60

Solvent effects on East Asian lacquer (Toxicodendron vernicifluum)

Carolyn McSharry, Rupert Faulkner, Shayne Rivers, Milo S.P. Shaffer and Tom Welton

Introduction

When newly manufactured, East Asian lacquer is typically (though not exclusively) characterized by a glossy, lus-trous surface. Over time and with exposure to light, lac-quer surfaces become progressively more matt, in large part due to the formation of micro-cracks in the surface; the effect of light on the durability of lacquer has previ-ously been reviewed (McSharry et al. 2007: 33). In the past, natural resin varnishes were often applied to East Asian lacquer surfaces by European restorers in an attempt to re-saturate colour and renew gloss, or to match in their repairs. Unfortunately, these varnishes do not replicate the original appearance of lacquer and tend to compromise the subtlety of the decoration. In time, many of these coatings in turn become cracked and discoloured, and removing them becomes a conservation issue.

Cleaning lacquer during a conservation treatment has been described as a high risk activity because of the poten-tial danger of damaging the original surface of the object (Webb 2000: 72). This situation is even more complex when varnish removal is considered. Alcohols and ketones are often very effective for removing degraded natural resin var-nishes and are usually the conservation solvent of choice for this purpose (Rivers 2002: 55). However, there is an inverse relationship between the exposure history of a lacquer sur-face to light and its resistance to water and solvents. The more photodegraded the underlying lacquer and the more degraded and oxidized the western coating, the more diffi-cult it is to remove the varnish without damaging the origi-nal surface (Rivers and Umney 2003: 763).

To date, conservators’ understanding of the effects of organic solvents on East Asian lacquer has been based on empirical observation. There is a need to investigate and ideally quantify the degree of solubility or swelling of East Asian lacquer in a range of solvents that includes non-polar solvents such as hexane, dipolar hydrogen-bond donating (HBD) solvents such as methanol and water, and dipolar non-hydrogen-bond donating (non-HBD) solvents such as

acetone. The aim of this investigation was to enable conser-vators to better understand the swelling or leaching effects of solvents used to remove degraded natural resin varnish from lacquer surfaces. Using freshly cured and artificially aged samples, the swelling characteristics of East Asian lac-quer with examples of each of the different solvent types were measured quantitatively using immersion and vapour sorption tests.

Experimental

Materials and solvents

Samples of uncured kijiro urushi1 were brushed onto glass microscope slides to an approximate thickness of 9–11 µm, cured over three days in 75–80% relative humidity (RH), and subsequently stored in the dark for one month under ambi-ent conditions before use. Board samples were prepared in 2006 on a hinoki (Japanese cypress) substrate using tradi-tional methods and similar curing conditions, with a layer of hemp cloth, two foundation layers and three layers of lacquer. The final layer was polished with charcoal, and then with powder and oil. AnalaR grade solvents2 were used with-out further purification, with the exception of the following: Exxsol DSP80/110 (ExxonMobil) is the European equivalent of the Japanese ligroine, which is used during urushi-gatame treatments for removing excess urushi from the lacquer sur-face (Rivers and Yamashita 2006: 289). Exxsol DSP80/110 is a de-aromatized aliphatic hydrocarbon, distilled between 85 and 112°C, and composed typically of heptane and isomers (71%) as well as smaller amounts of methylcyclohex-ane (14%), cyclohexane (8%), octane isomers (3%), hexane isomers (2%) and n-hexane (2%). HAN 8070 (ExxonMobil), used as a diluent for urushi-gatame, is an aromatic hydro-carbon mixture, typically consisting of kerosene (20–30%), naphtha (70–80%), trimethylbenzene (1–5%), mesitylene (0.1–15%) and naphthalene (5–10%).

EAL.indb 60 08/07/2011 09:31

61

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

Artificial ageing regimes

Three methods of ageing were compared against a freshly cured standard, listed in increasing order of the degree of photodegra-dation produced: 500 hours natural daylight (light/dark cycling), 3,500 hours ultraviolet (UV) daylight simulation with a poly- carbonate filter, and 2,000 hours mercury-tungsten lamp (HgW) source exposure without filter (Table 1). In previous work the drawbacks associated with various accelerated ageing methods were investigated (McSharry 2009: 125). A range of analytical methods was used to compare specific changes in gloss, colour, molecular structure and surface cracking of artificially aged lac-quers to damage that occurred naturally over a prolonged period on a moderately well-protected piece. Both mercury-tungsten lamp sources (HgW) and UV-daylight simulation produced sig-nificant and measurable changes, albeit at different rates and with different outcomes – neither method produces samples that exactly replicate all the changes observed in a naturally aged

specimen. Ageing by exposure to natural daylight behind glass under ambient conditions did not produce a sufficiently useful degree of change over the experimental period.

Methodology and results

Immersion: leaching and swelling

The swelling characteristics of unaged and artificially aged lac-quer were determined using methods adapted from Zellers et al. (1996a,b). Lacquer films of approximately 25 × 5.5 × 0.01 mm were weighed (W1) and then immersed in a range of solvents (listed in Table 2) for 24 hours. After this time, they were removed from the solvent and dried for six hours. The films were re-weighed (W2) before being returned to the sol-vent for a further 72 hours, following which they were again

nomenclature Method conditions temperature (°c) average lux reading

Daylight aged(500 hours)

natural ageing Samples kept on a windowsill, cycled light, ambient temperature, ambient rH

variable range (16–21) 4,000 (over a 24-hour period)

uv-daylight aged(3,500 hours)

uv daylight simulator Polycarbonate filter (cuts out <315 nm) 35 27,500 lux

HgW aged(2,000 hours)

Mercury-tungsten (HgW) source fadeometer

no filter, 366 nm (prominent wavelength), and 400–750 nm (20% of the source output)

39-44 30,000 lux

Table 1 Artificial ageing methods.

