Anal. Chem. 2009, 81, 7443–7447
Surface-Enhanced Raman Spectroscopy: A Direct
Method to Identify Colorants in Various Artist
Media
Christa L. Brosseau,† Kari S. Rayner,† Francesca Casadio,‡ Cecily M. Grzywacz,§ and
Richard P. Van Duyne*,†
Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois, 60208, Art Institute of
Chicago, Chicago, Illinois, 60603, and Getty Conservation Institute, Los Angeles, California, 90049
Surface-enhanced Raman spectroscopy (SERS) has been
developed as a direct, extractionless, nonhydrolysis tool
to detect lake pigments and colorants of various classes
used in a variety of artist materials. Presented first is the
SERS analysis of the natural colorant turmeric (Curcuma
longa L.), main component curcumin, as present in dry
lake pigment grains, dyed textile yarns, and reference
paint layers containing the lake pigment bound in animal
glue painted on glass. This experiment demonstrated that
it is possible to detect the chromophore in various
matrixes of increasing complexity, allowing its unambiguous identification in a wide range of artists’ materials, even
at very low concentration and in the presence of binders
such as glue. In addition, removal of the colorant from
the complex with the inorganic substrate or mordanted
yarn was not necessary for identification. This proof-ofconcept study was then extended to include analysis of
several pastel sticks from a historical pastel box and two
samples from a pastel artwork, both attributed to American painter Mary Cassatt (1844-1926). This study
represents the first extractionless, nonhydrolysis direct
SERS study of multiple artist materials, including identification of natural and synthetic colorants and organic
pigments contained in historic artists’ pastels spanning a
broad range of chemical classes: polyphenols, rhodamines,
azo pigments, and anthraquinones. Successful identification is demonstrated on samples as small as a single grain
of pigment.
Identification of pigments and colorants contained in artworks
and historical textiles is of fundamental importance for the study
of their manufacturing technology and to their conservation and
long-term preservation. Conventional methods for the identification of natural and synthetic organic dyes include high-performance liquid chromatography (HPLC) and UV-vis spectroscopy.
The former method is destructive and requires relatively large
amounts of sample (low microgram)1,2 not always available from
priceless works of art, even if chromatographic separations
* To whom correspondence should be addressed. E-mail: vanduyne@
northwestern.edu. Phone: 847-491-3516. Fax: 847-491-7713.
†
Northwestern University.
‡
Art Institute of Chicago.
§
Getty Conservation Institute.
10.1021/ac901219m CCC: $40.75  2009 American Chemical Society
Published on Web 07/28/2009
followed by online identifications of separated components may
sometimes be necessary to identify the biological source of the
colorant down to the species level.3 The latter method has poor
specificity4 and suffers greatly from matrix interferences. The ideal
analytical tool for the identification of artist dyestuffs would be
minimally destructive, highly diagnostic, sensitive and would be
capable of probing a broad range of samples in a variety of
matrixes. Among the vibrational spectroscopic techniques that
have been applied to the analysis of organic colorants (i.e., Fourier
transform infrared (FT-IR) and near-IR (NIR) spectroscopy),
normal Raman spectroscopy in theory provides the most promise
for identification of colorants in minute amounts of sample.
However, normal Raman spectroscopy suffers from inherently
weak signals and interference from fluorescence. As a result, this
tool is not easily applied to the identification of natural organic
dyes and pigments especially when the latter are embedded in
biomaterial matrixes such as traditional artists’ paints or historic
textiles. However, the use of surface-enhanced Raman spectroscopy (SERS) allows for both an enhanced Raman signal and
substantial quenching of fluorescence through use of noble metal
substrates.5-8 Recently a number of papers have reported success
in using SERS to identify organic, highly fluorescent dyes
commonly found in works of art.9-11 One of the most commonly
followed approaches involves samples (mainly textile yarns) where
the dye was extracted from the host material. A handful of studies
have reported on analysis done directly on the samples without
extraction;12,13 however, these studies were limited to dyed
reference textiles or paint mock-ups and did not include analysis
(1) Balakina, G. G.; Vasiliev, V. G.; Karpova, E. V.; Mamatyuk, V. I. Dyes Pigm.
