Analysis and Cleaning Tests of Zinc-type Haze on Oil-based Portrait Paintings of the Peranakans Mr And Mrs Tan Beng Wan

Abstract

Insoluble and stable crusts, haze, and efflorescence that have developed on the surface of paintings can be visually
disturbing and are generally difficult to remove during conservation. Highlighted here are a pair of oil-based portrait paintings of the Peranakans Mr and Mrs Tan Beng Wan, dating to the late 19th to early 20th century, that presents an interesting case study of various zinc-type hazes on paint surfaces. The thin whitish films and unsightly patches were analysed using a
digital microscope, scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS), and Fourier-transform
infrared (FTIR) micro-spectroscopy, and were identified as different mixtures of zinc oxalate, carbonate (hydrozincite),
hydroxychlorides, and sulfates. Gordaite (NaZn4Cl(OH)6SO4.6H2O) and zinc chlorosulfate (Zn4Cl2(OH)4SO4.5H2O), also known
to occur in marine or urban/industrial atmospheric corrosion of zinc, were characterised probably for the first time as
surface efflorescence on oil paintings. Zinc soaps migrating from the ground and paint layers were postulated as precursors
to the formation of the zinc-type haze localised in micrometer scale layers at the paint surface. Removal of these zinctype salts is challenging as they are insoluble in water and resistant to organic solvents used in conservation. Mechanical
removal of the thin haze has the risk of disrupting the underlying paint layers. The efficacy of cleaning the whitish haze on
Portrait of Mr Tan Beng Wan using chelating solutions was tested, and treatment considerations are discussed.

Full text

Analysis and Cleaning Tests of
Zinc-type Haze on Oil-based
Portrait Paintings of the Peranakans
Mr And Mrs Tan Beng Wan
LYNN CHUA
Conservation Scientist,
Heritage Conservation
Centre
lynn_chua@nhb.gov.sg
IRENE DOMINGUEZ
Conservator (Paintings),
Heritage Conservation
Centre
irene_dominguez_
jimenez@nhb.gov.sg
KEYWORDS
zinc oxalate, gordaite,
zinc chlorosulfate, haze,
efflorescence, cleaning
ABSTRACT
Insoluble and stable crusts, haze, and efflorescence that have developed on the surface of paintings can be visually
disturbing and are generally difcult to remove during conservation. Highlighted here are a pair of oil-based portrait paintings
of the Peranakans Mr and Mrs Tan Beng Wan, dating to the late 19th to early 20th century, that presents an interesting case
study of various zinc-type hazes on paint surfaces. The thin whitish lms and unsightly patches were analysed using a
digital microscope, scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS), and Fourier-transform
infrared (FTIR) micro-spectroscopy, and were identied as different mixtures of zinc oxalate, carbonate (hydrozincite),
hydroxychlorides, and sulfates. Gordaite (NaZn4Cl(OH)6SO4.6H2O) and zinc chlorosulfate (Zn4Cl2(OH)4SO4.5H2O), also known
to occur in marine or urban/industrial atmospheric corrosion of zinc, were characterised probably for the rst time as
surface efflorescence on oil paintings. Zinc soaps migrating from the ground and paint layers were postulated as precursors
to the formation of the zinc-type haze localised in micrometer scale layers at the paint surface. Removal of these zinc-
type salts is challenging as they are insoluble in water and resistant to organic solvents used in conservation. Mechanical
removal of the thin haze has the risk of disrupting the underlying paint layers. The efcacy of cleaning the whitish haze on
Portrait of Mr Tan Beng Wan using chelating solutions was tested, and treatment considerations are discussed.
209
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

