Fe(III) ratio in iron gall inks by potentiometry: A preliminary study

Journal of Electroanalytical Chemistry 650 (2010) 16–23

Contents lists available at ScienceDirect

Journal of Electroanalytical Chemistry journal homepage: www.elsevier.com/locate/jelechem

Determination of the Fe(II)/Fe(III) ratio in iron gall inks by potentiometry: A preliminary study Cédric Burgaud a,b, Véronique Rouchon a,b, Alain Wattiaux c, Jean Bleton d, René Sabot a, Philippe Refait a,⇑ a

Laboratoire d’étude des matériaux en milieux agressifs (LEMMA), EA 3167, Université de La Rochelle, Bât. Marie Curie, Av. Michel Crépeau, F-17042 La Rochelle cedex 01, France Centre de recherche sur la conservation des collections (CRCC), USR 3224, MNHN-CNRS-MCC, 36 rue Geoffroy Saint Hilaire, F-75005 Paris, France c Institut de chimie de la matière condensée de Bordeaux (ICMCB), Université de Bordeaux, CNRS, UPR 9048, 87 avenue du Dr. A. Schweitzer, 33608 F-Pessac Cedex, France d Laboratoire d’études des techniques et instruments d’analyse moléculaire (LETIAM), Université Paris Sud 11, IUT d’Orsay, Plateau du Moulon, F-91400 Orsay, France b

a r t i c l e

i n f o

Article history: Received 17 May 2010 Received in revised form 14 September 2010 Accepted 17 September 2010 Available online 22 September 2010 Keywords: Potentiometry Gallic acid Fe(II)/Fe(III) redox system Ancient manuscripts

a b s t r a c t Ancient iron gall ink manuscripts can be treated by immersion in water to dissolve excess iron compounds known to be involved in the degradation of such manuscripts. In this study, redox potential measurements were performed so as to follow the dissolution of iron gall inks from original degraded manuscripts over a period of time. Due to the complexity of the system, potentiometry was also applied to understand the interactions between the main components of iron gall inks, namely Fe(II) and Fe(III) sulphates, gallic acid and gum arabic. All these components were electrochemically active but the redox potential was mainly governed by the Fe(III)/Fe(II) redox couple. It could be demonstrated that one gallic acid molecule could readily reduce, in the experimental conditions considered here, up to four Fe(III) cations. Ó 2010 Elsevier B.V. All rights reserved.

1. Introduction Iron gall inks were commonly in use for writing or drawing until the beginning of the 20th century. The majority of these inks were composed of a mixture of vitriol, tannins and a binder [1]. The so called ‘‘vitriol” mentioned in the ancient recipes is close to iron (II) sulphate. As it was extracted from pyrites, it is probable that iron (III) was present along with other metallic impurities. The tannins are generally extracted from gallnuts by various means (cooking, decoction, etc.). They contain large quantities of polyphenolic acids that are decomposed by hydrolysis into smaller molecules. Gallic acid is one of the major constituents of these extracts when the hydrolysis is considered to be complete. The association of vitriol and tannins leads to the formation of ironbased solids and results in a dark colour which is characteristic of these inks. Such a compound was recently identified and characterised by Mössbauer and Raman spectroscopy [2–4]. A binder is generally associated in the recipes. It creates a suspension of dark particles and makes the ink texture more suitable for writing. In western countries, this binder was mainly gum arabic. When used on paper, iron gall inks may induce high damages, especially when the papers are stored in a humid and warm atmosphere. These degradation phenomena are considered to be the ⇑ Corresponding author. Tel.: +33 5 46 45 82 27; fax: +33 5 46 45 72 72. E-mail address: [email protected] (P. Refait). 1572-6657/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jelechem.2010.09.015

result of two major processes [5,6]. Firstly, the presence of tannins and iron in the ink leads to very acidic pH values (between 2 and 4) and the cellulose is very sensitive to acid hydrolysis. Secondly, iron (II) catalyses cellulose oxidation mechanisms initiated by Fenton-like reactions, thus contributing to its de-polymerisation. Iron is also expected to be one of the major actors of the cellulose degradation mechanisms. Yet little knowledge is available on its chemistry within original iron gall inks, mainly because of the experimental difficulties related to the analysis of original samples. Moreover, the particularly rich chemistry of iron allows a wide range of possible reactions: Fenton type reactions [7], chelation with polyphenols [8–11] or with sugars [12], redox reaction with gallic acid [13,14], oxidation induced by oxygen, formation of hydroxy compounds [15], etc. Several behaviours were forwarded but few of them were demonstrated. For instance, several iron– gallic acid complex structures are reported in literature [8,16,17], but the conditions of their synthesis differ drastically from that of iron gall ink and no experimental evidence proves their occurrence in original manuscripts. Despite this lack of knowledge, there is a need to define relevant conservation strategies on original manuscripts. A great effort of research was realized in the two last decades in order to investigate chemical treatments capable of delaying paper degradation induced by iron gall inks [6,18–22]. However, most of the conservation methods in use nowadays remain traditional. In particular, as water remains the preferred solvent used on paper objects, manuscripts

