Possible modifications of parchment during ion beam analysis – Heritage Science

PIXE detection limits as a function of the collected charge

PIXE showed the presence of Si, P, S, Cl, K, Ca, Ti, Fe and Zn, as minor and trace level constituents of the parchment samples. Si, P, S, Cl, K, Ca, Fe and Zn are present in skin tissue in general, additional amount of some of these (e.g. Si, S, Ca) and the appearance of titanium might be due to the making process. Silicon (4000 ppm) could only be detected with the highest accumulated charge (10 µC/cm², i.e. 1 µC on the irradiated spot), while phosphorus (300 ppm) appeared at 1 µC/cm², i.e. 100 nC on the irradiated spot. Sulphur (2000 ppm), chlorine (600 ppm), potassium (140 ppm), and calcium (2800 ppm) could be seen even at 0.25 µC/cm², i.e. 25 nC on the irradiated spot while titanium (30 ppm), iron (60 ppm) and zinc (20 ppm) could be easily identified from 1 µC/cm² charge density, i.e. 100 nC on the irradiated spot. Figure 1 shows the detection limits as a function of the accumulated charge for the applied detector and measurement parameters. For most of the elements, a marked decrease could be seen from 0.25 µC/cm² to 2 µC/cm², then no further considerable decrease was depicted. We note that the detection limits are related to the number of incoming particles, that is to the actual collected charge, while the possible modifications are related to the fluence (charge per area). The size of the irradiated spot was always the same, 0.1 cm², therefore we indicated the collected charge as deposited on a unit area on the x-axis of Fig. 1, as well as consistently at the other parts of this paper. We also note that the PIXE detector was an SDD detector with 200 nm silicon–nitride window and with a 4 cm long magnet in front of it to protect the detector from scattered protons hitting it, that is, a detector with small solid angle, optimized for lighter elements. When only heavier elements are of interest (such as metals in ink or paint on the parchment), detectors of far more efficiency can be used, and applying detector clusters is also possible [43], thus lower detection limits can be reached.

PIXE detection limits for the applied X-ray detector optimized for lighter elements. Beam spot size: 0.1 cm²

Discolouration occurring in the parchment

After irradiation, a yellowish colour was observed which faded rapidly for the lower doses, within minutes, and more slowly for the higher doses. The colour disappeared when the sample was kept under ambient temperature, pressure and humidity. Storage in dark or light conditions made no observable differences. However, a 3% H₂O₂ solution bleached it instantly. To test the role of oxygen, we irradiated a few test pieces in vacuum and kept the samples in the vacuum chamber, in this case the yellow colour was still there even after 3 days but disappeared when the samples were exposed to the atmospheric conditions. For the applied collected charges (0.25–10 µC/cm²) the colour change (visible from the surface) was mostly reversible and could have been easily missed for the lower doses. The reversibility and the role of oxygen may point towards the formation of free radicals which were neutralized subsequently, similarly to reversible colour formation in polymers [44]. Nevertheless, deep in the material irreversible colour changes also occurred, as it was seen in cross sections [31], possibly caused by double bond formations and the extended conjugation systems. For the highest dose, the yellow colour was found to be permanently visible even from the surface, without sectioning.

Infra-red spectroscopy

For collagen the most relevant bands are the amide I, II, III, A and B bands. Amide I corresponds to the C=O stretching vibration with some contribution from CN stretching. Amide II mainly related to NH bending and CN stretching, while amide III is a complex band of several modes. Amide A is risen from NH stretching vibration and amide B is coming from a Fermi resonance between the first overtone of amide II and NH stretching [45, 46].

In this study, we focus on the amide I and II bands, which are the most extensively discussed bands in the heritage science context (Fig. 2). Table 1 summarizes the positions (local maxima) and relative intensities obtained for the non-irradiated areas in six samples. Peak areas were calculated with simple integration of absorbance values, between 1480 and 1580 cm⁻¹ for amide II, and between 1580 and 1700 cm⁻¹ for amide I. We found the maximum of the amide I band around 1632 cm⁻¹, similarly to Sendrea [47] (1631 cm⁻¹), Boyatzis [48] (1629 cm⁻¹), or Derrick [49] (1629–1635 cm⁻¹). The maximum of the amide II band was at around 1538 cm⁻¹, also in accordance with literature data (Sendrea [47] 1540 cm⁻¹, Boyatzis [48] 1542 cm⁻¹, Derrick [49] 1540–1544 cm⁻¹). The variability of the distance of the two main amide peaks did not exceed the reproducibility of the device (4 cm⁻¹), while the peak area ratios differed by less than 2%. More detailed data alongside the corresponding data for the irradiated spots are available in Additional file 1: Table S1.

