Identification of darkened pigments in cultural objects by graphite

furnace atomic absorption spectroscopy and inductively coupled

plasma-mass spectrometryw

Douglas M. Goltz,*aKimberly Charleton,

abEdward Cloutis,

aPatricia Grinberg

c

and Cathy Collinsd

Received 10th August 2006, Accepted 3rd November 2006

First published as an Advance Article on the web 20th November 2006

DOI: 10.1039/b610488f

Inorganic pigments in paint have been used by artists throughout history, primarily because of

their resistance to chemical change over long periods of time. Many of the historically-important

pigments consist of transition metal oxides, sulfides or carbonates. The determination of metals in

pigments is a powerful tool that can be used by conservators and researchers to identify pigments.

In this study, pigments were sampled using a dry cotton bud that was contacted lightly with the

surface of painted objects to remove a small quantity of pigment (o1 mg). This sampling

approach causes no visible damage to a paint surface. Evidence of contact cannot be detected

visually, even with the aid of a magnifying glass. Prior to analysis, pigment metals were extracted

from the cotton with concentrated HNO3 in a 2 ml polystyrene beaker. The pigment metals were

determined using graphite furnace atomic absorption spectroscopy (GFAAS) and electrothermal

vaporization inductively coupled plasma mass spectrometry (ETV-ICP-MS). The small sample

size requirement and high sensitivity of these instruments make them very suitable for analyzing

pigment metals. Solution nebulization SN-ICP-MS was also used for obtaining a mass scan for

most elements in the periodic table. The application of both GFAAS and ICP-MS is described

for determining the identity of the metals in different pigments of two cultural objects (a painting

and a map), which had darkened over time. Raman spectroscopy was used for confirming the

identity of a darkened red pigment in a map as Red Lead (Pb3O4).

Introduction

Throughout time, inorganic pigments have been used success-

fully in paint for their opacity or for providing colour. One

property that most historically-important pigments share is

their resistance to changes in appearance or chemical degrada-

tion over time. The colour permanence of pigments is essen-

tially due to their chemical stability, as the most stable

pigments often occur naturally as minerals. Since the appear-

ance and chemical composition of most historically-important

pigments are well known, pigment identification is possible by

determining the metals that are present in them. Pigment

identification can be an important step prior to the preserva-

tion and conservation of historical objects. Pigment identifica-

tion may also be a tool that art historians can use for

attribution or dating of artwork.1 Numerous approaches are

available to the conservator for identifying metals or pigments

used in paintings, illuminated manuscripts and other cultural

objects, including textiles.

Spectroscopic techniques have become some of the most

important tools used by conservators for studying cultural

objects.2 For example, Raman spectroscopy has emerged as an

important technique for identifying the chemical structure of

pigments in paint. When combined with microscopy, Raman

spectroscopy is an extremely powerful tool for non-destruc-

tively examining paint, pigments and binding agents in cultur-

al objects; numerous papers have been published on its use.3–7

Libraries of Raman spectra of various painting materials,

including pigments, have been published by Clark et al.8,9

Atomic spectroscopy has also been used for identifying the

metals in pigments because most metals can be identified with

very high sensitivity. One potential drawback is that the

analysis may result in damage to the sample surface. Laser

induced breakdown spectroscopy (LIBS) has emerged as a

useful technique for identifying metals in pigments. A

Nd-YAG laser is used to excite the atoms on the surface of

a cultural object and the excited atoms emit an emission

pattern, which is indicative of the metals present.10 Damage

to the painted surface can be minimized by focusing the laser

on a small area (1 mm2).11–15 An added benefit of LIBS is that

depth profiling of different paint layers can be achieved.16,17

X-ray spectroscopy has also become established as one of the

most important tools for non-destructively identifying metals

a C-SCAPE-Centre for Scientific and Curatorial Analysis of PaintingElements, Dept. of Chemistry, University of Winnipeg, 515 PortageAve, Winnipeg, MB, Canada R3B 2E9

bDept. of Chemistry, University of Manitoba, Winnipeg, MB, CanadaR3T 2N2

cChemical Metrology, Institute for National Measurement Standards,National Research Council of Canada, Ottawa, Ontario, CanadaK1A 0R9

dWinnipeg Art Gallery, 300 Memorial Blvd, Winnipeg, MB, CanadaR3C 1V1

w Electronic supplementary information (ESI) available: photographsof painting Untitled (Arctic Spring) and a 1669 map. See DOI:10.1039/b610488f