1 The authors use the term
Peranakan here to denote
early Chinese immigrants
who married indigenous
women from the Malay
Archipelago in Southeast
Asia, and are therefore, of
mixed ethnic origins.
2 In 1871, Mr Tan Kim Tian
and his son Mr Tan Beng
Wan founded Tan Kim
Tian and Son Steamship
Company, one of the rst
Chinese rms in Singapore
to buy and build ships.
Introduction
Surface haze reported in painted works of art has been attributed to various sources, including
mould, pollutants, migration of free fatty acids and wax, physical disintegration of surface coatings,
saponification, and complex salt formation (Gridley 2019). Similar terms like elorescence,
encrustations, whitening, blanching, and blooming are also oen used in the conservation field
(Ordonez & Twilley 1998; Van Loon 2008). Although not all hazes on aected artworks are visible, some
can be visually impactful and compromise the artwork’s aesthetics (Burnstock et al. 2011; Puglieri et
al. 2016; Van Loon, Noble & Boon, 2011). In the case of mould, if not treated immediately, it could
propagate deep into the paint layers, and the damage caused may become irreversible in future.
In some instances, the haze reappears even aer its removal (Puglieri et al. 2016). Hence, even the
slightest occurrence would be a cause for concern in conservation.
A pair of portrait paintings in the Peranakan1 (Chong, Teo, Yoong & Joseph 2015) collection of the
National Heritage Board of Singapore (NHB) was brought to the authors’ attention, as the painted
surfaces showed an unusual haze that appeared to have eloresced to diering degrees (Fig. 1). These
paintings, dating to the late 19th to early 20th century, are important portraits of Mr Tan Beng Wan
(1850-1891) and his wife Madam Lim Imm Neo (1851-1925), who were highly respected pioneers in the
shipping industry of early Singapore2. It was suggested that the portrait paintings were created using
photographs as reference, and commissioned to a studio (Chong et al. 2015; Lee et al. 2015).
During the initial assessment of the paintings, it was uncertain if the whitish haze was part of the
artist’s technique or due to natural degradation, though the latter seemed to be a more plausible
reason since portrait paintings of this genre were usually painted with a background executed in one
uniform chroma. To return the paintings to their original appearance, it was desirable for the haze to
be removed. Despite several condition assessments made in the past, the nature of the haze had not
been analysed and identified chemically. Although a few attempts to clean the haze were tested in
the past, the tests were unsuccessful, leaving the haze mainly unimproved. Mechanical scraping of the
haze was possible, but this approach put the paint layer underneath at risk, whereas rolling cotton
swabs with deionised water was eective in some areas of Portrait of Mr Tan Beng Wan, but not in
others. Selected for a potential upcoming exhibition following renovation of The Peranakan Museum
(TPM), these portrait paintings were again brought to the conservation lab, providing an opportunity
to re-investigate the surface haze.
Fig. 1. Paintings aected by zinc-type haze: Portrait of Mr Tan Beng Wan and Portrait of Mrs Tan Beng Wan by an unknown artist,
donated to TPM by Mr and Mrs Tan Choon Hoe. Yellow arrows show locations of haze being sampled.
210

Background of the paintings
Both paintings were assessed for the first time at the Heritage Conservation Centre (HCC) in 1996, and
several times subsequently, over a span of two decades. Other than the haze, the condition of the
paintings was fair, and the pair mainly underwent treatments that improved their structural stability,
for example strip lining and repairs on the primary support. The paintings were executed thinly on a
thin cotton canvas (84.6cm x 63.1cm) with no tacking margin present. The information provided in
the condition report in 1996 described Portrait of Mr Tan Beng Wan as “framed and glazed, having an
excess moisture causing the paint layer to be disintegrated and the paint layer being directly in contact
with the glass plate.”
Understanding the media is important to comprehend the possible mechanisms for forming the hazy
degradation on the paintings. However, previous inspections had led to confusion regarding the type
of binding medium used. This doubt was highlighted in previous conservation reports, in which the
paint medium was recorded as pastel, tempera, gouache, and oil paint separately. To the authors’
surprise, this discrepancy was repeated in books, newsletters, and archival records that featured the
portraits. These sources have described the paintings using dierent media and support, such as
“tempera on cotton”3, “tempera on wood” (Tan 2003), “gouache on paper” (Chong et al. 2015), and
“oil”4. In 2012, analysis using FTIR spectroscopy airmed the binding media in Portrait of Mrs Tan Beng
Wan as oil (Chaplin 2012) and in 2017, Portrait of Mr Tan Beng Wan was similarly identified as an oil
binder with zinc soaps (Chua & Dominguez 2018). These results verified that the paintings are best
represented as “oil on cotton.”
It is diicult to accurately attribute the haze solely by visual observation. Taking reference from past
condition reports, the haze on the paintings examined by dierent conservators had been described
dierently. For Portrait of Mr Tan Beng Wan, the haze was listed as “severe media discoloration
especially at the upper part of greyish colour,” “extensive blanching in background appearing to be
degradation of top layer of paint,” and “fungi growth being covered by the varnish”. In contrast, for
Portrait of Mrs Tan Beng Wan, terms relating to “haze” were not mentioned; instead, the condition of
the painting was assessed as “discoloration especially at the top and bottom parts of the painting”.
The discrepancy in the description of haze shows the importance to conduct a scientific analysis to
determine the nature of the haze.
It was unclear what caused the haze and when it began to occur. One of the rare records was a photograph
taken in 1968 showing Portrait of Mr Tan Beng Wan displayed in the reception hall of the Botan house (Fig.
2(a)-2(c)). At that time, the house was located at Neil Road, close to the shipping area for which the family
was well-known. Portrait of Mrs Tan Beng Wan was assumed to be hung on the opposite wall. It is not
evident from the photograph whether the haze existed in the paintings at that time. From HCC’s records,
the earliest retrievable images of the paintings were taken in 2012, and those photos showed that the
haze was already present. Combined with previous condition reports from 1996, it could be postulated
that the haze on the paintings was already present at the point of donation.
3 Record in the National
Archives of Singapore
4 Record in the National
Heritage Board’s Roots.SG
platform.
211
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

Fig. 2. (a) Reception hall at the Botan House, 1968. The red arrow points to the Portrait of Mr Tan Beng Wan painting.
It is likely that the wife’s portrait was located on the opposite wall (Lee 2019). (b) Exterior of Botan House, 1968 (Lee
2019). (c) Map of the Singapore city and environs, c. 1914 (Baedeker 1914). The red dot refers to the Botan House
at Neil Road.
(a) (b)
(c)
212