unaged uv 3,500 h Daylight 500 h HgW 2,000 h

Solvent Δd δp δh leach (%) Swell (%) leach (%) Swell (%) leach (%) Swell (%) leach (%) Swell (%)

Pentane 14.5 0.0 0.0 0.0 3.2 (0.04) 0.0 1.0 (0.04) 0.0 2.1 (0.06) 0.0 0.0 (0.01)

Hexane 14.9 0.0 0.0 0.0 2.9 (0.01) 0.0 0.0 (0.00) 0.0 1.4 (0.02) 0.0 0.0 (0.01)

cyclohexane 16.8 0.0 0.2 0.0 1.7 (0.18) 0.0 1.3 (0.15) 0.0 1.0 (0.20) 0.0 0.0 (0.09)

Benzene 18.4 0.0 2.0 0.0 2.0 (0.08) 0.0 1.3 (0.09) 0.0 5.1 (0.13) 1.4 6.7 (0.17)

toluene 18.0 1.4 2.0 0.0 2.1 (0.20) 0.0 2.2 (0.15) 0.0 5.0 (0.30) 1.9 4.3 (0.42)

Xylene 17.8 1.0 3.1 0.0 3.1 (0.93) 0.0 3.2 (0.87) 0.0 3.6 (0.66) 0.0 5.4 (0.09)

Dichloromethane 18.2 6.3 6.1 0.0 2.5 (0.03) 2.0 6.0 (0.05) 0.0 5.9 (0.13) 2.1 6.2 (0.08)

chloroform 17.8 3.1 5.7 2.3 1.9 (0.40) 2.7 8.3 (0.44) 2.5 6.2 (0.35) 3.0 13.2 (0.41)

Diethyl ether 14.5 2.9 5.1 0.0 2.0 (0.90) 1.9 5.3 (0.70) 0.0 6.4 (0.70) 0.9 6.5 (0.80)

tetrahydrofuran 16.8 5.7 8.0 2.6 0.5 (0.20) 2.8 6.7 (0.13) 3.1 5.6 (0.14) 4.3 7.3 (0.25)

ethyl acetate 15.8 5.3 7.2 0.0 2.3 (0.65) 0.0 7.7 (0.13) 0.0 7.2 (0.20) 0.0 8.9 (0.23)

acetone 15.5 10.0 7.0 0.0 2.5 (0.50) 1.2 5.0 (0.70) 0.0 3.9 (0.45) 1.3 7.6 (0.80)

Butan-2-one 16.0 9.0 5.1 3.1 2.4 (0.07) 2.9 5.8 (0.08) 3.2 6.3 (0.07) 5.6 6.7 (0.07)

ethylene glycol 17.0 11.0 26.0 0.0 0.0 (0.01) 0.0 4.4 (0.03) 0.0 0.0 (0.01) 0.0 3.6 (0.01)

Methanol 15.1 12.0 22.3 0.0 0.4 (0.51) 0.0 5.2 (0.66) 0.0 0.7 (0.60) 0.0 5.6 (0.70)

ethanol 15.8 8.8 19.4 0.0 3.0 (0.10) 0.0 9.2 (0.56) 0.0 2.4 (0.50) 0.0 11.7 (0.71)

Propan-2-ol 15.8 6.1 16.4 0.0 0.3 (0.30) 0.0 5.1 (0.19) 0.0 1.0 (0.56) 0.0 5.7 (0.29)

Butanol 15.8 5.7 15.8 0.0 0.8 (0.24) 0.0 3.4 (0.44) 0.0 1.1 (0.30) 0.0 5.4 (0.40)

acetonitrile 15.3 18.0 6.1 0.0 0.9 (0.24) 0.0 0.3 (0.30) 0.0 0.5 (0.39) 1.5 1.3 (0.41)

Dimethylformamide 17.4 14.0 11.3 3.0 0.3 (0.12) 3.7 0.9 (0.20) 6.7 1.3 (0.25) 14.0 2.1 (0.20)

Water 15.6 16.0 42.3 0.0 2.9 (0.07) 0.0 4.2 (0.05) 0.0 3.2 (0.06) 0.0 4.4 (0.09)

Table 2 Summary of leaching and swelling effects of a range of solvents (standard errors shown in parentheses).

EAL.indb 61 08/07/2011 09:31

62

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

Figure 1 Plot of partial solubility parameters (a) 2δd vs δh, (b) 2δd vs δp, and (c) δh vs δp, of solvents with minimum swelling criteria (>5% wt.) for daylight aged (green circle), mercury-tungsten-aged (yellow circle) and UV-daylight aged (red circle).

a

b

c

EAL.indb 62 08/07/2011 09:31

63

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

re-weighed (W3). The first immersion stage allowed for leach-ing from the damaged lacquer. The lowest weight from W1 and W2 was subtracted from the final weight to determine overall swelling. Each film type was repeated six times. The standard errors range from 0.05 for water uptake, to 0.7 and 0.66 wt% for the faster evaporating solvents acetone and methanol respec-tively. The overall immersion swelling and/or leaching results for each lacquer type are shown in Table 2.

The overall swelling data were used to determine solubility parameters using Hansen’s graphical estimation method. This method was used by Zellers et al. (1996a) in the determination of solubility parameters for lightly crosslinked polymers when assessing the properties and responses to solvents of materi-als used in laboratory gloves. The approach only uses the high swelling solvent data to estimate Hansen’s 3D solubility para- meters (3DSPs) for the polymer system. Unaged lacquer films did not swell more than 3.2 wt% for any of the organic sol-vents used therefore graphical estimations of partial solubility parameters/swelling characteristics for these were not possi-ble. In the determination of 3DSPs for daylight-aged (500 h), mercury-tungsten-aged (2,000 h) and UV-daylight-aged (3,500 h) lacquer, the minimum solvent swelling criterion for recognition as a positive result was >5 wt%. The results for these immersion tests are shown in Figure 1 and the partial solubility parameters are assigned according to the centre of the smallest possible circle that encloses all solvents giving at least the minimum swelling (see Table 3). Qualitatively, the mercury-tungsten-aged and UV-daylight-aged circles almost coincide, suggesting similar changes in polymer chemistry, towards a polar, more hydrogen-bonding character, regard-less of the wavelength of the irradiation, as compared to the relatively unchanged daylight-aged sample.

