Study of the Degradation and Conservation of Historical Leather Book Covers with Macro Attenuated Total Reflection-Fourier Transform Infrared Spectroscopic Imaging.

The analytical study of historical collagen-based materials such as leather book covers is a complex task for conservation scientists. Historical leather presents a heterogeneous composition of both organic and inorganic materials that show an evident reactivity, particularly when exposed to specific environments. Its correct preservation in archival documents remains challenging as some aspects of its chemical composition, degradation, and effectiveness of conservation treatments are still not fully understood. Here, we describe a novel application of attenuated total reflection (ATR)-Fourier transform infrared spectroscopic imaging coupled with a macro ATR accessory to study historical leather book covers. This nondestructive and high spatial resolution approach has allowed the visualization of degradation phenomena affecting this fragile material, particularly the gelatinization of collagen and, for the first time, the detection of the formation of calcium stearate (metal carboxylates or soaps). In addition, the distribution of modified soybean oil used as a treatment to maintain properties such as elasticity and hydrophobicity of the leather was studied. The effect of anomalous dispersion on the strong IR bands obtained in the ATR mode and the resulting changes to the band positions are also discussed. This research addresses issues that are relevant to the conservation of archival materials of cultural heritage for future generations.

Introduction

The correct conservation of historical archival materials and documents is vital not only for their artistic value, but also because they represent physical records of human history. Leather, a collagen-based material combined with various organic and inorganic compounds, is frequently found as part of these documents. Since antiquity, leather has been a preferred material for bookbinding.[1] Generally made from mammal skins, the manufacture took several months to years, significantly differing depending on geographical and historical factors and the quality of the final product required. The manufacturing process has been extensively described elsewhere.[2] In brief, the animal skin was first soaked in an alkali bath with added lime and then scraped with a knife to remove the epidermis layer, hair, and fat to reach the corium layer of the skin. The carbonatation of calcium hydroxide residues from the liming bath with the CO2 present in the atmosphere causes the precipitation of calcium carbonate commonly found in leather. Successively, the skins were tanned. Vegetable tannins derived from plants were first used before metal tanning agents were introduced in the second half of the 19th century. Tannins prevent the decay of the skin by cross-linking the collagen fibers transforming the skin into durable leather.[3] The tanning also gives the leather its color from light brown-red to darker shades. When considered ready, after been washed and dried, the leather surface was treated with additional materials. Compounds, such as oils and waxes, were applied to give the leather specific physical/mechanical properties (i.e., flexibility and impermeability) and mixed with colorants and pigments for decorative purposes.[4]

Leather artifacts are prone to alteration and degradation because of their heterogeneous composition and sensitivity to environmental factors (humidity, temperature, and pollutants).[5−9] The decay of leather can ultimately end in the disintegration of the material and damage the documents. Different approaches are currently applied in the conservation of leather book covers. These approaches can comprise a regulation of the environment where the archival material is stored to create the most suitable conditions for a long-term conservation. In other cases, the level of decay of the leather requires a “physical” intervention, such as the use of oil emulsions to restore mechanical properties lost. The analytical characterization of the leather composition, alteration, and conservation state is a vital step to understand the best procedure to apply.

IR spectroscopy techniques are very suitable for investigating collagen-based historical materials, and they have been successfully used for studying leather wall covers in historical buildings,[10] for identifying tannins in leather,[11] and characterizing the composition and degradation of parchment, also made from animal skins.[12−14] Recently, Fourier transform infrared (FTIR) spectroscopy has been coupled with a focal plane array (FPA) detector, allowing thousands of FTIR spectra to be simultaneously collected in a short time. By plotting the integrated absorbance of the spectral bands as a function of all pixels, spectroscopic images representing the distribution in a micrometer scale of the components detected are obtained.[15] Compared to transmission, the attenuated total reflection (ATR) mode requires little sample preparation as the effective pathlength of the IR light in the sample is not limited by its thickness, potentially allowing the sample to be analyzed in a nondestructive way. ATR–FTIR spectroscopy is a flexible approach to collect spectra in the mid-IR region. Portable instruments for ATR–FTIR spectroscopy are available and used to analyze artifacts in situ. Sufficient contact between the sample/object studied and the ATR crystal is necessary. Therefore, it is important to proceed with an appropriate care in this step in order not to damage the sample/object. This can be achieved by considering the shape and dimensions of the ATR crystals and the physical characteristics of the material to be analyzed. Depending on the accessory used, macro ATR–FTIR spectroscopic imaging can be applied to large objects (if they can be moved to the lab) without the need of complex sample preparation. We have previously reported that micro and macro ATR–FTIR spectroscopic imaging is a powerful analytical technique for conservation science.[16−18] However, there are no studies reporting its application to leather book covers.