2006, 71, 54.
(2) Sanyova, J.; Reisse, J. J. Cult. Herit. 2006, 7, 229.
(3) Wouters, J.; Verhecken, A. Stud. Conserv. 1989, 34, 189.
(4) Karapanagiotis, I.; Valianou, L.; Daniilia, S.; Chryssoulakis, Y. J. Cult. Herit.
2007, 8, 294.
(5) Jeanmaire, D. L.; Van Duyne, R. P. J. Electroanal. Chem. 1977, 81, 1.
(6) Albrecht, M. A.; Creighton, J. A. J. Am. Chem. Soc. 1977, 99, 5215.
(7) Birke, R. L.; Lombardi, J. R. Surface Enhanced Raman Scattering. In
Spectroelectrochemistry: Theory and Practice; Gale, R. J., Ed.; Plenum: New
York, 1988; pp 263-348.
(8) Moscovits, M. Rev. Mod. Phys. 1985, 57, 783.
(9) Leona, M.; Lombardi, J. R. J. Raman Spectrosc. 2007, 38, 853.
(10) Chen, K.; Leona, M.; Vo-Dinh, T. Sens. Rev. 2007, 27 (2), 109.
(11) Leona, M.; Stenger, J.; Ferloni, E. J. Raman Spectrosc. 2006, 37, 981.
(12) Chen, K.; Vo-Dinh, K.-C.; Yan, F.; Wabuyele, M. B.; Vo-Dinh, T. Anal. Chim.
Acta 2006, 569, 234.
(13) Jurasekova, Z.; Domingo, C.; Garcia-Ramos, J. V.; Sanchez-Cortes, S. J.
Raman Spectrosc. 2008, 39, 1309.
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
7443
on actual historic works of art. Indeed, one of the main limitations
of SERS up to the present day, seriously hampering its use in art
conservation applications, has been that while it has proven greatly
successful with model reference materials, translation to actual
works of art has been difficult. This is because real artworks
present the challenge of incorporating the target dye molecule(s)
in a complex host matrix, and both dye and host matrix will have
generated byproducts as a consequence of aging. The first direct
SERS study of an actual archeological pigment containing purpurin
was reported by Van Elslande et al. in 2008.14
The present study used nonextractive, nonhydrolysis, direct
SERS for the identification of organic chromophores contained
in a variety of artist materials in a proof-of-concept fashion, using
reference materials containing the yellow organic dye turmeric
(Curcuma longa L.) main component curcumin (a polyphenol),
and this methodology was then used to analyze actual historic
pastel samples contained in a pastel box once belonging to
American painter Mary Cassatt (1844-1926), now part of the
collection of the Museum of Fine Arts, Boston, and samples from
a pastel artwork attributed to Cassatt in the collection of the Art
Institute of Chicago.
EXPERIMENTAL SECTION
Materials. Silver nitrate (99+%) and sodium citrate were
purchased from Sigma-Aldrich (St. Louis, MO). Dyed yarn
samples of turmeric, as well as turmeric pigment precipitated on
alumina and turmeric pigment bound in animal glue and painted
on glass, were prepared by the Getty Conservation Institute as
part of the Institute’s Asian organic colorants project (details of
manufactory are given elsewhere).15 Curcumin was obtained from
Sigma-Aldrich (St. Louis, MO, 99%), while madder root was
obtained from Kremer pigments (New York, NY). Samples from
the pastel box of Mary Cassatt were provided by the Boston
Museum of Fine Arts. Two minute samples were taken using a
tungsten needle from the pastel artwork “Sketch of Margaret
Sloane, Looking Right” attributed to Mary Cassatt and part of the
collection of the Art Institute of Chicago (pastel on tan wove paper;
410 mm × 330 mm; gift of Laura May Ripley, AIC 1992.158).