FTIR-ATR spectra of a sample irradiated with 2.3 MeV protons, applying 250 pA current until the accumulated charge reached 5 µC/cm² (red), as well as of a reference, i.e. non-irradiated area on the same piece of parchment (blue)

The three main natural degradation paths for collagen are denaturation, hydrolysis and oxidation, which are all irreversible [50]. Denaturation occurs when hydrogen bonds holding together the collagen molecule secondary structure are broken due to some effects such as addition of heat or water, or radiation. The weakening of the hydrogen bonds can be monitored in the FTIR spectrum through the shift of the amide II peak to lower wave numbers, thus increasing the separation of the two main characteristic peaks for collagen [49, 51, 52]. On the other hand, the amide I peak might shift to higher wave numbers since it corresponds to stretching vibration, while amide II is mostly a bending vibration, and it is a general characteristic of stretching and bending vibrations that hydrogen bonds decrease and increase their frequencies, respectively [53].

Figure 3 shows the change in the amide I–amide II distance relative to the reference value. The corresponding numerical data are available in Table S1. The FTIR-ATR spectra showed an increase of the distance of the amide I and amide II absorbance peaks for the 5 and 10 µC/cm² collected charges at all three currents, and also for the 3–4 µC/cm² in case of the lowest (250 pA) current. The amide II peak shifted from 1534–1540 cm⁻¹ down to 1522 cm⁻¹ at most, from a higher wave number associated with a more ordered structure to a lower one corresponding to a more disordered state. These findings are similar to that of Cappa et al. [23] who investigated the effects of mixed light-thermal ageing on parchment or to that of Vyskočilová [54] who applied a xenon lamp to leather, another collagen-based material.

Changes in the distance of amide I and amide II peaks, compared to the corresponding reference (non-irradiated) area on the same piece of parchment sample

According to Derrick, hydrolysis can be observed in the FTIR spectra through the amide I/amide II peak ratio. The idea is that an increase is caused by the OH⁻ vibrational band, originating from water molecules, overlapping with the amide I region. Derrick hypothesizes that this is bound water [49]. Mallamace et al., who studied lysozyme protein, found that bulk water which overlaps with amide I consists of the non-HB-network water molecules, i.e. those that do not directly participate in hydrogen bonding between peptide groups, while the HB-network water peak is situated between the amide I and II peaks [20]. Kudo et al. studied the adsorption of water to collagen, although the region around amide A is discussed in more detail in relation with the peak areas of different water contributions, free and bound [55]. In our case, the amide I/amide II peak area ratio increased by approx. 5% in the case of the highest collected charge (10 µC/cm²), while there is an indication of increase (2–3%) in the 2–5 µC/cm² region, especially for the lowest current condition. However, in our case the change in the ratio is caused rather by the relative decrease in amide II than by the increase of amide I. First, in case of archaeological parchments or in experimental studies hydrolysis is associated by a marked increase in the amide I/amide II ratio, in our case we found only a small one, similar to the xenon lamp irradiation study mentioned above on leather [54] (Fig. 4, the corresponding numerical data are available in Additional file1: Table S1). In case of FTIR spectrometry, the absolute intensities of absorbance peaks can only be considered with caution as the experimental condition affects these values greatly, so the relative intensities are used most of the time. Nevertheless, as we took reference measurements only a few millimetres away from the irradiated spots and only a few minutes apart in time, and used the same pressure, we most probably can consider the absolute values, too, not withholding the variability inherent in the measurements. Doing so, we can state that we could not see any indication for the growth of the amide I peak. However, apparently the contribution of the amide II peak tends to be somewhat smaller for the irradiated samples than for the non-irradiated ones. This phenomenon could be caused by the structural scission of the amide group, or by the loss of the triple helical or secondary structure of the collagen component. These chemical or structural changes lead to a reduced amount of intermolecular crosslinks [56], and thus can also cause the weakening of the material. We note that Vyskočilová explains the increase of the amide I/amide II peak area ratio observed for acid hydrolysis with the decrease of amide II, not with the increase of amide I, contrary to Derrick. However, acid hydrolysis was not characterized with the shift of the amide II peak [54].