Results and discussion
Assessment of the haze
Portrait of Mr Tan Beng Wan was most visually aected by the haze. Almost the entire grey background
was covered with a whitish haze. Other aected localised areas were the whitish patches on the black
jacket ma gua (馬褂), and whitish haze and violet haze along the cracks on the blue robe chang pao
(長袍). Haze on the paintings was observed directly using a digital microscope lens mounted to a
z-motor unit. The latter enables 3D images to be collected over an extended depth of focus. As seen
in Fig. 3, the haze was very thinly formed on the paint surface (the thickness measured between 2 to
6 µm), implying that mechanical removal using tools without risking damage of the underlying paint
would be inevitable. The fact that both portraits were created using very thin paint, and with a uniform
and flat painting technique, makes cleaning even more challenging, since any attempt to alter the
surface would be easily noticeable. The haze on Portrait of Mrs Tan Beng Wan was less apparent (Fig. 4),
but when viewed along the plane of the painting, the whitish haze could be seen tracing the painting’s
craquelure on the paint layer.
Fig. 3. Close-up images of the haze on Portrait of Mr Tan Beng Wan. (a) 1- Violet haze on blue cracks. (b) 2- White haze on blue. (c)
3- White haze on grey. (d) 4- White haze on black. Upper row shows camera images. Bottom row shows digital microscopic images
at 1000x magnification.
Fig. 4. Close-up images of the haze on Portrait of Mrs Tan Beng Wan. (a) 5- White haze at top le corner in background. (b) 6- White
haze at le of chair. Upper row shows camera images. Bottom row shows digital microscopic images at 500x magnification.
(a) (c)(b) (d)
(a) (b)
213
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

Characterisation of the haze
A microsample of the haze was collected from the paintings and characterised using SEM-EDS and
micro-FTIR spectroscopy. Table 1 summarises the FTIR results and indicates the compositions of the
haze.
Table 1. Summary of FTIR and SEM-EDS results of haze samples on the portraits.
Painting No Haze FTIR EDS5
Portrait of
Mr Tan Beng Wan
1 Violet haze on blue
cracks
Zinc oxalate, zinc chlorosulfate
(Zn4Cl2(OH)4SO4.5H2O)
Zn, S, Cl (Na, Fe, K, Mg, Al, Si)
2 White haze on blue Gordaite (NaZn4Cl(OH)6SO4.6H2O) Zn, S, Cl, (Na, Fe)
3 White haze on grey Zinc oxalate, hydrozincite Zn, (Na, Al, Si, S, Cl, Mn, Fe,
Ca, P)
4 White haze on
black
Zinc oxalate Zn, Ca, P, (Na, S, Cl, Mg, Al, Si, K)
Portrait of
Mrs Tan Beng Wan
5 White haze at
top le corner in
background
Gordaite (NaZn4Cl(OH)6SO4.6H2O),
zinc oxalate
Zn, S, Cl, (Na, Al, Si, Ca, Cu)
6 White haze at le
of chair
Zinc chlorosulfate
(Zn4Cl2(OH)4SO4.5H2O), zinc oxalate
Zn, S, Cl, (Na, Al, Si, Ca, Cu)
Zinc oxalate salts
Oxalate salts were indicated by the strong marker peaks in the FTIR spectrum at 1630-1640 cm-1, 1362 cm-1,
and 1320 cm-1. These peaks were observed in almost all the haze samples taken from both portraits (Fig.
5). The additional sharp band observed at 825 cm-1 is a good marker peak for zinc oxalate specifically.
This band dierentiated zinc oxalate from calcium oxalate – a common degradation product on
painted artworks which would otherwise show a sharp band at 782 cm-1 in the FTIR spectrum. In some
samples, the band at 825 cm-1 was masked by spectral bands of other compounds, hence elemental
verification with SEM-EDS was necessary. SEM-EDS of the haze showed major intensity for zinc, and
only trace intensity for calcium, confirming the presence of zinc oxalate. The white haze sample taken
from the black jacket of the man’s portrait showed small amounts of calcium (Ca) and phosphorus (P)
in SEM-EDS, which could be due to a whitening phenomenon of bone black in the underlying black
paint (Van Loon 2008).
The occurrence of zinc oxalate on painted surfaces is far less common compared to that of calcium
oxalates, with only a handful of the former mentioned so far (Dunkerton et al. 2013; Frøysaker, Liu &
Miliani 2013; Monico et al. 2013; Rosi et al. 2019). None of these studies included treatment strategies
for paintings aected by zinc oxalate elorescence. Although some cleaning tests were done on Edvard
Munch’s unvarnished oil paintings containing zinc oxalates, the main purpose of these tests was to
remove dirt and dust, rather than zinc oxalates (Frøysaker, Liu & Miliani 2013).
Although frequently found on artworks, the origin of metal oxalates remains unclear, and various
reasons have been postulated for the occurrence of metal oxalates on artworks. Biological activity
of microorganisms is capable of secreting oxalic acid (Gridley 2019), whereas chemical formation
of oxalates has been attributed to oxidative degradation of the organic binder or resinous coating,
catalysed by the presence of zinc white (Colombini et al. 2002). Recently, oxalic acid was reported as
the most abundant dicarboxylic acid (DCA) in organic atmospheric aerosols (Martinelango, Dasgupta
& Al-Horr 2007). It constitutes up to 50% of total atmospheric DCAs, especially in non-urban and marine
atmospheres.
5 Major, minor, (trace)
intensities
214

