·Home ·Table of Contents ·Conservation and Restoration in Art and Architecture

FTIR techniques applied to iron gall inked damaged paper REMAZEILLES Céline, QUILLET Véronique, BERNARD Jacky LEMMA, Université de La Rochelle, avenue Marillac, 17042 La Rochelle Cedex Phone : (33) 0546458217, Fax : (33) 0546458243 Email : vquillet@univ-lr.fr ou cremazei@univ-lr.fr Contact

Key words: ink - irongall - FTIR - degradation - cellulose.

INTRODUCTION

Iron gall ink corrosion of paper is one of the largest threat for our graphic patrimony. A great work has been done in this field to explain the possible mechanism of paper degradation and to propose curative methods [1,2,3,4]. The main degradation mechanism proposed in the literature is the following : iron gall ink prepared with different ingredients including tannins and vitriol causes both acidic hydrolysis and Fe²⁺ catalysed oxidation of cellulose. Paper turns brown and loses its mechanical properties. Yet the great variety of iron gall ink recipes [5,6], and the great variety of visual aspects of manuscripts suggest that many side effects could occur and contribute to the different aspects of paper degradation (colour changes, halos, mechanical properties).

Many analytical methods have been explored to analyse original iron gall inked manuscripts: among all of these, RAMAN makes it possible to identify the paper charges and the carbon inks, Gas Chromatography provides information on the evolution of gallic acid and/or the binder[7,8], SEM and/or PIXE analysis[2,9,10] are used to measure the elementary composition of the ink and the paper. But Fourier Transform Infra Red (FTIR) spectroscopy is not fully explored in this field. Infra Red techniques have been largely used in the paper industry for identification of different types of cellulose, hemicellulose and lignin. Research were mainly focused on paper made from wood and have been already reviewed elsewhere[11,12]. Yet, before the XIXth century, only hand paper sheets made from stuff were in use, and FTIR techniques may provide further information on this types of paper and/or on the writing itself. Some preliminary tests have been undertaken on original drawings, in order to identify the kind of ink used [13]. Yet the identification of metallogallic or organic ink via FTIR spectroscopy appears rather difficult, since the main signal is due to the cellulose absorption. Mosini and all[14] pointed out that FTIR spectroscopy provides informations on the cellulose oxidation stage. More recently, this technique has been used successfully to observe the paper carbonate content decrease which results from the transformation of calcium carbonate into calcium sulfate [15]. FTIR spectra provides also useful information on the evolution of paper. Still we observed that the limits and the possibilities of the different FTIR techniques for the identification of metallogallic inked paper have not been explored, mainly because the interpretation of spectra is rather difficult : indeed, the cellulose remains the major component of the sample even in inked and damaged paper and only small discrepancies are observed between inked or damaged paper spectra and pure cellulose spectra. The first point to test is also the reproducibility of the measurements. Then small influences can be pointed out. Despite all this work, the interpretation of discrepancies measured on original samples remains difficult because the ink composition, the ink degradation and the paper degradation may have combined influences.

In this work, we measured FTIR spectra on original and laboratory probes by different techniques : reflectance microscopy, diffuse reflectometry with KBr (DRIFT), and transmission microscopy throught a fibre. Advantages and drawbacks of each technique will be discussed. Interpretation of spectra will be discussed.

I - EXPERIMENTALS:

I-1 : FTIR APPARATUS:


FTIR reflectance microscopy:
the FTIR spectra are recorded directly on the sample via a Perkin Elmer Spectrum 2000. The main measurements features are the following : range 4000 to 600 cm^-1 , 100 scan, resolution : 4 cm^-1, gain 8. The analysed area is about 100x100 squaremicrons -

Transmission through a fibre:
a fibre of the sample is taken of and pressed between two diamond cells. The spectra are recorded with a Nicolet 740i and with the following conditions : range 6000 to 600cm^-1, 256 scan, resolution 8 cm^-1, gain 2, diaphragm 1.5mm.

Diffuse reflectance spectroscopy with KBr:
a small quantity of matter (a few milligrams) is grounded, then mixed with KBr and placed in a DRIFT cell . The spectra are recorded with a Perkin Elmer Paragon 1000 PC . The main measurements features are the following : range 4000 to 400 cm^-1, 35 scan, resolution 4 cm^-1, gain 1.

I-2 : SAMPLES :


Original sample :
An original sample was given to us by an antique dealer. It belongs to a small letter wich was sliped into an eighteenth century printed book. The ink has strongly corroded the paper, yielding to large brown halos around inscriptions and loss of matter in the inscriptions.