In this analysis, the swelling criterion is arbitrary and not all the data obtained in the immersion experiments are used. For instance, Figure 1 shows that some solvents within the circular ‘swelling’ regions do not meet the minimum swelling criterion. A more accurate analysis of the results can be applied that takes into account the effects of molar volume and uses all of the swelling measurements. To calculate the partial sol-ubility parameters in this way, a weight-averaged calculation

was used with weight-averaging factors as described by Zellers et al. (1996b):

δ (d2, p2, h2 )= ∑ ui Vz δ(d2, p2, h2) / ∑ ui V

z

where partial solubility parameters are determined from the fractional uptake (ui) of the solvent (weight and volume), and the molar volume (V) of the solvent, taking into account the effect of molar volume by using z, which is an exponential fac-tor (z = 0, 1, 0.5) varied to adjust the effect of molar volume, since molar volume influences the uptake of solvents in terms of rate of diffusion or capacity of the polymer. The 3D solubility parameters were calculated using the immersion test results for each lacquer type and are shown in Table 3.

Vapour sorption tests There are limitations to studying the responses of free films using immersion. In particular, the mass increase can be small compared to the excess surface solvent that must be removed and is thus difficult to measure precisely. The free films tend to curl, split and stick to the glass vessel surface, causing fur-ther breakage and sample loss; the most damaged films dis-integrate very easily once placed in the solvent reservoir. The experiments were, therefore, repeated using a vapour sorp-tion technique whereby the lacquer was suspended in a sealed vessel containing a saturated atmosphere of solvent, such that direct contact between the solvent reservoir and the film was avoided.

The solvents used in vapour sorption tests are listed in Table 4. The lacquer films were suspended on a perforated alu-minium stage in a saturated solvent atmosphere for 24 hours, after which the mass increase was determined immediately on removal from the vessel. The films were then exposed to ambient laboratory conditions in an open container for a fur-ther 24 hours to measure desorption of the solvent from the film. Preliminary experiments confirmed that 24 hours was sufficient time for water to reach equilibrium. The sorption cycle was repeated and measured. The 3D solubility parameters were determined as for the immersion tests and the results are shown in Table 5. For this method, standard errors for water were 0.07 wt% but for the faster evaporating solvents acetone and ethanol these errors were 0.5 and 0.16 wt% respectively, offering some improvement over the direct immersion tech-nique. Although this method is a good measure of 3DSP values of the pure lacquer, the response is not directly comparable to a real, multilayer lacquer surface in which each layer exerts con-straints on its neighbours, as discussed further below.

Solubility parameters on the Teas chart

The Teas chart is a practical and effective tool originally devel-oped in the 1960s for the coatings industry. It uses a set of fractional parameters derived from Hansen’s parameters to plot solvents according to their relative strength of disper-sion, hydrogen bonding and polarity; the three-component solvent data are plotted on a 2D graph in a manner similar to a ternary phase diagram and grouped according to their solvent class. The chart makes the assumption that all mate-

lacquer type Method used δd (MPa1/2) δp (MPa1/2) δh (MPa1/2)

unaged Graphical estimate

n/a n/a n/a

Weight averaged

16.30 5.72 8.07

Daylight aged 500 h

Graphical estimate

16.20 4.10 4.00

Weight averaged

16.60 6.19 1.42

HgW aged 2,000 h

Graphical estimate

16.40 6.40 11.40

Weight averaged

16.61 6.52 1.54

uv-daylight aged 3,500 h

Graphical estimate

16.20 1.20 12.90

Weight averaged

16.47 6.50 8.04

Table 3 3D solubility parameters obtained for immersed lacquers by graphical estimation and weight-averaged methods.

EAL.indb 63 08/07/2011 09:31

64

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

rials have the same Hildebrand value and therefore solubility behaviour is shown by relative contributions of the three forces. Despite its limitations, the chart is useful in the selection of a suitable solvent or blend of solvents that demonstrate material- specific behaviour.

The 3DSPs determined by vapour sorption uptake, and calculated using the contribution of molar volume by the weight-averaged model, are plotted in Figure 1. The Teas frac-tional solubility parameters are derived from the 3DSPs and calculated using the relationship (Burke 1984):

fd + fp + fh = 100

where:

fd = δd / (δd + δp + δh)

fp = δp / (δd + δp + δh)

fh = δh / (δd + δp + δh)

The swelling regions for East Asian lacquer, based on the vapour sorption tests, are plotted onto a Teas chart (Figure 2). The unaged lacquer’s peak swelling region lies within the ester family of solvents, and as ageing processes are applied, the swelling regions move towards the ketone and chlorinated solvent groups, with the most aggressively aged lacquer lying in the most polar region of the latter class of solvents.

Teas charts have been employed previously to understand the behaviour of varnishes as they age; for example, Figure 3 shows the solubility region of unaged mastic and dammar res-ins, all of which may be encountered in restoration coatings on lacquer (Horie 1987: 215). The Teas charts allow compari-sons of the solubility/swelling behaviour of different classes of materials to be compared; thus, Figure 4 shows the swell-ing region of the most damaged (HgW-aged) lacquer, deter-mined using the data obtained in the vapour sorption tests in comparison to the solubility region of aged natural resin varnishes. The interpretation of these Teas charts is discussed further below.