Zinc hydroxychloride and sulphate salts
SEM-EDS analyses of the haze from all sampled areas on both portraits consistently showed major
amounts of zinc, along with small to trace levels of sulphur and chlorine. A haze sample carefully taken
without mechanical disruption of morphology showed crystalline platelets in SEM. These results
suggest elorescence formation of zinc hydroxychloride and sulphate salts (Nasdala 1998; Odnevall
& Leygraf 1993; Odnevall & Leygraf 1994). To determine the type of zinc-coordinated salt with chloride
and sulphate, FTIR microscopy was used. It is interesting to highlight the lack of FTIR reference spectra
for zinc hydroxychloride and sulphate salts in conservation science literature, despite the prominence
of zinc-related products (e.g. zinc white pigment, zinc sulfide (in lithopone), zinc stearate additive) used
in artworks. The literature on atmospheric corrosion of zinc metal, however, was more forthcoming.
Zinc-derived salts of chlorides and sulphates are commonly encountered in zinc metalworks used
in the marine and automotive industries (Leygraf et al. 2016). Such zinc-derived corrosion products
commonly contain oxides and hydroxides (ZnO, Zn(OH)2), carbonates (hydrozincite Zn5(CO3)2 (OH)6),
chlorides (simonkolleite Zn5(OH)8Cl2·H2O), sulfates (zinc sulphate ZnSO4·nH2O and zinc hydroxysulfate
Zn4SO4(OH)6·nH2O), and chlorosulfates (gordaite NaZn4Cl(OH)6SO4.6H2O and zinc chlorosulfate
Zn4Cl2(OH)4SO4.5H2O) (Zhu et al. 2000).
In the FTIR spectra (Fig. 5), gordaite (NaZn4Cl(OH)6SO4.6H2O) was identified on the white haze on blue
(Portrait of Mr Tan Beng Wan) and also on the top le corner in background (Portrait of Mrs Tan Beng
Wan). Marker peaks diagnostic of gordaite were in agreement with literature values: two peaks at 1120
and 990 cm-1 (SO4 stretching modes), 1670 cm-1, and 1639 cm-1 (HOH bending), and three distinct
peaks at 3347 cm-1, 3401 cm-1, and 3505 cm-1 (OH stretching in the metal hydroxide layer, or structurally
bound water) (Jayasree et al. 2006; Nasdala 1998). As the X-ray emission energy of Na (Ka-1.04 kV) is very
close to Zn (La-1.01 kV) in the EDS spectrum, the detection of Na can easily be missed in the presence of
high Zn levels. In all the haze samples from both portraits, traces of sodium were likely present as the
X-ray line shied slightly to 1.02 kV, supporting the presence of gordaite.
Zinc chlorosulfate (Zn4Cl2(OH)4SO4.5H2O) was identified on the violet haze on blue cracks (Portrait
of Mr Tan Beng Wan) and also on the white haze at le of the chair (Portrait of Mrs Tan Beng Wan).
In the FTIR spectra containing zinc chlorosulfate, spectral bands attributed to oxalate were also
apparent. The strong oxalate stretching band and broad OH stretching of zinc oxalate overlap with
some diagnostic bands of zinc chlorosulfate (Jonsson 2012). Hence, other spectral bands were used
to identify zinc chlorosulfate; these include a strong band around 1000 cm-1, along with 1058 cm-1, 956
cm-1 (symmetric SO4 stretching), a strong band at 1138 cm-1 (asymmetric SO4 stretching), a sharp band
at 611 cm-1 (asymmetric SO4 deformation), and a weak band at 3601-3605 cm-1 (OH stretch) (Jayasree
et al. 2006; Jonsson 2012).
The ground and paint layers deep below the haze contained zinc carboxylate agglomerates. Zinc
soaps from the ground have likely migrated to the surface and remineralised as zinc carbonate or
hydrozincite (Van Loon 2008), which subsequently interacted with the environment (e.g. dirt, pollutants,
organic debris) to produce zinc hydroxychloride and sulphate. According to the atmospheric corrosion
sequence of zinc, gordaite and zinc chlorosulfate (identified on the portrait paintings) are corrosion
products of zinc in the final stages, which suggests that elorescence on the paintings have reached
stability, and are unlikely to undergo further chemical changes (Jonsson 2012). Given that gordaite
and zinc chlorosulfate on zinc metal are commonly found in marine (high Cl- to SO42- concentration)
and urban/industrial environments (high SO42- to Cl- concentration) (Jonsson 2012; Odnevall & Leygraf
1993; Odnevall & Leygraf 1994), the authors postulate that the origin of the zinc hydroxychloride and
sulphate haze characterised on the paintings could be traced to their historical location in the family
house at an urban area near the seaside (Fig. 2).
215
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

