Nuclear Instruments and Methods in Physics Research B 213 (2004) 723–728

www.elsevier.com/locate/nimb

Determination of localized Fe2þ/Fe3þ ratios in inks ofhistoric documents by means of l-XANES

K. Proost a, K. Janssens a,*, B. Wagner b, E. Bulska b, M. Schreiner c

a Department of Chemistry, University of Antwerp, Universiteitsplein 1, B-2610 Antwerp, Belgiumb Department of Chemistry, University of Warsaw, Pasteura 1, PL 02-093 Warsaw, Poland

c Institute for Sciences and Technologies in Art, Academy of Fine Arts, Schillerplatz 3, A-1040 Vienna, Austria

Abstract

An important part of the European cultural heritage is composed of hand-written documents. Many of these

documents were drawn up with iron–gall ink. This type of ink present a serious conservation problem, as it slowly

oxidizes (�burns�) the paper it is written on, thereby gradually disintegrating the historic document. Acid hydrolysis of

the cellulose and/or the oxidation of organic compounds promoted by radical intermediates that are formed due to the

presence of Fe2þ ions are considered to be the cause of the disintegration. l-XANES measurements were performed

with a lateral resolution of 30–50 lm in order to determine the local Fe2þ/Fe3þ ratio in 19th C. documents from the

Austrian National Archives and fragments of 16th C documents from the Polish National Library. In the 19th C

documents, no significant amount of Fe2þ was detected. On the other hand, in the 16th C fragments, significant

amounts of Fe2þ and appreciable differences in distribution of Fe2þ and Fe3þ within individual letters/ink stains were

observed.

� 2003 Elsevier B.V. All rights reserved.

Keywords: Ferro–gallic ink; Historic documents; Paper degradation; X-ray absorption near edge spectroscopy

1. Introduction

An important part of the European cultural

heritage is composed of hand-written documents.

In many of these documents, a particular type of

ink, called iron–gall ink was used. This ink is pre-

pared on the basis of a mixture of vitriol (FeSO4),gallic acids (extracted from the spherical protru-

sions formed in leaves, caused by local damage

inflicted by gall-wasps) and a viscous medium such

as arabic gum. The darkbrown/black color of the

* Corresponding author. Fax: +32-3-820-2376.

E-mail address: koen.janssens@ua.ac.be (K. Janssens).

0168-583X/$ - see front matter � 2003 Elsevier B.V. All rights reser

doi:10.1016/S0168-583X(03)01693-8

ink is caused by the oxidation of Fe2þ to Fe3þ and

the formation of Fe3þ–gallic acid complexes.

Historical documents prepared with iron–gall

ink present a serious conservation problem, as the

ink slowly oxidizes (�burns�) the paper it is written

on, thereby gradually disintegrating the document.

Two mechanisms were put forward to explain thisbehavior [1,2]:

(a) either the strongly acidic components of the

ink (sulphuric acid) hydrolyze the cellulose in

the paper and cause it to disintegrate,

(b) Fe2þ residues (left behind after incomplete

airoxidation of the original FeSO4) act

as catalysts for the oxidation of the organic

ved.

724 K. Proost et al. / Nucl. Instr. and Meth. in Phys. Res. B 213 (2004) 723–728

compounds present. This oxidation involves

the intermediate formation of radicals (O�

2,

HOO�). Reaction of Fe2þ with HOO� in acidicmedium leads to the formation of H2O2, which

in its turn can further oxidize the paper (and

the Fe2þ present in the ink):

1. Formation of organic radicals

Fe2þ þ O2 $ Fe3þ þ O�

2

R� þ O2 $ ROO�

Fe3þ þ O�

2 þ RH $ R� þ HOO� þ Fe2þ

ROO� þ R0H $ RCOOH þ R0�

2. Formation of hydrogen peroxide (Fenton

reaction)

Fe2þ þ HOO� þ Hþ $ Fe3þ þ H2O2

Fe2þ þ H2O2 $ Fe3þ þ HO� þ OH�

Whereas in reactions (1), Fe2þ truly acts as a cat-alyst (i.e. is not consumed during the reaction),

during reactions (2), each HOO� radical may lead

to the oxidation of two Fe2þ ions. Local mea-

surement of the Fe2þ/Fe3þ ratio in individual

hand-written letters of historical documents would

therefore allow to objectively monitor the local

�damage potential� of the ink in a document. If all

the Fe2þ has reacted away, it is no longer possiblefor radicals to formed by mechanism (1), de-

creasing the future damage potential the ink rep-

resents for the paper. If on the other hand, a

substantial part of the Fe is still present in the 2+

state, the ink is able to inflict further damage to the

cellulose substrate it is written on.

In order to be able to extract quantitative in-

formation on the Fe2þ/Fe3þ ratio from specificareas of individual written characters, lines or other

inked areas on paper, the method of l-XANES

(X-ray absorption near edge spectroscopy) was

employed. A microscopic beam of monochromatic

X-rays is used to irradiate the material at a specific

location and the intensity of the resulting Fe-Ka

fluorescent radiation is recorded.