In this work, an FTIR spectroscopic imaging apparatus coupled with a single reflection diamond ATR accessory for the macro mode is used to study samples from original book covers dated between the 16th and the 19th century from the collection of the Mesrop Mashtots Institute of Ancient Manuscripts, the Matenadaran, Yerevan, Armenia. The Matenadaran is one of the oldest and most precious depositories of ancient manuscripts in the world. The scope of this work is to help gain a deeper knowledge regarding the composition of historical leather in archival material by examining the distribution of the IR bands detected and linking it to the original materials, possible degradation products, and an oil emulsion used to treat the book covers. Different materials derived from manufacturing or used in later interventions can interact between them to form potentially dangerous products for the preservation of leather book covers. A striking example is the formation of calcium carboxylates (metal soaps) that has not been deeply investigated yet. Additionally, the protein collagen constituting the skin fibers is susceptible to oxidation, partial hydrolysis, biological attack, and acidification (red-rot), and it can undergo denaturation as early as during the manufacturing process.[19] When denaturation occurs, the native polypeptide chain, organized in a left-handed triple α helix, unfolds into a random coil structure. This leads to the formation of gelatin, a water-soluble degradation product found in collagen archival materials. The preservation state of collagen in leather and parchment and the presence of gelatin can be evaluated by investigating the amide I and II spectral bands. These two bands, found in the 1800–1500 cm–1 spectral region, are complex IR bands, and their shape and position are particularly informative as they correlate to the secondary structure of collagen and its interaction with the environment.[20−22]

Finally, when interpreting FTIR spectra measured in the ATR mode, to avoid misinterpretations, it is essential to consider the presence of anomalous dispersion, a distortion that can affect particularly strong spectral bands of materials with a high refractive index (n). Particularly with a diamond ATR crystal (n ≈ 2.4), significant red shifts with respect to the band positions in spectra collected in transmission mode can occur.[23] These band shifts are usually stronger when a diamond crystal instead of a crystal with a higher refractive index (i.e. germanium, n ≈ 4) is used as the difference between the refractive indices of the diamond and the sample is smaller. Hence, materials relevant to this work were measured with the same setup to evaluate the effect of using a diamond ATR crystal on their band positions and to correctly assign the bands found in the spectra of the leather samples analyzed.

Experimental Section

Macro ATR–FTIR Spectroscopic Imaging

An imaging Golden Gate ATR accessory (Specac, UK) hosted in a IMAC macrochamber attached to a Tensor FTIR spectrometer (Bruker Optics) and equipped with an FPA detector (Santa Barbara Focalplane, USA) was used for macro-ATR–FTIR spectroscopic imaging. The ATR accessory has specifically designed correcting lenses and a prism-shaped diamond ATR crystal with a contact surface of 2 × 2 mm2 (n ≈ 2.4, angle of incidence ≈ 44°). The samples were simply put into contact with the ATR crystal by gently pressing with a stainless steel flat anvil of the ATR accessory. An array of 64 × 64 pixels of the FPA detector was selected and 4096 FTIR spectra simultaneously collected. This setup provides an image size of ∼600 × 550 μm with a spatial resolution of 15–20 μm. A total of 128 or 256 scans were co-added with 8 cm–1 spectral resolution in the range of 3800–850 cm–1. The data were treated with the commercial software OPUS (Bruker). Spectroscopic images are obtained by plotting the distribution of the integrated absorbance of selected spectral bands. A false color scale represents how strong the absorbance is, where magenta stands for highest and blue for lowest absorbance. The spectra presented throughout this paper are extracted from a single pixel in measured spectroscopic images.