Laboratory probes :
An incomplete ink was prepared following a basic-recipe used by Neevel :distilled water (200 ml), monohydrated gallic acid (Aldrich, ref 39,822-5 - 0.21g) and heptahydrate iron sulphate (Aldrich ref 21,542-2 - 1.138g).
Laboratory probes were prepared with a 100% linen paper without any charge or sizing. The paper past was prepared by a French paper hand maker mill ( the «Moulin de Richard de Bas») with pure linen stuff. Then paper sheets were prepared with a leaf casting machine (average weight 90g/m2). Small 3x3cm2 squares were then immersed in the ink for a few seconds and dried in ambient conditions for a few days. Each probe is weighed before and after the inking process to determine the ink content. The average mass concentration of the probes was found to be 0,01% for the ink with a standard deviation of 0,0034.
Samples containing the ink n°1 were artificially aged in a programmable ageing oven (Vötsch VC 0020) for 7, 14, 21 and 42 days. Ageing conditions were similar to those used by Neevel : temperature remains constant at 90°C - Cyclic relative humidity (RH) conditions were programmed : 3 hours at 80% - decreasing within 1hour to 35% - 3 hours at 35% - increasing within 1 hour to 80% .

II- EXPERIMENTAL PROCESS - COMPARISON OF DIFFERENT FTIR TECHNIQUES

II-1 : FTIR REFLECTANCE MICROSCOPY :
The main advantage of this technique remains in its totally non destructive aspect. Measurements can be carried out in a few second directly on an original sample inked area. The reproducibility of the measurements is however rather poor : figure 1 shows different spectra which are recorded on different inked and blank areas of the same document. Analysed areas were very close to each other (less than 2mm), so that the composition of the ink and the paper degradation stage are reasonably supposed to be uniform. To compare several measurements, we choose to translate all the spectra to zero at 1900cm^-1, because no absorption band appears in this region. Since the dimension of the beam is comparable to the paper fibre diameter, the surface heterogeneity should induce some discrepancies for the background and/or for the relative intensities of the bands. In fact, the observed discrepancies were much larger that what was expected, especially in the cellulose high absorption region:

Fig 1: reflectance microscopy on an original sample On the top : FTIR reflectance microscopy spectra on inked areas. In the middle : FTIR reflectance microscopy on virgin paper. On the bottom : Diffuse reflectance spectroscopy of pure cellulose mixed with KBr

• the cellulose strong absorption region (1200 to 950 cm^-1)
Despite the cellulose being the major component of the sample, no strong absorption band appears in this region. Moreover, the absorption bands position are not reproducible. This behaviour is due to the competition between two opposite effects since the collected signal includes both regular and diffuse reflection. Indeed, the strong absorption frequencies correspond to a high reflective index too. Also strong absorption occurs simultaneously with regular reflection. The measured signal corresponds to the sum of these two opposite effects. Small sample heterogeneity induces also large discrepancies in the shape of the spectra and no reliable information can be drawn from the measurements (16)

• the cellulose low absorption region (950 to 600 cm^-1 and 1900 to 1200 cm^-1)
The low absorption region of the cellulose corresponds to a low reflective index. The regular reflection spectrum becomes also negligible. The light diffuses in the sample. Then it is reflected at the bottom side of the sample and comes back to the sample surface. The measured spectra are also similar to a transmission profile. The reproducibility of the signal is better than in the previous case, but it is still limited by both the surface heterogeneity and the small dimension of the spot.
Also even if the measurements are more reproducible in the cellulose medium-low absorption region than in the cellulose strong absorption region, one must admit that the reproducibility of reflectance microscopy is too limited to measure small effects with accuracy.

Despite this limitation, reflectance microscopy makes it possible to compare very quickly the inked and blank areas. We first notice that the paper profile is very close to the pure cellulose profile, except for the band at 1650 cm^-1 which corresponds to the presence of water in the fibre. One can notice furthermore that inked areas are more absorbent than blank areas in the range 1750 to 1600 cm^-1. This increase of absorption could be attributed to the cellulose oxidation process described by Neevel and all. Finally, one can think after a careful examination, that the inked areas spectra show an increase of absorbance around 800 cm^-1, 1600 cm^-1 and maybe 1400 cm^-1. Since these bands are rather small, we decided to explore other FTIR techniques to confirm these assumptions.

