The application of chemistry to conserve cultural heritage.
The preservation of material cultural heritage represents a set of values that began several centuries ago with the formation of museum collections. The value of objects as a way of informing people about other cultures than their own is highly significant and gives materials an unexpected transformational power. Individual outputs from a variety of cultural activities are often appreciated in a totally different context when placed in a museum collection. Princely gilded bronze slippers from Korean archaeological sites of the 8th century unfold the mysteries of an ancient civilisation for the viewer. At the time they would have been functional, decorative and a pragmatic response to a real need for keeping the royal feet clean. During the past 50 years the amount of chemistry applied to the preservation of all sorts of materials, from wood, to ceramics, glass and metallic objects has increased dramatically as materials conservation laboratories became established around the world. In Australia, the finding of a series of historic shipwrecks of ships from the Dutch Vereenigde Oostindische Compagnie (VOC) that operated in the 17th and 18th centuries catalysed activities in the Western Australian Museum. Since many of the forms of decay were peculiar to the well-oxygenated, warm sub-tropical waters of Western Australia, new methods had to be developed to successfully stabilise these historic objects. Having developed research skills in producing new methods to stabilise shipwreck materials, the museum work program was extended to include Aboriginal rock art sites and historical objects of value to the WA community. Part of the success of the WA Museum in managing these and related issues is summarised in the three case histories highlighted below, which reflect the diversity of cultural experiences found in this state.
CONSERVATION OF ABORIGINAL ROCK ART
The increased awareness of the value of culture that took place in the Australian bicentennial year in 1988 saw major funding relating to conservation research into the Aboriginal rock art in the West Kimberley region become available.
Sampling methodology was developed in agreement with the Bunuba people, the traditional owners, which included avoidance of any dating measurements on the trapped carbon in the layers of paint. Their lore told them it was their ancestors who did the paintings and that was all that was needed to be understood. Although the major emphasis of the research involved microclimate studies in both the wet and dry seasons of 1990 and 1992, the concomitant study of pigments was pivotal in developing a new management strategy for the stabilisation of these remote sites. One of the characteristics of these sites is the liberal use of the brilliant white pigment huntite, [Mg.sub.3]Ca[(C[O.sub.3]).sub.4], (Figure 1) which is found in rare deposits in the Kimberley region. In some instances, the images had been executed over a white background created by huntite being applied to the rock or some underlying paintings (Clarke, 1976). Owing to the susceptibility of carbonate mineralogy to acidic reactions produced by plant and microbiological metabolites, sampling focused in the white to grey spectrum of the apparent pigment samples. Some botryoidal accretions were also sampled to see if they could provide direct evidence of the complex interactions of plant and microbiological metabolites on the fate of the pigments.
Thirty-seven of the forty samples were taken from four sites within 30 km of each other at the base of the Napier Range near Fitzroy Crossing (Figure 2). The range is part of an exposed 370 million year old Devonian limestone reef. Three samples of the white pigments came from the Mitchell Plateau (Northern Kimberley) region of Western Australia. The frequency of the minerals identified was: fifteen samples of whewellite Ca[C.sub.2][O.sub.4].[H.sub.2]O; eleven huntite samples, nine of which also contained calcite CaC[O.sub.3]; five dolomite samples MgCa[(C[O.sub.3]).sub.2]; two samples of the dihydrate of calcium oxalate, weddellite, Ca[C.sub.2][O.sub.4].2[H.sub.2]O, which came from the walls and ceiling of a rock shelter in the Mitchell Plateau; and there were only single samples of antigorite [Mg.sub.3][Si.sub.2][O.sub.5][(OH).sub.4], a serpentine mineral, and the clay illite (K,[H.sub.30])[Al.sub.2][Si.sub.3]Al[O.sub.10][(OH).sub.2]. The iron minerals hematite [Fe.sub.2][O.sub.3] and goethite FeO.OH were the last two samples in the collection of pigments.
All seven samples of dark brown and black rock surface deposits were identified as whewellite, with minor amounts of ankerite, (Ca,Mg,Fe)C[O.sub.3], calcite, and in one case, jarosite, K[Fe.sub.3][(S[O.sub.4]).sub.2][(OH).sub.6] and pyrite Fe[S.sub.2]. In many of these samples from the limestone Napier Range, only whewellite was present as the major (>90%) constituent and many of these included one or more complete paint layers. Weddellite was found in sandstone country along the walls and ceiling of a shelter in the Mitchell Plateau (Figure 3). Dolomite samples were found both in the presence and absence of whewellite. Goethite and hematite were found to be components in samples of red pigment that also included kaolin and mica, which probably formed a background prior to the application of the red pigment. The aluminium phosphate mineral taranakite, [K.sub.3][Al.sub.5][H.sub.6](P[O.sub.4])8.18[H.sub.2]O, was a major component in a layer deposited on the surface of a weathered sandstone site at one of the sites in the Mitchell plateau, along with quartz and minor amounts of kaolinite and calcite. Although this deposit overlay one section of the paintings, it is not believed to have been a pigment. The material overlaying some of the images adjacent to the source of the taranakite consisted mainly of amorphous silica and some whewellite.
