Content uploaded by Jaap J. Boon
Author content
All content in this area was uploaded by Jaap J. Boon on Jan 02, 2016
Content may be subject to copyright.
37
AICCM Bulletin Vol 29, 2005
Introduction
A serious deterioration phenomenon in lead-
pigmented paint layers has been recognised in an
increasing number of paintings worldwide. Research
has developed in response to surface defects identi-
fied in paintings spanning the 13th–20th centuries.
Most published material to date has concentrated on
the phenomenology and mechanistic aspects of lead
soap formation in old master paintings. Rembrandt’s
The anatomy lesson of Dr Nicolaes Tulp
was among
the first to be studied because of the disfiguring
lumps and craters apparent across its surface (Noble
et al 1998). Boon et al (2002) discussed the phe-
nomenon of lead soaps at the International Council
of Museums Conservation Committee (ICOM-CC)
Triennial Meeting in Rio de Janeiro and a detailed
questionnaire for international distribution was pre-
pared by Noble [1]. Higgitt et al (2003) comprehen-
sively discussed soap formation in paintings where
red lead and lead-tin yellow are the cause of dete-
rioration. Boon et al (2004) propose a mechanism
of lead soap aggregation in lead-tin yellow paints.
A comparable process of deterioration has also
been identified in zinc-containing lead chromate
paint layers in paintings by Vincent van Gogh (van
der Weerd et al 2003, Keune 2005). These studies
prompted the investigation of an Australian painting
by R. Godfrey Rivers, which exhibits similar phe-
nomena related to zinc-containing paint.
Metal soaps, metal salts of fatty acids, are com-
monly formed as part of the normal ageing of oil
paint. Depending on the paint composition, fatty acids
present in the oil might form metal carboxylates with
pigments or other metal containing components (van
den Berg 2002). An excess of fatty acids in paint can
lead to aggregation or efflorescence. Palmitate and
stearic acids are two characteristic fatty acids in oil
paint which appear to preferentially separate from
the oil network and are most frequently associated
with surface ‘bloom’ (eg. Williams 1988, Koller and
Burmester 1990, Ordonez and Twilley 1997, Rimer et
al 1999). Palmitic and stearate are similarly significant
in soap aggregation
(
Higgitt et al 2003
(Higgitt et al 2003(
)
Higgitt et al 2003)Higgitt et al 2003
.
Inorganic material is also often observed at the
centre of soap masses, sometimes with a structure
suggesting formation via precipitation reactions
(Boon et al 2002, Higgitt et al 2003). Studies into lead
soaps show this material varies according to paint
composition, with lead carbonate the most common
mineral phase (Boon et al 2002, Higgitt et al 2003).
Zinc oxide was postulated in soap formations in
Van Gogh’s
Falling leaves (Les Alyscamps)
(van der
Weerd 2003), but subsequent research has identified
zinc carbonate as the mineral fraction (Keune 2005).
In industry it is known that zinc white reacts readily
with fatty acids in drying oils and the soaps which
A study of zinc soap aggregates in a late
A study of zinc soap aggregates in a late
19th century painting by R.G. Rivers at
19th century painting by R.G. Rivers at
the Queensland Art Gallery
the Queensland Art Gallery
Gillian Osmond
a*
, Katrien Keune
, Katrien Keune
b
and Jaap Boon
c
a Conservation Department, Queensland Art Gallery, PO Box 3686, South Brisbane, Qld 4101
a Conservation Department, Queensland Art Gallery, PO Box 3686, South Brisbane, Qld 4101
b Molecular Paintings Research Group, University of Amsterdam
b Molecular Paintings Research Group, University of Amsterdam
c Department of Analytical Mass Spectrometry, University of Amsterdam
c Department of Analytical Mass Spectrometry, University of Amsterdam
*Corresponding author email address <gillian.osmond@qag.qld.gov.au>
*Corresponding author email address <gillian.osmond@qag.qld.gov.au>
Abstract
Abstract
Metal soap formation in paintings has been implicated in a serious deterioration phenomenon. The present
Metal soap formation in paintings has been implicated in a serious deterioration phenomenon. The present
study documents zinc soap aggregates observed in a late 19th
study documents zinc soap aggregates observed in a late 19th
century painting by R. Godfrey Rivers. Optical
microscopy and scanning electron microscopy with energy dispersive X-ray analysis of paint cross-sections are
microscopy and scanning electron microscopy with energy dispersive X-ray analysis of paint cross-sections are
used to describe the appearance and elemental composition of affected paint layers. Fourier Transform Infrared
used to describe the appearance and elemental composition of affected paint layers. Fourier Transform Infrared
Spectroscopy (FTIR) is used to confirm the presence of zinc carboxylates and static secondary ion mass spec-
Spectroscopy (FTIR) is used to confirm the presence of zinc carboxylates and static secondary ion mass spec-
trometry (SIMS) results are given for one sample. Scanning electron microscopy energy dispersive X-ray (SEM-
trometry (SIMS) results are given for one sample. Scanning electron microscopy energy dispersive X-ray (SEM-
EDX) maps and spot analyses are used to examine aggregates in detail. In addition to zinc, carbon and oxygen,
EDX) maps and spot analyses are used to examine aggregates in detail. In addition to zinc, carbon and oxygen,
magnesium is frequently present. Paint composition and environment are discussed in terms of their potential to
magnesium is frequently present. Paint composition and environment are discussed in terms of their potential to
influence soap formation.