The open source software ImageJ was used to assign specific colors to the components of interest, the distribution of which was plotted in the FTIR spectroscopic images. First, from the FTIR images, single-color images representing the distribution of each component were obtained. Then, selected single-color images of the relevant components were recombined into a composite image. This helped to better visualize the localization and overlapping of the compounds detected in the same area.

ATR–FTIR Spectroscopy

Conventional FTIR spectra were recorded in the ATR mode with a portable spectrometer (Alpha-P, Bruker Optics) fitted with a single-reflection diamond crystal and a single-element detector.

Materials

Pure precipitated calcium carbonate (Sigma) and a pork gelatin sheet (AB World Foods) were used and measured with the macro ATR–FTIR spectroscopic imaging setup to evaluate the effect of anomalous dispersion on the strong IR bands.

The leather samples from historical book covers dated from the 17th to the 19th century (Table ) are part of the collection of the Matenadaran Institute, Armenia. These leather book covers, from which the samples were taken, were removed from manuscripts in the past (the date of removal is unknown) as damaged or degraded. The removed covers were kept in the depository of the Matenadaran Institute. The dating of the covers was done based on hand-written dates noted on the back of the covers. These notes were probably made by an archivist of the institute in the past. The dates reported are consistent with the composition and manufacturing type of the covers, supporting the dating. The samples were taken with scissors from the covers under a microscope to check the absence of biodegradation.

Table 1

Details of the Leather Samples from Original Book Covers Provided by the Matenadaran Institute, Armenia

archival material	date
manuscript	17th century
manuscript	1750
Bible printed in Amsterdam	1666
leather cover	1698
book of Lamentation, Narek	1812
Soybean Oil-Treated Samples
manuscript	17th century
manuscript	1750
Bible printed in Amsterdam	1666

Three samples were treated with soybean oil emulsified in a concentrated H2SO4 solution (in a ratio of 10% by weight of the soybean oil) and applied by a brush. This treatment, developed at the Matenadaran, aims to remove the humidity adsorbed by the collagen fibers and provide elasticity and softness to the covers. These historical cover samples were suitable for testing the feasibility of our approach in detecting the modified soybean oil treatment on leather covers. Except for the cover of the Bible printed in Amsterdam, the rest of the covers were most probably produced in Eastern Europe and Middle East.

Results and Discussion

All untreated leather samples presented at a visual examination a flaked surface with a network of cracks on both sides, probably caused by a loss of elasticity (visible images of the samples are reported in the Supporting Information, Figure S1).

Two main aspects were considered when studying the amide I and II bands of the FTIR spectra of the leather samples: the transformation of collagen in gelatin and the effect of anomalous dispersion on these two spectral bands. In the leather samples analyzed in this work, the amide I band was detected in the region from 1645 to 1623 cm–1. However, the amide I band has previously been reported at about 1658 cm–1 for native collagen, 1651 cm–1 for denatured collagen,[20] and at 1633 cm–1 for gelatin.[21,22] The significant shift of 10–15 cm–1 for this spectral band, compared to literature values, may originate from anomalous dispersion when a diamond ATR crystal is used.[23] A dry pork gelatin sheet (AB World Foods) measured with the same setup showed the amide I band at 1626 cm–1 (Supporting Information, Figure S2), thus confirming the effect of using a diamond crystal. The amide I band in the region 1645–1638 cm–1 was then assigned to the preserved collagen; when detected between 1637 and 1630 cm–1, it was attributed to the unfolding of the helices and, when below 1630 cm–1, it was assigned to gelatin. When gelatin was evident in the spectra, a shoulder at about 1655 cm–1 appeared, possibly resulting from remaining α-helix chains. As shown in Figure for the top surface of a leather cover of a manuscript dated 1750, spectroscopic images can be obtained by plotting the absorbance of a delimitated area of the amide I band assigned to collagen (spectral range 1710–1635 cm–1) and to gelatin (spectral range 1634–1600 cm–1). Therefore, the extension of the gelatinization affecting the surface of the leather samples can be revealed. In this case, the amide II band in the gelatin spectrum appeared upshifted, possibly as a consequence of hydration.[21] This was also visible in the FTIR spectra of pork gelatin where the amide II band shifted from 1536 to 1540 cm–1 after it was briefly soaked in water (Figure S2).