The only clay mineral used as a pigment was illite, which came from the Napier Range. This particular pigment also contained a trace of ankerite, Ca([Mg.sub.0.67][Fe.sub.0.33])[(C[O.sub.3]).sub.2], along with gypsum CaS[O.sub.4].2[H.sub.2]O which was detected in several pigment samples as a minor component. The element sulphur was ubiquitous in all of the pigments and occurred at higher levels in the whewellite samples in the form of gypsum. The gypsum present in the mineral samples is probably due to contamination of the material with varying amounts of weathered rock substrate.
The antiquity of some sites in the Napier Range is indicated by the number of layers revealed in cross-sections, in which many different coloured pigment strata are interspersed with what appear to be mineral accretions--most likely calcite, gypsum or oxalates. A typical section is shown in Figure 4. Many of these layers contain quite extensive deposits of charcoal, which occur both as a result of the deliberate use of charcoal, either alone or in mixtures, with other pigments. Deposition on exposed surfaces during bushfires must also occur since charcoal can be found embedded in non-pigment mineral accretions and rock skins not associated with art. Also revealed in cross-sections are black layers, which do not appear to be charcoal. They may be composed of calcium oxalates or gypsum, both of which have been previously described as black deposits on rock surfaces in tropical and other environments, or they may be organic deposits of some kind (Watchman, 1990). North and Clarke in 1998 noted that finely divided minerals associated with dissolution/reprecipitation phenomena will often trap incident light and so appear to be much darker than the massive forms of the same mineral.
MORPHOLOGY OF THE WHITE TO CREAM PIGMENTS
The huntite samples were of a vivid white appearance, and were all reasonably pure, only containing small quantities of quartz, charcoal, and yellow to red-brown particles of muskoxite, [Mg.sub.3][Fe.sub.4][O.sub.13].10[H.sub.2]O, in which the iron has a formal oxidation state of five. Electron microscopy revealed rhombohedral plate-like huntite crystals about 1-2 [micro]m in size. Another iron and magnesium containing impurity found in a sample of huntite from a "quarry" in the West Kimberley was the mineral amakinite, [Fe.sub.0.73][Mg.sub.0.22][Mn.sub.0.05][(OH).sub.2], which has helped to source the white mineral pigments used in the paintings. The calcite pigments were all grey, except for one, which was cream. Microscopically they are open, porous masses, in which smaller pigment particles (1-5 [micro]m) clump together in 10-30 [micro]m aggregates. Their grey colour is a result of the admixture of charcoal of highly variable particle size (typically 10-500 pm). Without charcoal, the paint is cream in colour and easily distinguishable from the brilliant huntite whites. All of the calcite pigment layers contained significant amounts of light red-brown particles ranging up to 100 pm, and studies showed that they were all iron-stained ortho-quartzites. The Mitchell Plateau antigorite pigment sample was more compact than typical calcite pigments, but it was more sugary and porous in appearance than huntite. The particles were generally around 1 [micro]m and of a flattened cubic shape.
