J O U R N A L O F C H E M I S T R Y Materials An investigation into the deacidification of paper by ethoxymagnesium ethylcarbonate Robin J. H. Clark,*a Peter J. Gibbsa and Raik A. Jarjisb aChristopher Ingold Laboratories, University College London, 20 Gordon Street, London, UK WC1H 0AJ bNuclear Physics Laboratory, University of Oxford, Keble Road, Oxford UK OX1 3RH Received 12th June 1998, Accepted 5th October 1998 Aspects of the procedure used since 1995 by the British Library whereby paper is deacidified with ethoxymagnesium ethylcarbonate (EMEC) in hexamethyldisiloxane (HMDS) have been studied using scanning electron microscopy (SEM) along with energy dispersive X-ray analysis (EDAX), X-ray fluorescence (XRF) spectrometry and particleinduced X-ray emission (PIXE).The treatment was applied to three types of test paper, and the analytical results compared with those for similar samples treated using a method which mimicked as closely as possible the library’s previous deacidification procedure; this involved the use of the now-banned solvent trichlorofluoroethane.The amount of EMEC distributed over the whole paper sample was fairly consistent for each paper type, but microscopically it was found to be congealed into large deposits, possibly hydrolysed by moisture in the paper, and not distributed as evenly among the paper fibres as is either desirable or possible.The depth of penetration of the EMEC particles obtained by spraying only one side of the paper is found to be poor for paper of heavy gauge and high moisture content, the thinnest samples retaining little alkaline buVer.The results obtained by the newer procedure were similar to those obtained by the older one, and no HMDS either reacted with, or remained in, the paper despite the relatively long drying time. The principal cause of the deterioration of cellulose-containing microscopy (SEM) and associated energy dispersive X-ray analysis (EDAX), and particle-induced X-ray emission materials, especially paper, was identified in the 1950s to be sulfuric acid generated by the hydrolysis of the common sizing (PIXE) to establish the uniformity of distribution and depth of penetration of the deacidification agent EMEC among agent aluminium sulfate1 (referred to, somewhat misleadingly, as ‘paper makers’ alum’).Aluminium sulfate is used in conjunc- paper fibres.Due to the large number of documents to be treated in the tion with natural wood resins to control water penetration in paper. Traditionally it was used to deposit saponified resins BL, a deacidification solution must be priced reasonably and, moreover, it must also penetrate each book uniformly in a onto the surface of paper fibres during manufacture so as to moderate period of time, and provide a suYcient alkaline achieve the desired firmness and to prevent the blurring of reserve for prolonged performance.10 Over the past two dec- dyes and inks.2 Other causes of acidity include the oxidation ades, the trend has been towards using non-aqueous solvents of lignin and cellulose and the absorption of atmospheric to deliver the deacidification agent, some of which involved pollutants;3 indeed, any acidic environment, regardless of the CFC chemicals that have now been phased out due to environ- source of the acidity, aVects paper in the same way, initiating mental legislation.In 1995, the BL changed their delivery ageing processes that aVect the lengths of the cellulose chains, solvent from trichlorofluoroethane to polydimethylsiloxane, resulting in brittleness and fragility.2 The library and museum but the latter was found to be slow to dry causing inks to be community are obliged to preserve printed matter that was fugitive and even tide-mark stains to develop. prepared on poor-quality (acidic), mass-produced paper; such The new paper preservation technique is based upon ethoxy- items, for example newspapers or the scribbled first drafts of magnesium ethylcarbonate (EMEC, 1), which is the deacidifi- now famous manuscripts, were often intended to be ephemeral cation agent, and the siloxane solvent, hexamethyldisiloxane but now have a high cultural value.The deacidification of (HMDS, 2).20 these items to preserve them for future generations is a major The EMEC reacts with the acid in the paper (in this concern of librarians and archivists world-wide.2–17 