxidation products of cellulose
2 6 C, O, Pb, Ca, Fe, Si calcium carbonate
are carboxylic acids, which can catalyse and promote the hydroly-
3 8 C, O, Pb, Ca
4 1 C, O, Pb, Ca, Si, K, Fe sis of cellulose [26]. However, the bres studied in this work have
5 11 C, O, Pb, Ca not suffered from high alteration as indicated by the presence of
6 4 C, O, Pb, Ca crystalline phases of cellulose in the XRD patterns.
 L.K. Herrera et al. / Talanta 80 (2009) 7183 77

Fig. 6. (a) FT-IR spectra corresponding to: top pictorial layer, that shows the presence of bands of silicates (1028 cm1 ), carbonates (1439 and 1466 cm1 ), and drying oils
(1733 and 1157 cm1 ). (b) Reconstructed ion chromatogram obtained from pyrolysis/methylation of Bocanegra painting sample. Numbers on top of peak refer to table (mass
spectra characteristics of compounds evolved from pyrolysis methylation and structural attribution to peak numbers).

3.2. Identication of preparation layers
The preparation layer of all 20 samples is composed of gypsum,
argillaceous earth, iron oxides and animal glue mixed in different
proportions. Gypsum was conrmed by FT-IR spectroscopy (Fig. 4)
showing, respectively, the stretching and bending absorptions of
SO4 2 groups at 11461116 cm1 , the water of hydration ( OH at
3403 and 3547 cm1 , and the OH at 1684 and 1622 cm1 ) and ani-
mal glue was suspected by the protein bands (3239 cm1 -stretching
 NH, 1652 cm1 -stretching  C O amide I and 1547 cm1 bending
NH amide II). Moreover, not all elements can be detected with
EDX, because the detection threshold is relatively high and the X-
78 L.K. Herrera et al. / Talanta 80 (2009) 7183

Fig. 7. (a) EDX spectra of the different three layers of the cross-section originating from red San Ignacios dress and (b) the X-ray diffraction pattern of powder total sample
obtained with under incidence angle 1 (), whose peaks can be attributed the presence of gypsum (Gyp), cerussite (Cer), hydrocerussite (Hyd), calcite (Cal) and cinnabar
(Cin).

ray energies of several elements overlap. In general, light elements
in low concentrations are hard to detect with EDX. EDX is not a
suitable technique for the detection of trace elements in pigments
because their concentration is often below the detection limit (the
detection limit is approximately 0.1 wt% for pure materials) [27]. In
Table 1, EDX analyses conrm the presence of C, O, Mg, Al, S, Si, K,
Ca, and Fe. However, this technique is not sufcient to characterize
the different phases.
The separation of different layers is very complicated. For this
reason, the samples (including all the layers) were ground for XRD
characterization. The XRD study is carried out in powder form
using the conventional BraggBrentano method with a focusing
geometry (2 coupled) that requires a very small quantity of sam-
ple. The XRD pattern of the preparation layer (Fig. 5a) conrmed
the presence of calcium sulphate in two phases (mainly gypsum
(CaSO4 2H2 O) and anhydrite (CaSO4 )).
The XRD and FT-IR techniques conrm that the preparation lay-
ers were carried out with some pieces of textile mixture with animal
glue associated to the support followed by the preparation of a lean
character (animal glue and gypsum) layer. In some of the prepara-
tion layers, Bocanegra employs calcite and animal glue. In addition,
a supercial colored layer in which iron compounds contribute to
the yellow to ochre-reddish tone variation is present. However, in
these samples, the iron compounds were only detected by EDX anal-
yses and not by conventional XRD because this instrument may not
be sensitive enough to detect minor phases of the cross-section.
The iron phases are present in a low proportion in the prepara-
tion layer. In addition, the powder reections are very small in
comparison with the other minerals present in the mixture. These
facts are responsible for the difcult characterization of the iron
compounds by conventional X-ray diffraction. Experiments on thin
cross-sections with combined SR XRF/XRD were performed. Flu-
orescence and diffraction data sets were simultaneously handled
with PyMCA. This set-up shows high potential by concomitantly
analysing element distribution (XRF) and structure information
(XRD) in the cross-sections.
The elemental mapping distribution of iron in the preparation
layer is shown in Fig. 5b. Over the same region, the diffraction
pattern is calculated (Fig. 5c) and shows the goethite phase mix-
ture with calcite and quartz in the preparation layer. These results
 L.K. Herrera et al. / Talanta 80 (2009) 7183 79