THE NATURE OF WHEWELLITE PIGMENT LAYERS
Our report of the widespread presence of oxalates as pigments was the first record of oxalates as pigments in their own right, or of them being the major alteration of other pigments to whewellite (Ford, MacLeod & Haydock, 1994). Since there are no major deposits of whewellite in the Kimberley region (Playford, 1992) its presence is most likely due to it being an alteration product. Oxalic acid can come from air pollution during lightning storms or from microbiological activity associated with fungi or algae within the shelters. Samples of rock and dust taken under sterile conditions were cultured and found to be as high as 850,000 bacteria per gram of substrate (MacLeod, Haydock, Tulloch & Ford, 1995). A series of laboratory experiments with huntite at the natural pH of 5.5-5.9, with and without oxalate sources, confirmed that the vivid white pigment is converted at room temperatures of 22 [+ or -] 2[degrees]C to whewellite. The reaction products identified by XRD included synthetic brucite, Mg[(OH).sub.2], vaterite CaC[O.sub.3] and dolomite MgCa[(C[O.sub.3]).sub.2]. This data is consistent with a stepwise dissolution and re-precipitation mechanism since vaterite was also found in association with whewellite at the rock art sites. The admixture of whewellite, dolomite and calcite found in the reaction products strongly indicates that these pigments are all alteration products of huntite. Equations describing the mineral changes are outlined below:
[Mg.sub.3]Ca[(C[O.sub.3]).sub.4] + 3[H.sup.+] + [H.sub.2]O + [C.sub.2][O.sup.2-.sub.4] [right arrow] Ca[C.sub.2][O.sub.4].[H.sub.2]O + 3[Mg.sup.2+] + 3HC[O.sub.3] + C[O.sup.2-.sub.3] (1)
[Mg.sub.3]Ca[(C[O.sub.3]).sub.4] + 2[H.sub.2]O [right arrow] Mg[(OH).sub.2] + MgCa[(C[O.sub.3]).sub.2] + [Mg.sup.2+] + 2HC[O.sub.3] (2)
MgCa[(C[O.sub.3]).sub.2] + [H.sup.+] [right arrow] CaC[O.sub.3] + [Mg.sup.2+] + HC[O.sub.3] (3)
The fact that whewellite was only found either as the major or a minor component of a pigment layer, never in between, may reflect the proximity of the pigments to sources of oxalate ions. Owing to the exposure of huntite to oxalate ions producing dolomite, it is contended that the dolomite is present as a result of the dissolution of huntite and not as a pigment in its own right. The traditional elders indicated that the old people had only used the one quarry source of almost pure huntite in the execution of the images. Thus the found dolomite is consistent with it being derived from a thermodynamically favoured route for the dissolution of huntite to produce dolomite and magnesite, MgC[O.sub.3] (Konigsberger & Gamsjager, 1987). Huntite has a very high surface area of 13.5 square metres per gram and has been shown to absorb multilayers of water after the initial monolayer has been adsorbed. Thunderstorms produce sulphuric acid via [H.sub.2]S oxidation and provide a ready source of S[O.sub.4.sup.2-] ions to precipitate the commonly found CaS[O.sub.4].2[H.sub.2]O gypsum mineral. The 1992 wet season measurements on 1987 repainted areas showed already some areas of significant conversion of huntite into dolomite in just five years of outdoor exposure. Regardless of whether whewellite in pigment layers has occurred as a result of alteration processes involving local biological activity or atmospheric oxalates, the resulting changes appear to preserve the art in terms of the retention of the shape and style of the images. The main form of conservation management is to prevent grazing stock from rubbing up against the painted rock surfaces and to introduce "drip lines" of silicon sealant that prevents water from coursing across the painted surfaces during heavy rain events.
THE DE VLAMINGH PLATE: A STORY FROM THE 17TH CENTURY
Dirk Hartog named the eponymous island off the most westerly part of the Australian mainland in 1616 and became the first European to leave evidence of his visit. His documentation took the form of a flattened pewter dinner bowl onto which he inscribed the names of himself, the crew and the ship that brought him there. This plate was found by fellow Dutch explorer, Willem de Vlamingh, when he landed at the same place in February 1697 and copied the Hartog text onto another flattened pewter plate before adding his own record. The Hartog plate was taken to Batavia (Jakarta) by de Vlamingh and it is now exhibited in the Rijksmuseum, Amsterdam, The Netherlands. The de Vlamingh plate was fixed to a wooden post of Rottnest Island pine with rectangular iron ship's nails and set into the ground. The French natural scientist Emmanuel Hamelin in 1801 found the plate lying face down on the windswept calcareous rocky landscape. He was so impressed with the plate that he attached it to a new post using square-shanked bronze sheathing nails from his vessel. The plate was ultimately recovered by Louis de Freycinet in 1818 who had witnessed Hamelin's commitment to heritage preservation in 1801. The rest of the history of the pewter plate and how it came from France to Australia in 1946 can be found in the excellent book by Playford and Hesselsz (1998). The award of a grant from the Australian Synchrotron in Melbourne allowed the museum team to spend four days minutely examining the surface of the historic plate to reveal its history (MacLeod, Thurrowgood, Pohl, Howard, Patterson, 2014). The X-ray fluorescence microscopic (XFM) analysis revealed an array of hidden conservation problems and the way in which the coastal environment had created a unique corrosion record as observed in the decay mechanism of the plate.