However, example, sulfuric acid) in a similar way to that discussed for the mass deacidification of paper is also controversial.The other organometallic deacidification agents [Scheme 2(a)].21 current method of evaluating the ‘pH of paper’ is known to MgO formed from the primary hydrolysis product be flawed, as it is really the pH of the solution used to moisten [Scheme 2(b)]21 reacts with CO2 to form the desired buVer, the paper surface that is being measured by a conventional MgCO3. The source of the CO2 may be the hydrolysis of the pH probe; there is thus the real possibility that millions of ethylcarbonate ligand, but comparison with the similar Batelle folios are being treated needlessly because they are being deemed, incorrectly, to be too acidic.This controversy will continue until a method is devised to measure the pH of moisture trapped inside the actual paper fibres, thereby yielding the true ‘pH of the paper’.18,19 The research reported here is an investigation into the eVectiveness of the delivery procedure of the mass deacidifi- cation agent EMEC used in the British Library (BL). Many C2H5OCO2MgOC2H5 H3C Si O CH3 CH3 Si CH3 CH3 CH3 2 1 treated and untreated samples have been examined by X-ray Scheme 1 Ethoxymagnesium ethylcarbonate (EMEC, 1) and hexamethyldisiloxane (HMDS, 2).fluorescence (XRF) spectrometry, scanning electron J. Mater. Chem., 1998, 8, 2685–2690 2685process2 (vide infra) suggests that it is probably atmospheric.Scanning electron microscopy and energy dispersive X-ray analysis C2H5OCO2MgOC2H5+H2SO4AMgSO4+CO2+2C2H5OH A study of 66 samples and 12 standards was performed on a (a) JEOL JSM-6400 scanning electron microscope with an accelerating voltage of 5 kV and an EDAX attachment. The samples C2H5OCO2MgOC2H5+H2OAMgO+CO2+2C2H5OH were mounted on 40 mm aluminium discs and sputter-coated with a 5–10 nm film of gold and palladium. MgO+CO2AMgCO3 (b) Scheme 2 (a) An example of EMEC deacidification and (b) a proposed X-Ray fluorescence spectrometry mechanism for the formation of the desired buVer.21 48 samples and 12 standards were analysed by XRF spectrometry using a Philips PW 1480 sequential X-ray spectro- There are three questions of interest: first, whether the photometer.The radiation source was a rhodium lamp with HMDS solvent permits a more uniform distribution of the wide-band excitation operating at 60 kV and 40 mA. The EMEC than does the CFC solvent; second, whether the crystals used were InSb (Si analysis) and thallium HMDS detaches from the cellulose fibres after application; azide phthalate (Mg analysis) and the number of counts was and third, whether the depth of penetration of the EMEC measured by a flow counter.Both sides of the samples were particles could be determined (the solution is normally applied studied and, for each, six recordings and background counts to one side of the paper only, and thus the EMEC particles (collection time=40 s) were performed for the Mg analysis, may not penetrate through to the other side before the solvent and four recordings and background counts (collection time= evaporates).Hereinafter, all Mg-containing particles related 16 s) were performed for the Si analysis. to EMEC will be described as EMEC particles; this is used as a generic term and no distinction is made as to whether the Particle-induced X-ray emission compound has reacted with acid, been hydrolysed, or perhaps not reacted, as only the presence of a magnesium-containing Elemental depth profiling was carried out by detecting X-rays compound is determinable.generated as a result of scanning a focused beam of 3.0 MeV The deacidification agent and the bulk solvent both contain protons over the cross-sections of the paper samples.The the elements Mg and Si not associated normally with organic results produced in the form of maps and line-scan plots were materials such as the cellulose fibres of paper, although they obtained using the scanning proton microprobe facility at the are associated with sizes, loads and impurities, and therefore University of Oxford. The experiments were performed using control samples were used extensively.Elemental analysis a 1 mm wide proton beam spot, and the X-rays produced were targeting these two elements in particular was considered to detected using a conventional Si(Li) detector. Further details