Preparation of Citrate-Reduced Colloids. Citrate-reduced
silver colloids were prepared using the standard Lee and Meisel
preparation,16 having a peak absorption wavelength of ∼500 nm
and a full width at half-maximum (fwhm) of ∼100 nm. After being
cooled, they were centrifuged 10 times (relative centrifugal force
) 36 000g, 15 min per cycle) to concentrate the colloid. The
colloids prepared this way were stable at room temperature
(stored in the dark) for up to 3 weeks.
Surface-Enhanced Raman Spectroscopy. All SER spectra
were collected on a custom-built macro setup. The 632.8 nm
excitation was obtained using a HeNe laser (12 mW output power,
9 mW at the sample, 0.8 mm beam diameter) (Research ElectroOptics, Boulder, CO). The SERS measurements employ 1 in.
interference and notch filters (Semrock, Rochester, NY), a singlegrating monochromator with the entrance slit set to 100 µm
(model VM-505, Acton Research Corporation, Acton, MA), a liquid
(14) Van Elslande, E.; Lecomte, S.; Le Hô, A.-S. J. Raman Spectrosc. 2008, 39,
1001.
(15) Grzywacz, C.; Bomin, S.; Yuquan, F.; Wouters, J. ICOM Committee Conserv.
2008, 1, 528.
(16) Lee, P. C.; Meisel, D. J. Phys. Chem. 1982, 86, 3391.
7444
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
N2 cooled charge-coupled device (CCD) detector (model
Spec10:400B, Roper Scientific, Trenton, NJ), and a data acquisition system (Photometrics, Tucson, AZ). The spectral positions
of the CCD pixels were calibrated using a neon lamp. The
spectral resolution was 4 cm-1. Although excitations of 532 and
785 nm were also evaluated for SERS of these dyes, 633 nm
excitation was found to give the most intense SERS signal, with
a minimum of interfering fluorescence.
Sample Preparation. For all of the samples in this study, 5 µL
of the centrifuged, citrate-reduced colloids was added directly to the
sample and mixed with a clean gold wire. The sample mixed with
the colloids was then applied to a clean glass microscope slide (if
not already present on a microscope slide), and the SER spectra
were recorded after the colloid paste had dried onto the sample.
Excellent SER spectra could be obtained from the same sample
after several months of storage.
RESULTS AND DISCUSSION
Proof-of-Concept Study on Curcumin-Containing Reference Materials. As proof-of-concept for the application of nonextractive, nonhydrolysis SERS for the direct detection and
identification of natural organic dyes in a variety of artists’
materials and media, turmeric, used as a dye or precipitated on
alumina to form a lake pigment, was studied. This colorant was
chosen because it was available in several different matrixes that
could be studied and also because, although several studies of
anthraquinones have been published,12,17-19 far fewer SERS
studies on yellow lake pigments are available.20 Parts a-c of
Figure 1 show the SER spectra for a painting layer containing
turmeric lake bound in animal glue (as could be suggested, for
example, in traditional Asian wall paintings)15 and painted on glass,
a curcumin-dyed silk yarn, and for turmeric precipitated out on
alumina, respectively. When compared to the SER spectrum for
the free dye (curcumin) (Figure 1d), the data show that all three
spectra are distinctive for curcumin, having bands arising at 1587,
1525, 1489, 1293, 1250, 1158, and 1125 cm-1.20 These results
indicate that SER spectra can be obtained without the need
for extraction/hydrolysis of the pigment from the host material.