II-2 : TRANSMISSION THROUGH A FIBRE :
This technique can be considered as almost non destructive, because of the small quantities of matter required (some paper fibres). More interesting, the paper absorption profile can be measured on the verso of an inscription by sampling a non inked, but corroded fibre. Deviation from the cellulose spectra gives then information on the paper damages and/or the migration of some ink constituent but not on the ink itself. This technique appears also particularly interesting when the paper turns brown on the verso of the sheet but not around inscriptions.
Figure 2 shows several spectra recorded on the same fibre. During the measurements, the IR light is transmitted through a squashed fibre, so that cellulose strong absorption bands do clearly appear. This technique is also more reproducible than the previous one, since absorption bands are at least located at the same frequency in all the spectra. Still, some discrepancies are observed, which can be attributed to several effects :

• the definition of cellulose absorption bands is generally poorer in the fibre's heart areas than in fibre's peripheral areas. The fibre heterogeneity is not negligible
• the thickness of matter over crossed is not controlled.
• light interferences at the diamond interfaces may create some distortions of the background.

Fig 2-a : transmission microscopy through a flax fibre (laboratory probe) Files 1 and 2 : virgin fibre (inner part). Files 3 and 4 : virgin fibre (peripheral part). File 5 : inked fibre (peripheral part)

Fig 2-b : Diffuse reflectometry with KBr on blank ground paper (DRIFT). Measurements are converted to Kubelka Munk units. Represented spectra correspond to a dilution of 3.5%. The k/s factor is linearly dependent on the paper dilution.

Fig 2 : reproducibility of the different FTIR techniques


We then tried to estimate if, whether or not, this technique was reproducible enough to follow the degradation of cellulose induced by iron gall ink. Spectra were also recorded on the laboratory probes prepared with the incomplete ink and artificially aged. Several spectra were collected on each sample, and among all of these, we selected on figure 3 the spectra corresponding to the best definition of the cellulose bands. The figures 2-a and 3 underline the following behaviours :
• no clear distinction can be formulated between the spectra of virgin and inked fibres. The influence of ink on the FTIR spectrum is also too small to be measured by this technique. The concentrations of sulphate and gallic acid appear to be too low to modify the cellulose spectrum.
• no real information can be extracted in the cellulose strong absorption region. Cellulose remains the main component even in strongly damaged samples when the paper has turn dark brown and lost its mechanical properties. In this region, the measurements are not reproducible enough to underline clearly a tendency.
• some information can be extracted in the cellulose low absorption region : cellulose oxidation appears clearly in the range 1750 to 1600 cm^-1 and small absorption bands seem to grow up at 780 cm^-1 and 1320 cm^-1.

Fig 3: FTIR transmission through a fibre. Laboratory probes prepared with paper immersed in ink and artificial aged

To confirm these tendencies, we made similar measurements by diffuse reflectometry which is a more reproducible FTIR technique.

II-3 : DIFFUSE REFLECTANCE SCPECTROSCOPY (DRIFT) WITH KBR .
This technique can not be considered as a non destructive technique since one measurement requires a few milligrams of paper. Its use is also mainly limited to laboratory probes in order to crosscheck the assumptions formulated through the previous techniques.
Diffuse reflectance spectroscopy with KBr was preferred at first to transmission spectrometry through a KBr pellet because our preliminary intention was to interpret diffuse reflectance spectra measured directly on original samples. We came further to the conclusion that the diffuse reflectance spectra were much more comparable to transmission spectra so far a Kubelka Munck treatment was undertaken.
In the Kubelka Munk theory, the light is supposed to be scattered isotropically and the regular reflection negligible. These assumptions appear reasonable for KBr containing a small quantity of paper. In theses conditions, the reflectance R (that is the reflected to the incident intensities ratio) can be converted to a factor k/s:
.....k/s = (1-R)²/2R
k being mainly proportional to the absorption coefficient of the sample and s being mainly proportional to the scattering coefficient of the sample.
The sample consists of a little amount of paper dispersed throughout a KBr medium. The s factor should also be very closeto 1 and the k factor very close to k=x.k_p , x being the paper concentration of the sample and k_p the absorption coefficient of the paper.
We first checked the reproducibility of measurements : virgin paper sheets were cut into small pieces, ground for 90 minutes in a zirconium oxide self-grinding unit (Retsch MM200) at low frequency (10Hz ) and then sifted to obtain particle diameters below 100 microns. A few milligrams of the resulting powder was then mixed with a large quantity of KBr in an agate bowl. Several paper concentration were also prepared from 1% to 5% and 10 spectra were recorded for each concentrations. Spectra measured this way proved to be much more reproducible than with the previous techniques (Figure 2-b), We noticed that with this technique, the cellulose absorption bands were not very nicely defined. Different sample's preparations were tested but none gave better results.
Then the proportionality between k and x was checked by plotting the average k/s factor versus x for different frequencies. Figure 2-b shows for example two plots respectively at 900cm^-1 and 1430 cm^-1. Despite the frequencies being very far from each other, the curves are quite linear for concentrations below 4% and the slopes are quite similar. We also decided to prepare samples using the concentration of 2,5%.