MECHANICAL AND STRUCTURAL DAMAGE ASSESSMENT
The naked plate was X-rayed at the National Gallery of Victoria, using a beam from a 60 kV X-ray tube for one minute. The image showed large gaps in the metal structure, from a combination of mechanical and corrosion damage. This damage was most profound in and around the original bowl depression. Without the X-ray analysis the conservation team would not have been aware of the advanced state of decay of the plate. The combined impact of physical and corrosion damage has made the plate very fragile (Figure 5). The reverse side of the plate showed up a series of striations, approximately 30 [micro]m in width, in the area which once formed the bowl. The crisscross pattern of scratches is consistent with them being made by crew members on de Vlamingh's ship de Geelvinck (Yellow Finch) before the plate was requisitioned for making the inscription. In the absence of conservation records Fourier Transform Infrared (FTIR) spectroscopic analysis confirmed it was a microcrystalline wax that had been used to stabilise the corroded pustular surface. Additionally, it confirmed that a damaged section at the rim had been reattached using an epoxy resin. Radiating cracks were found around the original iron nail holes (12.8 x 2.6 mm or 1/2 x 1/10 Amsterdam inch) were indicators of stress corrosion cracking of the pewter (Figure 6) where it had been hammered flat and the metal recrystallised. These particular cracks were absent from the nail holes in the outer rim, which retained its original and more pliable as-cast structure. The issue of greatest conservation concern is the rim structure of the plate as there is clearly no solid metal connecting approximately two-thirds of the central bowl structure to the outer rim. Before taking it to the Synchrotron, the de Vlamingh pevvter plate was supported on a hollowed-out sheet of Perspex braced with four spring steel holders that enabled it to be securely attached to the precision stage whose synchronous motors drove the plate up and down exposing all areas to the beam line that facilitated the surface scans.
XFM IMAGES AND INTERPRETATION OF DEGRADATION PATTERNS
The high intensity X-ray fluorescence beam line (XFM) at the Australian Synchrotron was used to examine the 32 cm diameter plate in 100 [micro]m steps and this took four days of continuous measurements to record both sides of the plate. A range of X-ray compositional maps made using the Maia 384 detector and the CSIRO GeoPixie software provided a unique insight into the decay mechanism. The X-ray maps of arsenic on the reverse side of the plate showed a clear halo of the metalloid around the 3.5 mm square-shanked French sheathing nails that had been used by Hamelin in 1801. Bronze sheathing tacks of that period typically had 8 [+ or -] 2% tin, with arsenic at 0.6 [+ or -] 0.3%--it is only through the sensitivity of the instruments at the Synchrotron facility that this observation could have been made. A micro-sample of plate was examined by laser ablation inductively coupled mass spectrometry (ICP-MS) and shown to contain 0.3-0.8 wt% zinc which explains the presence of zinc corrosion products distributed almost uniformly over the surfaces. The sample also had an average composition of 0.07 [+ or -] 0.02% arsenic. The microenvironment at the rear of the plate had promoted selective attack on the copper-rich a-phase, which has higher concentrations of arsenic and zinc than the ([alpha]+[delta]) eutectoid. Thus it was established that one of the past microenvironments of the plate was a highly oxygenated, salt-laden moist environment. The obverse side shows a concentration of zinc around the six o'clock position which is due to the mobility of the zinc corrosion products which accumulated at the foot of the plate before running to waste during rain events. The zinc X-ray maps also reveal the dendritic structure of the outer rim of the plate.
The general copper distribution map, not unexpectedly, shows higher concentrations around the French nail holes. Apart from these zones the only significant concentrations of copper are inside the rim of the former bowl at the six o'clock position and at the inner edge of the crack at the half past two position. The micro-sample of metal examined by ICP-MS showed 3.2-4.6 wt% copper, which is associated with copper-tin minerals such as stannite, [Cu.sub.2]FeSn[S.sub.4], which is found in Cornish tin mines. The two apparent "flow lines" from the centre towards the nine o'clock position may relate to a period when the post had fallen or was inclined and mobilised copper corrosion products were gravitating towards that position. The X-ray map for lead on the obverse side shows a series of straight line corrosion patterns between six and seven o'clock, which is consistent with a combination of soluble iron (III) chlorides etching the lead-rich phases of the pewter and copper corrosion products leading to galvanic corrosion couples, which exacerbated the reaction (Figure 6). This preferential corrosion of interdendritic phases in pewter alloys in a marine environment has been previously reported (MacLeod & Wozniak, 1997). The first of the lead depletion lines is oriented at an angle of -23[degrees] from the vertical which indicates that the post acquired a list which was stable for sufficient time to etch the lead corrosion line. The initial collapse was followed by two other episodes where the vertical orientation moved another 2[degrees] in each step to an ultimate angle of -27[degrees] after which the post collapsed. Wind-borne salts also caused depletion of lead at the leading edge closest to the sea.