In fact, although slight laser photodegradation of the mordanted
dye was observed, evidenced by the appearance of large, broad
background features at ∼1250 and ∼1550 cm-1 (due to
amorphous carbon),20 its extent is less significant compared
to the free dye, indicating enhanced photostability of the
complex. This experiment represents a fundamental step
forward in the ability to use SERS as an additional tool for the
identification of colorants in works of art especially when only
minute amounts of material are available, down to a single
particle of pigment. To extend this on-the-specimen SERS
analysis of artist materials further, analysis was preformed on
actual samples of historical pastel grains.
SER Spectra for Pastels from the Pastel Box of Mary
Cassatt. A few grains of colored powder were sampled from pastel
(17) Baran, A.; Wrzosek, B.; Bukowska, J.; Proniewicz, L. M.; Baranska, M. J.
Raman Spectrosc. 2009, 40 (4), 436.
(18) Cañamares, M. V.; Garcia-Ramos, J. V.; Domingo, C.; Sanchez-Cortes, S.
Vib. Spectros. 2006, 40, 161.
(19) (a) Cañamares, M. V.; Garcia-Ramos, J. V.; Domingo, C.; Sanchez-Cortes,
S. J. Raman Spectrosc. 2004, 35, 11–921.
(20) Cañamares, M. V.; Garcia-Ramos, J. V.; Sanchez-Cortes, S. Appl. Spectrosc.
2006, 60, 1386.
Figure 1. Proof-of-concept for SER spectra of various artist materials
containing the colorant turmeric: (a) SER spectra of turmeric pigment
bound in animal glue and painted on glass (multiplied by three for
ease of comparison); (b) SER spectra of a silk yarn dyed with turmeric;
(c) SER spectra of turmeric precipitated out on alumina; (d) SER
spectra of free curcumin dye. All samples were mixed with 5 µL of
citrate-reduced silver colloids to obtain the SERS signal. Characteristic
peaks for the colorant are labeled, and peaks which correspond to
citrate are labeled with an asterisk.
sticks contained in the pastel box of Mary Cassatt (Figure 2a).
The colors of the pastels chosen for this study ranged from bright
red to pale pink and from dark purple to mauve, reflecting
increasing dilution of the coloring agent with inorganic fillers
(such as calcium carbonate and kaolin clay or, occasionally,
barium sulfate and gypsum) and mixing with other pigments to
achieve the purple hues (Figure 2b). Figures 3-5 show the SER
spectra for the six pastels. The red pigment in pastel sticks nos.
9 (lilac), 10 (bright red), and 17 (light pink) was identified as
carmine lake21,22 (Figure 3), mixed with ultramarine blue to
achieve a purple hue in the case of pastel no. 9. The fuchsia
pigment in the pink pastel stick no. 14 was identified as a mixture
of rhodamine B23 and rhodamine 6G,24 indicated in Figure 4 as
the gray and black labeled peaks, respectively. In addition, the
strong band at ∼1070 cm-1 indicates the presence of hydromagnesite, a mineral binder, in the pastel sample. The purple
pastel sticks nos. 1 and 7 were found to give strong SER spectra
(Figure 5), confirming that they contain the same dyestuff. The
bands at 1603, 1537, 1350, and 1160 cm-1 are characteristic for
nitrobenzene,25,26 whereas the band at 1487 cm-1 indicates an
azo ring vibration.27 The band at 1389 cm-1 indicates the
presence of an NdN azo moiety, and bands at 484 and 716
cm-1 indicate the presence of a naphthol moiety.28 This
spectroscopic evidence seems to point toward the identification
of this early synthetic organic pigment as belonging to a
(21) Whitney, A. V.; Casadio, F.; Van Duyne, R. P. Appl. Spectrosc. 2007, 61,
994.
(22) Chen, K.; Leona, M.; Vo-Dinh, K.-C.; Yan, F.; Wabuyele, M. B.; Vo-Dinh,
T. J. Raman Spectrosc. 2006, 37, 520.
(23) Vo-Dinh, T.; Allain, L. R.; Stokes, D. L. J. Raman Spectrosc. 2002, 33, 511.