We then looked if we could confirm by this technique the previous observation we made by transmission microscopy. Figure 4 plots the evolution of DRIFT spectra with artificial ageing. It appears clearly that the definition of the cellulose absorption bands is rather poor for the non aged and the 7 days aged paper. Paradoxically, the cellulose absorption bands are well defined for high damaged paper, over 14 days of artificial ageing. We think that this evolution is due to the preparation process : after 7 days of artificial ageing the paper has turn brown and lost it's mechanical properties. It can be ground in an agate mortar by hand, and this is the way we did. The fibre is then not crushed, but broken, and the resulting spectra is also very similar to the transmission spectra measured in the peripheral areas of a fibre (figure 2).

Fig 4: Diffuse reflectometry with KBr (DRIFT). Laboratory probes prepared with paper immersed in ink and artificial aged.

The comparison between the high damaged samples and the pure cellulose suggests the following comments :

• the increase of absorption at 1320 cm^-1 which was suggested by the transmission spectra (figure 3) is not observed anymore. It is also supposed to be an artefact due to the fibre heterogeneity.
• the increase of absorption in the range 1750 to 1600 cm^-1 and at 780 cm^-1 is confirmed. Since the presence of ink doesn't create any deformation of the spectrum before the ageing process, we came to the conclusion that these absorption bands are relevant to the degradation of cellulose induced by the artificial ageing.

The region 1750-1600 cm^-1 reveals the superposition of different bands, suggesting the apparition of several carbonyl and/or alkene groups (17). The apparition of a small band at 780 cm^-1 is attributed to the C-H out of plane vibration in conjugated molecules. Still, further interpretation of theses absorption bands remains difficult. The oxidation mechanism proposed by Neevel can account for the growth of absorption around 1700 cm^-1 but not for the growth of absorption at 780cm^-1 (C-H out of plane vibration).

III- ANALYSIS OF AN ORIGINAL SAMPLE :

A strong corroded manuscripts was examined by DRIFT and by transmission microscopy. The presence of iron gall ink was checked by Electron Scanning Microscopy. Figure 5 shows the visual aspect of the sample and the different spectra measured with the two techniques. Significant differences between inked and non inked areas are observed: absorption bands appear respectively between 1750-1600 cm^-1, at 1400 cm^-1 and between 830-780cm^-1.

Fig 5 : Original manuscript 5-a : FTIR measurements on an original sample. Spectra (1) and (2) are recorded by transmission microscopy through a fibre. Spectra (3) and (4) are recorded by diffuse reflectometry . 5-b : visual aspect of the manuscript, both sides.


We first noticed that the absorption band at 1400 cm^-1 was not observed with artificially aged laboratory probes. This band is also attributed to the ink binding substance or to the effect this substance may have on the paper.
Several groups may correspond to the frequency1400 cm^-1:
• salts of cellulose oxidation products with inorganic cations ( R-COOH « R-COOMetal). These salts absorb in the regions 1350-1450 cm^-1 and 1550-1610 cm^-1 [11]. Carboxylic salts may also be present in gums.
• SO₂ asymmetric stretching of sulphates ((RO)₂SO₂) [17]. But no equivalent band was found in laboratory probes.
• Some azo compounds (ammonium salts, amides, ...). Proteinic binders may then account for the presence of nitrogen

One can further notice the regions 1750-1600 cm^-1 and 830-780 cm^-1 correspond to the superposition of several bands. Deconvolution of the spectrum in the region 830-780 cm^-1 leads to three bands situated respectively at 795 806 and at 819 cm^-1. The 780 cm^-1 band which appears with the artificially aged laboratory probes is also not observed.
Deconvolution of the spectrum in the region 1750-1600 cm^-1 leads to four bands situated respectively at 1710, 1680, 1662, and 1608 cm^-1.
At this step of the study, the FTIR measurements on original samples can not be discussed further because too few information is available on this sample, in particular, what is the ink binder made with? Does it contain proteins? lipids? sugars?
two directions of investigation should also be proposed: first, other techniques, such as gas chromatography, should be performed on the sample to drive some more information on the binder. Secondly, other laboratory probes including different binding components should be prepared, for comparison with the original sample.