Fig. 5. FTIR spectra of haze from Portrait of Mr Tan Beng Wan: (a) 1- Violet haze on blue cracks. (b) 2- White haze on blue. (c) 3- White
haze on grey. (d) 4- White haze on black; and from Portrait of Mrs Tan Beng Wan: (e) 5- White haze at top le corner in background.
(f) 6- White haze at le of chair. Marker peaks are annotated.
Cleaning tests of zinc-type haze
Inorganic salt deposits on paintings that are water-insoluble, resistant to organic solvents, and
intimately connected to the paint could not be removed by traditional methods (Burnstock 2011;
Sutherland et al. 2013; Van Loon, Noble & Boon 2011). For salt removal, the use of soluble chelating
agents, which can bind to cations in a salt and encourage the salt’s dissolution, appeared a viable
option (Van Loon, Noble & Boon 2011). Ethylenediaminetetraacetic acid (EDTA) was studied in the
removal of lead soap elorescence and calcium salts on paintings (Sawicka et al. 2014; Selva Bonino et
al. 2015), while triammonium citrate (TAC) has been used eectively to remove surface dirt on Turner’s
oil sketches (Carlyle, Townsend & Hackney 1990). Although removal of zinc salts on paintings using
chelating agents has not been published to the authors’ knowledge, this method seemed a promising
option for further testing.
216

Out of all of the hazy patches observed on the two paintings, the whitish haze on the grey background
of Portrait of Mr Tan Beng Wan, identified as a composite of zinc oxalate and hydrozincite, is the most
prominent, and therefore was the subject of testing for its removal. Before conducting the actual
cleaning tests on the painting, the solubility of these zinc salts was tested in chelating agents and also
using a simulated cleaning on sample mock-ups. TAC and EDTA in both 1% and 5% concentrations
were selected as potential chelating agents for removing the white deposits. In this case, the formation
of zinc-type haze on polished zinc metal foil as sample mock-ups for the cleaning tests6 was simulated.
Taking into account that the haze is very thin (measuring on the micrometre scale) and is attached
to the surface of the paint layer, the method of applying the chelating agent needs to be as least
disruptive to the paint layer as possible. In this case, the use of a rigid agarose gel was tested simply by
placing it over the area, to be cleaned without mechanical movement. This approach was preferred as
a gentler method of application.
Solubility test of zinc salt powder
Table 2 lists the solubility results for zinc oxalate and hydrozincite powders7 (Hales & Frost 2007) tested
in 1% and 5% TAC and EDTA solutions. The results clearly indicate that hydrozincite is soluble in both
chelating agents. The dissolution of zinc oxalate in the chelating agents is poorer, though it is possible
for a clear solution to result as time and concentration of chelating agent increases. As expected, EDTA
with an additional chelating site was found to be a stronger reagent than TAC in chelating the zinc
cations.
Table 2. Solubility of zinc oxalate and hydrozincite salts in chelating solutions.
1% EDTA (1mL) 1% TAC (1mL) 5% EDTA (1mL) 5% TAC (1mL)
Zinc oxalate (4mg) Insoluble Insoluble Aer vortex,
dissolves in half hour
Almost dissolves
aer a few hours
Hydrozincite (4mg) Slow dissolution Slow dissolution Immediate
dissolution
Immediate
dissolution
Cleaning tests of zinc-type haze simulated on zinc foil mock-ups
Table 3 lists the results of the cleaning tests of zinc-type haze simulated on the surface of zinc foil mock-
ups using 1% and 5% TAC and EDTA in an agarose gel. From the results, it is clear that hydrozincite can
be removed with chelating agents in agarose. This proved more eective at the higher concentrations
of 5%, and the results indicate that EDTA was stronger than TAC in this regard. However, for the
removal of zinc oxalate, more time and higher concentration was required to see a clear removal.
These results support the findings for the dissolution of the powder samples presented in Table 2.
Table 3. Cleaning tests of zinc-type haze simulated on the surface of zinc foil mock-ups.
1% TAC
(in 3% agarose)
1% EDTA
(in 3% agarose)
5% TAC
(in 3% agarose)
5% EDTA
(in 3% agarose)
Zinc oxalate
on Zn foil
No immediate change.
Aer 3 hours, partial
removal was observed.
No immediate change.
Aer 3 hours, partial
removal was observed.
No immediate change.
Aer 3 hours, complete
removal was observed
No immediate change.
Aer 3 hours, complete
removal was observed
Hydrozincite
on Zn foil
Partial removal. Partial removal. Immediate removal. Immediate removal.
6 Preparation of zinc-type
haze on zinc foil involved
polishing zinc foil with SiC
paper and immersing it
in separate solutions of
1M oxalic acid and 0.9%
NaCl solution over three
days. The corrosion lms
were conrmed with FTIR
spectroscopy as zinc
oxalate and hydrozincite
respectively.
7 Zinc oxalate was
purchased from Sigma
Aldrich. Hydrozincite salt
was prepared using the
synthesis method by Hales
& Frost.
217
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