about the experimental arrangements and new developments be the most perceptive approach to this project, via XRF in the application of ion-beam analysis (IBA) to historical spectrometry, SEM, EDAX analysis and PIXE techniques.materials (including paper) are presented elsewhere.23,24 A novel experimental procedure was developed for this study and will be published subsequently. Experimental Results and discussion Preparation of paper samples Scanning electron microscopy and energy dispersive X-ray Three diVerent types of paper studied: 20th century newsprint analysis (mechanical wood fibres, 50–60 g m-2, 6–9% moisture content, calliper 0.08 mm); general 1920s print (chemical/mechan- At a relatively high magnification (×6000) it was possible ical wood fibres, 80–90 g m-2, 5% moisture content, calliper among the paper fibres to detect white particles, which were 0.13–0.14 mm); and 18th century handmade (rag linen rarely greater than 1 mm in diameter and often much smaller: fibres, 100–120 g m-2, 5–8% moisture content, calliper an example of the type of image observed is shown in Fig. 1(a). 0.20–0.25 mm). The BL provided over 70 paper samples The particles were almost spherical and analysis using EDAX (9 cm×9 cm), including untreated controls, 27 samples treated revealed that they contained magnesium.The particles were with EMEC in HMDS, and 27 samples treated with EMEC small and of regular shape and their presence in all of the in the solvent 1,1-dichloro-2,2,2-trifluoroethane (DCTFE). As EMEC samples treated, their magnesium content, and the lack it is now impossible to obtain CFC solvents in the UK, of any other element with a relative atomic mass >23, proves DCTFE, which itself will be banned early next century, was that they are derived from the EMEC deacidification agent.substituted for trichlorofluoroethane. The samples were treated Other particles were found amongst the paper fibres but, in-house by the BL following their usual deacidification pro- due to their shape, size and analysis by EDAX, they could cedure. They were cut to size and then placed on a wire-mesh not be confused with the EMEC particles.The results of the treatment screen and held in place by a suction motor (air EDAX analysis are listed in Table 1. velocity ca. 1.4 m s-1). The freshly prepared deacidification Apart from the EMEC, the most important particles to be solution, propelled by nitrogen at a pressure of 3.85 kg cm-2, identified in the general and newsprint samples were irregular was sprayed by hand from approximately 10 cm in a sweeping crystals with both a high Al and, importantly, high Si content; motion until ‘full saturation’ was achieved.22 they are probably an aluminosilicate clay.The high Si content Sixteen 2 cm×2 cm sub-samples were removed from each of these particles will have an important bearing on the of 33 of the 9 cm×9 cm samples, including 6 controls, and interpretation of the results of the PIXE and XRF the positions of the sub-samples in the original samples were spectrometric analysis (vide infra).recorded. Four 40 mm circular sub-samples were removed In the test samples examined, there were generally signifi- from each of the remaining treated samples and six controls cantly more EMEC particles on the sprayed side than on the to take advantage of the optimum sampling area (XRF unsprayed side.An extreme example is illustrated in Fig. 1(a) spectrometry); the original positions of the sub-samples were and (b): one side of a general print/HMDS sample [the again recorded. In total there were 660 samples available for sprayed side, Fig. 1(a)] was well covered in what was, for this analysis, representing equally each paper type and each delivery study, a well above average number of EMEC particles, but on the reverse [unsprayed side, Fig. 1(b)] such particles are solvent, and with suYcient standard samples. 