38
AICCM Bulletin Vol 29, 2005
ied at London’s Slade School from 1877 to 1883. He
arrived in Australia in 1889 and in 1891 moved to
Brisbane to become Art Master at Brisbane School
of the Arts. In 1895, he became the Queensland Art
Gallery’s inaugural Curator. Despite being a ‘Slade’
man, Rivers maintained a strong association with
the Royal Academy and its more traditional paint-
ers. During return visits to England, he visited the
studios of many Academicians, including Frederick
Leighton (Strumpf 1996).
Woolshed, NSW
is conventionally painted on a
Woolshed, NSW is conventionally painted on a Woolshed, NSW
commercially primed canvas. It portrays the inte-
rior of a sheep-shearing shed, with glimpses of the
Australian landscape beyond (Figure 1). This paint-
ing was acquired by the Gallery within a few years
of being painted and showed little evidence of con-
servation intervention prior to its cleaning and var-
nish removal in the early 1990s. Fine, mechanical
cracking is widespread. Localised drying cracks are
also apparent. Under magnification, small, translu-
cent spots are distributed through some passages of
paint. In lighter paint these inclusions appear dark,
sometimes with a defined centre (Figure 2). In dark-
er (brown) paints, they give the surface a globular,
form contribute to the characteristic hardness of zinc
paints (Kuhn 1986). So it is not simply the formation
paints (Kuhn 1986). So it is not simply the formation
of soaps which leads to deterioration, but more the
extent to which soaps form and the way they aggre-
gate and ultimately distort the surrounding paint.
Investigation of paintings at the
Queensland Art Gallery
A significant number of British and Australian
paintings from the late 19th and early 20th century in
paintings from the late 19th and early 20th century in
the Queensland Art Gallery collection are suspected
to contain zinc soap aggregates based on light micro-
scopic examination of paint cross-sections. An initial
survey of archived cross-sections from fourteen
paintings of this general oeuvre has been undertaken.
paintings of this general oeuvre has been undertaken.
Table 1 details nine that appear to contain soaps; all
are catalogued as oil on canvas; however, no medium
analysis has been undertaken.
The present study describes analytical imaging
techniques used to examine the appearance and
composition of zinc soap inclusions found in one of
the listed paintings,
Woolshed, NSW
(1890) by R.
Woolshed, NSW (1890) by R. Woolshed, NSW
Godfrey Rivers.
R. Godfrey Rivers was born in England and stud-
Figure 1. R Godfrey Rivers,
Woolshed, New South Wales
1890. Gift of the artist 1895. Numbers denote sites from which
1890. Gift of the artist 1895. Numbers denote sites from which
1890. Gift of the artist 1895. Numbers denote sites from which
paint samples derive.
39
AICCM Bulletin Vol 29, 2005
bubbly texture. Soap formation has subtly altered
the appearance of the painting.
Sample preparation and examination
techniques
Three paint samples from Rivers,
Woolshed, NSW
(1890) were selected for detailed study to charac-
terise their inclusions. RWS5 was sampled from
blue-grey paint in the iron roofing of the shearing
shed, RWS4 contains yellow green paint from the
background landscape and RWS3 is from a shearer’s
dark brown trouser (Figure 1).
Paint samples were embedded in a polyester resin,
and microtomed with a tungsten blade to expose a
cross-section. Final (dry) polishing where neces-
sary was undertaken with Micro- Mesh™ cloth-
backed abrasive sheet (12 000 grit). Cross-sections
were documented using optical microscopy with
visible and ultraviolet (UV) light sources. RWS3
and 4 were then carbon coated to enable scanning
electron microscopy (SEM) imaging with energy
dispersive X-ray (EDX) analysis at the University
of Queensland Centre for Microscopy and
Microanalysis, using either a Philips LaB6 XL30 [2]
or JEOL JSM-6460 LA [3] scanning electron micro-
scope. Samples were examined in high vacuum
mode at 20 kV accelerating voltage. Points of inter-
est on the specimen were acquired in spot mode.
Larger regions of the sample were full spectrum X-
ray mapped to elucidate elemental distribution.
RWS5 was examined separately at the Foundation
for Fundamental Research on Matter, Institute for
Atomic and Molecular Physics (FOM-AMOLF).