Cleaning tests on actual painting
From the results of the previous two tests (Tables 2 and 3), the authors confirmed that the selected
chelating agents (TAC and EDTA) and the agarose gel method of application can be eective in
dissolution and removal of the specific zinc-type haze of interest (zinc oxalate and hydrozincite). For
cleaning tests done on the Portrait of Mr Tan Beng Wan painting, the authors decided to start with the
lower concentration (1%) and weaker chelating agent (TAC). The objective of the in-situ cleaning tests
on the painting was to determine if the chelating action can be limited to the haze without aecting
the underlying paint layer. A 3 mm diameter cylindrical agarose gel with 1% TAC was placed over
the test area on the painting. The images before and aer cleaning were recorded with the digital
microscope. The agarose gel was le at 1s to 30s intervals for up to a few minutes, or until the area of
interest showed a visible change. At the end of each test, a fresh agarose gel without chelating agent
was placed over the area of interest to clear o residual chelating agents on the paint surface.
An unobtrusive area with substantial amounts of white haze on the grey background of Portrait of Mr
Tan Beng Wan was selected for the cleaning test. 1% TAC in 3% agarose gel was applied to dierent
spots using dierent application methods (Table 4). Quite unexpectedly, the agarose gel showed the
tendency to cause whitening of the dark grey paint layer below the haze, similar to “blanching”8 (Genty-
Vincent et al. 2015), and this eect worsened the longer the gel was in contact with the paint surface
(Fig. 6). This whitening phenomenon could be a result of more soluble components leaching from the
lower layers to the upper layers due to poulticing. Tests using a higher concentration chelating agent
consisting 5% TAC in 3% agarose gel reduced the haze further, but these also showed blanching and
was undesirable. In addition, these caused some areas of the paint layer to be removed, revealing the
white ground.
Table 4. Cleaning tests of painted surface with TAC in agarose gel.
Spot Application method (1% TAC in 3% agarose gel) Result
1Tapping 1s interval up to 10s No change
2Leave 30s interval up to 30s Causes “blanching”
2Leave 30s interval up to 180s Partial removal of haze. Causes “blanching”
A cleaning test using cotton swabs with 1% TAC solution turned out to be more eective than the
agarose gel method (Fig. 7). The rate of cleaning was controllable, reaching the point of revealing
the original dark grey paint below the haze. No blanching was seen with this method. However, the
cleaning was uneven; the haze was not completely removed even aer four passes of swab rolling with
1% TAC. It was observed that the swabs had picked up some of the dark grey paint, hence the cleaning
tests did not continue using TAC at a higher concentration.
8 Blanching here refers to
whitening of a paint surface
arising from Rayleigh
scattering, commonly
associated with humidity
and porosity.
Fig. 6. Results of cleaning with 1% TAC in 3% agarose gel (spot 2).
218

Conservation considerations
The results obtained from the tests provided an overview of the cleaning possibilities that chelating
agents can oer when cleaning a haze composed of zinc oxalates and hydrozincite. While it is possible
to clear the haze with 1% TAC using cotton swabs, the authors have reservations about carrying out
this treatment. The action of using TAC to remove the haze has resulted in portions of the original
paint being removed, and the paint system inevitably aggravated. Considering that the treatment is
an extensive area that occupies almost one third of the whole painting, it is unlikely that the intended
cleaning can safely achieve a uniform background without causing further paint damage. Moreover,
the test cleaning spots appeared exceptionally matte and do not match the slight glossiness exhibited
in the rest of the original dark grey paint that had been unaected by the haze. There is also concern
that removal of the haze would encourage migration of more zinc soaps to the surface, thereby
regenerating the haze. On account of above reasons, it was decided the haze would not be removed.
Conclusion
The analysis and cleaning tests of zinc-type haze on the two oil-based portraits were carried out.
Dierent parts of the haze were characterised as primarily zinc oxalate, hydroxychloride, and sulphates,
as well as hydrozincite, which were likely formed at the paint’s surface upon migration of zinc soaps
from the thick ground. The presence of gordaite and zinc chlorosulfate on the portrait paintings,
identified as surface elorescence probably for the first time on oil paintings, could be traced to the
historical location of the house in an urban/ industrial area close to the sea.
The white haze on the grey background of Portrait of Mr Tan Beng Wan, characterised as a mixture of
zinc oxalate and hydrozincite, was the most disfiguring, and therefore the most desirable for removal.
As previous attempts to remove this haze using mechanical scraping, organic solvents, and water,
showed little success, the use of the chelating agents TAC and EDTA was studied. The cleaning tests on
zinc oxalate and hydrozincite, in powder form and simulated on zinc foil, showed that both chelating
agents (EDTA stronger than TAC) in a rigid agarose gel achieved gentle removal for a positive outcome.
However, when tested on the actual painting, the agarose gel with 1% TAC caused unacceptable
blanching of the paint surface and could only partially remove the haze. The agarose gel with 5%
TAC reduced the haze further, however, it led to paint loss and blanching. On the contrary, removal
of the haze with 1% TAC using cotton swabs was better than expected, although this method also
picked up some paint and caused uneven cleaning. Due to the diiculty in predicting cleaning eicacy
without causing paint damage, removal of the white haze using the above mentioned methods was
considered unacceptable for this painting. Nevertheless, the results of this study significantly improved
the authors’ understanding of the observed surface haze and justified the conservation treatment
decision.
Fig. 7. Results of cleaning an area with 1% TAC cotton swab.
219
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