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

17

may be exposed to water in a number of treatments ranging from humidification to immersion in a wash bath [23]. Immersion treatments for instance, lasting from a few minutes to more than half an hour, are often used to dissolve part of the excess Fe compounds present in degraded manuscripts. During these treatments, side effects, such as elemental migrations of iron and sulphur in the paper core, have been pointed out [24]. These migrations are competing with elemental extraction in the solvent and also depend on the water application mode. The removal of iron and sulphur seems predominant in long time immersion treatment, whereas their migration in the paper is favoured by short time immersion treatment or high humidity exposure. Elemental quantitative measurements performed on treated original samples [25] shows however that all the iron present in the manuscript is not soluble. This article aims to estimate the potentiality of potentiometric methods for the study of the behaviour of iron during aqueous treatments. Almost no information about these dissolution processes is available. In particular, one may wonder which elements, among the Fe(II) or Fe(III), are the most soluble. As the phenomena occurring during the immersion are not monitored, nor controlled, the measurements of the redox potential of the solution via (for instance) a platinum electrode could provide information on the Fe(II)/Fe(III) ratio, and thus indicate the dissolution processes occurring during the treatment of ancient manuscripts. In this article, a specific electrochemical cell was designed for that purpose. Potentiometric experiments were performed with synthetic solutions containing mixtures of the main compounds present in iron gall inks, namely Fe(II) and Fe(III) sulphates, gallic acid and gum arabic. They allowed us to precise the interactions between these various constituents. These interactions were confirmed by Mossbauer spectroscopy measurements performed on dried residues. Finally, the methodology was applied to two ancient manuscripts.

enhanced. The redox electrode was a platinum grid that could be placed horizontally above the manuscript. A saturated mercury/mercurous sulphate electrode (MSE) was used as reference, but all redox potential values will be given with respect to the standard hydrogen electrode (SHE) (ESHE = EMSE + 0.651 V). An Ametek PAR 501 potentiostat monitored by the Soft Corr III software was used for recording the potential vs. time curves. The potential of the platinum grid was followed, with one measurement performed every 9  104 s, this being the maximum measurement rate of the apparatus. In order to check that the cell was free of contaminants, the potential of ultrapure water (150 mL) was recorded before each experiment. Then, as the potential became stable, 50 mL of the solution to be tested was prepared (in 2 min), introduced into the cell, and thus diluted with the pure water. In this work, all results are presented regarding to the final concentration of each ingredient after this dilution.

2. Materials and methods

2.3. Laboratory solutions used for potentiometric measurements

2.1. Electrochemical methods

As dissolution phenomena were to be studied, the concentrations considered here were approximately one order of magnitude below those found in original inks. The final concentration used for gallic acid was set at 2.35  103 mol L1 for all experiments, which led to a solution with a pH between 3.4 and 3.5. The same concentration of iron was then chosen for the solutions of Fe(II) and Fe(III) sulphates. The concentration used for gum arabic suspensions was also fixed at a unique value of 3.93 g L1, so that the proportions of iron and gum arabic correspond to a binder-rich iron gall ink according to an ancient French ink maker’s book dating from 1927 [30]. A preliminary study was devoted to the Fe(III)/Fe(II) redox system. Solutions containing various proportions of Fe(II) and Fe(III) 3þ sulphates were prepared. The activities of Fe2þ aq and Feaq in these solutions were computed using the MINTEQA2 software [31]. The various Fe(II) and Fe(III) complexes likely to form in the solutions þ 2+ (FeSO04 , FeSOþ 4 , FeOH , FeðOHÞ2 ) were all taken into account. The corresponding data are given in Tables 1 and 2. The equilibrium pH of the solutions was also computed and found close to the measured values given in Table 1. Of course, the computed activity val3þ ues of Fe2þ aq and Feaq take the variations of pH into account.

Since dissolution phenomena occurring in conservation workshops, i.e. in an open air environment, were researched, it was not envisaged to work in anoxic conditions, using for instance a closed cell with an argon flow, even though reactive compounds towards dissolved O2 were involved. The specific cell used in this study was then designed as simply as possible without any requirements to shelter samples and electrolytes from air. A schematic representation of this cell is given in Fig. 1. A glass container (25 cm long, 10 cm wide, 7 cm deep) was set on a vibrating apparatus so that the electrolyte was stirred and the transport of the dissolved species from the manuscript to the redox electrode was

2.2. Chemicals The solutions were prepared with laboratory pure products used as received. The chemical formula of gallic acid, i.e. 3,4,5-trihydroxybenzoic acid, can be written as C6H3(OH)3COOH. In this work, gallic acid monohydrate (ALDRICH reference 39,822-5, 98% purity) was chosen as a tannin, and iron(II) sulphate heptahydrate FeSO47H2O (ALDRICH reference 21,542-2, 99% purity) represents the ‘‘vitriol”. Iron(III) sulphate pentahydrate Fe2(SO4)35H2O (ALDRICH reference 30,771-8, 97% purity) was also used to get some more information on the interactions between iron(III) and gallic acid. Finally, gum arabic was chosen as a binder. It is a polysaccharide constituted of four main sugars (galactose, arabinose, rhamnose and glucuronic acid), but it generally also contains organic (2–5%) and metallic (4–7%) impurities [26–29]. The product used in this study was purchased from ALDRICH (reference 26,077-0).

2.4. Dried residue

Fig. 1. Electrochemical cell designed for the study of iron gall ink manuscripts.