Preliminary data have been summarised in the PhD
thesis of Keune (2005). In addition to SEM-EDX,
FTIR imaging was undertaken using a Bio-Rad
Stingray [4], and static-SIMS experiments were per-
formed on a Physical Electronics TRIFT-II time-of-
flight secondary ion mass spectrometer (TOF-SIMS)
[5]. In all samples, optical and fluorescence charac-
teristics in cross-section and elemental composition
have been used to infer the likely presence of pig-
ments. Organic pigments have not been identified
due to the limitations of SEM-EDX.
Table 1. Paintings with zinc soap inclusions
Painting
RIVERS, R. Godfrey
England/Australia b.1859 d.1925
Woolshed, New South Wales
1890
Gift of the artist 1895
RIVERS, R. Godfrey
England/Australia b.1859 d.1925
Under the jacaranda
1903
Purchased 1903
RIVERS, R. Godfrey
England/Australia b.1859 d.1925
An alien in Australia
1904, reworked later
Gift of the Godfrey Rivers Trust 1940
HOBDAY, Percy Stanhope
Australia b.1879 d.1951
The western sky
c.1909
The western sky c.1909The western sky
Purchased 1909
FOX, E. Phillips
Australia/France b.1865 d.1915
Bathing hour (L’Heure de Bain)
c.1909
Purchased 1946
LISTER, W. Lister
Australia b.1859 d.1943
Sydney Harbour, overlooking Taylor’s Bay
c.1912
Sydney Harbour, overlooking Taylor’s Bay c.1912Sydney Harbour, overlooking Taylor’s Bay
Purchased 1949
HEMY, C. Napier
England b.1841 d.1917
The home wind
1901
The home wind 1901The home wind
Purchased 1903
CALDERON, W. Frank
England b.1865 d.1943
Crest of the hill
1898
Crest of the hill 1898Crest of the hill
Purchased 1899
LISTER, W. Lister
Australia b.1859 d.1943
Sea breeze
AGNSW collection
Results
Paint visible in cross-section is characteristic of
commercially prepared paint with typically small
and uniform particle size. The three samples have
complex pigmentation and distinctive inclusions are
distributed through certain layers of paint. Consistent
with the results described previously by others,
inclusions comprise a gel-like phase quite distinct
from the granularity of surrounding paint. The gel is
highly UV-fluorescent. Backscatter electron images
(BSI) are dramatic because of contrast between the
elements of high atomic mass dominant in the paint
Figure 2. Surface detail from passage of flesh (photo-
graphed at 50× magnification). Soap aggregates appear
as dark spots or small craters with defined centres.
40
AICCM Bulletin Vol 29, 2005
and the organic nature of the soap phase. Inorganic
centres within soapy inclusions are clearly resolved
in the SEM. EDX mapping illustrates the alignment
of elements within samples and allows more confi-
dent pigment attribution. Subtle variations in size and
distribution of pigment particles between layers are
also revealed; the significance of these differences is
unknown and beyond the scope of the present study.
RWS5
A double ground is evident in cross-sections taken
from
Woolshed
. The lower layer contains calcium
Woolshed. The lower layer contains calcium Woolshed
carbonate with a small amount of lead white. The
upper ground is consistent with lead white and a
minor chalk component. Three paint layers can be
discerned: a light brown paint (layer 2), a blue-grey
layer (3), and a purple-grey layer (4) (Figure 3). Iron
and earth-based pigments are present in all paint lay-
ers. Cobalt blue is evident in layers 2 and 4. Pigment
with fluorescence typical of madder is prevalent in
layer 4. All paint layers appear to be mixtures with
white. Two areas of unmixed lead white paint can
be discerned in layer 2 (carbonates are confirmed
Figure 4. Rivers WS5 ultraviolet fluorescence image.
Soap centres appear dark with a fluorescent surround.
Figure 3. Rivers WS5 visible light microscopic image with
outline denoting layers. Soap aggregates are concen-
trated in layer 4 and appear as dark, transparent voids,
many with defined, refractive centres.
Figure 5. Rivers WS5 backscatter electron image.
Soaps are most prevalent in the upper paint layer (4),
visible as rounded dark regions with brighter centres.
with FTIR). A few isolated spots of bright UV-fluo-
rescence in layer 4 suggest zinc oxide is also present.
Lead, zinc, carbon, oxygen, aluminium, silicon, mag-
nesium, and chlorine are widely detected with EDX
analysis of the paint matrix in each layer.
Layers 3 and 4 are affected by many small aggre-
gates (approximately 10 µm) with a greenish core
and a transparent, UV-fluorescent halo (Figure 4).
The small aggregates are finely distributed within
the paint matrix of both layers, although are more
prevalent and slightly larger in layer 4. The aggre-
gates are clearly visible in BSI; all aggregates consist
of a brighter centre and a dark rim (Figure 5). EDX
analysis shows only the presence of carbon, oxygen,
zinc, and magnesium (low) inside the aggregates.