Dynamic vapour sorption

Improving on the gravimetric analysis approach, lacquers were exposed to a range of solvent vapours and their uptake was measured by dynamic vapour sorption (DVS). This tech-nique is highly accurate and reproducible, although is typi-cally restricted to a more limited range of solvents; acetone and the aromatic solvents, for example, were not compatible with our DVS equipment. Unlike the manual vapour sorp-tion experiment described earlier, the sample mass is meas-ured continuously within an atmosphere of specified solvent vapour pressure, thereby eliminating the error introduced on removal from the saturated atmosphere for weighing, and allowing observations of absorption kinetics and the deter-mination of isotherms. The determination of vapour sorption and diffusion rates of organic solvents or water are useful in a range of applications such as in the food industry, pharma-ceuticals and polymers (Levoguer and Williams 1998; Buckton and Darcy 1995; Roman-Gutierrez et al. 2002). In the current

% wt. uptake of solvents

Solvent Molar volume (cm3/mol)

unaged Daylight aged (500 hours)

uv-daylight aged (3,500 hours)

HgW aged (2,000 hours)

Pentane 111.00 1.3 0.5 0.0 1.3

Hexane 131.31 2.7 0.0 2.0 2.9

Benzene 89.48 4.9 10.9 4.3 9.7

toluene 106.56 12.6 62.9 33.6 39.4

chloroform 80.66 40.3 83.1 51.3 78.0

Diethyl ether 105.50 4.2 13.0 22.4 9.1

tetrahydrofuran 82.44 41.2 68.4 48.0 92.0

ethyl acetate 98.54 79.0 29.1 76.1 76.0

acetone 73.93 13.3 33.0 29.2 33.3

Butan-2-one 90.10 12.8 26.1 45.6 51.0

ethanol 58.52 15.3 22.1 15.4 16.6

acetonitrile 52.68 4.3 6.3 5.2 16.7

Water pH 5.5 18.07 5.0 2.8 5.7 1.3

Table 4 Solvent uptake in vapour sorption experiments.

lacquer type

Method used δd (Mpa1/2) δp (Mpa1/2) δh (Mpa1/2)

unaged Graphical estimate

11.30 4.30 6.50

Weight averaged

16.53 5.56 1.85

Weight averaged (v)

16.53 5.05 6.76

Weight averaged (v1/2)

16.54 5.26 1.19

Daylight aged 500 h

Graphical estimate

16.70 5.50 8.50

Weight averaged

16.60 6.19 1.42

Weight averaged (v)

16.67 5.64 6.68

Weight averaged (v1/2)

16.64 5.89 6.99

HgW aged 2000 h

Graphical estimate

16.60 5.80 6.00

Weight averaged

16.61 6.52 1.54

Weight averaged (v)

16.71 5.64 6.20

Weight averaged (v1/2)

16.67 6.03 6.72

uv-daylight aged 3500 h

Graphical estimate

16.70 5.70 4.80

Weight averaged

16.47 6.50 8.04

Weight averaged (v)

16.62 5.20 5.79

Weight averaged (v1/2)

16.56 5.76 6.66

Table 5 3D solubility parameters obtained for lacquer exposed to vapour sorption by graphical estimation and weight-averaged methods.

EAL.indb 64 08/07/2011 09:31

65

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

Figure 3 Region of solubility for an unaged (purple) and aged (yellow) dammar, and an unaged (red vertical lines) and aged (green vertical lines) mastic varnish.

Figure 2 Teas chart showing the position of the swelling regions of the unaged (blue), daylight-aged (green), mercury-tungsten-aged (yellow) and UV-daylight aged (red) films as determined by calculated solubility parameters from vapour sorption data.

EAL.indb 65 08/07/2011 09:31

66

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

Figure 4 Graphical summary of swelling/solubility region of HgW artificially aged lacquers and natural resin varnishes, including dammar (yellow ellipse), shellac (pink ellipse) and mastic (green ellipse). The high swelling region for HgW artificially aged lacquers lies within the blue circle.

context, DVS provides a very good indication of the response, and therefore susceptibility to damage, of the lacquer to a par-ticular solvent type.

In DVS, the vapour partial pressure surrounding the lac-quer film is controlled by combining saturated and dry carrier-gas streams. The temperature within the sample chamber is kept constant to ± 0.1°C since it is enclosed in a temperature- controlled chamber; the following experiments were carried out at 30°C using the DVS-HT High Throughput system,3 which has ten stainless steel sample pans (10 mm dimen-sion). Typical samples consisted of around 0.1 g of (aged) lac-quer, approximately 9 µm thick. The fundamental data (see for example Figures 5 and 6) related the mass change observed at a given target relative vapour pressure of solvent; the shape of the mass curve at each vapour pressure step gives an indication of how well the system has equilibrated. For the current study, samples were conditioned at 0% RH for 1500 min (not shown), and then subjected to two absorption/desorption cycles, with the relative vapour pressure adjusted in 10% steps.

Figure 7 shows the uptake of octane, ethanol, ethyl ace-tate, acetone and water measured for lacquer exposed to UV- daylight ageing for 20 weeks (~3,500 hours; longer in the case of water and ethanol).

DVS isotherms, derived from the absorption/desorption curves, show the equilibrium solvent uptake as a function of vapour partial pressure. The isotherms for five different sol-vents were obtained for a range of photodegraded lacquers, and the sorption properties, including desorption and resorp-tion cycles, of the films determined. However, the isotherm

is only well defined if the DVS data indicate an equilibrated response, as in the case of water (see Figure 8). The change in mass (%) plot shows the percentage change in mass rela-tive to a dry mass established during the initial conditioning at 0% P/P0.