In order to confirm some of the results obtained on Fe(III) sulphate and gallic acid solutions, we undertook to measure by Mössbauer spectroscopy the percentage of Fe(II) species present in dried residues. As these residues were additionally prepared for a better understanding of the chemistry of iron in iron gall

18

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

Table 1 Composition and thermodynamic data of the Fe(II)–Fe(III) sulphate solutions used for the study of the Fe(III)/Fe(II) redox system. The activities were computed with the MINTEQA2 program [31] using the equilibrium constants given in Table 2. Solutions

99% Fe(II)–1% Fe(III) 75% Fe(II)–25% Fe(III) 60% Fe(II)–40% Fe(III) 50% Fe(II)–50% Fe(III) 33% Fe(II)–67% Fe(III) 25% Fe(II)–75% Fe(III) 14% Fe(II)–86% Fe(III) 1% Fe(II)–99% Fe(III)

Concentrations (mol L1)

Computed activities

Measured pH

Fe(II)

Fe(III)

Fe2þ aq

Fe3þ aq

4.66  103 3.52  103 3.52  103 2.35  103 2.35  103 1.18  103 1.18  103 4.70  105

4.77  105 1.18  103 2.35  103 2.35  103 4.70  103 3.53  103 7.05  103 4.66  103

2.3  103 1.7  103 1.6  103 1.2  103 1.0  103 5.9  104 5.1  104 2.4  105

6.3  108 2.5  105 5.2  105 6.0  105 1.1  104 1.0  104 1.6  104 1.4  104

3.9 3.1 2.9 2.9 2.8 2.8 2.8 2.8

Table 2 Considered chemical equilibria and corresponding equilibrium constants [31]. Equilibria

log K

FeSO04 M Fe2+ + SO2 4 FeOH+ + H+ M Fe2+ + H2O

2.25

3+ FeSOþ + SO2 4 M Fe 4

9.5 3.92

3+ + 2SO2 FeðSO4 Þ 2 M Fe 4 FeOH2+ + H+ M Fe3+ + H2O þ + 3+ FeðOHÞ2 + 2H M Fe + 2H2O

5.42

FeðOHÞ03 + 3H+ M Fe3+ + 3H2O

13.6

+ 3+ Fe2 ðOHÞ4þ + 2H2O 2 + 2H M 2Fe + 3+ + 4H2O Fe3 ðOHÞ5þ 4 + 4H M 3Fe

2 + HSO 4 M SO4 + H

2.19 5.67 2.95 6.3 1.987

ink solutions, concentrated solutions of Fe(III) sulphate and gallic acid only were considered. The concentration of gallic acid used for this experiment (10 g L1) was close to saturation, and the chosen concentration of Fe(III) sulphate corresponded to a large molar excess of iron with respect to gallic acid, i.e. 5.5 iron atoms for one molecule of gallic acid. Original recipes frequently advised to leave the ink for several weeks before using it [30]. For this reason, we choose to leave the solution for 14 days at 60 °C in a closed bottle. After this period, it was left for several days in a large open container at ambient temperature to allow the evaporation of water. The dried residues remaining in the beaker at the end of the process were grounded roughly into powder for analysis. 2.5. Ancient manuscripts Two French valueless original samples, dating from the XVIIIth century, were considered for this work (see Fig. 2). They were both highly damaged by the ink, and offered a large quantity of matter for analysis. Yet their aspect was rather different. On the sample called ‘‘M03”, large brown halos could be observed around the inscriptions. In these halos, the paper was very fragile but could be manipulated. On the sample called ‘‘C04”, no large halo was observed. The degradation of the cellulose was located in the inked areas, and led to crumbling of the paper. 2.6. Mössbauer spectrometry and Proton Induced X-ray Emission (PIXE) analysis

Fig. 2. Visual aspect of the original samples.

Proton Induced X-ray Emission (PIXE) analysis was used in order to determine the elemental composition of the original manuscripts. These measurements were carried out on the AGLAE experimental plateform (Centre de Recherche et de Restauration des Musées de France, Paris), using a 3 MeV proton beam. The equipment and the process of PIXE spectra are depicted elsewhere [25,32]. Original manuscripts were also analysed by gas chromatography coupled with mass spectrometry in order to check the presence of gallic acid in the ink. A detailed description of the technique used is reported in the literature [33]. 3. Results and discussion

Mössbauer measurements were performed at room temperature using a constant acceleration HALDER type spectrometer, with a 57Co source (Rh matrix) in transmission geometry. The spectra were recorded at room temperature. The velocity was calibrated using pure iron metal as reference material. The experimental data were resolved into symmetric doublets with Lorentzian lineshapes using an iterative least-squares fit program.

3.1. The Fe(III)/Fe(II) redox system in aerated sulphate-containing solutions The first point that had to be clarified was the effect of dissolved oxygen on the measured values of the redox potential. In particular, Fe(II) is sensitive to the oxidising action of O2, and the resulting