CONCLUSION :

Our first purpose was to investigate the possible application of FTIR techniques for the analysis of original manuscripts. The main difficulty of this kind of analysis consists in the great variety of possible ingredients and the influence these ingredients may have on paper degradation. The work presented above on laboratory probes indicates that the paper profile is very close to the pure cellulose profile even if the paper is highly damaged. Diffuse reflectance microscopy performed directly on the sample is of limited use, because of regular reflection. DRIFT spectroscopy with KBr offers a better reproducibility than transmission microscopy and is also preferentially used for laboratory probes. On the contrary, transmission microscopy is preferred for original samples, because of it's non destructive aspects. In this work, we showed that the two techniques give similar results on an original manuscript, which presents great discrepancies between inked and non inked areas.
Still the interpretation of the measured effects remains difficult and should be related to the comparison of laboratory probes and original manuscripts. We showed that the presence of iron gall complex can not be detected by FTIR The great differences observed in the spectra should also be attributed to the ink binding substance, and/or to the degradation state of cellulose in the paper.

BIBLIOGRAPHY :

1. Neevel J.G., « Phytate : a potential conservation agent for the treatment of ink corrosion caused by irongall inks », Restaurator, 16, (1995), 143-160.
2. Neevel J.G Mensch C. T.J., « the behaviour of iron and sulfuric acid during iron gall ink corrosion », ICOM CC, 12th triennal meeting, Lyon, 1999, vol II, 528-533.
3. Reissland B., « Ink Corrosion Aqueous and Non Aqueous Treatment of Paper Objects - State of the Art », Restaurator 20, (1999), 167-180.
4. Reissland B., De Groot S., « Ink corrosion : comparison of currently used aqueous treatments for paper objects », IADA, Copenhagen, 1999, 121-129.
5. Zerdoun M. « Les encres noires au Moyen Age (jusqu'à 1600) » Editions du C.N.R.S. (1983).
6. Wunderlich C.H., « Geschichte und Chemie des Eisengallustinte », Restauro, 6, (1994) , 414-421.
7. Darbour M., Bonnassies S., Flieder F. « Les encres métallogalliques : étude de la dégradation de l'acide gallique et analyse du complexe ferrogallique » ICOM CC, 6th Triennal Meeting, Ottawa (1981) 3-14.
8. Bleton J., Coupry C., Sansoulet J. « Approche d'étude des encres anciennes » Studies in Conservation 41 (1996) 95-108.
9. Sistach M.C., Espadaler I. « Organic and Inorganic Components of Iron Gall Inks », ICOM CC, 10th Tiennal Meeting, 1993, Vol II, 485-489
10. Duval A. « PIXE Analysis of Pisanello's drawings » 5th International Conference on non-destructive Testing Microanalytical Methods and Environmental Evaluation for Study and Conservation of Works of Art (Budapest, 24-28 Sept 96)
11. Zhbankov R.G, « Infrared Spectra of Cellulose », Consultant Bureau, New York, 1966
12. Havermans J.B.G.A. « environmental influences on the deterioration of paper », barjesteh, Meeuwes and Co, Rotterdam, 1995, 43-76.
13. Burandt J., « An Investigation toward the identification of traditional drawing inks », The Book and Paper Group, (1994), 9-16.
14. Mosini V., Calvini P., Mattogno G. Righini G. « Derivative infrared spectroscopy and electron spectroscopy for chemical analysis of ancient paper documents », Cellulose Chemistry and Technology, 24, (1990), 263-272.
15. Sistach C., Ferre N., Romera M.T., « Fourier Transform Infrared Spectroscopy applied to the Analysis of ancient Manuscripts », Restaurator n°19, 173-186, 1998.
16. Pandey K., « A Study of Chemical Structure of Soft and Hardwood and Wood Polymer by FTIR Spectroscopy », Journal of Applied Polymer Science, 71, (1999), 1969-1975.
17. Socrates G., « Infrared Chracteristic Group Frequencies », Wiley and sons New York, (1994).

© AIPnD , created by NDT.net

|Home|    |Top|