Zinc soap formation aligned with wood grain
pores on two late 19th century oil paintings
Marta Félix Campos, Leslie Carlyle and Maria João Melo
Abstract Landscape with apple trees in blossom and peasant woman (c.1881-93)
and Beach at Póvoa de Varzim (1881) are two oil paintings on panel by the
Portuguese artist Silva Porto (1850-1893). Both paintings exhibit paint losses
revealing the underlying wood where crumbled ground residue surrounds worm-
like raised lines of zinc soap following the wood grain. In addition, the Landscape
painting exhibits raised paint in horizontal ridges aligned with the panel’s wood
grain. These deformations are associated with 2-12 mm paint losses which show
no relation to the paint’s colour or thickness. Both paintings have the same ground
construction consisting of a first layer of zinc white directly on top of the panel,
and a second layer consisting of a mixture of lead white, barium sulphate and
calcium carbonate. The panels are thought to be commercially prepared. Materials
were identified with a combination of optical microscopy, micro-Raman, and
scanning electron microscopy with energy dispersive x-ray spectroscopy on
embedded cross sections, and with micro-Fourier transform infrared spectroscopy
on micro-samples. The presence of zinc soaps concentrated along the wood grain
is discussed in relation to a previous study identifying lead soap concentration
aligned with the wood grain on a much earlier panel. The zinc soap containing
material in the two 19th century paintings appears to be contributing to paint losses
due to a volume change associated with the concentrated zinc soap. We aim to
provide analytical evidence that the paint losses and associated surface
deformations are being caused by zinc soap formation within the ground layers
and disseminate this form of metal soaps degradation. A scheme illustrating
possible steps leading to deformation and paint loss is provided. This problem is
causing extreme instability and risk of further loss. Currently, treatment options
Keywords Zinc soaps . Ground layers . Paint loss . 19th century . Oil paint .
Infrared spectroscopy (µFTIR) . Optical microscopy (OM) . Scanning electron
microscopy and energy dispersive x-ray spectroscopy (SEM-EDS)
António Silva Porto (1850-1893) was a Portuguese Naturalist painter and a key
figure in the introduction and dissemination of outdoor painting in Portugal. While
studying a collection of 19 of his works in oil paint from Casa Museu Dr.
Anastácio Gonçalves (in Lisbon, Portugal), it was discovered that two of the
paintings exhibited a peculiar form of paint loss.
Landscape with Apple Trees in Blossom and Peasant Woman (c. 1881-93) on a
wooden panel (36.5 x 55.5 cm) showed extensive paint losses ranging from 2 to
12 mm wide (Fig.1 top). The losses included both paint and ground layers, leaving
the wooden panel almost bare with the exception of worm-like raised lines of
aggregated material along the wood grain and crumbled residue. These losses
were associated with horizontal deformations throughout the entire surface which
appear to follow the wood grain and are independent of the paint colour or
Beach at Póvoa de Varzim (1881), also on panel (31.5 x 55cm), showed the
same type of paint loss, all the way to the wood panel with raised lines and
crumbled white material left behind, but in this case the losses were restricted to
just two areas each approximately 3 mm wide (Fig. 1 bottom). Beach at Póvoa de
Varzim did not show horizontal deformations in the paint.
Fig. 1 Top: Landscape with Apple Trees in Blossom and Peasant Woman with locations of detail
images. Details 1 & 2 show paint losses along the wood grain with whitish material left behind.
Also visible is the crumbled residue from the ground. Bottom: Beach at Póvoa de Varzim with
location of detail images. Detail 3 shows paint loss along the wood grain with raised worm-like
material left behind as found in 1 & 2 above. Detail 4 shows the crumbled ground residue.
Fig. 2 Details from the Landscape: normal light (top) compared with raking light (bottom): the
extensive paint deformations following the wood grain are not evident in normal light
photography, but can easily be seen in raking light. These deformations occur throughout the
entire surface of Landscape. No surface deformations were evident on the Beach.