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

formation of dissolved Fe(III) species could have a non-negligible impact on the potential taken by the platinum grid. To evaluate the importance of such effects, the redox potential of various Fe(II) and/or Fe(III) sulphate solutions was followed with time. Fig. 3 shows the potential vs. time curves obtained with Fe(II) (Fig. 3a) and Fe(III) (Fig. 3b) sulphate solutions. In each case, three curves were plotted, corresponding to distinct experiences performed in similar conditions, so as to illustrate the variations observed from one experiment to another. The redox potential measurements began before addition of Fe(II) or Fe(III). So the beginning of the curves describes the evolution of the redox potential of the initial distilled water, where the main redox couple can be assumed to be O2/H2O (E° = 1.23 V/SHE). After a rapid increase during the first minutes, the potential somewhat stabilised at a value of 0.73 ± 0.02 V/SHE. The iron sulphates were added as small volumes of concentrated solutions when the potential could be considered as stable. The addition of Fe(II) produced an immediate decrease of the potential (Fig. 3a) and the addition of Fe(III) an immediate increase of the potential (Fig. 3b). The potential of the platinum electrode is then linked to the redox couple Fe(III)/Fe(II) (E° = 0.77 V/SHE). Due to the addition of Fe(II), the redox potential decreased to 0.54 ± 0.01 V/SHE, but then rapidly increased before stabilizing at 0.61 V/SHE. This phenomenon may be attributed to a partial oxidation of Fe(II) by the dissolved O2. The redox potential increased very rapidly in the first few minutes as (i) the initial Fe(III) concentration was very low and its increase necessarily produced important variations of potential and (ii) dissolved O2 was available. The redox potential then stabilised, which may be explained by the consumption of the initial amount of dissolved O2. Note that the increase of the redox potential may also be, like that of the equilibrium potential given by Nernst’s law, logarithmic vs. the Fe(III) concentration. The addition of Fe(III) induced an increase of the redox potential up to 0.93 ± 0.09 V/SHE. In contrast with what was observed with Fe(II), the potential remained stable afterwards, showing little decrease. This is of course due to the fact that Fe(III) does not react with O2. Note that the stability of the potential of Fe(III) solutions confirms that the increase observed with Fe(II) solutions is due to the oxidation of Fe(II) by dissolved O2. Similar experiments (not shown) were performed with the solutions of Fe(II) and Fe(III) sulphates described in Table 1. The resulting redox potential increased with the Fe(III)/Fe(II) ratio, as illustrated by Fig. 4. The experimental variation can be compared with the theoretical one deduced from the equilibrium conditions 3þ between Fe2þ aq and Feaq [34]:

Fe2þ ¼ Fe3þ þ e

ð1Þ

Eeq ¼ 0:775 þ 0:059 logðaFeIII =aFeII Þ

ð2Þ

19

2þ aFeIII and aFeII are the activities of Fe3þ aq and Feaq , respectively. They were computed using the MINTEQA2 program [31], which allowed us to draw the Eeq = f[log(aFeIII/aFeII)] line in Fig. 4. It can be seen that the experimental values are found close to this theoretical curve, except for the smallest log(aFeIII/aFeII) value. In fact, the measured redox potential is always larger than the theoretical equilibrium potential. It must be recalled that the potential of the platinum grid is a mixed potential, mainly related to the predominant redox couple, in our case Fe(III)/Fe(II), but also depending on the other redox couples present and in particular O2/H2O. However, the discrepancy between measured and computed values is important for the high Fe(II) proportion, i.e. 99% Fe(II). This illustrates the effect of the oxidation of Fe(II) into Fe(III) by dissolved O2: The Fe(III)/Fe(II) concentration ratio is in this case significantly larger than expected. A curve fitting led to the equation (note that the point corresponding to 99% Fe(II) was omitted):

Eredox ¼ 0:789 þ 0:058 logðaFeIII =aFeII Þ

ð3Þ

The slope of the curve differs only slightly from the expected value of 0.059 V and the main discrepancy with the theoretical curve is a shift of 14 mV towards more positive values. This is less than the differences observed between various identical experiments, as it can be seen in Fig. 3. This slight difference has various origins: (i) the redox potential is a mixed potential, as explained above, and (ii) the electrolyte is at room temperature (21–23 °C) and not exactly at 25 °C. In conclusion, even if potentiometric experiments are performed in aerated conditions, as it could be the case in restoration workshops, the Fe(II)/Fe(III) ratio of the dissolving part of the ink impregnating a manuscript could be estimated. Such estimation would require a calibration curve such like that drawn from our experiments in Fig. 5, representing the measured redox potential as a function of the molar fraction of Fe(II) of the iron sulphate solution, that is x(FeII) = [FeII]/([FeII]+[FeIII]). This representation illustrates how the potential is sensitive to the Fe(II) proportion at the two extreme situations, i.e. for pure Fe(II) and pure Fe(III) solutions.

3.2. The redox systems involving gallic acid and gum arabic Gallic acid exhibits the characteristics of a reducing agent and it was actually observed that it could reduce Fe(III) [13,14]. It was proposed that two adjacent –OH groups linked to the aromatic cycle are oxidised in such a process. In this case, one gallic acid molecule could reduce two Fe(III) cations. Similarly, gum arabic is a polysaccharide constituted of four main sugars (galactose, arabinose, rhamnose and glucuronic acid) with –OH groups that can be considered as potential reducing agents.

Fig. 3. Potential vs. time curves obtained when adding a solution of Fe(II) sulphate (a) or Fe(III) sulphate (b) to distilled water. The experiments were performed in the specific cell described in Fig. 1. Fe sulphate concentration = 2.35  103 mol L1.

20

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

Fig. 6. Potential vs. time curves obtained when adding successively a solution of gallic acid and a suspension of gum arabic to distilled water. In one case gallic acid was added first, in the other case gum arabic was added first. The experiments were performed in the specific cell described in Fig. 1. [Gallic acid] = 2.35  103 mol L1; [gum arabic] = 3.9 g L1. Fig. 4. Variation of the redox potential with the activity ratio a(FeIII)/a(FeII).