A specialised Vortex detector was used to map the chloride and sulphate ions, which involved bringing the detector perilously close to the highly profiled buckled plate surface. There was approximately 2 wt% chlorine and 1.5 wt% sulphur, which gives a molar ratio of chlorine to sulphur 16 times less than in salt spray. Owing to the much higher solubility of Pb[Cl.sub.2] compared with PbS[O.sub.4], most of the soluble lead chloride corrosion products have washed away overtime (MacLeod & Wozniak, 1996). The accumulation of chloride ions also explains the localised pitting corrosion phenomenon, which results in a few high tin areas on the plate surface. The other tin pitting areas are found in the lower section of the line around the inner rim and in the area below the second largest iron nail. A chloride content of 2 wt% means that without storage in a controlled inert gas environment the plate will be at risk from further chloride assisted corrosion.
SCANNING ELECTRON MICROSCOPE (SEM) ANALYSIS OF THE CORRODED METAL FROM THE CENTRE OF THE PLATE
The SEM images of the tiny fragment of the pewter plate showed up a corrosion surface that was a mixture of amorphous and crystalline corrosion products. A small sand grain that had been embedded in the upturned surface of the plate after collapse of the post was partly encapsulated by tin corrosion products (Figure 7). The surfaces are dominated by Sn (IV) oxides, while the second major component consists of Sn (II) oxides with a 1:1 for the ratio of tin to oxygen. The morphology of the surface is also consistent with corroded lead being present as PbS[O.sub.4]. The polished and un-etched sample shows that the fabrication techniques of the plate changed its microstructure in that the centre of the plate has a recrystallised structure while the non-hammered rim retains its as-cast dendritic structure. The XFM analysis, after corrections for elements present in the supporting structure, gave 84% Sn, 15% Pb, 0.6% Cu and 0.2% Zn which is consistent with 17th century pewters, which were approximately 81% tin and 19% lead. The lead-rich islands in the solid metal have a composition that places them just on the lead-rich side of the eutectic point at 61.9% tin. The ICP-MS analysis of the microscopic fragment of the pewter showed it had between 2-3% bismuth, which is a common impurity found in Derbyshire (England) lead mines. This all indicates that the primary minerals for the tin components came from Cornwall and the lead from Derbyshire, so the only Dutch component was in the actual manufacture of the pewter bowl which was fashioned into a plate. The rest is history.
CHEMICAL REVERSAL OF CENTURIES OF CORROSION IN TEN MINUTES
A mid-18th century northern Italian ecclesiastical semi-circular cope, some 2.92 metres wide was the focus of a major conservation initiative associated with the 200th anniversary of the 1814 birth of Dom Rosendo Salvado, who founded the Benedictine community at New Norcia, Western Australia. The cope includes the orphrey, which is a significant band of decoration on the leading edge of the garment. This orphrey is richly decorated with silk embroidered floral motifs and massive amounts of gilded silver alloy threads, which were analysed by X-ray fluorescence. The textile conservation of this magnificent work, sponsored by the Copland Foundation, has recently been reported (MacLeod and Car, 2014a). What this project focused on was the use of neutral dithionite solutions to reverse the centuries of corrosion, so it is instructive to review the electrochemistry of this reducing agent.
DITHIONITE OXIDATION: A CONCOMITANT REACTION WITH REMOVAL OF CORROSION
The electrochemistry of sulphur in aqueous solution is perhaps the most complex of all the elements, with a vast range of oxidation states ranging from -2 for [S.sup.2-] +2 for [S.sub.2][O.sup.2-.sub.3], +3 for [S.sub.2][O.sup.4-.sub.2], +4 with S[O.sup.2-.sub.3], +5 as seen in [S.sub.2][O.sup.2-.sub.6] to +6 in S[O.sup.2-.sub.4], which is the most commonly occurring form in aerobic environments.
The method chosen for this study was cyclic voltammetry. This involved recording the current at a platinum electrode as the potential (voltage) is cycled from an initial cathodic potential to an anodic turning point then progressively returned to the starting point. Modern equipment using computer-controlled, voltage-time curves and digital recording of the resultant current greatly facilitates the collation of the data. This method often provides a great insight into complex chemistry. Scan rates varied between 10 and 200 mV.sec1 with a starting voltage of -1.1 V and a concluding potential of +1.0 V vs. Ag/AgCI in 3M KCl.