Figure 1

Leather sample from a manuscript dated 1750, Matenadaran. (a) FTIR spectra of the amide I and II bands extracted from the chemical images (black arrows). Amide I and II bands appear at 1640 and 1539 cm–1, respectively (red line), for collagen and at 1625 and 1547 cm–1, respectively, for gelatin (blue line), possibly as the result of hydration. A shoulder at ∼1655 cm–1 in the gelatin spectrum is assigned to the remaining α-helix chains. The red and blue areas represent the integrated regions of the amide I band selected for plotting the (b) spectroscopic images showing the distribution of collagen and gelatin. The image size is approximately 600 μm × 550 μm.

As the amide I and II bands overlapped with the bands of other components present, such as lipids and tannins, the second derivative of selected spectra extracted was calculated, as shown for the leather sample of a Bible printed in Amsterdam, dated 1666 (Figure ). This method helped to discern the peak positions of the overlapping IR bands of lipids, collagen/gelatin, and tannins in the spectral range 1800–1200 cm–1.

Figure 2

FTIR spectrum (black line) in the 1800–1250 cm–1 region extracted from the macro ATR–FTIR image of the external surface of a leather cover sample from the Bible printed in Amsterdam, dated 1666, and its second derivative (gray line). The overlapping spectral bands were assigned to the following: collagen/gelatin amide I (1655 and 1623 cm–1) and amide II (1535 cm–1), vegetable hydrolysable tannins (1705, 1336, and 1310 cm–1), and a lipid component (1734 and 1462 cm–1). The spectral bands assigned to tannins of vegetable origin, common for both condensed and hydrolysable tannins, are visible at about 1605, 1508, and 1444 (ν C=C aromatic ring) cm–1. The distinctive bands of hydrolysable tannins are present at 1705, 1336, and ∼1310 cm–1.

In the FTIR spectra of the leather samples, the spectral bands assigned to tannins of vegetable origin were found at about 3310 (ν OH), 1605 and 1508 (ν C=C aromatic ring), and 1180 (ν COH) cm–1. These bands are common for both condensed and hydrolysable tannins. However, hydrolysable tannins show distinctive IR bands in the region 1710–1700 cm–1, arising from the stretching of the phenolic ester groups, and at about 1325 and 1310 cm–1 assigned to the δ OH and the νs C–O–C, respectively.[11,24] In the spectra presented in Figure of the leather cover from the Bible printed in Amsterdam, dated 1666, these bands are visible at 1705, 1336, and ∼1310 cm–1 along with the other bands typical of vegetable tannins.

Calcium carbonate is commonly found in historical leather resulting from the carbonatation of calcium hydroxide residues of the liming bath with the CO2 present in the atmosphere or added during manufacturing. In the leather samples analyzed, the anti-symmetric stretching (νas) mode of CO3 groups appeared in the region 1416–1405 cm–1 (Figure b), along with a second sharp band at about 875 cm–1 (δ CO3). In the literature, this band is reported at 1435–1420 cm–1 in the transmission mode.[25] The position of the νas CO3 band could have indicated that other carbonate species were present, uncommon for the leather covers analyzed. Furthermore, the position of this band appeared correlated to the absorbance: as the absorbance increased, the band shift appeared stronger, a consequence of the anomalous dispersion effect. ATR–FTIR spectra of pure CaCO3 powder (Sigma) collected with the same setup confirmed that the νas CO3 band shifted from 1413 to 1396 cm–1 as the absorbance increased (Supporting Information, Figure S3).