2686 J. Mater. Chem., 1998, 8, 2685–2690Fig. 1 Electron micrographs of (a) EMEC particles on a general print samples ×6000, (b) EMEC particles on the reverse of general print samples in (a) ×6000, (c) ‘strings’ of EMEC particles ×6000, (d) ‘bunches’ of EMEC particles ×6000.represented, but on a much lesser scale. This was generally paper is well covered in EMEC particles, there are many fewer EMEC particles on the reverse. true for all of the samples examined, regardless of whether the deacidification agent was delivered in the DCTFE or The conclusions drawn above are dependent in a large way on how many particles were deposited on the sprayed side of siloxane solvent.Using SEM/EDAX it is not possible to judge the depth of the paper samples in the first place. The type of coverage illustrated in Fig. 1(a), for which the sample had been sprayed penetration of the particles but only possible to examine each side of a sample to see whether EMEC particles are rep- with the siloxane-based solvent, was greater than that of any other sample examined, and in most cases it was much greater.resented. By this method, the following conclusions could be made as to the eVectiveness of the penetration of the particles More representative examples are shown in Fig. 1(c) and (d), which show isolated groups of EMEC particles in paper for the diVerent paper types: (1) some EMEC particles were always observed on both sides of the newsprint samples, but samples treated with the siloxane solvent. Generally, for all paper types and both solvent delivery systems, these groups for the thicker handmade paper very few if any penetrated to the reverse side.Newsprint paper is thin, and this is the likely of particles were few and far between. Two significant observations can be made from the images reason for the better penetration; (2) in the case of the general print samples, EMEC particles were always observed on both observed in this experiment: (1) even on the samples (those with the sides sprayed) with the most EMEC particles, the sides, but Fig. 1(a) and (b) indicate the relative ineVectiveness of the penetration to the reverse. Furthermore, the presence distribution is very uneven for all of the paper types, with isolated groups of particles dotted around the samples; (2) the of ink on the sprayed side of the samples blocks the penetration of the deacidification agent.particle size is generally larger on all of the newsprint, handmade samples and the general print samples examined which Generally, therefore, the depth of penetration achieved by spraying on one side of the paper depends only upon the had been treated with the DCTFE-based agent, than it is for the general print samples treated with the HMDS-based solu- thickness of the paper and whether or not there is ink on the sprayed side.However, even where one side of the thinnest tion [Fig. 1(a)–(d) are all at the same magnification, ×6000]. Table 1 EDAX analysis of crystals found between the paper fibres Elements identified (relative atomic mass <23) Identity of compound Paper typea and notes Mg EMEC or magnesium salt product Treated G, H & N Ca and S Gypsum,CaSO4·2H2O G and H.Probably added as a load Ca Chalk, CaCO3 G Fe, K and S Alum, AlK(SO4)2·12H2O G. Probably added as a size Al and Si Aluminosilicate clay G & N. Probably added as a load aG=1920s general print, H=18th century handmade and N=20th century newsprint. J. Mater. Chem., 1998, 8, 2685–2690 2687It was common for the EMEC particles to form either ‘strings’ general print standards is about three times more than on the newsprint standards, and over 40 times that detected on the [Fig. 1(c)] or ‘bunches’ [Fig. 1(d)] on the samples that displayed larger EMEC particles (the ‘string’ between the particles handmade samples. From the EDAX analysis, it is known that aluminosilicate clays are present in significant quantities in Fig. 1(c) is formed by the EMEC). The smaller particles in some general print examples treated with the HMDS-based in the general print samples, and to a lesser degree in the newsprint samples: they are probably added deliberately to fill solution do also coagulate, but they tend to distribute much more widely.the paper, and are the source of the high Si content. The very low relative amounts of Si on the handmade paper indicates The particles on most of the paper samples examined are up to ten times the diameter of the smallest observed on the that little or no Si-containing compounds were added deliberately at manufacture.general print samples treated with the siloxane-based solution. The smaller particles obviously give better coverage where Higher levels of Mg were detected on all of the treated paper samples relative to the standard samples.For the general they are deposited, although they can still