Rapid electron transfer reactions, such as hexacyano-ferrite-ferrate redox couple Fe[(CN).sup.-3.sub.6]/ Fe[(CN).sup.-4.sub.6], produce well-defined anodic and cathodic branches on the current-voltage curves with the peaks separated by 59/n mV, where n is the number of electrons involved in the oxidation and reduction steps. When the electrochemical reaction involves two or more electrons and where the ionic structures change, the kinetics are much slower and consequently the separation of the anodic and cathodic peaks increases. Since sulphate and sulphite ions cannot be readily reduced in aqueous solution i.e., the kinetics of sulphate accepting electrons to form dithionite are exceptionally slow (Pourbaix, 1974, p. 552), the subsequent cathodic currents are poorly defined. Since the kinetics of the oxidation of dithionite are involved in treating corroded metals, it was decided to focus on this half of the chemical reactions. Inspection of the cyclic voltammograms for a 1.5x[10.sup.-3] M dithionite solution at pH 7 (Figure 8) shows anodic peaks at -0.345, +0.541 and +0.992 volts vs. Ag/AgCl while the corresponding cathodic peaks are at -0.424, -0.136 and +0.323 volts which gives voltage separations of 79, 677 and 639 mV. The first electrode process is associated with a rapid reaction (not associated with dithionite oxidation) while the second and third oxidation steps are characteristically slow and are associated with the oxidation of [S.sup.+3] to [S.sup.+5], which is then oxidized up to S+6.
Since the nature of the ionic species in solution changes as a function of both the pH and the voltage, a series of experiments were conducted at operational pH values of 4, 7, and 10 using standard phosphate buffer solutions and in 1 M NaOH at a pH of 13.6. Owing to the electrochemically slow oxidation kinetics, the anodic peak potentials varied with scan rate, becoming increasingly anodic with more drawn-out peaks as the scan rate was increased. By plotting the peak voltages as a function of scan rate, the equilibrium values representing a closer approximation to the thermodynamic properties were obtained from extrapolation of the data to zero scan rate ([sup.zero][E.sub.[alpha]]), which helped enormously in deciphering the processes that were controlling the reactions. A typical plot of the anodic peak potentials for a 1.5x[10.sup.3] M dithionite solution at pH 7 is shown in Figure 9. The three oxidation peaks seen in Figure 8 have intercept values of -0.482, +0.433 and +0.846 volts vs. Ag/AgCI at zero scan rates. The [R.sup.2] values for the regression analysis ranged from 0.8882 to 0.9972 which gave values of [+ or -] 0.014, [+ or -] 0.005 and [+ or -] 0.002 volts respectively in the errors of the intercept values.
The first oxidation peak was moderately sensitive to changes in pH and this was described by equation 4,
[sup.zero][E.sub.al] = -0.608 + 0.0203 pH (4)
The slope of 0.0203 [+ or -] 0.0008 volts/pH between 4 and 10 does not correspond to any known single solution oxidation process and it is likely that the anodic reaction is a mixture of anodic oxidation of hydrogen that was adsorbed onto the platinum electrode at -1.0 volts vs. Ag/AgCl and oxidation of a more reduced form of dithionite that was produced at the resting cathodic voltage. Between pH 10 to 13.6 the first anodic peak was pH independent, which is likely to represent oxidation of the electrochemically formed tetrathionate ion [S.sub.4][O.sup.2-.sub.6] to dithionite,
[S.sub.4][O.sup.2-.sub.6] + 2OH [right arrow] 2[S.sub.2][O.sup.2-.sub.4] + 2[H.sup.+] + 2e (5)
The tetrathionate ion has a formal oxidation state of +2.5 while the oxidation state of sulphur in dithionite is +3. The first peak is essentially an artefact of the cyclic voltammetry experiment and is not part of the oxidation of dithionite, which is involved in it working as a chemical reducing agent in overcoming corrosion products formed on gilded metal threads.
The second anodic peak, [sup.zero][E.sub.a2], is essentially constant as the pH increases from pH 4 to 7 and it becomes less anodic with increasing concentration of hydroxide ions by 85 [+ or -] 2 mV per pH between neutral pH and 1M sodium hydroxide. The oxidation peak shifts cathodically by 0.542 V in moving from pH 7 to 13.4 which makes the dithionite a much more powerful reducing agent in caustic solutions. The chemical reaction is oxidation of the hydrogen dithionite ion to hydrogen sulphite as shown in equation 6 (Pourbaix, 1974, p. 548)
H[S.sub.2][O.sup.-1.sub.4] + 2[H.sub.2]O [right arrow] 2HS[O.sup.-1.sub.3] + 3[H.sup.+] + 2e (6)
The pH dependence of the second anodic peak (equation 6) matches the experimentally observed slope for the reaction. Although the active form of dithionite is only 3% at pH 4 it still controls the kinetics of the oxidation process.