Materials and methods
Materials used:
Agarose (low gelling temperature), EDTA (ethylenediaminetetraacetic acid disodium salt dihydrate),
TAC (ammonium citrate tri-basic, 97%), zinc oxalate, Zn foil.
Equipment used:
Keyence VHX6000 digital microscope mounted on a mobile stand, Agilent micro-FTIR spectrometer,
and Hitachi SU5000 SEM coupled with a Bruker EDS.
Acknowledgements
The authors are grateful to Dr Gregory Dale Smith, Senior Conservation Scientist of the Indianapolis
Museum of Art at Newfields, and Lee Swee Mun, Senior Assistant Director and Conservator at the
HCC, for reviewing this paper. Much thanks to Jackie Yoong, Curator at TPM, for providing historical
information of the paintings, and Dominic Low, Assistant Curator at TPM, for providing the historical
images. Appreciation to Maria del Mar Cusso Solano, Paintings Conservator of National Gallery
Singapore, for her advice on the preparation and use of agarose gels, and Seah Zi Quan, NUS Chemistry
Undergraduate for the synthesis of hydrozincite.
Authors' biographies
Lynn Chua is the conservation scientist at the Heritage Conservation Centre (HCC), National Heritage
Board. Chua’s research specialises in technical analysis of painted artworks and their degradation. She
graduated with a Masters in Science (research) from the University Technology of Sydney.
Irene Dominguez Jimenez is a Paintings Conservator at the Heritage Conservation Centre (HCC),
Singapore. She obtained her Bachelor of Fine Arts (Painting), followed by a Master’s degree in Cultural
Heritage Management from the University of Barcelona. In 2012 she received her Master of Arts (M.A.),
Conservation of Fine Art (Easel Paintings) from Northumbria University. Irene has completed several
internships and fellowships in conservation in dierent countries including Spain, U.S., Hong Kong
and UK.
References
Baedeker, K 1914, India, Guide for Travelers.
Burnstock, A, Hinde, L, Jan van den Berg, K, & de Groot, S 2011, Characterisation of surface whitening in
twentieth-century European paintings at Dudmaston Hall, United Kingdom, ICOM-CC International Council
of Museums Conservation Committee 16th Triennial Conference, ICOM-CC, Lisbon, pp. 1-10.
Carlyle, L, Townsend, J, & Hackney, S 1990. Triammonium citrate: an investigation into its application for
surface cleaning, Dirt and Pictures Separated, UKIC and Tate Gallery, pp. 44-48.
Chaplin, T 2012. Analysis of paint samples from the Portrait of Mrs Tan Ben Wang, Analysis Report for the
Heritage Conservation Centre (HCC).
Chong, A, Teo, J, Yoong, J, & Joseph, MK 2015. Great Peranakans: Fiy Remarkable Lives, Asian Civilisations
Museum, Singapore.
Chua, L, Dominguez, I 2018. Micro-characterisation of haze and degradation on zinc white oil-based
painting Portrait of a Peranakan Gentleman Mr Tan Beng Wan, Singapore [Poster], Conference on Modern
Oil Paints, Amsterdam.
Colombini, MP, Modugno, F, Fuoco, R, & Tognazzi, A 2002. A GC-MS study on the deterioration of lipidic paint
binders, J Microchem 73, pp. 175-185.
220