Fig. 5. Curve showing the variations of the measured redox potential with the molar fraction of Fe(II), x(FeII) = [FeII]/([FeII] + [FeIII]).

Fig. 6 shows the potential vs. time curves obtained when adding gum arabic and gallic acid to water. As it was the case for the study of the Fe(III)/Fe(II) system, a small volume of concentrated solution containing the considered compound was added to water once the redox potential was stabilised. The curve in solid line was obtained by adding the gum arabic first, and the gallic acid later. The curve in dotted line was obtained by adding the gallic acid before the gum arabic. First, it can be seen that the addition of gallic acid or gum arabic induced a decrease of the redox potential. So, both species are electrochemically active and their presence can then be revealed by potentiometric experiments. Secondly, it is interesting to note that both curves stabilised at the same potential value, 0.58 V/ SHE, when both species are present. This value differs from those reached with gum arabic and gallic acid alone, equal to 0.68 and 0.62 V/SHE, respectively.

for Fe(II) sulphate solutions (Fig. 3). The fact that the same value is obtained whether gum arabic is present or not shows that the redox potential is mainly controlled by the Fe(III)/Fe(II) redox system. As observed in Fig. 3a with solutions of Fe(II) sulphate, the redox potential increases as Fe(II) is oxidised into Fe(III) by dissolved O2. It can be seen that this increase of potential is less important in the presence of gum arabic. It reached 0.58 V/SHE after 0.4 h, whereas it reached 0.61 V/SHE without gum arabic. In contrast, when gum arabic was added to the solution of Fe(II) sulphate, the redox potential increased again, reaching 0.65 V/SHE. This effect may be attributed to dissolved O2 present within the solution containing gum arabic and consequently introduced into the system together with gum arabic. If the redox potential is controlled by the Fe(II)/Fe(III) redox system, the proportion of Fe(III) can be estimated from the potential value via the calibration curve of Fig. 5. It would be close to 1% when the gum arabic was added first and about 10% when the gum arabic was added secondly. The two curves obtained with Fe(III) sulphate are shown in Fig. 7b. The addition of gum arabic decreased the potential of 50 mV, that of Fe(III) increased the potential of 200 mV, and in both cases the redox potential stabilised at the same value, 0.86 V/SHE. This redox potential value is typical of Fe sulphate solutions containing less than 1% of Fe(II) as it can be seen in Fig. 5. These observations show that some species associated with gum arabic interact with dissolved Fe species. For instance, it seems that a slight reduction of Fe(III) occurred as expected (see Section 3.2). This would explain (i) the decrease of the redox potential when gum arabic was added to Fe(III) sulphate (Fig. 7b) and (ii) the smaller increase of the potential when Fe(II) sulphate was added to gum arabic (Fig. 7a). Moreover, since K+ and Ca2+ ions can be released from the gum into the solution, small amounts of gypsum CaSO42H2O and jarosite KFe3(SO4)2(OH)6 could precipitate. Actually, yellow particles appeared after some time in the solutions where Fe(III) sulphate and gum arabic were both present. This illustrates the complexity of the Fe(III)/Fe(II) – O2/H2O – gum arabic system.

3.4. Effects of gallic acid on the Fe(III)/Fe(II) redox system 3.3. Effects of gum arabic on the Fe(III)/Fe(II) redox system Fig. 7 shows the potential vs. time curves obtained with iron sulphate and gum arabic. The first set of curves (Fig. 7a) was obtained with Fe(II) sulphate. It shows that the two curves, obtained by changing the order of addition of the species in water, are different. However, in both cases, the addition of Fe(II) led to a decrease of the redox potential down to 0.54 V/SHE, the characteristic value

Note that the solutions containing Fe(II) and gallic acid were light blue and tended to darken with time. The solutions containing Fe(III) and gallic acid were initially dark blue but rapidly turned green. This variety of colours shows that several reactions are occurring, each of them influencing the colour of the ink. Fig. 8 shows the potential vs. time curves obtained with iron sulphate and gallic acid. The first set of curves (Fig. 8a) was

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

21

Fig. 7. Potential vs. time curves obtained when adding successively a solution of gum arabic and a solution of Fe(II) sulphate (a) or Fe(III) sulphate (b) to distilled water. In one case the Fe sulphate solution was added first, in the other case gum arabic was added first. The experiments were performed in the specific cell described in Fig. 1. [Fe sulphate] = 2.35  103 mol L1; [gum arabic] = 3.9 g L1.