be spread inconsistently across the general print samples treated with the HMDS. print samples, there was a higher level of Mg on the sprayed side, but considerable amounts were detected on the unsprayed The large particle size and ‘bunching’ of the EMEC particles on all of the other sample types reinforces the poor distribution sides as well.There was little diVerence between the average amounts of Mg detected whether the EMEC was delivered in of the deacidification agent. A possible cause is that the EMEC/HMDS solution is very water sensitive, and a white the HMDS or DCTFE. The range of the relative amounts of Mg detected does suggest that the EMEC/HMDS has slightly product, MgCO3, drops out of solution if the deacidification solution is contaminated with moisture.22 A similar problem better penetration, which would be explained by the smaller particle size.was noticed in the development of the Battelle process, employed by the Deutsche Bibliothek,2 which uses the same For the handmade samples, there was a vast diVerence between the amount of Mg detected on the sprayed and solvent as the BL but a diVerent deacidification agent, Mg(OC2H5)2.This alkoxide also hydrolyses to the desired unsprayed sides. On average, more Mg was detected on the sprayed sides of the samples treated with EMEC/DCTFE than buVer, MgCO3, but this can occur rapidly in humid conditions, aVecting the uniformity of distribution.2 The Battelle process those sprayed with EMEC/HMDS.The most important observation, however, is that the amount of Mg detected on the therefore includes a pre-drying of the samples from their stored humidity of 5–7% by weight to a water content of less reverse of the handmade samples was very low, comparable to that on untreated samples. In agreement with the SEM than 1%.2 The BL do not pre-dry their samples, and this may be the observations, this suggests that very little EMEC penetrates the thick-gauged handmade paper, regardless of the carrying reason for the clumping of particles observed generally in the handmade and newsprint samples, both of which have a solvent.The relatively high moisture content of the handmade paper may also contribute to this observation due to the relatively high moisture content.The general print samples, however, have the lowest moisture levels, as is expected for EMEC particles being hydrolysed before they can penetrate deeply into the paper. mechanical papers of such an age with degraded cellulose fibres. It is possible that the good spread of EMEC particles The most surprising observation was the relatively low levels of Mg detected on the treated newsprint samples.The amounts in some of the general print samples examined is attributable to a lack of moisture, implying that pre-drying will assist the were higher than detected on the untreated standards, but not by as much as had been expected. Similar amounts of Mg distribution and so the eVectiveness of the deacidification agent.It seems that the distribution of the EMEC particles were detected on the sprayed and unsprayed sides, suggesting that the thin gauge of the paper aided the even distribution of among the paper fibres is typically less eYcient than is either desirable or achievable. the particles; but, as the samples of the three papers were treated identically, the thin gauge of the newsprint does seem to inhibit its potential to retain a suYcient alkaline reserve X-Ray fluorescence spectrometry before becoming saturated. Both sides of four circular sub-samples, 40 mm diameter, The mean of the number of counts (minus the background count) for Si and Mg was calculated and a standard deviation were examined for each 9 cm×9 cm original.The samples covered 62% of the whole area of each sprayed sheet.Strikingly obtained for each sample. The normality of distribution of the number of counts for each sample set was established for both variable relative amounts of Mg were sometimes found in sub-samples from the same original (e.g. a ratio of the Mg and Si analysis (3s error test). However, the mean number of counts between similar samples (e.g.the unsprayed 0.4250.5150.8351.0 on four diVerent samples from the sprayed side of one general print/HMDS sample), but the relatively sides of EMEC/HMDS general print samples) were always found to diVer significantly using the statistical