Non-polar solvents: alkanesOctane (vapour pressure 1.5 kPa, 30°C) was used as a repre-sentative non-polar alkane solvent. As expected, the uptake was minimal for all the lacquers tested (an example is shown in Figure 6). For a lacquer film aged for 3,500 hours under UV-daylight simulation, there was a maximum uptake of 0.17% octane, which desorbed completely, followed by an uptake of 0.14% in the second cycle. The small uptake of alkane by the lacquers, including the aged samples, is consistent with immersion and vapour sorption measurements described above for hexane, cyclohexane and pentane (see Tables 2 and 4). Limited swelling of the crosslinked lacquer is seen in the case of long-chain, non-polar alkane solvents, since these have a low affinity for the saturated and polar constituents of the urushiol molecule or the other water-soluble components of the lacquer.

AlcoholAlcohols are a common conservation choice for the removal of natural resin varnishes. An example of the DVS response of a UV-daylight-aged lacquer to ethanol (vapour pressure 14kPa, 30°C) is shown in Figure 5; clearly, the initial swelling of the dry film is kinetically limited and is not manifested until relatively

EAL.indb 66 08/07/2011 09:31

67

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

Figure 5 Uptake of ethanol by UV-daylight-aged lacquer aged for 3,500 hours, measured by DVS.

Figure 6 Uptake of octane by UV-daylight-aged lacquer for 3,500 hours, measured by DVS.

EAL.indb 67 08/07/2011 09:31

68

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

Figure 7 Summary of mass changes of UV-daylight-aged lacquers on exposure to a range of solvents over increasing ageing time measured by DVS.

Figure 8 DVS isotherm for water sorption, desorption and resorption cycles on UV-daylight-aged lacquer (20 weeks, 3,500 hours).

high vapour pressures, but it eventually reaches around 24% mass uptake. In the timescale of the desorption cycle, not all solvent is removed from the film (about 4% is retained), allowing a more rapid and better equilibrated response in the second adsorption cycle, but reaching similar maximum/

minimum values. The response for the different (aged) lac-quer samples is summarized in Table 6. Consistent with the observed changes in solubility parameters, the uptake of ethanol (dipolar HBD) increases gradually with increasing ageing time.

EAL.indb 68 08/07/2011 09:31

69

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

WaterDVS experiments were carried out using water at pH 5.5 (vapour pressure 4.3 kPa, 30°C). The unaged lacquer exhib-ited a varied uptake of between 2 and 4.1% water. The uptake of water by UV-daylight-aged lacquer increased as the ageing time (and therefore degree of photodegradation) increased up to 12% for lacquer UV-daylight aged for one year, as shown in Table 7. Unlike the ethanol, in all the water experiments kinetic effects were minimal with the DVS showing a well-equilibrated response, returning to 0% uptake after both desorptions, and the same maximum uptake for each adsorption cycle. The derived isotherms (Figure 8) can therefore be considered reliable.

Ethyl acetateA member of the ester solvent group, ethyl acetate was chosen to confirm the results of the vapour sorption tests, which sug-gested a greater response to ester solvents from unaged lacquer in comparison to aged lacquer. An unaged sample absorbed 25 wt%, the highest increase seen from any of the test sol-vents; the desorption cycle ended with the lacquer retaining 5 wt% solvent. For UV-daylight-aged (20 weeks, ~3,500 hours) lacquer, the uptake was 15 wt%. After the initial desorption cycle, the lacquer had a weight increase of 6% and the second cycle produced an increase of another 15 wt%. The mercury- tungsten-aged sample followed a very similar pattern to the UV-daylight-aged sample – a weight increase of 16% was observed, 5% was retained after desorption, and a further 12% was absorbed in the second cycle. The results confirm the suscepti-bility of the freshly cured lacquer to the relatively less polar non-HBD ethyl acetate, and highlight a similar slow kinetic response, as seen with ethanol.

Scanning electron microscopy

In addition to the solvent-swelling tests, lacquers were examined using scanning electron microscopy (SEM) (Hitachi S3400-N variable pressure, acceleration voltage 15.0 kV) to evaluate

lacquer type 1st sorption cycle mass uptake (%)

retention after desorption cycle (%)

2nd sorption cycle mass uptake (%)

unaged 18 3 18.0

uv-daylight aged (3,500 hours)

19 4 13.5

HgW aged (2,000 hours)

20 4 14.0

Table 6 Ethanol uptake by aged lacquers for two cycles and retention of solvent following desorption.

length of ageing time (hours) uptake of water (%)

0 2–4.1

1350 4.3

3500 5.4

8700 12

Table 7 Percentage mass uptake of water by UV-daylight-aged lacquer measured by DVS. Note that for these samples, the first and second adsorption cycles reached the same maximum uptake and returned to 0% uptake after the desorption in both cases.

changes in surface morphology. Samples were taken from an unaged board, boards artificially aged using HgW and xenon-arc light sources (12 and 20 weeks, or 2,000 and 3,500 hours, respectively), and a naturally aged sample from the lacquered frame of a nineteenth-century Japanese screen. A small area (~2 mm2) of each sample was examined by SEM before and after exposure to solvent; where recognizable, the same region of the sample was photographed. The solvent (~0.2 ml) was intro-duced by rolling a swab over the surface, drying for 24 hours, re-swabbing, and then storing in the dark for seven days at ambient temperature and RH. Images captured at ×1000 magnification before and after solvent application are included in the follow-ing discussion of the results for different solvents. The freshly cured samples presented a smooth surface, whereas all the aged surfaces displayed a network of cracks. The cracks produced by artificial aging were wider, straighter and more sharply defined than the natural cracks, but on a similar length scale, forming plateaus on the order of tens of microns. Provisional comments for various solvents are discussed below; statistically rigorous analysis would require further study of additional sample pairs under more controlled imaging conditions (note that the before and after images below were collected using different second-ary electron modes).

Non-polar solvents: alkanesSEM analysis confirmed that alkanes (hexane and Exxsol DSP 80/110) had little effect on the morphology of the surface. No exacerbation of larger pre-existing cracks was observed after swabbing the aged lacquers with these solvents. Surface debris seen on the before-swabbing image of the artificially aged lac-quer was removed by swabbing with Exxsol DSP 80/110 but the original surface appeared largely unchanged. No leaching effects, indicated by weight loss of the film after immersion, were observed following immersion of aged and unaged free-film samples in hexane.