Fig. 8. Potential vs. time curves obtained when adding successively a solution of gallic acid and a solution of Fe(II) sulphate (a) or Fe(III) sulphate (b) to distilled water. In one case the Fe sulphate solution was added first, in the other case gallic acid was added first. The experiments were performed in the specific cell described in Fig. 1. [Fe sulphate] = [gallic acid] = 2.35  103 mol L1.

obtained with Fe(II) sulphate. Both curves stabilised at the same potential value, about 0.55–0.57 V/SHE, typical of Fe(II) sulphate solutions. As expected, when Fe(II) sulphate was added before gallic acid, the potential dropped down to 0.55 V/SHE and increased to 0.62 V/SHE as Fe(II) was oxidised into Fe(III) by O2. The addition of gallic acid then induced a decrease of the potential back to 0.57 V/SHE. When Fe(II) sulphate was added in the gallic acid solution, the potential dropped down to 0.55 V/SHE and stabilised at this value, as if Fe(II) was not oxidised by dissolved O2. These results suggest that dissolved Fe(III) species are reduced by gallic acid, as it was previously observed [13,14]. The two curves obtained with Fe(III) sulphate (Fig. 8b) confirmed this assumption. The addition of gallic acid to the solution of Fe(III) sulphate induced a drop of the redox potential from 0.90 V/SHE, a value typical of Fe(III) solutions, down to 0.62 V/ SHE, a value that would correspond to a solution containing less than 5% of Fe(III) (see Fig. 5). Similarly, the addition of Fe(III) sulphate to a gallic acid solution led to a sharp increase of the potential immediately followed by a decrease down to 0.62 V/SHE. Additional experiments were performed to estimate the number of Fe(III) cations that one molecule of gallic acid could reduce. An example of the potential vs. time curves obtained is shown in Fig. 9. One unit (2.35  103 mol1) of gallic acid was added first and the redox potential dropped down to 0.60 V/SHE. Then, two units of Fe(III) sulphate were added, inducing a sharp increase of the potential up to 0.70 V/SHE. But Fe(III) was rapidly reduced by gallic acid and the potential went back to 0.60 V/SHE. Another unit of Fe(III) was added 0.4 h after. The corresponding Fe(III) cations were also reduce to Fe(II), but less rapidly. A fourth unit was added again, and reduced again. Thus, one gallic acid molecule can reduce

Fig. 9. Potential vs. time curve obtained when adding successively a solution of gallic acid and three solutions of Fe(III) sulphate to water. Two, one and one units of Fe(III) were added successively to one unit of gallic acid. One unit = 2.35  103 mol L1.

at least four Fe(III) cations. Finally, up to six units of Fe(III) were added (curve not shown). The redox potential stabilised at that time at about 0.70 V/SHE, a value corresponding to 65% of Fe(II), i.e. four of the six units of Fe(III) were reduced, and two remained as Fe(III). Only a few studies were devoted to the electrochemical behaviour of gallic acid. The most recent [35] showed polarisation curves obtained with a platinum electrode dipped in gallic acid solutions. Four oxidation peaks were obtained, which implies that electrons were transferred from each of the three –OH groups as well as the COOH group. This confirms that one gallic acid molecule is able to transfer four electrons to Fe(III) cations, i.e. to reduce four Fe(III) cations into Fe(II). We then researched whether similar observations could be formulated on solid state materials. We therefore considered the dried residues depicted in Section 2.3 and analysed them at room

22

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

temperature by Mössbauer spectrometry. The resulting spectrum (Fig. 10) is composed of three doublets of peaks. Their Mössbauer parameters are listed in Table 3. The two main doublets are composed of two peaks separated by D values of 2.73 and 3.28 mm s1 and centred on CS values close to 1.28 mm s1. This is typical of Fe(II) compounds. More precisely, both doublets correspond to Fe(II) sulphates. The main doublet can be attributed unambiguously to szomolnokite, FeSO4H2O [36], while the second Fe(II) doublet may correspond whether to rozenite FeSO44H2O or melanterite FeSO47H2O [37]. The third doublet is composed of two peaks separated by D = 0.29 mm s1 and centred on CS = 0.25 mm s1, which is typical of Fe(III) compounds. It may correspond to one of the various Fe(III) sulphates that have similar Mössbauer parameters [38]. The atomic proportion of Fe(III) can be estimated from the relative area of the corresponding doublet, that is 22%. Assuming that one gallic acid molecule can reduce four Fe(III) cations, there would have remained 1.5 Fe(III) of the 5.5 initially present, which corresponds to 27%. This is consistent with the estimation given by Mössbauer spectroscopy. This consistency between Mössbauer analysis, focused on Fe species, and potentiometric results confirm that the redox potential seems to be mainly controlled by the Fe(III)/Fe(II) redox couple. Of course, the presence of gallic acid and its oxidised states implies that the redox potential of the platinum electrode is a mixed potential. The presence of gallic acid has necessarily an influence on the redox potential, even if it seems to be masked by that of the Fe(III)/Fe(II) redox couple. Moreover, the presence of gallic acid modifies the ionic strength of the solution, induces the formation of iron–gallic acid complexes, which in turn modifies the activities 3þ of Fe2þ aq and Feaq species. So, the Fe(II) proportion as given by the calibration curve of Fig. 5 must be considered as a rough estimate. 3.5. Study of ancient manuscripts The original manuscripts depicted in Section 2.4 were tested. Their elemental compositions, reported in Table 4, were deter-

Table 4 Characterization of the original manuscripts.