technique of low readings in this sample were extreme. The variations tended to be greater between original samples than within the analysis of variance (ANOVA, P=0.05) whereby the variance of each element within the same sample group and between same sample.Overall, the least amount of Mg detected was rarely less than half the maximum amount detected on similar similarly treated samples is compared. In other words, the amount of Mg and Si detected in samples treated identically sub-samples, and this suggests that the spraying technique with either solvent is depositing the EMEC fairly evenly on a varies significantly.As the number of counts is relative for each sample, it is macro-scale; but as the SEM analysis revealed, the distribution on a micro-scale within the paper fibres is poor. possible to determine the range of the relative amounts of each element detected on samples treated in an identical way Hexamethyldisiloxane has a relatively low boiling point (100 °C), a relatively high vapour pressure (20 mbar at 20 °C) (Table 2).Though there is no statistical similarity between the relative amounts of each element detected on similar samples, and a low enthalpy of vaporization (186 kJ kg-1), which means that treated papers will dry fairly quickly.However, the tabulated data convey important information regarding the distribution of each element in the paper samples. Thus this solvent dries much more slowly than the trichlorofluoroethane used previously, giving rise to concern that the pro- the amount of Mg present in the standard (untreated) samples is consistently low for each type of paper, and probably results longed exposure of the paper to the siloxane solvent may have adverse eVects and that the solvent may even react with species from impurities that entered the paper during manufacture, rather than from substances added deliberately.By contrast, within the paper or the paper fibres themselves, depositing unknown silicon-containing compounds within the paper. the level of Si detected in the standard samples varies considerably for the three paper types.The level of Si detected on the The results of the XRF analysis displayed in Table 1 suggest 2688 J. Mater. Chem., 1998, 8, 2685–2690Table 2 Range and weighted mean of relative number of counts determined from similar samples for Si and Mg by XRF spectrometry Range of relative number of counts and weighted mean (in parentheses) Deacidification Sprayed (S) or Paper type solution solvent unsprayed (U) side Si Mg General print HMDS U 21.5–32.9 (28.7) 0.17–0.36 (0.26) S 29.7–44.5 (37.6) 0.19–0.49 (0.39) DCTFE U 22.1–30.6 (28.1) 0.16–0.28 (0.24) S 21.9–40.3 (33.3) 0.32–0.46 (0.41) Standard 28.4–31.9 (29.9) 0.06–0.09 (0.08) Handmade HMDS U 0.45–0.59 (0.49) 0.02–0.04 (0.03) S 0.41–0.82 (0.54) 0.14–0.20 (0.17) DCTFE U 0.31–0.42 (0.37) 0.08–0.09 (0.09) S 0.50–0.67 (0.60) 0.42–0.55 (0.48) Standard 0.67–0.80 (0.71) 0.07–0.09 (0.08) Newsprint HMDS U 8.63–12.66 (10.1) 0.06–0.08 (0.07) S 8.60–12.71 (10.1) 0.06–0.08 (0.07) DCTFE U 8.01–10.97 (9.71) 0.06–0.09 (0.08) S 8.21–11.52 (9.53) 0.07–0.10 (0.08) Standard 10.1–12.42 (11.2) 0.02–0.04 (0.03) that the HMDS solvent is not retained by the treated paper.There is no significant diVerence between the ranges of the amounts of Si detected, regardless of whether the paper types were standard samples, or ones treated with either the siloxanebased or the DCTFE solution. This is particularly apparent in the handmade samples, which already contain a relatively low amount of Si, and where those samples treated with EMEC/HMDS actually contained the lowest average amount of Si.Particle-induced X-ray emission spectrometry The similar profiles of the line-scan plots obtained for Al and Si from general print and newsprint samples treated with EMEC/HMDS [e.g. Fig. 2(a) and (b)] confirm that they predominantly co-exist, which corresponds to the observation by SEM/EDAX that aluminosilicate clays are present. This supports the observation that Si present on the paper samples is due neither to the HMDS solvent nor to any reaction products, but to additives or impurities in the manufacturing process.The similar profiles of the plots obtained for Ca and S [e.g. Fig. 2(c) and (d)] from the handmade and general print also confirm that these elements co-exist as gypsum.The depth profile of