(24) Dieringer, J. A.; Lettan, R. B., II; Scheidt, K. A.; Van Duyne, R. P. J. Am.
Chem. Soc. 2007, 129 (51), 16249.
(25) Integrated spectral database for organic compounds (SDBS). http://
riodb01.ibase.aist.go.jp/sdbs (accessed April 1, 2009).
(26) Zhang, D.; Lan, G.; Hu, S.; Wang, H.; Zheng, J. J. Raman Spectrosc. 1993,
24, 753.
(27) Hacker, H. Spectrochim. Acta 1965, 21, 1989.
Figure 2. (a) Detail of Mary Cassatt’s pastel box, courtesy of the
Boston Museum of Fine Arts, Conservation Department. (b) Photomicrographs of the pastel sticks from Cassatt’s pastel box described
in this study (with a small reproduction of the actual pastel stick in
the inset).
Figure 3. SER spectra of red pigment contained in (a) pastel no.
17, (b) pastel no. 10, and (c) pastel no. 9. Pigment was identified as
carmine lake; labeled peaks are characteristic for this pigment. The
asterisk indicates peak due to citrate.
β-naphthol and/or monoazo class of dyes. Although attempts
have been made to more precisely identify this dye by visual
comparison with available databases and direct SERS compariAnalytical Chemistry, Vol. 81, No. 17, September 1, 2009
7445
Figure 4. SER spectra of the red pigment contained in pastel no.
14. Bands which are consistent with rhodamine B are labeled as
black, whereas bands consistent with rhodamine 6G are labeled in
gray. Citrate bands are indicated by an asterisk.
Figure 6. Mary Cassatt, pastel study: “Sketch of Margaret Sloane,
Looking Right” (pastel on tan wove paper, measuring 410 mm × 330
mm; gift of Laura May Ripley, AIC 1992.158).
Figure 5. SER spectra of the red pigment contained in (a) pastel
no. 7 and (b) pastel no. 1. Labeled peaks are characteristic for the
pigment, and peaks labeled with an asterisk are from citrate.
son with reference samples of PR3 (1-[(4-methyl-2-nitrophenyl)azo]-2-naphthalenol), PR4 (1-[(2-chloro-4-nitrophenyl)azo)]2-naphthalenol), PO5 (1-[(2,4-dinitrophenyl)azo]-2-naphthol),
and PR23 (3-hydroxy-4-[(2-methoxy-5-nitrophenyl)azo]-N-(3nitrophenyl)-2-naphthalene carboxamide), all of which contain
the functional groups described above, a perfect match has
remained elusive. This observation underscores the necessity
for the development of more extensive SER spectral databases
of dyes and colorants, including these early synthetic dyes.
Highly specific identification of a dye is especially relevant for
synthetic organic dyes developed in the late 19th and early 20th
century, as the finding of a specific dye may have important
dating implications for the work of art on which it is used or
for timing of an intervention such as an overpaint. For example,
many different monoazo red pigments have been used as
artists’ pigments, but while the β-naphthol and naphthol reds
were introduced in the early 1900s, benzimidazolones were first
(28) Vandenabeele, P.; Moens, L.; Edwards, H. G. M.; Dams, R. J. Raman
Spectrosc. 2000, 31, 509.
7446
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
Figure 7. SER spectra of (a) pastel stick no. 7 and (b) sample no.
10 “fleshtone in face” from Mary Cassatt’s “Sketch of Margaret Sloane,
Looking Right” (multiplied by 10 for ease of comparison). Peaks which
are characteristic for the unidentified red pigment, which also appears
in pastel stick no. 7, are labeled. Peaks due to lead white (LW) and
chrome yellow-orange (C Y-O) are also labeled.
patented in the 1960s29 and would be anachronistic materials
for a Cassatt pastel, leading to questions about authenticity.