Dunkerton, J, Spring, M, Billinge, R, Kalinina, K, Morrison, R, Macaro, G, Peggie, D, & Roy, A 2013. Titian’s
Painting Technique to c. 1540, National Gallery Technical Bulletin 34, pp. 1-136.
Frøysaker, T, Liu, M, & Miliani, C, 2013. Extended Abstract—Noninvasive Assessments of Cleaning Tests on
an Unvarnished Oil Painting on Canvas by Edvard Munch, New Insights into the Cleaning of Paintings
Universidad Politécnica de Valencia and Museum Conservation Institute, Washington. D.C., pp. 119-123.
Genty-Vincent, A, Eveno, M, Nowik, W, Bastian, G, Ravaud, E, Cabillic, I, Uziel, J, Lubin-Germain, N, & Menu,
M 2015. Blanching of paint and varnish layers in easel paintings: contribution to the understanding of the
alteration, Applied Physics A 121, pp. 779-788.
Gridley, M 2019. White Surface Hazes, AIC.
Hales, MC, & Frost, RL 2007. Synthesis and vibrational spectroscopic characterisation of synthetic
hydrozincite and smithsonite, Polyhedron 26, pp. 4955-4962.
Jayasree, RS, Mahadevan Pillai, VP, Nayar, VU, Odnevall, I, & Keresztury, G 2006. Raman and infrared
spectral analysis of corrosion products on zinc NaZn4Cl(OH)6SO4·6H2O and Zn4Cl2(OH)4SO4·5H2O, Materials
Chemistry and Physics 99, pp. 474-478.
Jonsson, S 2012. Corrosion of zinc in the automotive environment, Institutet för metallforskning.
Lee Kip Lin Collection. Accessed on 1 June 2019. National Library Board.
Lee, P, Ang, D, Ng, SW, & Foo, SL 2015. Inherited & Salvaged: Family Portraits from the NUS Museum Straits
Chinese Collection 1st ed., NUS Baba House, Singapore.
Leygraf, C, Wallinder, I, Tidblad, J, & Graedel, T, 2016. Appendix J: The atmospheric corrosion chemistry of
zinc, pp. 348-359.
Martinelango, PK, Dasgupta, PK, & Al-Horr, RS 2007. Atmospheric production of oxalic acid/oxalate and
nitric acid/nitrate in the Tampa Bay airshed: Parallel pathways, Atmospheric Environment 41, pp. 4258-
4269.
Monico, L, Rosi, F, Miliani, C, Daveri, A, & Brunetti, B 2013. Non-invasive identification of metal-oxalate
complexes on polychrome artwork surfaces by reflection mid-infrared spectroscopy, Spectrochim Acta A
116, pp. 270-280.
Nasdala, L 1998. Gordaite [Zn4Na(OH)6(SO4)Cl·6H2O]: Second occurrence in the Juan de Fuca Ridge, and
new data, American Mineralogist, pp. 1111-1116.
Odnevall, I, & Leygraf, C 1993. Formation of NaZn4Cl(OH)6SO4 · 6H2O in a marine atmosphere, Corrosion
Science 34, pp. 1213-1229.
Odnevall, I, Leygraf, C 1994. The formation of Zn4Cl2(OH)4SO4 · 5H2O in an urban and an industrial
atmosphere, Corrosion Science 36, pp. 1551-1559.
Ordonez, E, & Twilley, J 1998. Clarifying the haze: Elorescence on works of art, WAAC Newsletter 20.
Puglieri, TS, Lavezzo, AS, dos Santos, IFS, & de Faria, DLA 2016. Investigation on the hazing of a Brazilian
contemporary painting, Spectrochim Acta A 159, pp. 117-122.
Rosi, F, Cartechini, L, Monico, L, Gabrieli, F, Vagnini, M, Buti, D, Doherty, B, Anselmi, C, Brunetti, BG, &
Miliani, C 2019. Tracking Metal Oxalates and Carboxylates on Painting Surfaces by Non-invasive Reflection
221
COLLECTIONS CARE: Staying Relevant in Changing Times, ASEAN & Beyond

Mid-FTIR Spectroscopy in: Casadio, F, Keune, K, Noble, P, Loon, AV, Hendriks, E, Centeno, SA, & Osmond, G
(Eds.), Metal Soaps in Art, pp. 173-193. Springer.
Sawicka, A, Burnstock, A, Izzo, FC, Keune, K, Boon, JJ, Kirsch, K, & van den Berg, KJ 2014. An Investigation
into the Viability of Removal of Lead Soap Elorescence from Contemporary Oil Paintings in: van den
Berg, KJ, Burnstock, A, de Keijzer, M, Krueger, J, Learner, T, Tagle, dA, & Heydenreich, G (Eds.), Issues in
Contemporary Oil Paint, pp. 311-332. Springer International Publishing.
Selva Bonino, VE, Tegoni, M, Mucchino, C, Predieri, G, & Casoli, A 2015. Model study of the constituents of
wall painting degradation patinas: The eect of the treatment with chelating agents on the solubility of the
calcium salts, J Microchem 118, pp. 62-68.
Sutherland, K, Price, B, Lins, A, & Passeri, I 2013. Extended Abstract—Oxalate-Rich Surface Layers on
Paintings: Implications for Interpretation and Cleaning, New Insights into the Cleaning of Paintings
Universidad Politécnica de Valencia and Museum Conservation Institute, pp. 85-87. Washington. D.C..
Tan, H 2003. ‘Peranakan Legacy’ at the Asian Civilisations Museum, Singapore, ASEMUS Newsletter, p. 50.
International Institute for Asian Studies.
Van Loon, A 2008. Color changes and chemical reactivity in seventeenth-century oil paintings, Faculty of
Science, Swammerdam Institute for Life Sciences.
Van Loon, A, Noble, P, & Boon, J 2011. White hazes and surface crusts in Rembrandt’s Homer and related
paintings, Preprints of the ICOM committee for conservation 16th triennial meeting, Lisbon.
Zhu, F, Persson, D, Thierry, D, & Taxen, C 2000. Formation of Corrosion Products on Open and Confined Zinc
Surfaces Exposed to Periodic Wet/Dry Conditions, Corrosion 56, p. 10.
222