Figure 3

Macro ATR–FTIR images and spectra of the external surface of two leather cover samples (Matenadaran) reveal the correlated distribution of CaCO3, gypsum, and Ca stearate (soap). (a) Leather cover, 17th century manuscript: visible microscopy image at the top of the area analyzed (scale bar is 500 μm), chemical images of gypsum (1110 cm–1) and Ca stearate (1575 cm–1), and FTIR spectra of gypsum (red line) and Ca stearate (gold line). (b) Leather cover, manuscript dated 1750: visible microscopy image at the top of the area analyzed (scale bar is 500 μm), chemical images of CaCO3 (1415 cm–1), Ca stearate and FTIR spectra of CaCO3 (red line) and Ca stearate (gold line). The composite images, processed with ImageJ, show the correlation in the presence of soaps (assigned color yellow) and the original calcium-based materials (assigned color red).

A strong band (Figure a) at about 1110 cm–1 along with a sharp band at 1620 and two bands at 3533 and 3394 cm–1 indicated the presence of gypsum bihydrated (CaSO4·2H2O). Powdered gypsum was sometimes used in animal skin processing to remove the excess fat.[26] The presence of calcium sulfate bihydrated in leather has been linked also to the reaction of sulfate ions with residual calcium milk used in the unhairing process of the animal skin.[27] Here, gypsum was detected in the 17th century leather cover sample only, along with calcium carbonate, but its concentration was far less strong than gypsum. This suggested that gypsum could be an original material of the cover added during its manufacturing and not a degradation product.

Spectral bands of lipids (Figures and 3), likely to originate from a fat substance used to treat the leather surface or fat residue of the animal skin, were observed at ∼2920 and 2850 cm–1 corresponding to the νas and νs CH2 groups and at ∼1735, 1468–60, 1240, and 1175 cm–1 assigned to the νas C=O bonds, the bending (δ) of CH2 groups, the νas and νs of C–O–C bonds.

Two sharp bands at 1575 and 1540 cm–1 along with two more bands at 1465 and 1420 cm–1 were detected, a typical spectral pattern of metal soaps (metal carboxylates). A distinctive match with the spectrum of calcium stearate from the IRUG database strongly supports this assignment.[28] The doublet at 1575 and 1540 cm–1 was assigned to the νas COO– and the band at 1420 cm–1 to the νs mode.[29] The bands at 1465, 2920, and 2850 cm–1 were assigned to the δ, the νas, and νs modes of CH2 groups of the typical soap aliphatic chain, respectively. Interestingly, the images obtained by spatially plotting the absorbance of Ca stearate, Ca carbonate, and gypsum revealed a correlation in their distribution (Figure ).

In oil paintings, metal soaps can spontaneously form because of the interaction of the fatty acid fractions of drying oils used as binders with the cations in the inorganic pigments and are considered responsible for the formation of protrusions and deformations in the paint layers.[17,30−32] Similarly, the Ca2+ of the carbonate and sulfate in the leather samples analyzed may have reacted with the fatty acids of an oil used to treat the cover surface via a saponification reaction to give Ca soaps. On closer examination, it appeared that the carbonate and sulfate absorbance decreased as the stearate absorbance increased (Figure ), which is the evidence of saponification of CaCO3 and gypsum.

Figure 4

Macro ATR–FTIR spectroscopic imaging of the top external surface of two leather cover samples: (a) 17th century manuscript and (b) the manuscript dated 1750. The images of the integrated absorbance of CaCO3 (1415 cm–1), gypsum (1110 cm−1 and Ca stearate (1575 cm–1) and the FTIR spectra extracted from pixel 1 to 5 (p1 to p5) clearly show that absorbance of the Ca stearate increases as absorbance of the carbonate and sulfate decreases. Spectra are vector-normalized.