140 | J. Anal. At. Spectrom., 2007, 22, 140–146 This journal is �c The Royal Society of Chemistry 2007

PAPER www.rsc.org/jaas | Journal of Analytical Atomic Spectrometry

in pigments. Unfortunately, portable instruments that use X-

ray fluorescence (XRF) spectroscopy can suffer from poor

sensitivity for some elements. To overcome this problem, total

reflection XRF (TXRF) has been developed and applied to the

analysis of pigments in the paint and pigments of illuminated

manuscripts18 and oil paintings.19–22 The most important

advantage of TXRF is that it is a very sensitive, non-destruc-

tive technique which requires very small quantities of sample.

Central to the success of TXRF is the sampling approach used

for works of art such as paintings. Typically, a very small

amount of pigment is gently removed using a cotton bud by

rubbing the paint surface. Since only mg-quantities of pigment

are required, there is minimal, if any, physical damage to the

paint surface. When approaches using atomic spectroscopy are

combined with molecular spectroscopy (i.e. Raman spectro-

scopy) they become extremely powerful tools for identifying

most inorganic pigments that are used in paint.18,19 Although

most of the previous work on pigment sampling with a Q-tips

used XRF for metal analysis, some work has been carried out

using graphite furnace atomic absorption spectroscopy

(GFAAS).23 The disadvantage of using GFAAS is that ana-

lysis can be slow as only one element at a time can be

determined. Fortunately, for most samples, there are a limited

number of elements that correspond to a specific pigment.

Furthermore, one potential advantage of GFAAS over TXRF

is that a greater sensitivity of measurement may be achieved

for lighter elements.

One objective of this work was to test the feasibility of

sampling paint using a dry cotton bud prior to analysis by a

number of approaches including: GFAAS, solution nebuliza-

tion (SN-ICP-MS), electrothermal vaporization (ETV-ICP-

MS) and Raman spectroscopy. One application of this meth-

odology is demonstrated for identifying pigments in two

different cultural objects that have darkened over time.

Experimental

The general procedure for extracting pigment metals is to

gently contact a dry cotton bud (similar to a Q-tips) to

the surface of a painted object. Since only a small quantity

(B1 mg) of pigment is required for analysis, it is usually

impossible to detect any pigment on the dry cotton bud with

the unaided eye. Furthermore, if too much pigment is removed

with the cotton bud, a 10–100-fold dilution step is required

prior to analysis with GFAAS or ICP-MS. After sampling, the

cotton bud was placed in a 2 ml polystyrene beaker and 500 mlof high purity HNO3 (Mallinckrodt, AR Select Plus) was

added. The beaker was then placed on a hotplate at a low

temperature to evaporate most of the acid. After 1 h, de-

ionized water (Milli-Q 18 MO cm�1) was added and the

undigested cotton was removed from the beaker.

To avoid potential problems with undissolved cellulose

fibres, the sample could have been filtered or centrifuged:

however for simplicity and to minimize the risk of contamina-

tion, the sample was allowed to sit for 1 h. This allowed the

undigested material to settle to the bottom of the beaker. The

choice of acid used in the digestion of the cotton may have a

significant impact on the analysis. For example, if HCl is used

the presence of large amounts of Cl� in the sample can cause

matrix effects for some elements with GFAAS. If SN-ICP-MS

analysis is performed, Cl� in the sample can also cause spectral

interferences. For example Cl� can recombine to form40Ar35Cl+, which overlaps the only isotope of 75As+. Since

GFAAS and ICP-MS analyses (both SN and ETV) were

performed on the same solutions, compromise conditions were

required for sample preparation. For this reason only HNO3

was used for the digestion of all the samples.