In the two paintings, the paint losses involve both the ground and paint layers,
with material left behind: a crumbled residue and lines of material oriented along
the larger pores of the wood grain. Raking light shows that the material along the
grain has considerable volume and appears raised (worm-like) (Fig.3).
Paint losses in the Landscape are ongoing since 1990 when the painting was
last restored. There were many fresh losses visible in 2013 (Fig.1). Evidence of
previous losses which had been restored indicate that the Landscape has had a
history of flaking and loss. On the contrary, previous restoration on the Beach is
associated with contact and abrasions at the inside edge of the frame and not due
to paint deformation in ridges. However, although these losses are due to abrasion,
delamination has occurred at the ground/panel interface suggesting lack of
cohesion within the ground layers.
In order to investigate the cause of the paint losses and deformations in the
Landscape’s paint, various analytical techniques (OM, SEM-EDS, µRaman,
µFTIR) were used to identify the material found in the paint losses on both
paintings and the intact double ground on both panels. Within the paint losses,
micro-samples were taken of the material which was concentrated at the wood
grain and that are present as a crumbled residue on the wood surface. For the
grounds, micro-samples were taken of the paint and ground composites from areas
around existing paint losses.
Materials and Analytical Methods
Micro-samples taken from the paint and the ground (separately) as well as the
material found in the wood grain (raised lines) and the crumbled material on the
surface of the wood, and were analysed with micro-Fourier transform infrared
spectroscopy (µFTIR). A Nicolet Nexus spectrophotometer was coupled to a
Continuum microscope (15x objective) with a MCT detector cooled by liquid
nitrogen. Spectra were obtained in transmission mode, between 4000-650 cm-1,
with a resolution of 4 cm-1 and 128 scans, using a Thermo brand diamond anvil
compression cell. The spectra for the material in the wood grain and reference
sample (see below) are shown (Fig.3 right) as acquired, without corrections or any
further manipulations, except for the removal of the CO2 absorption at ca.
For stratigraphical analysis, a selection of 13 micro samples were collected,
both in areas of paint loss and intact paint, and were embedded in polyester resin
(Resina Cristal, MR Dinis dos Santos©) for cross-section examination with
Optical Microscopy (OM), µRaman spectroscopy and Scanning Electron
Microscopy with Energy Dispersive Xray Spectroscopy (SEM-EDS). OM was
carried out on a Zeiss Axioplan 2® microscope equipped with a HAL100 halogen
lamp (under crossed-polar illumination) and an HBO100 mercury ultraviolet lamp
(Zeiss UV Filter set 2; excitation G365; emission 420). Images (Fig.4) were taken
with a Nikon DXM1200F digital camera coupled with the microscope. Image
acquisition and post-processing was done using the ACT-1 software.
Micro-Raman spectroscopy was carried out on the previously mounted cross-
sections for pigment identification in the different layers. A Labram 300 Jobin
Yvon spectrometer was used, equipped with a He-Ne laser of 17mW power
operating at 632.8 nm. Spectra were recorded as an extended scan. The laser beam
was focused with a 50x Olympus objective. The laser power at the surface of the
samples was varied with the aid of a set of neutral density filters (optical densities
0.3, 0.6, 1).
SEM images were obtained with a FEI Quanta 400 SFEG ESEM, which uses a
Schottky emitter field emission gun, operating at low vacuum conditions and at
15 kV, equipped with an EDAX Genesis X4M detector. Images were acquired
using secondary (SE) and backscattered (BSE) electron detectors, but only the
latter are shown here (Fig. 4, bottom). EDS point analysis was used to determine
the composition of the ground and paint layers in the cross sections.
With the help of an infrared reference database of metal carboxylates developed in
the Department of Conservation and Restoration at Universidade Nova de Lisboa
(Otero et al. 2014), it was possible to confirm that the material concentrated at the
wood grain consists of zinc soap, either a zinc stearate or a zinc palmitate or a
mixture of both (Fig. 3, right), characterised by the sharp peak at 1540cm-1
a(COO) (Robinet and Corbeil 2003; Hermans et al. 2015).