Fig. 8. (a) Reconstructed ion chromatogram, (b) single ion monitoring for m/z (198 + 200 + 202) obtained from direct pyrolysis of the Bocanegra painting sample, and (c) mass
spectrum of Hg atoms.

conrm that Bocanegra used goethite in the mixture of support (a) and single ion monitoring for m/z (198 + 200 + 202) are reported
preparation like other famous Baroque painters such as Zurbaran (Fig. 8b); the mass spectrum in Fig. 8c presents the isotopic pattern
and Murillo [28], and indicates that Bocanegra used a dark back- of mercury that probably results from cinnabar. As reported in a
ground due to the inuence of the Tenebrism style. previous paper [19] HgS, at high temperature and in helium atmo-

3.3. Identication of pictorial layers

Above the preparation layers, different colored layers were

detected in the paintings such as red, blue, ochre, and white.

3.3.1. White pigment

The EXD analyses (Table 2) show that the white layers are
constituted of C, Pb and O. The XRF map shows the elemen-
tal distribution of lead (Fig. 5b). In the white layer, mainly lead
and calcium can be distinguished, with lead being the major con-
stituent. In the lead region, XRD conrms the presence of lead
white (Fig. 5c) and shows the mineralogical phases hydrocerus-
site, Pb3 (CO3 )2 (OH)2 and cerussite, Pb3 CO3 . The presence of the
organic binding media was revealed through FT-IR spectrome-
try (Fig. 6a) carried out on the cross-section. The bands at 1738,
1162 and 1114 cm1 could be attributed, respectively, to the C O
stretching and COC stretching of the drying oils. PYGC/MS con-
rms the presence of siccative oils, which gave a chromatogram
(Fig. 6b) with high concentrations of palmitic, oleic, stearic and
dicarboxylic acids (azelaic and suberic acids). These data show that
the different pictorial layers were painted using drying oil as a lig-
and.

3.3.2. Red pigment

SEM/EDX and XRD give an approximation of the composition of
the red pictorial layers in the different paint cross-sections. Accord-
ing to optical microscopy (Fig. 7a) observations, the red pigments
in the pictorial layers are distributed as random aggregates of very
thin particles. The EDX elemental analyses of these thin particles
(Fig. 7b) identify Hg and S. The XRD diffraction pattern of the pow-
Fig. 9. (a) XRF elemental mapping of red particles in the layer: iron elemental
der sample conrmed the presence of cinnabar (Fig. 7c). Pyrolysis
mapping of the preparation layer: iron. (b) SR XRD of red particles: phases of mainly
shows the use of a siccative oil as ligand and, at the same time, hematite with a low contribution of goethite. SR XRD of red preparation layer:
the presence of Hg; in Fig. 8 the reconstructed ion chromatogram phases of goethite, calcite and quartz.
80 L.K. Herrera et al. / Talanta 80 (2009) 7183

Fig. 10. (a) Paint cross-section originating from ground. The cross-section shows the blue particles in the upper layer; the color of the pigment has faded, causing yellowing
of the oil medium. (b) SEM/EDX analyses show the presence of Co, As, and K in the blue particles. (c) XRF elemental mapping of the blue particles: arsenic and cobalt. (d)
SR XRD of the amorphous phases.