Zinc is predominant in the core of the aggregate,
while carbon is predominant in the rim. A very small
amount of magnesium is found inside the aggregate.
FTIR imaging of RWS5 detects metal soaps at
1542 cm
-1
in paint layers 3 and 4 (Figure 6). The
dark circular areas observed in layers 3 and 4 in BSI
are interpreted as zinc soap aggregates. Although
the aggregates are too small in size to identify them
Figure 6. Rivers WS5 FTIR image. Zinc carboxylate distri-
bution is illustrated by absorbance intensity at 1542cm
-1
(peaking in layer 4).
41
AICCM Bulletin Vol 29, 2005
with imaging FTIR (spatial resolution limit approxi-
mately 7 µm), the position of the asymmetric metal
carboxylate vibration peak is very informative. The
position of this absorption peak is dependent on the
type of metal attached. The peak at 1542 cm
-1
in the
spectrum of layers 3 and 4 is indicative for zinc car-
boxylates (Robinet and Corbeil 2003).
The paint cross-section was also analysed with
static-SIMS. Figure 7 shows the positive secondary
ion images of lead (
m/z
208) and zinc (
m/z 208) and zinc (m/z
m/z
65), and
m/z 65), and m/z
the negative ion image of deprotonated palmitic acid
(
m/z
255). The deprotonated palmitic acid ion image
m/z 255). The deprotonated palmitic acid ion image m/z
represents the distribution of oil in the paint. Lead
is present in all the layers, but absent in areas cor-
responding to the aggregates in layers 3 and 4. Zinc
is present in the aggregates (mainly in layer 4) and in
lower intensity in the other parts of layer 4. Palmitic
acid observed with SIMS is present in all four paint
layers, but the intensities are too low to correlate its
distribution to the light microscopic or backscatter
images. The P/S ratios determined with SIMS give a
reliable indication to distinguish the types of oil in the
paint (Keune et al 2005). The P/S ratio of layer 1 is
1.7, layer 3 is 2 and layer 4 is 3. The ratio of palmitic
to stearic acid varies between layers, indicating differ-
ences in the types of oil or possibly mixtures of oils
in these layers (Mills and White 1994).
RWS4
Ground layers are consistent with RWS5. Two
paint layers are present: light purple underlying a
yellow-green. Pigments considered present in both
layers include green earth, barium chromate, and
a fluorescent aluminium-based red assumed to be
madder. The lower layer also contains cobalt blue
and a second red pigment. Colours in both layers are
mixtures with white.
Characteristic features of soap aggregation are
present in both layers. UV-fluorescent masses are
concentrated in the lower layer (Figure 8). Inclusions
in the top layer appear darker with a narrow fluo-
rescent halo. BSI (Figure 9) distinguishes large
regions dominated by elements of low atomic mass
in the bottom layer, whereas the top has smaller dark
regions, many with bright centres; zinc is concentrat-
ed in these areas (Figure 10). Cobalt and magnesium
occur in alignment with zinc concentrations. Lead
dominates in the surrounding matrix. Carbon concen-
tration is low in the top layer, higher in the bottom.
RWS3
Ground layers are consistent with RWS5. Paint
appears very dark with a translucent central band;
blue, red, and brown pigmentation was observed
at the time of sampling. UV fluorescence distin-
Figure 7. Rivers WS5 SIMS images representing lead
(+;m/z 208), zinc (+;m/z 65) and deprotonated palmitic
acid (-;m/z 255). The three palmitic/stearic acid ratios (P:
S) representative for layer 1 (P:S 1.7), layer 3 (P:S 2) and
layer 4 (P:S 3) are shown in the ion image of deproton-
ated palmitic acid.
42
AICCM Bulletin Vol 29, 2005
Figure 10. Rivers WS4 (detail) Backscatter electron image
(top) and false colour zinc map (bottom) show correla-
tion between soaps and zinc concentration. Zinc is most
abundant at the centre of inclusions. Box denotes region
detailed in Figures 14 and 15.
Figure 8. Rivers WS4 ultraviolet fluorescence microscopic
image. Sample consists of (double) ground and two paint
layers. Soaps are present in both paint layers with those
in the lower layer visible as round, fluorescent masses.
Box denotes region detailed in Figure 10.
Figure 9. Rivers WS4 backscatter electron image. Soaps in
the lower paint layer are visible as round, dark regions.
Soaps in the top layer have defined centres.
Box denotes region detailed in Figure 10.
guishes three ‘layers’, the central band containing a
mass of fluorescent soap (Figure 11). Madder, ver-
milion, cobalt blue and green earth occur through-
out the paint.
Paint layers appear dark and have low contrast
in BSI (Figure 12), indicating a predominance of
elements of low atomic mass. Zinc corresponds
with soap aggregation through the central band
of the sample (Figure 13). Cobalt is not observed
within fluorescent masses. Although the colour of
the sample and BSI suggest little use of lead white,
mapping indicates some lead is present through the
central layer, in the regions not dominated by zinc.