Alkenes/aromaticsAlkene/aromatic-based solvents, as exemplified by HAN 8070 and xylene, had similarly little impact on the surface morphol-ogy, leaving cracking patterns undisturbed and partially delami-nated lacquer fragments in place. In the vapour sorption tests, toluene uptake, after 24 hours exposure, was considerable in the aged lacquers, indicating that aromatic solvents do inter-act with the surface over long exposure periods; however, over limited exposure times associated with swabbing, no problems occurred. Significantly, no leaching effects from aromatic sol-vents were observed during immersion tests.

AlcoholAfter swabbing with ethanol, some cracks become less distinct or disappear while others lengthen or appear; the effects are observed in both naturally aged (Figures 9a and b) and arti-ficially aged samples but are most severe in the HgW regime (Figures 10a and b). The change in appearance could be attrib-uted either to an erosion of the existing crack/plateau structure or to deposition of material in the existing cracks, in either case with fresh cracks appearing in the new upper surface; possibly both processes could be involved. The immersion tests do not show any leaching in ethanol, but rather a high degree of swell-ing for degraded lacquer.

EAL.indb 69 08/07/2011 09:31

70

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

a

b

a

b

Clearly, alcohols have an effect on the surface of the pho-todegraded lacquer. While these solvents might have the abil-ity to dissolve or swell aged natural resin varnishes, they also have a detrimental effect on the underlying photodegraded lacquer.

WaterThe pH of water has a significant effect on the swelling of lac-quer. The effect of altering the pH of water used in cleaning moisture-sensitive, aged lacquer has been reported (Schellmann and Rivers 2005). While it was not possible to alter the pH of water in the DVS or vapour sorption experiments, water was swabbed over the surface of naturally and artificially aged lac-quer samples using pH values of 3, 5.5 and 8. The results are shown in Figure 11. Although pH may affect the extent, broadly similar effects were observed in all cases: an extension of the existing crack network and the appearance of additional pin-holes in the film, sometimes accompanied by the removal of debris from the surface. The effects occur in both the naturally (Figures 11c, d, g and h) and artificially (Figures 11a, b, e and f ) aged samples, but most strongly in the HgW regime. Since

Figure 9 Naturally aged lacquer (a) before and (b) after swabbing with ethanol.

Figure 10 Artificially aged (HgW) lacquer (a) before and (b) after swabbing with ethanol.

the immersion experiments do not indicate leaching in water, the changes in morphology may be attributed to the swelling and subsequent contraction of the upper layers, though to a lesser extent than is seen with ethanol.

Polar solvents with carbonyl groupsAcetone: The naturally aged lacquer samples (Figures 12a and b) show relatively little change following acetone swabb ing, other than the removal of some surface debris. In con-trast, the artificially aged samples, particularly the HgW regime (Figures 12c and d) show significant change, similar to that observed with ethanol. The artificially aged samples have a significant leachable component in acetone (see Table 2) and there was substantial leaching from all lacquer types when exposed to butan-2-one (ranging from 2.9 to 5.6%).Ethyl acetate: Swabbing with ethyl acetate appeared to have minimal effects on the naturally aged piece (Figures 13a and b) but accentuated the hairline cracks on the artificially aged surface (Figures 13c and d). No leaching effects were observed through immersion tests for this solvent on any lacquer type.

EAL.indb 70 08/07/2011 09:31

71

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

Figure 11 Artificially aged (HgW) lacquer (a) before and (b) after swabbing with water of pH 3; naturally aged lacquer (c) before and (d) after swabbing with water of pH 3 (c and d are unmatched images); artificially aged (HgW) lacquer (e) before and (f ) after swabbing with water at pH 5.5; naturally aged lacquer (g) before and (h) after swabbing with water at pH 8.

a b

c d

e f

g h

EAL.indb 71 08/07/2011 09:31

72

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

Figure 12 Naturally aged lacquer surface (a) before and (b) after acetone swabbing; artificially aged (HgW) lacquer (c) before and (d) after swabbing with acetone (c and d are unmatched images).

Figure 13 Naturally aged lacquer (a) before and (b) after swabbing with ethyl acetate; artificially aged (HgW) lacquer (c) before and (d) after swabbing with ethyl acetate.

a b

c d

a b

c d

EAL.indb 72 08/07/2011 09:31

73

S o lv e n t e f f e c t S o n e a S t a S i a n l a c q u e r ( T o x i co d e n d r o n v e r n i c i f lu u m )

Discussion

The purpose of these investigations was to determine whether a solvent or solvent mixture could be used to dissolve or swell unwanted materials – specifically photodegraded natural resin varnishes – from photodegraded lacquer without damaging the original surface. The behaviour of photodegraded lacquer exposed to individual solvents was studied.

Solvents interact with urushi lacquer by diffusing into the extensively crosslinked system and swelling it to some degree. Some fragments, such as low molecular weight components or degradation products, may be leached or dissolved away. The solvent–lacquer interaction is dependent on the charac-teristics of the solvent (solvent type, molecular size and the functional groups it contains) and the condition of the lacquer (degree of crosslinking within the lacquer, scission of lacquer and other components present). Urushi becomes more acidic as it degrades, and contains increasingly more polar groups within the surface as a result of oxidation processes that cause the formation of ketone groups and acids (Hong et al. 2000). The use of polar solvents would, therefore, be expected to cause greater swelling of the uppermost lacquer layers after ageing and associated photodegradation.