Changes in the ratio of amide I and amide II peak areas, expressed as percentage of the ratio of the amide I and amide II peak areas for the corresponding reference (non-irradiated) area on the same piece of parchment sample

Oxidation of the polypeptide chain can result in the formation of carbonyl compounds, causing the appearance of a shoulder in the 1700–1750 cm⁻¹ wavenumber region of the FTIR spectrum. Nevertheless, we could not observe such changes after the proton beam irradiation.

FTIR analysis is an important tool in degradation studies but the information depth is quite small, less than 2 microns in our case. However, the deposited energy of ion beams can be several times bigger in depth of the material than at the surface, so FTIR on the sample surface underestimates the potential harm. Figure 5 shows the deposited energy along the path of the particles calculated with the SRIM programme package [57]. At the beginning of their path, the 2.3 MeV protons lose approx. 16 keV/micron, while before stopping they lose up to 100 keV/micron. When we see some changes on the surface, there is most probably already a damage deep below, similarly to irradiated polymers [58]. In our case, we found definite changes in the FTIR spectra at high doses, such as 5 and 10 µC/cm², but the situation at lower doses was not as univocal. In this study, FTIR was chosen for its non-invasiveness. Nevertheless, FTIR microscopy on cross-sections would be a valuable tool to track changes along the path of protons.

The Bragg curve: the energy loss of protons per unit distance (blue). The actual energy along the path is represented with the red line

In our case, we can clearly state that reducing the current did not lead to lesser changes observed in the FTIR spectra. Both the shift and relative decrease in amide II could be observed at lower accumulated charges for 250 pA. Indeed, irradiation with lower currents, thus longer in time, may increase the chance for changes mediated by oxygen or intermediary agents, as it was observed in case of certain polymers [59].

Bath test

The H₂O₂ solution was originally used to check the effect of an oxidizing agent on the discoloration caused by the irradiation. However, a gentle stroke with a cotton bud resulted in removing some material from the surface of samples irradiated with higher doses, which was obvious for painted samples (the effect of paint layer is discussed in our previous work [31]). Immersion of the samples in the solution eliminated the additional mechanical interference of stroking. The immersion in liquid resulted in clearly visible marks for higher doses, 2 µC/cm² and upwards, as shown in Fig. 6. Furthermore, addition of H₂O₂ was found to be not necessary, distilled water was sufficient to remove some of the irradiated collagen from the parchment pieces.

Transmitted-light photograph of a parchment sample which underwent a short bath [5 min] in 3% H₂O₂ solution. Accumulated charge: 2, 3 (upper left and right spots), 4 and 5 (lower left and right spots) µC/cm², current 1 nA). The photo was taken after drying

Optical coherence tomography

The OCT technique was employed for a limited set of samples (Table 2) to semi-quantitatively examine the modification of the samples caused by the irradiation. This technique could reveal the properties of the “crater” formed in the irradiated parchment samples which suffered the short immersion in water, either with or without H₂O₂.

Samples were scanned with OCT to estimate the width and depth of the “crater” and examine the scattering properties of the parchment in the irradiated spot. It is worthwhile to note that in case of samples not subjected to the short immersion in water (W1 in Table 2) neither crater formation nor structure alteration were observed by OCT. In the case of samples treated with water or H₂O₂ solution after being irradiated with 2 µC/cm² and upwards, craters were formed with lateral dimension somewhat smaller than the irradiated area, between 1.8 and 2.7 mm, and increasing in size with the applied charge, while the depth varied between 72 and 121 µm (Fig. 7). For higher fluences, the depth of the crater seems to be more related to the penetration depth of protons, which is the same for all fluences, approx. 100 microns. We see craters about this depth. Meanwhile, the width of the crater increases with the deposited charge in all but one instance (T1 spot 1 and 2—but here the depth is informative). These results are in accordance with the SEM cross-section data [31]. Besides the crater formation, a lowered scattering could be observed in the irradiated spot after the short bath in most cases. A lowered scattering indicates more homogeneous, therefore more deteriorated collagen structure.

OCT cross-sectional image of a parchment area subjected to 3 µC/cm² proton fluence at 500 pA, and immersed in 3% H₂O₂ solution for 5 min (sample T2, spot 4 in Table 2). The image is strongly expanded in a vertical (in depth) direction and shows a window of 7 mm × 0.70 mm size

As for the comparison of the high and low current conditions (1000 pA vs 250 pA, 2–5 µC/cm²), it must be noticed that the low current (i.e. longer irradiation) resulted in a more pronounced crater formation after the immersion in the hydrogen peroxide solution. This indicates more pronounced changes in the irradiated area, in accordance with the FTIR results.