Magnesium follows a pattern similar to lead. Carbon
is closely aligned with zinc.
Inclusions
BSI is particularly useful for distinguishing fea-
tures of samples at high magnification. Variations in
image density across small areas confirm the com-
plexity of inclusions.
Areas with characteristic soap aggregation were
selected from samples for more detailed study using
SEM-EDX. The centres of a variety of inclusions
were analysed in spot mode.
The defined centres contained in soaps from
samples RWS4 and RWS5 comprise mostly zinc,
oxygen, and carbon together with minor components
from the surrounding paint (Figure 14). Magnesium
is represented in every spectrum (including those
from RWS3) and is associated with higher oxygen
intensity (Figure 15). Spot analyses coupled with
stronger back scattering confirm a higher density
of zinc at soap centres, possibly in mineral form.
Clear zinc-based crystals have been identified in
soap aggregates of another Australian painting in the
Queensland Art Gallery’s collection (Lister
Sydney
Harbour …
c 1912) (see Table 1) (Keune 2005).
Heavily mineralised aggregates examined by Keune
(2005) in Van Gogh’s painting,
Falling leaves
, were
demonstrated to be zinc carbonate crystals.
In
Woolshed
, the phenomenon of mineralisation
Woolshed, the phenomenon of mineralisation Woolshed
of soap aggregates is not uniformly present. Soap
aggregates in RWS3 have no discernible centre
and spot analyses consistently detected carbon in
amounts exceeding 70 atomic percent. Together with
zinc, this result appears consistent with the presence
43
AICCM Bulletin Vol 29, 2005
of zinc carboxylates identified in a similar sample
using FTIR [6]. In a number of cases, zinc is again
concentrated in the centre of the zinc soap aggregate
but crystal formation cannot be observed. The pres-
ence of an organic layer surrounding the zinc min-
eral may be difficult to discern at 20kV accelerating
voltage in BSI mode, but would make a significant
contribution to the X-ray signal. The process of min-
eralisation is typically underway in the Rivers paint-
ings and may become more pronounced over time.
Further investigation is required to confirm the com-
position of inclusions from this and other paintings.
Discussion
A variety of factors are likely to influence the
formation of soap aggregates in paint films and their
potential to cause defects. These include the inherent
reactivity of zinc in oil, the quality and source of the
zinc, the type and preparation of oil, the combina-
tion of components in the paint, and the environ-
mental conditions and treatment interventions the
painting might have experienced.
Soap aggregates documented here occur in lay-
ers with consistently high concentrations of zinc and
carbon. Carbon concentration might derive from paint
with a high oil to pigment ratio. Slow-drying pig-
ments, and pigments which require high percentages
of oil to produce a workable paint, such as madder,
cobalt blue, and zinc white, are also often present in
affected layers. High oil concentration is likely to pro-
duce paint with a significant mobile fraction; this may
be particularly pronounced where poorly drying pig-
ments or media are used. Mobile components in paint
films have been the subject of considerable research.
The initial cross-linked oil in paint has been found to
hydrolyse over a period of a century (van den Berg et
al 2001) leading to reactive free fatty acids. Normally,
sufficient metal ions are present to form metal car-
boxylates which stabilise the oil paint. However,
in some cases insufficient metal ions are present to
trap the acids and blooming may result (Koller and
Burmester 1990; Rimer et al 1999). A significant
mobile fraction in the paint similarly enhances oppor-
tunities for metal carboxylates to form and aggregate.
Van der Weerd et al (2003) further describes the
Figure 13. Rivers WS3 (detail). Backscatter electron image
(top) and false colour zinc map (bottom) show correla-
tion between soap formation and zinc concentration.
Zinc is most abundant through the central layer.
Figure 11. Rivers WS3 ultraviolet fluorescence microscop-
ic image. Sample consists of (double) ground and three
paint layers. Strongly fluorescent soaps are distributed
through the central layer. Their gel-like appearance con-
trasts with surrounding paint.
Figure 12. Rivers WS3 backscatter electron image. Soaps
aggregated through the central layer lack the granular-
ity of surrounding paint. Box denotes region detailed in
Figure 13.
44
AICCM Bulletin Vol 29, 2005
mobility of zinc carboxylates themselves, observed in
van Gogh’s
Falling leaves
. Zinc carboxylates present
in a sample from Lister,
Sydney Harbour…
(Table
1) were also clearly shown to be mobile when SEM
imaging of a prepared cross-section revealed soaps
emerging from the sample (Keune 2005).
Increased mobility of components in paint has also
been discussed in relation to temperature and relative
humidity (Ordonez and Twilley 1997; Rimer et al
1999). The Queensland Art Gallery is located in sub-
tropical Brisbane. The Gallery’s collection has only
been housed in purpose-built facilities since 1982.