To investigate the effects of a range of solvent types on aged lacquers, experiments were conducted on lacquer films (including unaged, natural daylight, HgW- and UV- daylight-aged lacquers) in two parts: first, by complete immer-sion of a free film, and second, by exposing the film to a solvent-saturated atmosphere. In these tests, the solvent uptake was not restricted in any direction as the film was not mounted on any substrate, allowing solvent-induced strains (i.e. curl-ing). The effect of brief solvent exposure to damaged lacquer surfaces was also studied by SEM using lacquer boards, as described earlier. The combined results from the immersion, vapour sorption and SEM tests show that photodegraded lac-quer surfaces are most severely affected by more polar solvents such as ethanol, acetone and ethyl acetate, as demonstrated by the large uptake of solvent by mass, while the tendency of the freshly cured lacquer to swell in non-polar solvents dis-appears with ageing. Solvent immersion experiments showed that material (around 2–3 wt%) leaches from even the fresh lacquer in chloroform, dimethylformamide, tetrahydrofuran and butanone (see Table 2); however, after ageing, the leach-ing effects increase and are manifested in a broader range of solvents. The leaching tendency (relative to swelling) is par-ticularly strong after the unfiltered HgW exposure, including uniquely in aromatics, potentially suggesting a differing deg-radation mechanism to the other ageing processes, involving a high degree of chain scission; these effects may not, therefore, accurately reflect the damage that occurs for objects stored under museum conditions.

The use of solvent vapour exposures rather than immersion provides more reliable swelling data but no information about leaching. Burke (1984) suggested that the Teas chart of frac-tional parameters can be useful in predicting the behaviour of a mixture of solvents, enabling the selection of ‘safe’ solvents in terms of toxicity and evaporation rate, as well as an ability to swell one material and not another. The advantages of using a mathematical or graphical determination of the position on the chart of a particular solvent blend include the reduction of

trial and error tests required to determine solvent behaviour. The swelling regions and solubility parameters calculated from the swelling behaviour indicated that prolonged or harsh age-ing tends to move the solubility region in the direction of an increasingly polar solvent, particularly towards the chlorinated solvent region. The unaged lacquer seemed to be most respon-sive to ester solvents such as ethyl acetate (a result confirmed by the subsequent DVS experiments) and tetrahydrofuran; it is worth noting that the fresh lacquer contains predominantly ester linkages due to the acid–alcohol reactions during the curing process, which may explain this affinity. As the lacquer ages, presumably by photo-induced oxidation, it becomes less responsive to the ester class of solvents and more responsive to more polar, chlorinated or ketone solvents. These trends are very similar to the effects observed for aged natural resin var-nishes. This result suggests that it will be virtually impossible to select a solvent to remove a given varnish, especially one of uncertain identity or condition, which will intrinsically not also potentially interact strongly with the underlying lacquer, at least under equilibrium conditions. On the other hand, the practical response to cleaning solvents depends not only on the thermodynamic solvent character of the lacquer or var-nish, but also on the kinetics of the solvent uptake by the vari-ous layers, and the resulting changes in solubility or resistance to mechanical abrasion. The kinetics of the swelling process may be significant, both in controlling differential strains and the rate of softening or leaching relative to any swabbing or other conservation process. In other words, technique may be as important as solvent choice.

The use of DVS, with uptake measured under continuous environmental/atmospheric conditions, provides a more accu-rate indication of the lacquer films’ equilibrium tendencies to take up, and retain, different solvent types, as well as pro-viding an indication of the kinetics for these processes. For the compatible solvents, DVS confirmed the earlier swelling experiments, showing minimal impact on the lacquer by an alkane, a moderate impact by water, and a significant impact by ethanol and ethyl acetate. For comparison, unaged lac-quer exhibited an uptake of 18% and 15.3% by DVS and the lab-based vapour uptake, respectively, for ethanol, and 4.1% and 5%, respectively, for water. Interestingly, ethanol uptake in lacquer, although significant, displayed relatively very slow kinetics, especially for the initial swelling of a dry film. This observation may help to explain the success of the common use of ethanol as a cleaning solvent in conservation, despite the clear evidence in this study of the potential for damage. In terms of conservation practice, acetone and alcohol are used both for cleaning and varnish removal from lacquer. However, this research indicates that exposure to these sol-vents can have a noticeable effect on photodegraded lacquer. Immersion and vapour sorption tests indicated significant swelling, while the SEM experiments, carried out before and after the lacquer surface was swabbed with high risk (in terms of their retention and swelling properties) solvents (ethanol, toluene, chloroform, ethyl acetate), show that alter-ations of the surface occur as a result of removal of the photo- degraded surface near the cracks, leading to the appearance of deeper cracks and, in some cases, initiating the generation of new hairline cracks. It is worth noting that visible dam-age to the surface is not caused by the swelling directly, but

EAL.indb 73 08/07/2011 09:31

74

c a r o ly n M c S H a r r y, r u P e r t fau l k n e r , S H ay n e r i v e r S , M i lo S . P. S H a f f e r a n D to M W e lto n

by a variety of mechanisms, including subsequent shrinkage, cracking, leaching and abrasion and possibly redeposition. The mechanical effects are themselves complicated depending on differential strain, softening and loss of interlaminar adhe-sion; micro-scale damage, such as cracking, may itself alter the critical kinetics by allowing rapid solvent access to vul-nerable underlayers, encouraging delamination, blooming or other problems. The mechanical effects are most evident in the SEM studies that show crack widening, new crack formation, leaching and/or removal of small fragments. Qualitatively, the greater the degree of ageing of the lacquer, the greater the ten-dency for microscale damage during swabbing. More detailed, quantified studies could be performed in the future.