sphere, is degraded into atomic mercury which is volatile and can
be detected by GCMS. The most important source of mercury is
normally brilliant red, which has been used through our history as
a pigment for paints. Cinnabar and vermilion are names assigned to
a single chemical composition of different origin. The rst is a nat-
ural compound while the second refers to the synthetic pigment. In
both cases, there were no differences in structure and composition.
In the Bocanegra samples, it was difcult to distinguish between
cinnabar and vermilion. In Spain, the cinnabar mines in Almaden
were already known from the Roman times, and the use of vermil-
ion (articial mercuric sulphide) was described by the Arabs in the
8th-century. In China, vermilion has been used since antiquity [29].
Using XRF/XRD, it was possible to distinguish between the
composition of the iron phase present in the red layers of both the
pictorial layer and the preparation layer. The ruby red color, similar
to vermilion [30] of the pictorial layer, is constituted of hematite
 L.K. Herrera et al. / Talanta 80 (2009) 7183
in a higher proportion of goethite. The presence of only iron oxides
suggests an articial pigment. The mixture of goethite in a higher
proportion together with silicates in the preparation layer suggests
natural pigments with the calcite and quartz that Bocanegra used
in the preparation layer. Bocanegra used different ochre pigments
that contained varying amounts of octahedral iron oxides, namely
hematite (Fe2 O3 ) or/and goethite (FeOOH), mixed with quartz
and calcium. When hematite is the main iron oxide, a red color is
observed, whereas the ochre is yellow when the goethite domi-
nates. These colors are due to the ion Fe3+ contained in both oxides,
and more precisely to the charge transfer between Fe3+ and its lig-
ands, O2 or OH [31]. In Fig. 9a, the scheme of the stratigraphy
is superimposed with the iron mapping and diffraction patterns
(Fig. 9b) calculated in the same areas.
3.3.3. Blue pigments
The use of cobalt as a blue pigment in the Iberian Peninsula was
reported in the 13th [32] and 14th centuries [33]. In Fig. 10a, the
cross-section is shown and it is possible to observe the blue par-
ticles in the dark brown layer. In Fig. 10b, the blue particles were
examined using SEM/EDX, which showed the presence of potas-
sium, calcium, iron, cobalt, silicon and arsenic in low proportions.
The thin cross-section was analyzed using XRF/XRD to conrm
that this blue pigment was associated with a blue colored potassium
glass (smalt). The scheme of the distribution is superimposed with
the arsenic and cobalt mappings (Fig. 10c), and the diffraction pat-
terns were calculated in the same areas. The XRD pattern shows an
amorphous phase with a small contribution of silicon oxide that is
associated with a vitreous material (glass), as shown in Fig. 10d. The
presence of arsenic in the blue pigment is reported from the 16th
century [34]. The uses of the benet of the brightness of synchrotron
radiation in combination with both techniques (XRD/XRF) con-
Fig. 11. (a) Thin cross-section micrographs and elemental map of Cu obtained in the pictorial layer, of sample from the dress in the San Ignacio Herido en las Puertas de
Pamplona, (b) SR XRF spectrum, and (c) SR XRD pattern of blue grains.
81
rm that this blue pigment was prepared from cobalt-rich minerals,
such as cobaltite and smaltite to form cobalt oxide (Zaffer) [35].
Added as cobalt oxide during the manufacturing, with much of the
arsenic being volatilized [34]. When this cobalt blue glass pigment
degrades [36], it not only loses its color, but also causes yellow-
ing of the oil so that the paintings become brownish-yellow. When
smalt containing oil paint deteriorates, the alkaline component of
the glass, which in the case of smalt is potassium, is lost from the
pigment particles and migrates into the binding medium. There it
has the potential to cause concurrent alteration and degradation of
the oil, with possible processes including reactions with carboxylic
acids in the oil to form potassium fatty acid soaps, or with anion
sources in the environment to form species such as carbonates and
sulphates [37]. Bocanegra used this blue pigment that consists of a
potassium silicate glass containing cobalt oxide and a low quantity
of arsenic, which is still present in the cobalt matrix.
XRD has been proven to be a valuable tool for the clear identi-
cation of copper inorganic pigments. However, with conventional
XRD, it is rather difcult to acquire all of the information from these
samples due to the thinness of the color layers, which are in the
range of several tens of microns. Furthermore, the contributions of
the small quantities of the copper pigments to the total sample are