These results are in keeping with the discovery by Osmond et al. (2012) who have
identified a preferential pattern in the formation of zinc soaps on reconstructed
reference samples. Exposed top surfaces exhibit a range of different carboxylates
with an infrared absorption band between 1640-1505 cm-1, whereas unexposed
areas tend to form zinc stearates and palmitates, characterized by the asymmetric
COO- stretch at 1540 cm-1. This latter case corresponds to the zinc soaps identified
in Silva Porto’s Landscape and Beach paintings.
Fig. 3 Detail in a paint loss in the Landscape, top left in normal light, with location of the sample
(red arrow), versus bottom where raking light clearly shows the volume of the zinc soap lines on
top of the wood grain. Right: the zinc soap (in red) was characterised using infrared spectroscopy
compared with a reference sample of zinc stearate (in black).
The crumbled residue at the wood surface on both paintings was identified by
µFTIR and µRaman as a mixture of zinc white oil paint and zinc soap.
As detailed above, the lines of material present in the wood grain on both
paintings were identified as zinc soap. In the Landscape, the lines appear to be
associated with the extensive and disturbing raised horizontal deformations in the
paint, implying that this feature is related directly to the zinc soaps. Since the
Beach does not show deformations, only paint loss with similar zinc soap presence
in the wood grain, it is thought that this painting may not be at the same stage of
Zinc soap is the only material which remains in bulk on the surface of the
panels within the paint losses. It is concentrated in areas directly above the large
pores of the wood grain (Fig. 3 left). The extensive horizontal deformations on the
Landscape painting appear to have resulted from the presence of the zinc soap
below, since these raised ridges continue on either side of losses in alignment with
the material in the wood grain (Fig. 2). The current thinking is that the
deformations result from a change in volume during the formation of the zinc
soaps, see discussion below.
Analyses (OM, µRaman and SEM-EDS) of cross-sections from both paintings
revealed the same stratigraphy with two distinct oil grounds: first, a layer of zinc
white alone; and over it, a layer composed of mixture of lead white, barium
sulphate and calcium carbonate.
Fig. 4 Two cross sections from the Landscape; At the top: OM normal light, in the middle: UV
light, at the bottom: SEM. Left image: a cross-section from a green area where a regular layer
structure is easily visible: a translucent zinc-based layer on the bottom (1), with its characteristic
fluorescence under UV light, followed by a second ground layer consisting primarily of lead
white mixed with smaller quantities of barium sulphate and calcium carbonate (2); and finally a
layer of light green oil paint (3). Right image: a cross-section taken next to a paint loss with
equivalent ground layers (1 and 2), several paint layers (3) and a zinc soap aggregate forming
within the first ground layer (arrows).
Fig. 4, shows two cross-sections from Landscape: on the left, a sample taken from
an unaffected area; and on the right, a sample taken close to a paint loss. The first
and more translucent ground layer (1) shows the typical UV fluorescence of zinc
white. On the cross-section on the right, this fluorescence is less visible suggesting
a saponification of the metallic pigment.
In the embedded cross-section on the right, an inclusion of material is visible
in normal (crossed polar) and ultra violet light in the zinc layer (red arrows).
Based on appearance, which conforms to images in the literature (e.g. Keune
(2005), Osmond (2005), Helwig (2014)) this is thought to be a zinc soap
aggregate. In Fig. 4 bottom right, the SEM backscatter image (BSE) shows a
corresponding area of what appears to be predominantly organic material which
may represent zinc carboxylate.
The second ground layer (2) on the right appears to already have been distorted
by this aggregate formation. This could then lead to deformations in the paint
layers (3), similar to the ones in Fig. 2.
Discussion and Conclusions
A similar phenomenon of metal soap concentration along the wood grain of a
panel painting has been described (Noble et al. 2005). In that case, lead
carboxylate was found along the wood grain on a 17th century panel. This had
resulted in dark lines of greater transparency in the paint following the wood
grain. Nobel et al. proposed that the lead carboxylate concentration was related to
the larger pores of the wood being filled with more chalk during the application of
the ground. It was proposed that the larger proportion of binder associated with the
deeper chalk ground-filled pores supplied the fatty acids needed to form lead
carboxylates in the imprimitura.