The ICP-MS used in this work was a PerkinElmer Elan 6000

that was equipped with a quartz concentric nebulizer and a

cyclonic spray chamber for solution nebulization. The

ICP-MS was also interfaced to a PerkinElmer HGA 600

graphite furnace using 50 cm of PTFE tubing. Small volumes

of sample (10–40 ml) were pipetted into the electrothermal

vaporizer (ETV) using a PerkinElmer AS 60 autosampler. The

graphite furnace AAS was a PerkinElmer AS800 that was

equipped with an AS800 autosampler and Zeeman effect

background correction. Experimental parameters for the

ICP-MS are indicated in Table 1. For GFAAS, the recom-

mended temperature program for each element was used.

However, with ETV-ICP-MS, the multi-elemental analysis

required a temperature program, as shown in Table 2, which

required compromise conditions for vaporizing more than one

element. To reduce spectral interferences from ArO, Fe was

also analyzed in DRC mode using NH3 with a gas flow rate of

0.3 ml min�1.

A portable Raman spectrometer was used for confirming

pigment identification. A Delta Nu Raman spectrometer

equipped with a 3 mW 633 nm HeNe laser was used. The

spectral range of this instrument was 200–3400 cm�1 with a

resolution of 10 cm�1. Spectra were typically obtained by

placing the paint surface 1 cm from the source and the

measurement time for each spectrum was 1–5 accumulations

of 10–30 s each.

Table 1 Instrumental operating and data acquisition parameters

ICP mass spectrometerRF power 1100 WCoolant Ar flow 15.0 l min�1

Auxiliary Ar flow 1.4 l min�1

Carrier Ar flow 1.0 l min�1

ICP-MS data acquisitionDwell time 50 msScan mode Peak hoppingNumber of masses (m/z) monitored 3–10Signal measurement Average counts (SN)

Peak area (ETV)Resolution 0.7 amu at 10% peak height

Table 2 Instrumental operating and data acquisition parameters ofthe ETV-ICP-MS

T/1C Ramp/s Hold/sAr flowrate/ml min�1

StepDry 120 10 60 300Char 400–800 10 30 300Atomization 2000–2400 0 5 150a

Clean out 2650 0 5 300

a Read step.

This journal is �c The Royal Society of Chemistry 2007 J. Anal. At. Spectrom., 2007, 22, 140–146 | 141

Discussion

Perhaps the most important property associated with sam-

pling a painted surface using a cotton bud is the mass of metal

or pigment removed. Clearly, there is a relationship between

the amount of pigment removed and the amount of damage

sustained by the painted surface. For painted surfaces of

valuable artwork, any damage to the surface would be unac-

ceptable. In a previous study,23 painted surfaces were sampled

manually with a cotton bud and it was estimated that the

amount of pigment removed with a cotton bud is approxi-

mately 1 mg. To acquire a better estimate of the mass of

pigment that is typically removed with a cotton bud during

sampling, an experiment was designed to quantify the mass of

pigment removed with respect to the contact force. In this

experiment a series of microscope slides were painted using a

stainless steel spatula and allowed to dry overnight. The paint

was sampled by placing the microscope slide on a pan balance.

A cotton bud was secured with a clamp and carefully lowered

onto the paint surface. The force on the paint surface was

determined by mass. The pan of the balance was rotated for a

predetermined number of times, in both clockwise and counter

clockwise directions. This allowed the pigments on the paint

surface to be removed in a reproducible manner. Periodically,

the clamp securing the cotton bud was adjusted to ensure that

the mass remained within 10% of the desired value. Some

fluctuations were inevitable, since the cotton bud would com-

press somewhat as more force was exerted on it from the

painted surface.

Fig. 1 summarizes the relationship between contact force of

the cotton bud and the mass of pigment metal extracted for

different pigments. Since a fresh paint surface was sampled for

each determination, error bars on the vertical axis represent

the measurement uncertainty of the instrument (i.e. � 10%).