Paintings from the earliest years have experienced
many hot, humid summers. A study of metal soap
formation in comparable paintings located in cooler
climates offers potential to explore this theory.
There is some speculation regarding likely sources
of zinc in the affected paintings. The paint in sample
RWS3 is dark and unlikely to contain significant
quantities of white. White paint present in RWS5
appears to be lead-based. Paint layers in RWS4 and
RWS5 are mixtures with white; both zinc and lead are
present but almost completely contained in separate
‘phases’. It is not known to what degree mobility
of components in the paint might have contributed
to this separation. The absence of zinc (and soap
aggregates) from some layers and entire samples from
the same paintings suggests zinc is most likely present
as a component in selected commercially prepared
paints. Kuhn (1986) reports detection of zinc white
in coloured paints more than whites, suggesting its
addition to various pigments by manufacturers as
a lightening agent. Carlyle (2001) refers to various
19th century sources describing the addition of zinc
compounds to manipulate the drying or handling
properties of oil paints. In addition to zinc oxide, pos-
sible zinc-containing compounds include lithopone, (a
mixture of barium sulphate and zinc sulphide) or white
vitriol (zinc sulphate and zinc stearate). Lithopone is
unlikely to be the zinc source in this painting as barium
was not detected with EDX in two of the three paint
samples (barium chromate was detected in RWS4). No
sulphur was detected in the EDX spectra of affected
layers either, so it is unlikely that white vitriol was
used. Zinc stearate could have been added to the paint,
but the ratios of palmitate to stearate detected in affect-
ed layers of RWS5 are normal and do not give any
indications for additional stearic acid. However, in the
19th century, zinc stearate was not a pure compound,
Figure 14. Rivers WS4 (detail). SEM-EDX spectra derived from points indicated in backscatter electron image show high
Figure 14. Rivers WS4 (detail). SEM-EDX spectra derived from points indicated in backscatter electron image show high
zinc intensity associated with soap aggregation. Carbon intensity is significantly higher in the perimeter of the aggre-
zinc intensity associated with soap aggregation. Carbon intensity is significantly higher in the perimeter of the aggre-
gate, consistent with organic (carboxylate) composition. Reduced carbon intensity towards the centre of the aggregate
gate, consistent with organic (carboxylate) composition. Reduced carbon intensity towards the centre of the aggregate
might indicate a degree of mineralisation.
Figure 15. Rivers WS4 (detail). SEM-EDX backscatter elec-
tron image and maps for zinc, magnesium and oxygen
suggest some correlation between magnesium and oxy-
gen associated with aggregates.
45
AICCM Bulletin Vol 29, 2005
but more likely a mixture of stearic and palmitic acids
(Mayer 1951). So zinc soaps as an addition to the paint
cannot be excluded. In any event, information on the
origin of the zinc is lost due to the chemical reactivity
of the various paint components.
An interesting feature of the soaps documented
here is the presence of magnesium. Trace amounts
of cobalt are also detected more commonly than not.
The form and source of these constituents has not
been established; however, magnesium carbonate
was a commonly used extender in nineteenth cen-
tury paint preparations [7]. Magnesium correlates
strongly with zinc where soaps contain defined
centres. In RWS3, where no centres are discernible,
magnesium is present in the soap-affected layer
but does not coincide with the highest zinc con-
centrations. The presence of magnesium and cobalt
together with zinc, carbon, and oxygen is a trend
also observed in preliminary SEM-EDX analysis of
inclusions in the majority of paintings listed in Table
1. Further research is required to determine the
composition of these regions and what, if any, sig-
nificance the minor components play in soap aggre-
gation. A systematic study of comparable paintings
from Australian collections would greatly assist
ongoing work to determine the causes and mecha-
nisms of soap formation. Comparison with zinc
containing paintings where soap aggregation is not
evident may also give insight into the phenomenon.
Conclusion
Paintings from the Queensland Art Gallery collec-
tion identified with zinc soap aggregation tell a con-
sistent story and one which corresponds well with
research undertaken by others into lead soap forma-
tion. This has been demonstrated by a detailed study
of paint samples from a representative painting by
R. Godfrey Rivers. Zinc, carbon, and oxygen were
reliably detected in affected regions and the pres-
ence of zinc carboxylates confirmed. Minor amounts
of magnesium and cobalt also regularly occur.
Aggregates have formed in layers with high carbon
concentration, indicative of paint with high oil con-
tent. This supports the theory that soap aggregation
occurs in paints with a high mobile component.
Slow-drying paints, pigments requiring a high ratio
of oil, and paintings which have experienced hot,
humid conditions, are all possible influences. The
documentation of paintings with metal soap forma-
tion is crucial to further understanding the phenom-
enon and to managing its long-term implications.
Notes
1. Petria Noble and Jaap Boon, Mauritshuis/AMOLF
collaboration.