The third anodic peak, [sup.zero][E.sub.a3], corresponds to the oxidation of sulphite ions to sulphate ions. For the pH 4 solutions there was a strong smell of S[O.sub.2] as the nitrogen degassed the test vessel. Some cloudiness from suspended elemental sulphur was observed, which is due to the disproportionation of some of the dithionite. Under these conditions the kinetics of oxidation of dithionite is the slowest, with a peak potential of 0.943 [+ or -] 0.005 volts. In neutral pH solutions the oxidation of dithionite is considerably easier as the [sup.zero][E.sub.a3] decreased by 97 mV to 0.846 [+ or -] 0.003 V, which equates to a fall of 32 mV per unit change in pH. The peak potential was constant as the pH was increased to 10. By increasing the hydroxide concentration to 1M NaOH solution, the [sup.zero][Ea.sub.3] value fell by 0.478 volts to +0.370 V vs. Ag/AgCl. The change in voltage with pH in the alkaline region is consistent with the oxidation of sulphurous acid in a two-electron step, as shown in equation 7.
[H.sub.2]S[O.sub.3] + [H.sub.2]O [right arrow] S[O.sup.2-.sub.4] + 4[H.sup.+] + 2e (7)
Since the [sup.zero][E.sub.a3] value fell by 140 mV per pH it can be seen that there are small kinetic effects that increase the thermodynamic pH dependence from the theoretical 119 mV slope of equation 7 to a value of 140 millivolts.
DITHIONITE REMOVAL OF CORROSION PRODUCTS FROM GILT SILVER ALLOY EMBROIDERY
Inspection of the cope showed that it was very dirty after centuries of use and areas around the shoulders and the central clasp were worn, with chloride ion measurements showing up significant amounts of sweat. A significant amount of time was spent in stabilising the heavily fragmented jumble of gold embroidery before the cope could be washed and the corroded wire be conserved. Tests on the silk embroidery showed that all the old dyes were colour fast so the fibres were relaxed by washing the object with gentle hand palpation in a large shallow stainless steel bath which prepared the garment for chemical reduction treatment. The active ingredients of the weak washing solution included some sodium tripolyphosphate ([Na.sub.5][P.sub.3][O.sub.10]) and two surface-active agents, Lissapol N and Hostapon T, with sodium carboxy-methylcellulose, which helped to open up the fibres and allowed the release of the entrapped dirt. A Bruker portable X-ray fluorescence instrument established that the majority of the gold wires were a gold-coated, silver-copper (silver-gilt) alloy wound around a central core of yellow-coloured silk fibres. Areas on the edge of the hood and at the fringe showed much heavier tarnishing and the XRF analysis showed that this material was gilded brass. The use of a "gilding alloy" of approximately 95% copper and 5% zinc meant that there was an underlying golden colour on which the gold was deposited. The amount of arsenic and lead in these metallic threads appears to be from associated impurities found in the original copper ore from which the metal was refined: the lead is six times higher and arsenic is 25 times higher than in all the other Cu-Ag-Au metal threads.
The washed garment was immersed in a buffer solution of 0.01 M Na[H.sub.2]P[O.sub.4]/NaHP[O.sub.4] (pH 6.5) before dissolving 65 grams of dithionite (5.7x[10.sup.-3] M) in the bath or 0.1 wt% dithionite. Once the dissolved oxygen had been removed from solution the corrosion potential ([E.sub.corr]) of the gilded threads was recorded at -0.604 volts vs. Ag/AgCl (-0.400 volt vs. NHE) which was sufficient to reduce all the corrosion products on the metal threads in 10-15 minutes. After draining the reducing agents fresh deionised water was used to rinse the textile. Monitoring the [E.sub.Corr] values of the gilded threads during the rinsing program showed it takes four full rinses to remove all traces of the reducing agent from the surface of the textile (MacLeod & Car 2014b). The effectiveness of the treatment can be seen in comparison of typical sections of the cope in Figure 11 (before) and Figure 12 after treatment.