The absence of a good fit of the continuous curve in these

experiments can probably be attributed to variations in the

paint surfaces between samples, as well as between different

cotton buds. In Fig. 1(a) the relationship between the mass of

pigment and contact force is shown for the watercolour paint

‘olive green’. This paint contains chromium oxide, Cr2O3, as

the pigment and therefore, the metal used to estimate the

amount of pigment sampled was Cr. This pigment dried to a

smooth, hard finish, which meant that smaller amounts of

pigment were removed when contacted with the cotton bud,

especially when compared with the oil paint. For contrasting

surface properties a commercial oil paint ‘flake white’ was

chosen, which is shown in Fig. 1(b). This paint contains basic

lead carbonate, 2PbCO3 �Pb(OH)2, as the pigment and there-

fore the metal used to estimate the quantity of pigment

sampled was Pb. This paint required a longer period of time

(47 days) to dry because of the linseed oil. Compared to the

watercolour paint, the oil paint surface was much rougher and

more susceptible to damage when contacted with the cotton

bud. As a result, much higher masses of Pb were obtained,

even when gentle pressure (o0.5 N) was applied to the paint

surface.

Although this experiment differs in many ways from the

manual approach to pigment sampling of paint, it is useful for

understanding how the surface of a painting may be affected.

Recording the mass with a pan balance made it possible to

estimate the amount of pressure applied to the paint surface.

The amount of pressure applied is certainly subjective, how-

ever it is possible to make an estimate of the pressure exerted

on the surface of a painting. With a microscope slide on the

pan balance, for comparison, the estimated force on the paint

surface is approximately 0.05–0.1 N of force per 0.25 cm2. For

most paints this amount of force will have a negligible effect on

the surface. A larger amount of force (40.3 N) can cause a

dent to the paint surface, which is both harmful to the

appearance of the paint and unnecessary for obtaining a

representative paint sample.

The first object that was examined in this study was the

painting ‘untitled—Arctic Spring’ (1950) by George Swinton.

This abstract painting is a watercolour and it has been

described as a ‘‘hopeful presentation of springtime in the

Canadian Arctic’’. A colour image of this painting is available

as Electronic Supplementary Information (ESI)w. Swinton

(1917–2002) was a respected Canadian artist and collector of

Inuit artwork in the province of Manitoba, Canada. The

problem with ‘Arctic Spring’ is that it has darkened noticeably

since it was created. Areas of paint that were originally white

in appearance have become either grey or dark grey over time.

Therefore, identifying the white pigments in this painting is a

logical first step towards understanding why darkening has

occurred.

Fig. 1 (a) Relationship between the mass of Cr extracted from the

pigment ‘olive green’ and the applied force of the cotton bud. (b)

Relationship between the mass of Pb extracted from the pigment ‘flake

white’ and the applied force of the cotton bud.

142 | J. Anal. At. Spectrom., 2007, 22, 140–146 This journal is �c The Royal Society of Chemistry 2007

A number of areas in ‘Arctic Spring’ were sampled using a

cotton bud, with the goal of detecting the pigment metals for

the purpose of identifying both the white pigment and possibly

the cause of the darkening of the paint in this piece. Fig. 2(a)

shows transient signals obtained from a graphite furnace

atomic absorption spectrophotometer (GFAAS). This can be

a time-consuming approach because the signals for each metal

must be collected separately. Absorbance signals for a number

of metals that are typically associated with white pigments

were measured, including: Ti, Ba, Zn, Pb and Ca. Of these, the

largest signals were for Ba and Zn, which indicates that the

important white pigment in this piece is probably lithopone

(BaSO4, ZnS).

A second approach that was explored for metal identifica-

tion was electrothermal vaporization (ETV) inductively

coupled plasma mass spectrometry (ICP-MS). With this tech-

nique, the sample is vaporized in much the same manner as

GFAAS. The vaporized pigment metals are ionized in the ICP

prior to entering the mass spectrometer. The advantage of

using this approach is that multiple signals can be acquired in

a single firing of the furnace. Although we typically monitored

6–7 masses per ETV firing, other workers have shown that as

many as 20–25 masses can be measured with quadrupole based

instruments.24,25

Fig. 2(b) shows the transient signals of metals from a

darkened area of ‘Arctic Spring’. As with the GFAAS deter-

minations, the presence of large amounts of 64Zn+ and 138Ba+

were detected using ICP-MS, again indicating that the dar-

kened pigment is lithopone. With the ETV-ICP-MS, trace

amounts of other metals such as Cd and Ti could be detected.