2. Philips LaB6 XL30 high vacuum scanning electron
microscopy (SEM), equipped with EDAX brand,
super ultra thin window (SUTW), 138 eV nominal
resolution, SiLi crystal, Energy Dispersive X-ray
Spectrometer (EDX). EDAX Phoenix (V3.10)
software used to collect and analyse data.
3. JEOL JSM-6460 LA low vacuum analytical scan-
ning electron microscopy (SEM) equipped with
an integrated JEOL hyper mini-cup, 133eV reso-
lution, ultra thin window (UTW), SiLi crystal,
energy dispersive X-ray spectrometer (EDX).
Integrated JEOL Analysis Station (V3.2) software
used to collect and analyse data.
4. Bio-Rad Stingray (Bio-Rad, Cambridge, MA,
USA), combining the Bio-Rad FTS-6000
spectrometer equipped with a Bio-Rad UMA 500
infrared microscope with a 64 × 64 mercury-
cadmium telluride (MCT) focal plane array camera
was used to record the FTIR images. Analysis of
the embedded cross-section was carried out in
reflection mode recorded with a 16 cm
-1
spatial
resolution, a step scan frequency of 1 Hz, and
an UDR of 4. The reflection measurements were
corrected by the Kramers-Krönig transformation.
5. The static SIMS experiments were performed on
a Physical Electronics (Eden Prairie, MN, USA)
TRIFT-II time-of-flight SIMS (TOF-SIMS). The
surface of the sample was scanned with a 15 keV
primary ion beam from an
115
In
+
liquid metal ion
gun. The pulsed beam was non-bunched with a
pulse width of 20 ns, a current of 600 pA and the
spot size of ~120 nm. The surface of the sample
was charge compensated with electrons pulsed in
between the primary ion beam pulses. To prevent
large variations in the extraction field over the
large insulation surface area of the paint cross-
section a non-magnetic stainless steel plate with
slits (1 mm) was placed in top of the sample. The
paint cross-section was rinsed with hexane to
reduce contamination of polydimethyl siloxanes.
6. FTIR-imaging analysis of a sample from the
painting not included in this paper, RWS1 (FOM-
AMOLF). A peak at 1530 cm
-1
was assigned to the
presence of metal carboxylates. SEM/EDX analy-
sis confirmed the presence of zinc. The combined
results point to the presence of zinc soaps.
7. Winsor and Newton archive, Leslie Carlyle per-
sonal communication.
Acknowledgements
Special thanks to Ron Rasch, University of
Queensland Centre for Microscopy and Micro analysis,
for his generous assistance and considered advice.
Heleen Zuurendonk (AMOLF 2003–2004) under-
took FTIR analysis of RWS5.
46
AICCM Bulletin Vol 29, 2005
Leslie Carlyle (ICN) made the connection
between images of paint samples in Brisbane and
relevant MOLART research.
Deborah Lau (CSIRO) has made helpful com-
ments at various stages of the manuscript.
Thanks also to Conservation colleagues at the
Queensland Art Gallery, particularly Nicola Hall and
Alyssa Aleksanian for shared sessions on the scan-
ning electron microscope, and also Anne Carter and
John Hook, for advice, encouragement and support.
References
Boon, J. van der Weerd, J. Keune, K. Noble, P. and Wadum,
J. 2002, ‘Mechanical and chemical changes in Old
Master paintings: dissolution, metal soap formation and
remineralization processes in lead pigmented ground/
intermediate paint layers of 17
th
century paintings’,
th century paintings’,
th
Preprints of the ICOM Committee for Conservation 13
th
Triennial Meeting,
Rio
de Janeiro
,
ed. R Vontobel, James
and James, London, pp. 401–06.
Carlyle, L. 2001, ‘
The artist’s assistant: oil painting
instruction manuals and handbooks in Britain 1800-1900
with reference to selected eighteenth century sources’
,
Archetype Publications, London.
Higgitt, C. Spring, M, and Saunders, D. 2003, ‘Pigment-
medium interactions in oil paint films containing red lead
or lead-tin yellow’,
National Gallery Technical Bulletin
vol 24, pp. 75–95.
Keune, K. 2005, ‘
Binding medium, pigments and metal
soaps characterised and localised in paint cross-sec-
tions’
, MOLART 11, FOM-AMOLF, PhD Thesis,
University of Amsterdam, Amsterdam (http//www.amolf.
nl/publications/theses/).
Keune, K. Ferreira, E. and Boon, J.J. 2005,
‘Characterisation and localisation of the oil binding
medium in paint cross-sections using secondary ion mass
spectrometry’,
Preprints of the ICOM Committee for
Conservation 14
th
Triennial Meeting, The Hague
, James
and James, London, pp. 796-802
Koller, J. and Burmester, A. 1990, ‘Blanching of
unvarnished modern paintings: a case study on a painting
by Serge Poliakoff’,
Cleaning, retouching and coatings,
IIC Preprints of the contributions to the Brussels
Congress, 3-7 September 1990
, pp. 138-143.