In Silva Porto’s two paintings, it is notable that the zinc soaps have also
formed along the wood grain, perhaps in a similar mechanism with the deeper
wood pores providing the source of triglycerides from the zinc white ground build
up in the pores. Alternatively, it is possible that the wood panel was initially
sealed with a drying oil prior to the ground application. Although this method was
seldom described (Carlyle 2001: 180; Stols-Witlox 2015: 99), it would
presumably also leave a residue of oil in the wood grain as a source of fatty acids.
We propose that these aggregates are forming in place and provide a scheme
illustrating a possible sequence for the resulting paint loss (Fig.5). The first ground
layer fills the large pores of the wood grain; the oil binder, and possibly oil already
present from sealing the wood, provides a source of fatty acids which leads to the
formation of zinc soap aggregates. In the next step, the increased volume of the
aggregates creates raised deformations and cracks in the upper layers. Cracks in
the paint/ground are potential sites for greater ingress of water/moisture possibly
further encouraging the expansion of the aggregates. The increased volume of the
zinc soap aggregates forces the more brittle paint/ground composite upwards
leading to detachment and paint loss with the softer aggregated zinc soap material
Fig. 5 Scheme of the proposed degradation process. Wooden support followed by the first layer
of zinc ground (light grey); next a layer of lead white with barium sulphate and calcium
carbonate in oil (darker grey); finally, the top layer (yellow) representing the paint.
Unlike the usual manifestation of metal soaps that can aggregate, migrate to the
surface and erupt in the form of scattered white protrusions, the zinc soaps studied
in these paintings form worm-like structures (raised lines) along the wood grain of
the panel and do not protrude through the paint surface. Instead, it is the zinc
white ground layer itself that is detaching from the support leaving lines of zinc
soap (with small amounts of ground residue surrounding). Recent research on the
chemical and physical properties of zinc white oil paint and zinc soaps could help
to account for the low cohesion within the zinc ground and its poor adhesion to the
wooden support. Rogala et al. (2010: 105) state that zinc white pigment
compromises the structural stability of the oil paint matrix, “rather than a well
formed paint layer consisting of a uniform cross-linked network, a paint
containing zinc oxide consists of plate-like layered ‘islands’ held together by only
a few cross-links”. Following up on a study by Jacobsen and Gardner (1941),
Maines et al. (2011) found that the lamellar structure of zinc white in oil paint
keeps the hydrocarbon chains tightly packed, consequently making them more
difficult to oxidize. This leads to unusually high amounts of unsaturated oleic acid
(C18:1) in fully cured zinc white oil paints, which would normally have been
oxidized into azelaic acid (C9). This zinc–hydrocarbon–zinc layering, that is
restricting the cross-linking of the oleic acid, likely contributes to interlaminar
failure (Rogala et al. 2010: 105)
The lamellar structure of zinc white in oil which traps the oleic acid is also
thought to restrict the movement of zinc soaps (Rogala et al. 2010: 111), which
may help to explain the concentration of zinc soap at the panel surface. The
reduced oxidation of the paint system with its laminar morphology and unreacted
oleic acid could account for the presence of the crumbled ground residue at the
surface of the panel since this points to a lack of internal cohesion in the zinc
ground layer. Internal cohesion issues in zinc white oil paint were also referred to
by Helwig et al. (2014: 169).
None of the remaining 17 oil paintings studied exhibit the same kind of paint loss
or this form of metal soap degradation. When zinc grounds are present, they were
found only on canvas supports with no other ground layers on top and exhibited
the more common white protrusions on the surface. Other paintings on panel had
lead white ground layers.
Why one of these two painting is more affected than the other remains
uncertain and requires further study. In the case of Landscape, this problem is
causing extreme instability and there is a risk of further and/or total loss should
this condition progress. Unfortunately, at present, treatment options are unclear.