Surprisingly large signals for Pb were also observed using both

GFAAS and ETV-ICP-MS. The higher sensitivity with which208Pb+ can be detected using ETV-ICP-MS is also illustrated

in the figure.

Solution nebulization SN-ICP-MS is extremely useful for

obtaining information regarding the elemental abundance for

most of the elements in the periodic table. For example, a

more complete estimation of the elemental distribution of

metals in different pigments is easily achieved by performing

a mass scan. This can provide useful information on both

pigment metals and other metals which may be present either

as impurities or as a result of mixing two or more pigments.

Fig. 3 shows the elemental composition of different areas

(dark, light, paper) of ‘Arctic Spring’. Concentrations of each

metal were quantified with an external calibration curve.

Compositions are expressed as a mole % of the detected

elements only, and therefore the actual concentrations of

metals in the pigment would be lower than those shown in

Fig. 3. All metals were background-subtracted from blanks to

correct for traces of metals that were present in the cotton

buds. The darkest areas of lithopone are labelled as ‘dark’ and

areas that had darkened, but were lighter in appearance are

labelled as ‘light’. For comparison, a small area on the

painting that was white in appearance is also shown as ‘paper’

in this figure. This particular area did not appear to have a

visible layer of paint on the surface, except possibly a thin

ground layer of white paint. In Fig. 3 the presence of large

amounts of 64Zn+ and 138Ba+ were detected in all areas of the

darkened white pigment, which is consistent with the pigment

lithopone. The proportion of Ba:Zn was approximately 2:1 in

each of the sampled areas. Low concentrations of 114Cd+,88Sr+ and 208Pb+ were also detected in the lithopone, possibly

as impurities. The small area that was white in appearance

(‘paper’) contained significant amounts of 44Ca+ and much

less 64Zn+ or 138Ba+, which may indicate that a Ca-containing

pigment (e.g. CaCO3) was used as a ground layer.

Fig. 2 (a) GFAAS signals of pigment extracted from darkened area

of ‘Arctic Spring’. (b) ETV-ICP-MS signals of pigment extracted from

darkened area of ‘Arctic Spring’.

Fig. 3 Mass scan of white and darkened areas of Swinton’s ‘Arctic

Spring’.

This journal is �c The Royal Society of Chemistry 2007 J. Anal. At. Spectrom., 2007, 22, 140–146 | 143

The questionable permanence of lithopone has been known

for a long time and its problems as a modern pigment are well

documented.26,27 Exposure to sunlight and atmospheric levels

of oxygen are known to darken this pigment. The darkening of

lithopone is attributed to the photochemical reduction of the

ZnS to elemental Zn, which results in a darkened appearance

on the paint surface. Lithopone was originally developed

primarily as a pigment for household paint. Early in its

history, manufacturers became aware of this problem, as well

as its reversibility when stored in the dark. A number of

approaches have been developed to prevent the darkening,

such as adding magnesium or even coating it with varnish.28

Apart from positively identifying the darkened pigment, an

interesting outcome of this work is the identification of other

metals such as Pb and Fe in the lithopone. The concentrations

of Pb and Fe were further quantified using GFAAS, to

determine if the concentrations of these metals varied with

the extent of darkening. The concentrations of both Pb and Fe

were found to be approximately the same in areas of the

painting that were darker as shown in Table 3. In this

experiment, the concentrations of the metals were quantified

using the method of standard addition. Because of their low

concentrations relative to Ba and Zn, it would be difficult to

suggest that Pb or Fe would contribute significantly to the

darkening of the lithopone. At present, we have no experi-

mental evidence to confirm why the lithopone is darkened in

‘Arctic Spring’ other than the photochemical reduction of Zn.

The ‘Mapp or Generall Carte of the World’ (circa 1669)

provided a similar challenge in terms of pigment identification.