Metal soaps have been detected previously in historical leather objects in areas where the leather was in contact with metal elements.[33] However, the leather covers from which the samples were taken did not have metal features, and only calcium soaps were detected. It has also been suggested that metal soaps could have been the result of a mixture of fat and wood ash applied to the leather surface during manufacturing.[34] In the case of the leather book covers studied in this work, it is known that the covers have been not treated with this type of lubricant substances in the past. To the best of our knowledge, only one publication has demonstrated the presence of calcium soaps in leather covers, suggesting they could originate from the reaction of metal oxides with fat residues retained by the skin from the manufacturing process.[35] However, the formation mechanism of these soaps was not investigated. As evident in Figures and 4, the distribution of the Ca soaps with respect to CaCO3 and gypsum strongly indicates that a saponification reaction is responsible for Ca stearate formation. This is of major concern for conservators as soap formation could result in damages to the leather artefacts, for example, by attracting humidity in the structure of the cover.

Treated Leather Book Cover Samples

The soybean oil emulsion was applied by a brush to restore flexibility to the leather covers and remove humidity that can damage the collagen structure. The distribution of the oil emulsion in the leather is mainly carried out with a visible microscope with the aid of special dyes that react with fat molecules. However, this can be misleading as the dyes may react with other fat substances already present in the leather cover. Here, we tested the effectiveness of locating the oil treatment and discriminating it from original lipid materials in the leather samples provided by exploiting the differences in the molecular structure that give different IR bands and band position. In Figure , the macro ATR–FTIR spectroscopic imaging results of the cover samples from the manuscript dated 1750 and the Amsterdam Bible dated 1666 are reported. The spectroscopic images depict the distribution of the carbonyl stretching mode ν (C=O) at 1742 cm–1 of the treatment well discernible from the distribution of the carbonyl band at 1732 cm–1 associated with a possible lipid material applied in the past on the leather surface. The spectroscopic image (not reported) related to the weak band at 3008 cm–1 (indicated with an arrow in Figure b) attributed to the stretching of unsaturated CH groups also confirmed the distribution of the soybean oil treatment in the area of the samples analyzed.

Figure 5

(a) ATR–FTIR spectrum of the soybean oil emulsion acquired with the Alpha spectrometer and a single element detector and (b) FTIR spectra extracted from the macro ATR–FTIR images of the top side of the treated leather cover sample from the manuscript dated 1750. The spectra show the partial overlapping of the bands related to a lipid component of the leather (p1) and the treatment (p2). (c,d) Left, chemical images of the ν C=O (1732 cm–1) of a possible lipid (oil) material; center, chemical images of the ν C=O (1742 cm–1) of the oil treatment; and right, the composite images of the samples from the manuscript dated 1750 (c) and the Amsterdam Bible dated 1666 (d). In the composite image, purple is assigned to the soybean oil emulsion and green is assigned to the lipid component of the leather cover.

Conclusions

In this work, it is demonstrated how the sensitivity, chemical specificity, and spatial resolution of ATR–FTIR spectroscopic imaging can be used to successfully study particularly fragile materials such as leather book covers. Macro-ATR–FTIR spectroscopic imaging with a diamond ATR accessory without using a microscope has been applied to nondestructive investigations of relatively large areas of samples of historical leather book covers with a micrometer-scale spatial resolution. No damage is caused to the samples by the selected measurement mode. Therefore, the samples can be further analyzed with other techniques. For the first time, clear evidence has been obtained for the formation of calcium soaps (Ca stearate). These soaps were found in all the leather samples analyzed, suggesting that there is no evidence to link/correlate their formation to/with the provenance or manufacturing technology. Further analysis would improve our understanding of the formation mechanism and if a correlation exists between the soaps and the preservation state of these objects. Spectroscopic images allowed us to evaluate the presence of the soybean oil emulsion and to differentiate it from other fat or oil substances present on the leather covers prior to the treatment. This provided a more realistic evaluation of the presence of the soybean oil in the leather sample treated, compared to the observation with a visible microscope only. Furthermore, we demonstrated that discrimination between preserved collagen and the gelatin present on the surface of the studied samples is possible with this setup.

Finally, the origin of significant shifts of strong spectral bands of the studied samples caused by anomalous dispersion was verified by measuring pure calcium carbonate and gelatin with the same setup. This aided assignment of the spectral bands of the samples from historical leathers and, more generally, it could benefit the field of spectroscopic imaging.