Acknowledgments The authors would like to thank Casa-Museu Dr. Anastácio Gonçalves for
the opportunity to study their collection of Portuguese naturalist oil paintings. We would also
like to acknowledge Centro de Materiais da Universidade do Porto (CEMUP) for the SEM-EDS
analysis. Marta Félix would like to thank designer Ana Pedro for the scheme presented in Fig. 5.
This work was supported by the PhD grant SFRH/BD/75123/2010 from Fundação para a Ciência
e Tecnologia (FCT-MCTES).
Boon JJ, Weerd J van der, Keune K, et al (2002) Mechanical and chemical changes in Old
Master paintings: dissolution, metal soap formation and remineralization processes in lead
pigmented ground/intermediate paint layers of 17th century paintings. In: ICOM-CC 13th
Triennial Meeting Preprints, 22-27 September, Rio de Janeiro. pp 401-406
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
Centeno SA, Mahon D (2009) The Chemistry of Aging in Oil Paintings: Metal Soaps and Visual
Changes. Metrop Museum Art Bull 67:12-19.
Helwig K, Poulin J, Corbeil MC, Moffatt E, Duguay D (2014) Conservation Issues in Several
Twentieth-Century Canadian Oil Paintings: The Role of Zinc Carboxylate Reaction Products,
In: Issues in Contemporary Oil Paint,Springer
Hermans J J, Keune K, van Loon A, Iedema P D (2015) An infrared spectroscopic study of the
nature of zinc carboxylates in oil paintings. Journal of Analytical Atomic Spectrometry 30:
Jacobsen AE, Gardner WH (1941) Zinc soaps in paints: Zinc oleates. Ind Eng Chem 33:1254-
Keune K (2005) Binding medium, pigments and metal soaps characterised and localised in paint
cross-sections, PhD thesis, Amsterdam, University of Amsterdam
Keune K, Boon JJ (2007) Analytical Imaging Studies of Cross- Sections of Paintings Affected
by Lead Soap Aggregate Formation. Stud Conserv 5:161-176
Maines C, Rogala D, Lake S, Mecklenburg M (2011) Deterioration in Abstract Expressionist
Paintings: Analysis of Zinc Oxide Paint Layers in Works from the Collection of the
Hirshhorn Museum and Sculpture Garden, Smithsonian Institute. In: Materials Research
Society Symposium Proceedings. Vol. 1319, pp 275-284
Noble P, Loon A van, Boon JJ (2005) Chemical changes in old masters paintings II: darkening
due to increased transparency as a result of metal soap formation. In: ICOM-CC 14th
Triennial Meeting Preprints, 12-16 September, The Hague. pp 496-503
Osmond G, Boon JJ, Puskar L, Drennan J (2012) Metal stearate distributions in modern artists’
oil paints: surface and cross-sectional investigation of reference paint films using
conventional and synchrotron infrared microspectroscopy. Appl Spectrosc 66:1136-1144.
Osmond G, Keune K, Boon J (2005) A study of zinc soap aggregates in a late 19th century
painting by R.G. Rivers at the Queensland Art Gallery. AICCM Bulletin 29:37-46
Otero V, Sanches D, Montagner C, et al (2014) Characterisation of metal carboxylates by Raman
and infrared spectroscopy in works of art. J Raman Spectrosc 45:1197-1206
Robinet L, Corbeil M-C (2003) The Characterization of Metal Soaps. Stud Conserv 48:23–40.
Rogala D, Lake S, Maines C, Mecklenburg M (2010) Condition Problems Related to Zinc Oxide
Underlayers: Examination of Selected Abstract Expressionist Paintings from the Collection
of the Hirshhorn Museum and Sculpture Garden, Smithsonian Institution. J Am Inst Conserv
Shimazu Y, Keune K, Berg KJ van den, et al (2008) The effects of lead and zinc white
saponification on the surface appearance of paint. In: ICOM-CC 15th Triennial Meeting
Preprints, 22-26 September, New Delhi. pp 626-632
Stols-Witlox M (2015) Historical Recipes for Preparatory Layers of Oil Paintings in Manuals,
Manuscripts and Handbooks in North West Europe, 1550-1900: Analysis and
Reconstructions, Volume I. Archetype Publications, London