A colour image of this map is also available as Electronic

Supplementary Information (ESI)w. This map of the world is

part of the holdings in the Hudson Bay Archives (Accession

Number: 1987/199) and is housed in the Archives of Manitoba

(Winnipeg, Canada). This map is an engraved map that was

executed in black ink on laid paper during the time of King

Charles II (1660–1685). It is a map of the world, drawn in ink

by Monsieur Sanson as two hemispheres. It was rendered into

English and illustrated with figures by Richard Blome. As

such, some portions of the map are coloured in blue, yellow

and red pigments. The problem of darkening of the red

pigments became apparent while the map was undergoing

cleaning by a conservator. Pigments that were initially red in

appearance, such as the decorative flowers, appear to have

darkened to a brown colour. In some places the red appear-

ance of this pigment is still evident, but most areas containing

the red pigment have a darkened appearance, to some degree.

Solution nebulization-ICP-MS was used to obtain mass scans

of the blue and red pigments found in the map. Electrothermal

vaporization-ICP-MS was again used to detect a number of

metals in the blue pigment of the oval-shaped seal, as well as in

the red pigment used to colour the decorative flowers. Tran-

sient signals for the most important metals detected are shown

in Fig. 4(a). This figure shows the ETV-ICP-MS signals

obtained from the red pigment. Other than trace amounts of52Cr+, the element that was in the highest concentration was208Pb+. In this experiment, the large 208Pb+ signal, relative

to the blank, clearly indicates that this is a Pb-containing

pigment. Fig. 4(b) shows transient signals obtained from

ETV-ICP-MS for the blue-coloured pigment. It is apparent

from this experiment that relatively high concentrations of63Cu+ were detected, as well as other metals including 138Ba+,52Cr+ and 75As+. The transient signals obtained using ETV-

ICP-MS are useful for identifying the metals present: however,

quantifying the relative concentration of metals is more useful

for pigment identification.

The quantitative results of the metals in the blue and red

pigments are summarized in Fig. 5. Analysis of these pigments

was carried out in the same manner as the pigments in Arctic

Spring (Fig. 3). The relative purity of the pigments used to

colour portions of this map becomes evident, by the absence of

metals other than the pigment metals. Apart from some degree

of fading with the blue pigment, the colour of this pigment is

still apparent and has not degraded. The relative absence of Cr

and the high concentration of Cu, suggests that this pigment is

most likely to be copper carbonate. The basic form of copper

carbonate, CuCO3 �Cu(OH)2 is a pigment obtained from the

mineral Azurite.

Table 3 Summary of quantitative analysis of selected elements in‘Arctic Spring’

Concentration/nM

Pb Ba Zn Fe

Dark grey 0.04 � 0.01 6.77 � 3.77 2.61 � 0.45 0.18 � 0.02Light grey 0.06 � 0.06 7.77 � 10.1 3.22 � 2.68 0.28 � 0.12

Note: � denotes the average deviation (n = 3).

Fig. 4 (a) ETV-ICP-MS signals of darkened red pigment in the

‘Generall Carte of the World’. (b) ETV-ICP-MS signals of blue

pigment in the ‘Generall Carte of the World’.

144 | J. Anal. At. Spectrom., 2007, 22, 140–146 This journal is �c The Royal Society of Chemistry 2007

The absence of Fe or Hg indicates that neither Red Ochre

(Fe2O3) nor Cinnabar (HgS) were used in this object. Since it

had a very high concentration of Pb, the most likely identity

of the red pigment could be Chrome Red (PbO �PbCrO4),

Molybdate Red (7PbCrO4 � 2PbSO4 �PbMnO4) or Red Lead

(Pb3O4). A portable Raman spectrometer was used to confirm

the structure of the darkened red pigment. This was necessary

because both Cr and Mn were detected in measurable quan-

tities with the ICP-MS. Fig. 6 shows a Raman spectrum of the

darkened red pigment. For comparison, a Raman spectrum of

pure Pb3O4 is also shown and indicates that the darkened

pigment is consistent with Red Lead. The absence of CrO42�

or MnO4� in the Raman spectrum indicates that Cr and Mn

are present as impurities in the darkened pigment.