SER Spectra of Samples Taken from the Mary Cassatt
Pastel “Sketch of Margaret Sloane, Looking Right”. Two
minute samples were taken for analysis from the pastel artwork
by Mary Cassatt, shown in Figure 6. The first sample was taken
(29) Lomax, S. Q.; Learner, T. J. Am. Inst. Conserv. 2006, 45, 107.
a number of peaks of unknown origin, underscoring the
necessity to develop comprehensive databases of SER spectra
of modern synthetic organic pigments for positive identification
of the materials.
Figure 8. SER spectra of (a) madder root (Rubia tinctorum L.) and
(b) sample no. 12 “mauve” from Mary Cassatt’s “Sketch of Margaret
Sloane, Looking Right”. Dashed lines indicate peaks that are
consistent with madder root dye. Solid lines indicate unidentified
bands due to a second component in the sample. In the inset is a
photomicrograph of sample no. 12.
from the fleshtone color used in the face of the sitter in the sketch,
and the second sample was taken from mauve-colored strokes in
her white ruff. Figures 7 and 8 show the SER spectra for these
samples, respectively. When compared with the pastel sticks
sampled in this study, it is apparent that the fleshtone color used
for the face was achieved using the same pigment found in pastel
sticks nos. 1 and/or 7, exemplified by bands at 1537, 1350, and
1487 cm-1. In addition, the presence of bands at ∼1050 and
∼825 cm-1 indicate the presence of lead white and chrome
yellow-orange, respectively.30 The mauve color sampled from
the artwork contains a pink colorant that provides a SER
spectrum which is similar to madder with bands at 452, 807,
1158, 1274, 1398, and 1468 cm-1 and a second component with
bands at 998, 1112, 1248, 1340, 1635 cm-1, for which a good
match could not be found. Since the colors used in the artwork
were achieved by mixing multiple pastel sticks and were
layered on the silica-coated paper used as support, it is not
surprising that these spectra contain more than one component.
The spectra for this second artwork sample in particular contain
(30) Bell, I. M.; Clark, R. J. H.; Gibbs, P. J. Spectrochim. Acta, Part A 1997, 53,
2159.
CONCLUSIONS
This work demonstrates nonextractive, nonhydrolysis, direct
SERS applied for the first time to the identification of both
synthetic and natural organic dyes present in a wide variety of
artist materials in various media, including a pastel study attributed
to the American artist Mary Cassatt. Positive identification was
possible on a sample as small as a single grain of pigment (particle
size 1-6 µm). The fact that multiple samples coming from
different sources could be analyzed, and the chromophores could
be detected directly on a minute sample taken from a precious
artwork, highlights the usefulness and versatility of this technique
for art conservation studies. In the majority of cases, the pigment
could be readily identified. In some cases it was not possible to
unambiguously identify the colorant, and this exemplifies the need
for the collection and dissemination of more extensive SERS reference libraries of pigments and the need for a closer collaboration
with theoreticians who would be able to construct libraries of
theoretical SERS spectra for chromophores of interest.
ACKNOWLEDGMENT
Conservation science at Northwestern University and the Art
Institute of Chicago is made possible by grants from the Andrew
W. Mellon Foundation and the National Science Foundation
(Grants CHE-0414554, CHE-0911145, and DMR-0520513). The
authors thank Jan Wouters from the Getty Conservation Institute
for the turmeric-dyed reference materials. Richard Newman, head
of Scientific Research, Museum of Fine Arts, Boston, is gratefully
acknowledged for sharing samples from Cassatt’s pastel box and
allowing reproduction of its images and useful discussions
concerning the identity of the colorant in pastel sticks nos. 1 and
7. In addition, the authors acknowledge Tom Learner (Getty
Conservation Institute, CA) for samples of PR3, PR4, and PO5
and Microtrace LLC (Elgin, IL) for the sample of PR23.
Received for review June 4, 2009. Accepted July 17, 2009.
AC901219M
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
7447