The suitability of solvents for cleaning (degraded) urushi lacquers can be briefly summarized as follows. Alkanes are potentially useful for cleaning dust and some dirt from the surfaces without causing swelling, however, oxidized sub-stances such as aged varnish bonded to the original lacquer would not be dissolved or swelled by the solvent. Aromatic solvents exhibit similar behaviour, except for strongly UV-daylight-degraded lacquers for which leaching becomes significant. Polar solvents with fast evaporation and slow swelling rates, such as acetone or ethanol, may remove polar varnishes, while minimizing the opportunity for penetration below the uppermost lacquer layers, providing some reduc-tion of the potential risk despite their strong intrinsic swell-ing character; these solvents also have relatively low toxicity and good applicability in practical terms. However, the inev-itable presence of cracks in aged artefacts allows access to the underlayers and limits the value of this kinetic protec-tion. Butan-2-one is a moderate sweller and a high-leaching solvent; the slow evaporation rate means that the exposure time is prolonged, thereby exacerbating the potential risk. Ethyl acetate was seen to accentuate the existing cracks on an artificially photodegraded lacquer, but did not cause any meas-urable leaching. It has a particularly strong swelling effect on freshly cured lacquer but the effects remain significant after ageing.

In conclusion, there are no solvents that can be considered completely safe for removing a photodegraded natural resin varnish from a photodegraded lacquer surface without causing some degree of damage to the original surface. Conservators will need to use trial and error to determine the effects of sol-vents on each object as well as assess and balance the poten-tial risk of loss of original surface and decoration against the perceived benefits of removing an unwanted coating.

Acknowledgements

The authors would like to thank Dr Daryl Williams and Dr Majid Naderi of Surface Measurement Systems Ltd. for the use of their spe-cialist dynamic vapour sorption equipment. They are also grateful to Dr Ambrose Taylor of the Department of Mechanical Engineering, Imperial College London, for providing access to the scanning elec-tron microscope, and acknowledge Judith Thei, Department of Me-chanical Engineering, Imperial College London, for her involvement with the SEM analysis.

Notes

1. Obtained from Watanabe Shoten 6-5-8 Ueno, Taitō-ku, Tokyo 110-0005, Japan (http://www1.odn.ne.jp/j-lacquer/home_eng.html).

2. Obtained from VWR International Ltd., Hunter Boulevard, Magna Park, Lutterworth, Leics LE17 4XN (http://uk.vwr.com).

3. Designed by Surface Measurement Systems (SMS) Ltd., 5 Wharfside, Rosemont Road, Alperton, Middx HA0 4PE (http://www.thesorptionsolution.com).

References

Buckton, G. and Darcy, P. 1995. ‘The use of gravimetric studies to assess the degree of crystallinity of predominantly crystalline powders’, International Journal of Pharmaceutics 123: 265–71.

Burke, J. 1984. ‘Solubility parameters: theory and application’, AIC Book and Paper Group Annual 3: 13–58.

Hong, J-W., Park, M-Y., Kim, H.-K. and Choi, J.-O. 2000. ‘UV- degradation chemistry of oriental lacquer coating containing hin-dered amine light stabilizer’, Bulletin of the Korean Chemical Society 21(1): 61–4.

Horie, C.V. 1987. ‘Appendix 3: Solubility charts of polymers’, in Materi-als for Conservation: Organic Consolidants, Adhesives and Coat-ings. Oxford: Elsevier Butterworth-Heinemann, 193–223.

Levoguer, C. and Williams, D. 1998. ‘Determining permeability and diffusion rates in polymer films and packaging materials’, Medical Device Technology 9: 14–18.

McSharry, C. 2009. ‘Conserving Tangible and Intangible Cultural Her-itage: Cleaning Degraded East Asian Lacquer’, PhD thesis. Depart-ment of Chemistry, Imperial College London.

McSharry, C., Faulkner, R., Rivers, S., Shaffer, M.S.P. and Welton, T. 2007. ‘The chemistry of East Asian lacquer: a review of the sci-entific literature’, Reviews in Conservation 8: 29–40.

Rivers, S. 2002. ‘Removal of varnish from japanned and lacquered surfaces: principles and practice’, in The Meeting of East and West in the Furniture Trade: Proceedings of the 6th International Sym-posium on Wood and Furniture Conservation, P. van Duin and H. Piena (eds). Amsterdam: Stichting Ebenist, Rijksmuseum, 52–8.

Rivers, S. and Umney, N. 2003. ‘Oriental lacquer’, in Conservation of Furniture. Oxford: Butterworth-Heinemann, 760–70.

Rivers, S. and Yamashita, Y. 2006. ‘A cross-cultural approach to lacquer conservation: consolidation of metal foil decoration on the Mazarin Chest’, in The Object in Context: Crossing Conservation Boundaries. Preprints of the 21st IIC Congress, Munich, 28 August–1 September 2006. London: IIC, 286–92.

Roman-Gutierrez, A.D., Guilbert, S. and Cuq, B. 2002. ‘Distribution of water between wheat flour components: a dynamic water vapour adsorption study’, Journal of Cereal Science 36: 347–55.

Schellmann, N. and Rivers, S. 2005. ‘Aqueous cleaning of photo-degraded East Asian lacquer: a preliminary investigation of the effect of pH and ionic concentration on gloss and colour’, Zeitschrift für Kunsttechnologie und Konservierung 19(2): 369–76.

Webb, M. 2000. Lacquer Technology and Conservation: A Comprehen-sive Guide to the Technology and Conservation of both Asian and European Lacquer. Butterworth-Heinemann Series in Conserva-tion and Museology. Oxford: Butterworth-Heinemann.

Zellers, E.T., Anna, D.H., Sulewski, R. and Wei, X.-R. 1996a. ‘Criti-cal analysis of the graphical determination of Hansen’s solubility parameters for lightly crosslinked polymers’, Journal of Applied Polymer Science 62(12): 2069–80.

Zellers, E.T., Anna, D.H., Sulewski, R. and Wei, X.-R. 1996b. ‘Improved methods for the determination of Hansen’s solubility parameters and the estimation of solvent uptake for lightly crosslinked poly-mers’, Journal of Applied Polymer Science 62(12): 2081–96.

EAL.indb 74 08/07/2011 09:31