very low. Due to the small grains present in blue sample, it was dif-
cult to identify the pigments that appeared mixed within a single
layer.
To solve the difculties involving the characterization of small
quantity of pigments, experiments using XRF/XRD on the thin
cross-sections of this sample were conducted to illustrate the
potential of the combination of these techniques. Fluorescence and
diffraction data sets were simultaneously handled with PyMCA.
There is clear redundancy in analyzing the parallel elemental
and structure information provided by XRF and XRD. The average
82 L.K. Herrera et al. / Talanta 80 (2009) 7183
diffraction patterns were calculated over the same identied areas
(see Fig. 11a). In Fig. 11b, an XRF spectrum shows the K lines for
Cu, the thin cross-section is superimposed on some of the elemental
mapping (Cu). The use of this technique allowed for the determi-
nation that the blue grains had an azurite pattern (Fig. 11c). Azurite
is a natural blue pigment that is derived from the mineral azurite,
a basic copper carbonate. Bocanegra used it in the mixture of blue
and red pigments. Using XRF/XRD, it was possible to obtain only
the contribution of azurite without any interference of the other
pigments. XRD enables a precise identication of the crystalline
phase of azurite.
3.4. Organic binding medium
Analytical pyrolysis shows that all of the samples of the six paint-
ings have, as an organic ligand, siccative oils. This gives them a
high content of dicarboxylic acids, particularly azelaic acid. These
acids are considered, in high concentrations, markers of this type of
material. Additionally, the egg used in tempera painting has glyc-
erides with long chain aliphatic acids, but due the scarce content
of polyunsaturated acids, reactions with oxygen and the polymeri-
sation reaction of the siccative process is practically absent, and
only a low concentration of azelaic acid is observed. Chromato-
graphic peaks indicative of proteins are absent but their presence
is not excluded because the pyrolytical degradation of proteina-
ceous materials generally gives insignicant fragments. Only in the
presence of a high concentration of protein it is possible to identify
markers such as diketopiperazines, as reported in [38].
4. Conclusions
The analysis carried out in Bocanegra paintings of the San Igna-
cio of Loyola life using different physicalchemical techniques and
analytical methodology, and evaluating the relationships between
them, gave useful information about the employed materials and
painting techniques in the cultural heritage eld.
Scanning electron microscopy, especially in combination with
energy dispersive X-ray microanalysis, has been an indispensable
tool and contributed extensively to the identication of the ele-
ments that constituted the different inorganic pigments of the
conventional cross-sections of the paintings studied in this work.
However the phases present can not be detected, and is not a suit-
able technique for the detection of trace elements in pigments
because their concentration is often below the detection limit. FT-IR
supplied information on the organic components that the painter
used in the preparation layers. The advantages of PYGC/MS as a
tool for the rapid screening of the organic composition of art objects
is apparent from the results of this study; it permitted the identi-
cation of siccative oils like those used in the organic medium in
these paintings. This technique is capable of providing information
on the presence of inorganic compounds such as HgS.
Synchrotron-based XRD and XRF are particularly suited to
the non-destructive study and micro-characterization of the dif-
ferent materials present in multi-layered paintings. This study
demonstrated that synchrotron radiation micro-imaging tech-
niques have yielded the ability to analyze pictorial layers, including
grains, at the micrometer scale. XRD is very useful for the identi-
cation of phases that are present in low proportion in multi-layer
paintings with different inorganic pigments such as cerussite and
hydrocerussite, and those with low powder reection like azurite,
cobalt blue, hematite, and goethite. The conventional XRD is not
sensitive enough to detect these pigments in the multi-layer paint-
ings; only cinnabar that has a high powder reection was identied
by this technique.
The non-destructive conventional technique of XRD has been
applied in the study of ancient textile bres of Bocanegra paintings.
Using a glass capillary, it was possible to use only one bre of the
material. The presence of crystalline parts of the cellulose possibly
means that the bres are only partially degraded.
Acknowledgements
This work was supported by the European Commission,
Marie Curie Action MEST-CT2004-513915, MAT2004-04498 and