Before immersion Concentration in Fe (lg cm2) Other elements (concentration in (lg cm2) % Fe(II)

M03

C04

Technique used

120 ± 30

115 ± 30

PIXE analysis in inked areas

Al(0.2), Si(0.9), S(18), K(10), Ca(10), Cu(1.7) 90%

Na(0.9), Al(1.3), Si(0.9), P(0.8), S(9.6), K(22), Ca(4), Cu(0.4) 10%

At the end of the immersion Potential of the 0.63 ± 0.02 solution (V/SHE) % Fe(II) in the 90% solution

0.53 ± 0.01

Mössbauer spectroscopy Electrochemistry

100%

mined by PIXE analysis considering the average of five measurements performed on inked areas. Both manuscripts contained a large amount of iron, close to 120 lg cm2 and some traces of copper (less than 1 lg cm2). No other transition metals were detected. The iron salts used for the manufacture of the inks were also very pure and the contribution of other transition metal on the potential can reasonably be neglected. As GC/MS analysis performed on these manuscripts indicated very low gallic acid content (not shown), an eventual reduction of Fe(III) in the solution is also improbable. We can also suppose that the calibration curve of Fig. 5 is relevant for the interpretation of the potentiometric data recorded on the two manuscripts. Several fragments of each manuscript were introduced in the cell and immersed in water for 30 min. For a same manuscript, the potential vs. time curves obtained from one fragment to another were very similar. The different values of potential measured on the two manuscripts (see Fig. 11) are also attributed to the intrinsic composition of the manuscripts and not to their heterogeneity. The Mössbauer analysis of the manuscripts (not shown) performed on other fragments, containing both inked and non inked areas, revealed that the manuscript M03 was initially very rich in Fe(II) (approx. 90%) whereas C04 was rich in Fe(III) (approx. 90%). However, the potential vs. time curves obtained with the two manuscripts behave similarly (see Fig. 11). In each case the redox potential dropped after the immersion of the sample then tended

Fig. 10. Mössbauer spectrum of the compound obtained by mixing Fe(III) sulphate and gallic acid solutions, after 14 days of ageing at 60 °C, once water has evaporated.

Table 3 Mössbauer spectral parameters of the compound obtained by mixing Fe(III) sulphate and gallic acid solutions, after 14 days of ageing at 60 °C, once water has evaporated. CS = centre shift with respect to metallic a-iron at room temperature; D = quadrupole splitting; RA = relative area; FWHM = full widths at half maximum. Standard deviation is given in brackets. CS (mm s1)

D (mm s1)

FWHM (mm s1)

RA (%)

Fe oxidation no.

Compound

1.279(9) 1.28(1)

2.734(9) 3.28(1)

0.262(6) 0.231(6)

47 31

+2 +2

0.25(3)

0.29(3)

0.34(2)

22

+3

Szomolnokite Melanterite or rozenite Fe(III) sulphate

Fig. 11. Potential vs. time curves obtained during the immersion of two ancient manuscripts in water. The experiments were performed in the specific cell described in Fig. 1. The pH of the solutions was measured at 4.0 ± 0.1 for M03 and 4.92 ± 0.02 for C04.

C. Burgaud et al. / Journal of Electroanalytical Chemistry 650 (2010) 16–23

to decrease more slowly as part of the solids associated to the manuscript dissolved in water. At the end of the experiments, samples from manuscript C04 have reached a potential of 0.53 ± 0.01 V/ SHE, and samples from manuscript M03 a potential of 0.63 ± 0.02 V/SHE. If the cellulose and ink degradation products released in the solution have a negligible impact on the potential, the calibration curve of Fig. 5 may be used to determine the percentage of soluble Fe(II). It could then be forwarded that the manuscript C04 released almost exclusively Fe(II) whereas the manuscript M03 released 90% Fe(II) and 10% Fe(III). The Fe(II) and Fe(III) compounds of the manuscript M03 would then present a similar solubility, whereas on manuscript C04, Fe(II) compounds only would be soluble. This preliminary result should however be considered with care, as no information is available on the possible influence of the other soluble products released by the manuscripts. Moreover, the effect of dissolved oxygen on the redox potential may depend on the concentration of Fe(II) released in solution. If the Fe(II) concentration is negligible, comparable to that of dissolved O2, a major part of Fe(II) would be oxidised and the redox potential would be high and typical of a Fe(III) solution. If the Fe(II) concentration is much larger than that of dissolved O2, then the effects of dissolved O2 on the redox potential should be small and similar to those observed in Section 3.1.

4. Conclusions The laboratory study performed on synthetic solutions mixing the main compounds present in iron gall inks revealed that the redox potential was mainly controlled by the Fe(III)/Fe(II) redox system. This allowed us to establish that Fe(III) could be readily reduced by gallic acid, and that one gallic acid molecule could reduce up to four Fe(III) cations in the experimental conditions considered here. The interactions between gum arabic and Fe species proved to be more complex, probably due to the fact that gum arabic is itself a complex mixtures of organic molecules, metallic and organic impurities. The use of potentiometric experiments in aerated conditions had of course an impact on the redox potential measurements as dissolved O2 could oxidise part of the Fe(II) species present in solution. Significant discrepancies between expected redox potentials and measured ones proved however to be significant only for the solutions containing mainly Fe(II), i.e. 99% Fe(II) and 1% Fe(III). Potentiometric experiments could finally be successfully applied to monitor the release of dissolved species from ancient manuscripts. This procedure must now be developed so that the information obtained could be correlated with some of the properties of the manuscript.

Acknowledgements This study was performed within the PhD work of C. Burgaud co-financed by the ‘‘Centre National de la Recherche Scientifique”, and the ‘‘Conseil Régional de Poitou-Charentes”.