Kühn, H. 1986, ‘Zinc white’,
Artists’ pigments
vol. 1, ed.
R.L. Feller, Cambridge University Press, Cambridge, pp.
169–186.
Mayer, R. 1951, ‘
The artist’s handbook of materials &
techniques’
, first edition, Faber and Faber Limited,
London.
Mills, J.S. and White, R. 1994, ‘
The organic chemistry
of museum objects’
, Butterworth Heinemann Limited,
London.
Noble, P, Wadum, J, Groen, K, Heeren, R, and van den
Berg, K.J. 2000, ‘Aspects of 17
th
century binding
th century binding
th
medium: inclusions in Rembrandt’s ‘Anatomy lesson of
Nicolaes Tulp’’,
Art et chimie, La couleur, Paris, Actes
du Congrès (1998)
, pp. 126–129.
Ordonez, E. and Twilley, J. 1997, ‘Clarifying the haze:
efflorescence of works of art’,
WAAC Newsletter
20 (1)
WAAC Newsletter 20 (1) WAAC Newsletter
1998 pp. 11 (reproduced from
Analytical Chemistry
vol.
Analytical Chemistry vol. Analytical Chemistry
69, 1997 pp. 416A–422A).
Rimer, B. Fiedler, I. Miller, M. Cunningham, M. and van
den Berg, J. 1999, ‘Investigation of fatty acid migration in
alizarin crimson oil paint in two works by Frank Stella’,
AIC Paintings Specialty Group Postprints 1999
, pp. 1–14.
Robinet, L. and Corbeil, M.C. 2003, ‘The characterisation
of metal soaps’,
Studies in Conservation,
48, pp. 23-40.
Strumpf, J. 1996,
Looking beyond biography;
interpretations of two works by Richard Godfrey Rivers.
BA (Honors) thesis, University of Queensland.
van den Berg, J.D.J. van den Berg, K.J. and Boon, J.J.
2001, ‘Determination of the degree of hydrolysis of oil
paint samples using a two-step derivatisation method and
on-column GC/MS’,
Progress in Organic Coatings
, vol.
41, pp. 143-155.
van den Berg, J.D.J. 2002,
Analytical chemical studies
on traditional linseed oil paints
, MOLART 6, FOM-
AMOLF, PhD Thesis, University of Amsterdam,
Amsterdam.
van der Weerd, J. 2002,
Microspectroscopic analysis of
traditional oil paint
, MOLART 7, FOM-AMOLF, PhD
Thesis, University of Amsterdam, Amsterdam.
Van der Weerd, J. Gelddof, M. van der Loeff, L.S. Heeren,
R. and Boon, J. 2003, ‘Zinc soap aggregate formation in
‘Falling leaves (Les Alyscamps)’ by Vincent van Gogh’,
Zeitschrift für Kunsttechnologie und Konservierung
vol.
Zeitschrift für Kunsttechnologie und Konservierung vol. Zeitschrift für Kunsttechnologie und Konservierung
17, no. 2, pp. 407–416.
Williams, R. Scott 1988, ‘Blooms, blushes, transferred
images and mouldy surfaces: what are these distracting
accretions on art works?’
Proceedings of the 14
th
annual
IIC-Canadian Group conference, Toronto, Canada, May
27-30, 1988
pp. 65–84.
Gillian Osmond graduated from the University
of Canberra (CCAE) with a Bachelor of Applied
Science, Conservation of Cultural Materials
(Paintings Conservation) in 1988. In 1991-92 she was
a research intern at the Tate Gallery, London. She is
currently Conservator, Paintings at the Queensland
Art Gallery, where she has worked since 1988.
Katrien Keune graduated in chemistry from the
University of Amsterdam (2000). She subsequently
joined the De Mayerne project team in the Molecular
joined the De Mayerne project team in the Molecular
Paintings Research group of Prof. J. J. Boon. She
completed her PhD thesis, titled ‘Binding medium,
pigments and metal soaps characterised and
pigments and metal soaps characterised and
localised in paint cross-sections’ at the University of
Amsterdam in 2005.
Jaap J. Boon was trained in geology and chem-
istry and was awarded a PhD from Delft Technical
University in 1978. He became research associate
at the FOM Institute for Atomic and Molecular
Physics in 1983 and coordinated the NWO Priority
project MOLART (Molecular aspects of age-
project MOLART (Molecular aspects of age-
ing in Art) granted in 1995. Boon is Professor of
Analytical Mass Spectrometry at the University of
Amsterdam and Head of the Molecular Paintings
Research group at AMOLF performing research on
the molecular chemistry and chemical microscopy of
the molecular chemistry and chemical microscopy of
pigments and binding media and their interactions
pigments and binding media and their interactions
in paintings.
Please address correspondence to Gillian
Osmond at <gillian.osmond@qag.qld.gov.au>.