the distribution of Mg, as illustrated by line-scan spectra, show that the maximum concentration of Mg is at a depth of roughly 50–70 mm (i.e. about one-third to one-half the way through) into the paper for the general print samples [e.g. Fig. 2(e) and (f )]. This observation is similar for the handmade papers which, due to their thicker gauge, means that less of the paper is protected. As an immediate result of this study, the BL has altered its procedure to include spraying each page from both sides to ensure maximum protection.The Library is now considering whether reducing the moisture content of the paper to aid Fig. 2 PIXE line-scan plots (150 mm) of (a) Al and (b) Si on a general distribution of the agent is practical and desirable.print sample treated with EMEC/HMDS, (c) Ca and (d) S on a general print sample treated with EMEC/HMDS, (e) and (f ) Mg on Conclusions two diVerent samples treated with EMEC/HMDS indicating the lack of penetration through the paper. The amount of EMEC distributed over the whole paper sample was fairly consistent for each paper type, but microscopically, however, the EMEC was found to be congealed in large deposits, possibly hydrolysed by moisture in the paper, and reacted with, or remained in, the paper despite the relatively long drying time.not distributed as evenly among the paper fibres as is either desirable or possible. The depth of penetration of the EMEC particles obtained by spraying only one side of the paper is Acknowledgements poor for heavy gauged papers and those with a high moisture content, and the thinnest samples retained a relatively low We are indebted to the Leverhulme Trust for the award of a fellowship (PJG), to ULIRS and the von Clemm foundation amount of alkaline buVer.The results obtained with the new method of application were similar to those for the previous for financial support, to Dr G.Mariner at Royal Holloway and Bedford New College (XRF), to the electron microscope method, and there was no evidence that the HMDS either J. Mater. Chem., 1998, 8, 2685–2690 268912 A.-C. Brandt, Mass Deacidification of Paper: A Comparative Study unit at The Queen’s University of Belfast, and to Shad Mehmet of Existing Processes, Bibliothe`que Nationale, Paris, 1992. and Dr.Mirjam Foot at the British Library. 13 E. M. Kaminska and H. D. Burgess, Natl. Libr. News, 1994, 26, 11. 14 R. Areal Guerra, J. M. A. Gibert Vives, J. M. A Daga` Monmany References and J. Fernandez Garrido, Restaurator, 1995, 16, 175. 1 W. J. Barrow, Manuscripts and Documents: Their Deterioration 15 V. D. Daniels, Chem. Soc. Rev., 1996, 25, 179. and Restoration, University of Virginia Press, Charlottesville, 16 M. C. Sistach Anguera, Restaurator, 1996, 17, 117. 1955. 17 S. R. Middleton, A. M. Scallan, X. Zou and D. H. Page, Tappai, 2 J. Wittekind, Restaurator, 1994, 15, 189. 1996, 79, 187. 3 L. R. Green and M. Lesse, Restaurator, 1991, 12, 147. 18 A. Kennedy and K. R. Seddon, personal communication. 4 J. S. Arney, A. J. Jacobs and R. Newman, J. Am. Inst. 19 V. Bukovsky�, Restaurator, 1997, 18, 25. Conservation, 1979, 19, 34. 20 PDTA solution from Particle Technology Limited, Hatton, UK. 5 M. Hey, The Paper Conservator, 1979, vol. 4, p. 66. Solvent includes 10% ethanol to assist solution. 6 R. D. Smith, Can. Libr. J., 1979, 325. 21 A. N. MacInnes and A. R. Barron, J. Mater. Chem., 1992, 2, 1049 7 D. Mihram, Restaurator, 1986, 7, 81. and references therein. 8 D. Mihram, Restaurator, 1986, 7, 99. 22 S. Mehmet, personal communication. 9 L. Santucci, V. Grosso, M. Hey and L. Rossi, in New Directions in 23 R. A. Jarjis, in Application of Particle and Laser Beams in Paper Conservation, Institute of Paper Conservation, Leigh, 1986, Materials Technology, ed. P. Misalides, Kluwer Academic D68–D69. Publishers, Dordrecht, 1995, pp. 443–461. 10 G. B. Kelly, in Conservation of Library and Archive Materials and 24 R. A. Jarjis, in Conservation and Preservation of Islamic the Graphic Arts, ed. G. Petherbridge, Institute Of Paper Manuscripts, ed. Y. Ibish and G. Atiyeh, Al-Furqan Publications, Conservation/Society Of Archivists, London, 1987, pp. 117–123. London, 1996, pp. 93–117. 11 P. Schwerdt, in Sauvegarde et Conservation des Photographies, Dessins, Imprime�s et Manuscrits, ARSAG, Paris, 1991, pp. 213–216. Paper 8/04453H 2690 J. Mater. Chem., 1998, 8, 2685–