By varying the energy of the primary beam insmall (typically 1 eV) steps across the K-absorption

edge of Fe, the shape and position of the absorption

profile is obtained. Since these absorption profiles

are subject to a chemical shift, depending on the

oxidation state of the Fe-atoms that are irradiated,

it is possible to extract average oxidation state in-

formation of Fe in the irradiated spot. By recording

XANES profiles at different locations, the spatialvariation of the Fe2þ/Fe3þ ratio within, e.g. an in-

dividual hand-written character can be visualized.

Alternatively, this type of �chemical state� maps can

also be obtained by collecting a series of l-XRF Fe-

Ka intensity distributions at different energies

around the Fe-K absorption edge.

In what follows, we report the results of exa-

mining a number of individual iron–gall writtencharacters on historic documents of the 16th and

19th C and demonstrate the feasibility of the

chemical state mapping.

2. Experimental

The XANES experiments were carried out atbeamline L of HASYLAB (Hamburg, Germany),

where a 1.2 T bending magnet provides poly-

chromatic synchrotron radiation. For l-XANES

measurements, a narrow energy-band is selected

from the continuum by means of a hybrid fixed-

exit Sih1 1 1i monochromator [3]. After passing

cross-slits, the monochromatic beam is demagni-

fied by a polycapillary lens, mounted on a XYhugimbal holder (see Fig. 1). The lens is an assembly

of hollow glass tubes that all transport the X-ray

towards a common focal point [4]. Since the lens is

achromatic, the position of the focus does not

change when the energy of the primary beam is

changed. At the Fe-Ka edge energy keV (around

7.1 keV), the lens produces a focused beam of �30

lm diameter [4]. Alternatively, the polycapillarylens can be removed from the beam and a broad,

unfocussed beam (of �100 lm to 1 mm in diam-

eter) can be used for the irradiation.

In order to be able to extract quantitative in-

formation on the Fe2þ/Fe3þ ratio from the fluo-

rescent mode Fe-K XANES profiles, reference

profiles were recorded of FeSO4 � 7H2O and

Fe2(SO4)3 � 5H2O. For this purpose, the finelyground sulphate powders were diluted to 1% with

boron-nitride powder and pressed into �0.5 mm

thick pellets. Per XANES spectrum, 250 data

points with an energy spacing of 1 eV were re-

corded, using a 1 s collection interval per point.

Fig. 1. Schematic drawing of the experimental set-up at beam line L (HASYLAB, Hamburg, Germany) [3,4].

Fig. 2. Typical XANES profile of an historic iron–gall ink

sample, described as a linear combination (curve marked �re-

gression�) of the XANES profiles of FeSO4 � 7H2O and

Fe2(SO4)3 � 5H2O.

K. Proost et al. / Nucl. Instr. and Meth. in Phys. Res. B 213 (2004) 723–728 725

3. Results and discussion

3.1. Reference profiles and profile analysis

The K-XANES profiles of various Fe-minerals

and compounds have been extensively documented

[5,6]. In all profiles obtained from the iron–gall

samples, XANES profiles that strongly resemble

the profiles of either FeSO4, Fe2(SO4)3 or linear

combinations thereof were obtained, consistentwith the assumption that both the free and or-

ganically bound Fe-ions (either in the 2+ or the 3+

state) have oxygen atoms as closest neighbors.

Accordingly, as illustrated in Fig. 2, XANES

profiles obtained from specific locations on the

historic material could be converted into a Fe2þ/

Fe3þ ratio by regressing the (background corrected

and normalized) XANES profile against the (sim-ilarly pretreated) FeSO4 and Fe2(SO4)3 profiles.

This procedure allows to one determine Fe2þ/

Fe3þ ratios situated in the range 10%/90% to 90%/

10% with an uncertainty of �5%, the latter value

deriving mainly from the uncertainty on the re-

gression coefficients.

3.2. Examination of 19th C hand-written documents

A series of 19th C documents from the Austrian

national archives, including bills-of-goods, per-

sonal letters, ink-colored photograph borders andadministrative papers, were examined. By making

use of a card board holder, all documents could be

presented to the X-ray beam without sampling orother damage (see Fig. 3). Since a primary X-ray

beam was employed that was considerably smaller

than the written characters, measurements could

be performed on the edge and in the center of

inked areas; also locations with clearly visible, less

visible and no visible damage (see Fig. 4) were

analyzed. In all of the examined 19th C letters,

irrespective of the location inside the writing,XANES profiles strongly resembling that of

Fe2(SO4)3 were obtained, indicating that more

than 90% (if not all) of the Fe in these documents

is present as Fe3þ-compounds.

Fig. 3. Photograph taken during the l-XANES measurement

of a 19th C bill-of-goods (Vienna, Austria, 1851).

Fig. 4. (a) 19th C bill-of-goods (Vienna, Austria, 1851). Details

of the front side (b) and the back side (c) of the document show

locations where a large amount of ink was applied, resulting in

the complete disintegration of the paper at those positions.

Fig. 5. Optical photograph of fragments of a 16th C prayer

book. The circles indicate the areas where l-XANES mea-

surements were performed.