Red lead is a pigment that is known to darken over time as a

result of oxidation of the Pb(II) atoms in Pb3O4 to PbO2, which

is dark brown in appearance.29 Platernite (PbO2) appears to

have been detected in wall paintings using X-ray diffraction:

however, a more recent study used the laser of Raman spectro-

scopy for identifying degradation products of lead pigments.30

This work presented a reasonable degradation pathway for

Red Lead, Pb3O4, to PbO (litharge, tetragonal), which will

ultimately form PbO (massicot, orthorhombic). Unfortunately

the detection of PbO2 in painted objects using Raman spectro-

scopy is difficult at best, since PbO2 is a weak Raman scatterer.

The most important advantage of using the cotton bud is

the relative ease with which a surface layer of paint is sampled,

and it works equally well when either oil or water is used as the

binding agent. This approach is limited to sampling only the

surface, and a protective layer of varnish would certainly

diminish the usefulness of microsampling with a cotton bud.

For older cultural objects that have not been stored properly,

dirt or soot accumulated on the surface would certainly affect

the elemental composition that is observed by the analyst. The

magnitude of the AAS or ICP-MS signals, obtained from the

metal-containing pigments, are usually large enough that small

amounts of dirt or soot should not interfere with pigment

identification. This may not be the case for metals that are

present in pigments in low concentrations, such as when

pigments are mixed with one another.

Conclusion

Sampling with a cotton bud allows the possibility of analyzing

very small quantities of paint, with minimal damage to the

surface of an object. When compared to solution nebulization

(SN), the small sample size requirement (1 mg or less) of

GFAAS and ETV-ICP-MS makes them very powerful tech-

niques for detecting small quantities of metals. In order to

perform a mass scan, samples had to be diluted to provide a

sufficient volume prior to analysis by solution nebulization. As

a result, sensitivity is sacrificed for some metals compared with

ETV-ICP-MS. In many ways the multi-technique approaches

used in this study were complimentary to each other. The most

important advantage of using SN-ICP-MS was obtaining a

survey of the elements present in the pigment samples quickly.

The GFAAS was useful for determining As, Fe and Cr, which

have spectral interferences with 40Ar16O+ and 40Ar12C+, and

can be problematic for the ETV-ICP-MS user. Whilst GFAAS

is capable of measuring only single elements, the most im-

portant advantage of ETV-ICP-MS is its capability to deter-

mine multiple elements simultaneously. The Raman

spectrometer was excellent for identifying the primary pigment

but lacked the sensitivity to determine if another pigment was

present as well. The high sensitivity of ETV-ICP-MS and

GFAAS, for most metals, allows the user to determine if more

than one pigment is present. Since only detecting the metals

present is often not sufficient for identifying a pigment, the use

of Raman spectroscopy can be essential for confirming the

chemical structure of the pigment.

Acknowledgements

Funding for this research was provided by the Natural

Sciences and Engineering Research Council of Canada. The

authors gratefully acknowledge the Winnipeg Art Gallery and

the Manitoba Archives for access to material. R. Ridgen and

A. Walton of the Manitoba Archive are also thanked for their

technical assistance. The National Research Council of

Fig. 5 Mass scan of blue and darkened red pigments in the ‘Generall

Carte of the World’.

Fig. 6 Raman spectrum of darkened pigment of the ‘Generall Carte

of the World’.

This journal is �c The Royal Society of Chemistry 2007 J. Anal. At. Spectrom., 2007, 22, 140–146 | 145

Canada is also thanked for their expertise and use of their

research facilities in Chemical Metrology. K. Charleton is a

graduate student at the University of Manitoba.

References

1 R. Newman, W. S. Taft, D. Stulik, J. W. Mayer and P. I.Kuniholm, The Science of Paintings, Springer-Verlag, New York,2000, ch. 1, p. 10.

2 E. Ciliberto and G. Spoto, in Modern Analytical Methods in Artand Archaeology, Chemical Analysis Series, ed. J. D. Winefordner,John Wiley and Sons, Inc., New York, vol. 155, 2000.