Tests done on part of the discarded mid-18th century silk fibres from the hood lining confirmed that the 15-minute treatment in the dithionite solution resulted in no measurable difference in the mechanical properties of the tensile strength and elongation at breaking point. The mean tensile strength of the longitudinal samples was 101 [+ or -] 18 N/25 mm and the mean transverse value was 84 [+ or -] 28. The higher values of strength for the longitudinally cut test sections is simply a reflection of the greater tensile strength of the warp threads compared with the weft. Tarnishing of the treated metal threads is prevented by scavenging pollutants with ZnC[O.sub.3] impregnated blotting paper in the exhibition case and in the storage box. This method has prevented dithionite treated silver from the Pelsaert silver collection on board the Batavia (1629) from tarnishing for 18 years at the WA Museum Shipwreck Galleries.
I would like to thank the Bunuba traditional owners who accompanied Philip Haydock and Bruce Ford to their rock art sites. Financial assistance from the Australian Synchrotron facility in providing access to their beam line for four days enabled the work to be done on the de Vlamingh plate. David Thurrowgood from the Queen Victoria Museum and Art Gallery in Launceston initiated the Synchrotron study program and was assisted by Gwynneth Pohl from the University of Melbourne Centre for Cultural Materials Conservation. The assistance of the Benedictine community at New Norcia and of Dom Christopher Power is readily acknowledged.
Clarke, J. (1976). Two Aboriginal rock art pigments from Western Australia: their properties, use and durability. Studies in Conservation, 21, 134-142.
Ford, B., MacLeod, I.D., & Haydock, P. (1994). Rock art pigments from the Kimberley region of Western Australia: identification of the minerals and conversion mechanisms, Studies in Conservation 3?(l), 57-69.
Konigsberger, E., & Gamsjager, H. (1987). Solid-Solute Phase Equilibria in Aqueous Solution. I. Solubility Constant and Free Enthalpy of Formation of Huntite. Ber. Bunsenges. Phys. Chem. 91 [8), 785-790.
MacLeod, I.D., Haydock, P., Tulloch, D., & Ford, B. (1995). Effects of microbiological activity on the conservation of aboriginal rock art. AICCM Bulletin, 21(1), 3-10.
MacLeod, I.D., & Wozniak, R. (1996). Corrosion and Conservation of Lead in Sea Water, ICOM-Committee for Conservation Preprints 11th Triennial Meeting, Edinburgh, Sept 1996. James & James, London, 884-890.
MacLeod, I.D., & Wozniak, R. (1997). Corrosion and conservation of tin and pewter, Metal 95-Proceedings of the ICOM--CC Metals Working Group Conference, Semur-en-Auxois, France 1995, James & James, London, 118-123.
MacLeod, I.D., & Car, R.J. (2014a). Determining treatment priorities for ecclesiastical textiles using significance and conservation assessments, Journal of Cultural Heritage, 15(6), 628-636.
MacLeod, I.D., & Car, R.J., (2014b). Oxidation of dithionite treatment solutions and the effects of pH on fading textile dyes, AICCM Bulletin 35(1), 69-78.
MacLeod, I.D., Thurrowgood, D., Pohl, G., Howard, D., Patterson, D. (2014) Centuries of decay revealed by synchrotron analysis of the de Vlamingh 1697 pewter plate. In ICOM-CC 17th Triennial Conference Preprints, Melbourne, 15-19 September 2014, ed. J. Bridgland, art. 0903, 6 pp. Paris: International Council of Museums, (ISBN 978-92-9012-410-8)
North N. A., and Clarke J., 1988. Conservation of post-estuarine rock art in Kakadu National Park: pigment identification. Unpublished report, Australian National Parks and Wildlife Service.
Playford, P. Personal Communication, (1992).
Playford, P.E., & Hesselsz, W. (1998). Voyage of Discovery to Terra Australis by Willem de Vlamingh in 1696-97, Western Australian Museum, Perth, Western Australia.
Pourbaix, M. (1974). Atlas of Electrochemical Equilibria in Aqueous Solutions, NACE-Cebelcor, Houston, Texas, Chapter IV Section 19.2, 545-553
Watchman, A., (1990). A summary of occurrences of oxalate-rich crusts, Australia' Rock Art Research 7(1) 44-50.
Dr Ian D MacLeod is an electrochemist who has pioneered in-situ conservation and corrosion studies on historic shipwrecks off the Western Australian coast. In this article, he provides an overview of his fascinating work in preserving some of Australia's rich cultural heritage through applied chemistry.
Dr MacLeod will be delving further into his work at the CONASTA 64 Stanhope Oration.
|Printer friendly Cite/link Email Feedback|
|Author:||MacLeod, Ian D.|
|Date:||Jun 1, 2015|
|Previous Article:||CONASTA brings teachers a kaleidoscope of science.|
|Next Article:||Using data to improve learning: a practical guide for busy teachers.|