726 K. Proost et al. / Nucl. Instr. and Meth. in Phys. Res. B 213 (2004) 723–728

3.3. Examination of fragments of a 16th C prayer

book

Next to the 19th C documents, also fragments

of a 16th C prayer book �Meditationes, passionis

Domini nostri Iesu Christi� from the Polish Na-

tional Library were investigated. The ink-induced

damage inflicted to this artifact was so extensive

that a number of loose fragments were available

for analysis (Fig. 5).

Here, quite different Fe2þ/Fe3þ ratios could be

observed, as evidenced by the shift of the white line

of the XANES spectra collected at different loca-

tions (Fig. 6). The spectra were collected using aunfocussed X-ray beam of �100 lm diameter in

order to avoid damage to the material. The re-

sulting Fe2þ/Fetot and Fe3þ/Fetot ratios are listed in

Table 1; it is clear that substantial amounts of Fe2þ

are present, suggesting that reactions (1) and (2)

are still ongoing in the paper, possibly leading to

more damage. In fragments of other 16th C doc-

uments, relative Fe2þ abundances in the range 10–20% were observed; these numbers are consistent

with (average) results obtained by means of

M€oossbauer spectrometry on the same samples.

3.4. Chemical state mapping

In a rectangular area around position 4 (Fig. 5),

the Fe-Ka intensity distribution was recorded attwo closely spaced excitation energies. At a pri-

mary energy of 7.115 keV (see Fig. 7(a)), pre-

dominantly (but not exclusively) Fe2þ is excited

since the cross-section for photo-ionization of

iron in the 3+ state is still quite low; this energy

Fig. 6. XANES profiles derived from positions 1 to 6 of a

fragment of a 16th C prayer book shown in Fig. 5. The vertical

line indicates the white line position for Fe3þ.

Table 1

Relative percentages of Fe2þ and Fe3þ derived from the

XANES spectra shown in Fig. 6

Position Fe2þ (%) Fe3þ (%)

1 46 53

2 50 50

3 55 45

4 30 70

5 52 48

6 55 45

Fig. 7. l-XRF Fe-Ka intensity images (a,b) obtained from an

inked area (around positions 4 in Fig. 5) and corresponding

qualitative Fe2þ (c) and Fe3þ (d) distributions.

K. Proost et al. / Nucl. Instr. and Meth. in Phys. Res. B 213 (2004) 723–728 727

corresponds to the origin of the relative energy

scale of Fig. 6. The distribution of Fig. 7(b) was

collected at 7.125 keV; at this energy, the cross-

section for photo-ionization of Fe3þ is maximum

(see vertical line in Fig. 6) while also Fe2þ isefficiently excited. By using the mathematical

technique proposed by Iida [7], the Fe2þ and Fe3þ-

specific distributions of Fig. 7(c) and (d) can be

calculated. Clearly, the two spatial distributions

are quite different from each other, with more Fe2þ

present in the inner parts of the inked area. This

may be indicative of an Fe2þ-oxidation front that,

starting from the boundary of the ink-stain,gradually moves inwards.

4. Conclusions

It can be concluded that l-XANES is a tech-

nique with which it is possible to study the distri-

bution of Fe2þ and Fe3þ in historic documents

written with ferro–gallic inks and that within sin-

gle ink-stains, dissimilar distributions of Fe2þ and

Fe3þ can be discerned. In follow-up work, we in-

tend to use this method to monitor the effect ofdifferent conservation treatments on the Fe2þ/Fe3þ

balance in this type of artifacts [8].

Acknowledgements

The authors acknowledge support from the EU

provided within contract HPRI-CT-1999-00040/2001-00140 and within contract G6RD-CT-2001-

00602.

References

[1] G. Banik, F. Mairinger, H. Stachelberger, Restaurator 1–2

(1981) 71.

[2] J.G. Neevel, B. Reissland, in: H. van der Windt (Ed.),

Proceedings of a Workshop on Iron Gall Ink Corrosion,

Amsterdam, 1997, p. 37.

728 K. Proost et al. / Nucl. Instr. and Meth. in Phys. Res. B 213 (2004) 723–728

[3] G. Falkenberg, O. Clauss, A. Swiderski, Th. Tschentscher,

X-ray Spectrom. 30 (2001) 170.

[4] K. Proost, L. Vincze, K. Janssens, N. Gao, E. Bulska, M.

Schreiner, G. Falkenberg, X-ray Spectrom. 32 (2003) 215.

[5] S. Bajt, S.R. Sutton, J.S. Delaney, Geochim. Cosmochim.

Acta 58 (1994) 5209.

[6] M. Wilke, F. Farges, P.E. Petit, G.E. Brown, F. Martin,

Am. Mineral. 86 (2001) 714.

[7] A. Iida, in: K. Janssens et al. (Eds.), Microscopic X-ray

Fluorescence Analysis, Wiley, 2000 (Chapter 5).

[8] B. Wagner, E. Bulska, T. Meisel, W. Wegscheider, J. Anal.

Atom. Spectrom. 16 (2001) 417.