Low irradiation (at 500 pA), below 2 µC/cm² was less harmful to the sample: for 1 µC/cm² we only found a small hole (approx. 200 microns wide and 70 microns deep: T2 spot 2 in Table 2; Fig. 8) or a slight indentation (T1 spot 2 in Table 2) after the bath test. Nevertheless, the altered scattering properties could still be observed. The 0.5 µC/cm² accumulated charge did not cause observable alterations.

OCT cross-sectional image of a parchment area subjected to 1 µC/cm² proton fluence at 500 pA, and immersed in 3% H₂O₂ solution for 5 min (sample T2, spot 2 in Table 2). The image is strongly expanded in a vertical (in depth) direction and shows a window of 7 mm × 0.70 mm size

To summarize, the irradiation with 2 µC/cm² was too intense for avoiding modifications at all three currents applied, while 1 µC/cm² accumulated charge caused some changes, not observable with electron microscopy but revealed by means of the “bath”, at 500 pA current. Unfortunately, low current and low charge conditions were not sufficiently and conclusively examined with OCT. Conditions below 2 µC/cm² were investigated only at 500 pA applied current. Therefore, digital 3D microscopy was additionally used for further investigation.

Digital 3D microscopy

Investigation of the low current (250 pA) condition, after a short H₂O₂ bath, showed an altered spot at the 1 µC/cm² accumulated charge but no signs of damage were detected at 0.25 or 0.5 µC/cm². On the other hand, for the high current (1 nA), i.e., shorter measurement, the 1 µC/cm² was found to be harmless, the removability of the irradiated material did not increase. Figure 9 shows the digital 3D image of an area subjected to 10 µC/cm² accumulated charge (proton fluence) using 1 nA current. The observed picture is very similar to deliberate micromachining when the material is weakened through radiation and then removed, thus creating a patterned structure.

Digital 3D image (a) and depth profile (b, c) for a parchment area subjected to 10 µC/cm² proton fluence at 1 nA, and immersed in 3% H₂O₂ solution for 5 min. The image was taken at the border of the non-irradiated and irradiated areas at one side of the irradiation spot

As for the 1 µC/cm² accumulated charge at 250 pA condition, the damaged surface is smaller by 60% and the created crater is much shallower but still observable (Fig. 10.) We note that the two irradiation conditions were applied to two different areas on the same piece of parchment.

Digital 3D image (a) and depth profile (b, c) for a parchment area subjected to 1 µC/cm² proton fluence at 250 pA, and immersed in 3% H₂O₂ solution for 5 min. The image was taken at the border of the non-irradiated and irradiated areas at one side of the irradiation spot

Combining OCT and digital 3D microscopy results we can state that, depending on the current, 1 µC/cm² might or might not create damage, 2 µC/cm² is undoubtedly damaging, and 0.5 µC/cm² seems to be safe for the test material used in our study.

Organic objects are not in the main focus of ion beam analysis since this technique is dedicated for elemental analysis of inorganic materials, primarily [60]. However, when quantitative elemental information, depth profile, or information on lighter elements are of interest, IBA might be a good choice [26,27,28,29]. A simple “bath test”, revealing a small hole, showed that some change was caused already at 1 µC/cm², which is not an extraordinarily high dose in ion beam analysis [28]. The changes in the structure of parchment caused the weakening of the material, resulting in increased removability. When the measured objects are relatively large and thick, such as a painting, a tiny hole might not jeopardize the integrity of the object itself. However, for materials such as parchment or paper, with thickness comparable to the penetration depth of ions, it is very important to keep the dose as small as possible or to consider alternative techniques. In this study, we investigated fresh parchment which might be more sensitive than other organic objects. On the other hand, traditionally prepared paper might be even more susceptible to damage. Indeed, Xuan paper was found to be deteriorated at 0.4–0.6 µC/cm² [30]. Mechanical properties, water content and hydrothermal stability can change during the ageing of parchment [61,62,63,64]. Combining irradiation with artificial ageing [65], either exposing irradiated and non-irradiated samples to artificial ageing and comparing their behaviour or irradiating artificially aged samples to simulate the behaviour of old parchments would be an interesting next step for this research.

Of course, it is not just IBA which might cause harm, Raman spectroscopy or other laser-based techniques and intense X-ray radiation are also able to deposit large amount of energy in a small volume [1, 11, 13, 66]. All of these techniques should only be used for heritage materials when they are safe enough and the information they provide cannot be gathered otherwise.