3 H. G. M. Edwards, D. W. Farwell and C. J. Brooke, Anal. Bioanal.Chem., 2005, 383, 312–321.

4 H. G. M. Edwards, J. Mol. Struct., 2003, 661, 271–277.5 R. Mazzeo, P. Baraldi, R. Lujan and C. Fagnano, J. RamanSpectrosc., 2004, 35, 678–685.

6 B. Gilbert, S. Denoel, G. Weber and D. Allart, Analyst, 2003, 28,1213–1217.

7 L. Burgio, R. J. H. Clark, L. Sheldon and G. D. Smith, Anal.Chem., 2005, 77, 1261–1267.

8 I. M. Bell, R. J. H. Clark and P. J. Gibbs, Spectrochim. Acta, PartA, 1997, 53A, 2159–2179.

9 L. Burgio and R. J. H. Clark, Spectrochim. Acta, Part A, 2001,57A, 1491–1521.

10 R. J. H. Clark, C. R. Chim., 2002, 5, 7–20.11 M. Castillejo, M. Martin, M. Oujja, D. Silva, R. Torres, C.

Domingo, J. V. Garcia-Ramos and S. Sanchez-Cortes, Appl.Spectrosc., 2001, 55, 992–998.

12 K. Melessanaki, V. Papadakis, C. Balas and D. Anglos, Spectro-chim. Acta, Part B, 2001, 56, 2337–2346.

13 L. Burgio, K. Melessanaki, M. Doulgeridis, R. J. H. Clark and D.Anglos, Spectrochim. Acta, Part B, 2001, 56, 905–913.

14 M. Bicchieri, M. Nardone, P. A. Russo, A. Sodo, M. Corsi, G.Cristoforetti, V. Palleschi, A. Salvetti and E. Tognoni, Spectro-chim. Acta, Part B, 2001, 56, 915–922.

15 L. Burgio, R. J. H. Clark, T. Stratoudaki, M. Doulgeridis and D.Anglos, Appl. Spectrosc., 2000, 54, 463–469.

16 D. Anglos, K. Melesanaki, V. Zafiropulos, M. J. Gresalfi and J. C.Miller, Appl. Spectrosc., 2002, 56, 423–432.

17 D. Anglos, S. Couris and C. Fotakis, Appl. Spectrosc., 1997, 51,1025–1030.

18 G. Van Hooydonk, M. De Reu, L. Moens, J. Van Aelst and L.Milis, Eur. J. Inorg. Chem., 1998, 639–644.

19 N. Civici, O. Demko and R. J. H. Clark, J. Cult. Herit., 2005, 6,157–164.

20 L. Moens, W. Devos, R. Klockenkamper and A. Vonbohlen, J.Trace Microprobe Tech., 1995, 13, 119–139.

21 N. Civici, O. Demko and R. J. H. Clark, J. Cult. Herit., 2005, 6,157–164.

22 W. Devos, L. Moens, A. von Bohlen and R. Klockenkamper, Stud.Conserv., 1995, 40, 163–162.

23 D. M. Goltz, J. Coombs, C. Marion, E. Cloutis, J. Gibson, M.Attas, L. P. Choo-Smith and C. Collins, Talanta, 2004, 63, 609–616.

24 J. D. Venable, D. Langer and J. A. Holcombe, Anal. Chem., 2002,74, 3744–3753.

25 M. Resano, M. Verstraete, F. Vanhaecke and L. Moens, J. Anal.At. Spectrom., 2001, 16, 1018–1027.

26 E. Maass and R. Kempf, Angew. Chem., 1923, 36, 293–7.27 J. H. Goshorn and C. K. Black, Ind. Eng. Chem., 1929, 21, 348–9.28 A. Eibner, Chem. Ztg., 1923, 47, 13–6.29 R. J. Gettens, H. Kuhn andW. T. Chase, ‘‘Lead White’’, in Artists’

Pigments—A Handbook of Their History and Characteristics,ed. A. Roy, Oxford University Press, New York, 1993, vol. 2,ch. 3, p. 367.

30 L. Burgio, R. J. H. Clark and S. Firth, Analyst, 2001, 126, 222–227.

146 | J. Anal. At. Spectrom., 2007, 22, 140–146 This journal is �c The Royal Society of Chemistry 2007