23

References [1] A. Stijnman, in: J. Kolar, M. Strlic (Eds.), Iron Gall Inks: on Manufacture, Characterisation, Degradation, and Stabilisation, National and University Library, Ljubljana, 2006, pp. 25–67. [2] C. Burgaud, V. Rouchon, Ph. Refait, A. Wattiaux, Applied Physics A 92 (2008) 257–262. [3] A.S. Lee, P.J. Mahon, D.C. Creagh, Vibrational Spectroscopy 41 (2006) 170–175. [4] V. Rouchon, J. Bleton, C. Burgaud, K. Janssens, P. Refait, A. Wattiaux, Techné (2008) 60–67. [5] G. Banik, H. Weber, Tintenfrassschäden und ihre Behandlung, Kohlhammer, Stuttgart, 1999. [6] J. Kolar, M. Strlic, in: J. Kolar, M. Strlic (Eds.), Iron Gall Inks: on Manufacture, Characterisation, Degradation, and Stabilisation, National and University Library, Ljubljana, 2006, pp. 181–194. [7] M. Strlic, T. Radovic, J. Kolar, B. Pihlar, Journal of Agricultural and Food Chemistry 50 (2002) 6313–6317. [8] J. Gust, J. Suwalski, Corrosion Science 50 (1994) 355–366. [9] R.C. Hider, A.R. Mohd-Nor, J. Silver, I.E.G. Morrison, L.V.C. Rees, Journal of the Chemical Society, Dalton Transactions (1981) 609–622. [10] P.K. Migal, V.A. Ivanov, Russian Journal of Inorganic Chemistry 18 (1973) 536– 540. [11] M.J. Hynes, M.O. Coinceanainn, Journal of Inorganic Biochemistry 85 (2001) 131–142. [12] A. Kongdee, T. Bechtold, Carbohydrate Polymers 56 (2004) 47–53. [13] H. Kipton, J. Powel, M.C. Taylor, Australian Journal of Chemistry 35 (1982) 739–756. [14] V. Rouchon-Quillet, C. Rémazeilles, J. Bernard, A. Wattiaux, L. Fournès, Applied Physics A 79 (2004) 389–392. [15] A. Kongdee, T. Bechtold, Dyes and Pigments 60 (2004) 137–142. [16] C. Krekel, Chemische Untersuchungen an einsengallustinten und anwendung der ergebnisse bei der begutachtung einer mittelalterlichen handschrift, Diplomarbeit, Institut für Anorganishe Chemie, Georg August Göttingen University, 1990. [17] C.H. Wunderlich, Geschichte und Chemie der Eisengallustinte und Struktur des Eisen(III)-Gallussäure-Komplexes, Diplomarbeit, Fach Chemie an der Rheinishen Friedich-Whilhems, Universität Bonn, 1990. [18] J.G. Neevel, Restaurator 16 (1995) 143–160. [19] J. Kolar, A. Mozir, M. Strlic, G. de Bruin, B. Pihlar, T. Steemers, e-Preservation Science 4 (2007) 19–24. [20] B. Wagner, E. Bulska, Analytical and Bioanalytical Chemistry 375 (2003) 1148– 1153. [21] O. Hahn, M. Wilke, T. Wolff, Restaurator 29 (2008) 235–250. [22] U. Henniges, A. Potthast, Restaurator 29 (2008) 210–234. [23] B. Reissland, Restaurator 20 (1999) 167–180. [24] V. Rouchon, B. Durocher, E. Pellizzi, J. Stordiau-Pallot, Studies in Conservation 54 (2009) 236–254. [25] V. Rouchon, C. Burgaud, T.P. Nguyen, M. Eveno, L. Pichon, J. Salomon, Papier Restaurierung 9 (2) (2008) 18–28. [26] O.H.M. Idris, P.A. Williams, G.O. Phillips, Food Hydrocolloids 12 (1998) 379– 388. [27] D.M.W. Anderson, F.J. McDougall, Food Additives and Contaminants 4 (1987) 125–132. [28] D.A. El Kader, E. Molina, G. Molina, L. Montero, G. Léon de Pinto, Food Hydrocolloids 17 (2003) 251–253. [29] C. Rémazeilles, V. Rouchon-Quillet, J. Bernard, T. Calligaro, J.C. Dran, L. Pichon, J. Salomon, M. Eveno, Restaurator 26 (2005) 118–133. [30] Champour, Malepeyre, Villon, Chaplet, Fabrication des Encres, Encyclopédie Roret, Mulo, Paris, 1927. [31] MINTEQA2 Version 3.11, J.D. Allison, USEPA, Athens, Georgia. [32] C. Rémazeilles, V. Quillet, T. Calligaro, J.C. Dran, L. Pichon, J. Salomon, Nuclear Instrument and Methods B 181 (2001) 681–687. [33] J. Bleton, C. Rémazeilles, V. Rouchon-Quillet, A. Tchapla, in: International Conference of Chemical Technology on Wood, Pulp, and Paper, Section ‘‘Paper in Archives and Libraries”, Bratislava, 17–19 September 2003, pp. 387–392. [34] G.H. Kelsall, R.A. Williams, Journal of the Electrochemical Society 138 (1991) 931–940. [35] M.K. Carter, Journal of Molecular Structure 831 (2007) 26–36. [36] J.P. Eymery, F. Ylli, Journal of Alloys and Compounds 298 (2000) 306–309. [37] A.A. Olowe, J.-M.R. Génin, Hyperfine Interactions 68 (1991) 253–256. [38] J.A. Hurowitz, N.J. Tosca, M.D. Dyar, American Mineralogist 94 (2009) 409–414.