MAT2005-04838. The authors are very grateful to ESRF (ID 21 and
ID 18F) for their assistance during our experiments (experiment
EC-101). The authors thank the restoration team (PREDELA SL, Con-
servacin y Restauracin de Obras de Arte). A part of this study has
been funded by the Andalusian Government (Proyecto Andaluca
Barroca 2007).
References
[1] E. Ciliberto, G. Spoto (Eds.), Modern Analytical Methods in Art and Archaeology,
WileyInterscience, New York, USA, 2000, p. 763.
[2] D.C. Creagh, D.A. Bradley (Eds.), Radiation in Art and Archaeometry, Elsevier
Science, Amsterdam, The Netherlands, 2000, p. 297.
[3] M. Cotte, J. Susini, A. Sol, Y. Taniguchi, J. Chillida, E. Checroun, P. Walter, J. Anal.
At. Spectrom. 23 (2008) 820.
[4] J. Cassar, C. Degrigny, in: A. Adriaens, C. Degrigny, J. Cassar (Eds.), Benets of
Non-destructive Analytical Techniques for Conservation, Ofce for the Ofcial
Publications of the European Union, Luxembourg, 2005, pp. 912.
[5] K. Keune, Binding Medium, Pigments and Metal Soaps Characterized and Local-
ized in Paint Cross Sections, Dissertation, Molart Report, Archetype, London,
2005.
[6] E. Dooryhee, Ph. Colomban, Phase Transit. 81 (23) (2008) 139.
[7] L. Bertrand, D. Vantelon, E. Pantos, Appl. Phys. A 83 (2006) 225.
[8] E. Pantos, in: M. Uda, G. Demortier, I. Nakai (Eds.), X-rays for Archaeology,
Springer, London, 2005, pp. 199208.
[9] M. Uda, G. Demortier, I. Nakai, X-rays for Archaeology, Springer-Verlag, Nether-
lands, 2005, p. 459.
[10] D.C. Creagh, Radiat. Phys. Chem. 74 (2005) 426.
[11] L.K. Herrera, Physico-Chemical Research of Cultural Heritage Materials Using
Microanalytical Methods, PhD Thesis, Seville University and Material Science
Institute, Seville, Spain, 2009, ISBN 846920906-X.
[12] A. Von Haartman, Metodologa y tcnicas en la restauracin de obras pictricas
del Siglo de Oro espanol de la catedral de Almera, Universidad de Granada,
1992, p. 85.
[13] E. Orozco Diaz, Pedro Atanasio Bocanegra, Boletn Facultad de Letras, Granada,
1937, p. 24.
[14] F. Garca Gutierrez, Boletn de las Artes 9 (1991) 49.
[15] M. Cotte, E. Checroun, V. Mazel, V.A. Sol, P. Richardin, Y. Taniguchi, P. Walter, J.
Susini, e-PRESERVATIONScience (2009) 19.
[16] N. Khandekar, Rev. Conserv. 4 (2003) 52.
[17] A. Duran, L.K. Herrera, Ma .D. Robador, J.L. Perez-Rodriguez, Color Res. Appl. 32
(6) (2007) 489.
[18] L.K. Herrera, M. Cotte, M.C. Jimenez de Haro, A. Duran, A. Justo, J.L. Perez
Rodriguez, Appl. Clay Sci. 42 (12) (2009) 57.
[19] G. Chiavari, S. Prati, Chromatographia 58 (2003) 543.
[20] J. Boon, S. Asahina, Microsc. Microanal. 12 (2) (2006) 1322.
[21] A. Somogyi, M. Drakopoulos, L. Vincze, B. Vekemans, C. Camerani, K. Janssens,
A. Snigirev, F. Adams, X-ray Spectrom. 30 (2001) 242.
[22] V.A. Sol, E. Papillon, M. Cotte, P.H. Walter, J. Susini, Spectrochim. Acta Part B 62
(1) (2007) 63.
[23] A.P. Hammersley, C. Riekel, Synchrotron Radiat. News 2 (1989) 24.
[24] G. Chiavari, S. Prati, G. Lanterna, C. Lalli, A. Cagnini, Microchim. Acta 159 (34)
(2007) 357.
[25] A. Satta, R. Hagege, M. Sotton, Bull. Sci. ITF 15 (1986) 3.
[26] G.F. Davidson, T.P. Nevell, J. Textile Inst. 47 (1956) 439.
[27] J.I. Goldstein, D.E. Newbury, D.C. Joy, C.E. Lyman, P. Echlin, E. Lifshin, L.C. Sawyer,
J.R. Michael, Scanning Electron Microscopy and X-ray Microanalysis, 3rd ed.,
Kluwer Academic/Plenum Publishers, New York, 2003, p. 40.
[28] A. Duran, Metodologa de Estudio y Anlisis de diferentes tipos de obras
pertenecientes a la Escuela Sevillana de los siglos XVII y XVIII, PhD Thesis,
Universidad de Sevilla, 2006, 480 pp.
[29] R.J. Gettens, R.L. Feller, W.T. Chase, in: A. Roy (Ed.), A Handbook of their His-
tory and Characteristics, vol. 2, National Gallery of Art, Washington and Oxford
University Press, Oxford, 1993, pp. 159182.
[30] R.J. Gettens, G.L. Stout, Painting Materials: A Short Encyclopaedia, Dover Publi-
cations, New York, 1966, pp. 157159.
[31] M. Elias, C. Chartier, G. Prvot, H. Garay, C. Vignaud, Mater. Sci. Eng. B 127 (2006)
70.
[32] C. Roldan, J. Coll, J. Ferrero, J. Cult. Herit. 7 (2) (2006) 134.
 L.K. Herrera et al. / Talanta 80 (2009) 7183 83

[33] J. Prez-Arantegui, M. Resano, E. Garca-Ruiz, F. Vanhaecke, C. Roldn, J. Ferrero, [36] M. Spring, C. Higgitt, D. Saunders, Natl. Gallery Tech. Bull. 26 (2005)
J. Coll, Talanta 74 (5) (2008) 1271. 56.
[34] G. Pappalardo, E. Costa, C. Marchetta, L. Pappalardo, F.P. Romano, A. Zucchiatti, [37] M. Spring, N. Penny, R. White, M. Wyld, Natl. Gallery Tech. Bull. 22 (2001)
P. Prati, P.A. Mand, A. Migliori, L. Palombo, M.G. Vaccari, J. Cult. Herit. 5 (2004) 54.
183. [38] G. Chiavari, N. Gandini, P. Russo, D. Fabbri, Chromatographia 47 (1988)
[35] B. Mhlethaler, J. Thissen, in: A. Roy (Ed.), Artists Pigments: A Handbook of 420.
their History and Characteristics, vol. 2, National Gallery of Art, Washington,
DC, 1993, pp. 113130.