Effects of leather burial on collagen shrinkage activity by MHT method
It was shown that the hydrothermal stability of new, undamaged leather depends on the tannin type, animal species and processing method, and influences leather resistance against deterioration [24,25,26,27]. However, both the effect of animal species and tannin type levels out as the damage progresses and hydrothermal stability becomes increasingly influenced by the mechanism of deterioration. The MHT method was purposely set up for quick and simple evaluation of the hydrothermal stability of historical leather and parchment using very tinny sample. It combines thermal treatment and stereomicroscopy  allowing conservators to visualise how collagen fibres contract on heating using an inexpensive equipment, easy to use and providing apparently easy to interpret results . The shrinkage (contraction) activity of collagen fibres is described by a sequence of temperature intervals: no activity–A1–B1–ΔC–B2–A2–complete shrinkage. Interval A is characterized by shrinkage activity in individual fibers, one fibre at a time and with pause in between the individual movements. When shrinkage activity in one fibre is immediately followed by shrinkage activity in another fibre without pause, interval B is reached. Interval ΔC, known as the main shrinkage interval, is defined by the simultaneously and continuously shrinking of at least two fibres. Shrinkage temperature Ts refers to the starting temperature of the main shrinkage interval, while Tfirst and Tlast are defined as the temperatures at which the first and last shrinkage activity of an individual fiber is observed. The total length of the shrinkage process is expressed as ΔT = (Tlast − Tfirst).
In the last two decades, shrinkage temperature has been widely used as a metric for damage in old leather and parchment [23, 30,31,32,33] as well as to study the effects of various ageing or conservation treatments on leather and parchment [34,35,36,37,38]. Some of us have reported how leather hydrothermal stability, expressed as Tf and Ts, and structural cohesivity and heterogeneity, expressed as ΔC and ΔT, respectively, correlates with natural or artificially induced ageing patterns [18, 25, 33].
For the investigated vegetable tanned leathers, the variation of shrinkage parameters with burial times is reported in Fig. 1 where the cumulated (A1 + B1) and (A2 + B2) intervals are illustrated. The most important changes we observed were the following:
Ts increased for all leather samples, regardless of the type of tannin (hydrolysable or condensed), animal species (calf or sheep) or age (calf and cattle) and duration of treatment, except for the CaC4 sample which no longer shows contraction activity in the main shrinkage interval ΔC.
Tf decreased for all samples, except for SC2. A dramatic decrease occurred for the CH leather samples tanned with hydrolysable tannin.
The shrinkage interval ΔC generally increased, except for the CaC4 sample.
The total shrinkage interval ΔT generally increased as a result of increasing either (A1 + B1) or (A2 + B2) or both intervals.
The shrinkage interval ΔC disappears for CaC4.
The most important change refers to an increase in the degree of leather structural inhomogeneity. The increase in structural inhomogeneity was correlated to the key-steps of leather deterioration: thermal destabilization of chemically modified collagen (collagen–tannin matrix), de-tanning (breakdown of the interactions between collagen and tannin), thermal destabilization of chemically unmodified collagen (de-tanned collagen) and gelatinization . Thus, the length of (A1 + B1) interval was related with formation of pre-gelatinized collagen and gelatin, while de-tanning was associated to the length of (A2 + B2) interval . For example, in the case of CH1 and CH2 leather samples, Tf sharply decreases below 30 °C causing the (A1 + B1) interval lengthening. This could be explained by thermal destabilisation and gelatinisation of collagen microfibres [18, 25, 39] and attributed to the neutral soil pH and humidity which favour de-tanning process. It is known that vegetable tannin fixation on collagen is at minimum at pH 5 and decreases with pH increasing. Once the bonds between collagen and tannin are broken, collagen hydrolyses is promoted by soil moisture, while tannins and their possible degradation compounds are washed out by rainwater. On the other hand, de-tanning is in contrast the Ts increase measured for CH samples, from 62 °C to around 70 °C. In general, thermal stabilisation is related to a stronger attraction between the collagen fibres that makes the fibrous network more cohesive (indicated by a shorter ΔC interval) and less susceptible to gelatinisation.
In the case of CC1 and CC2 leather samples, the increase of structural inhomogeneity is equally due to the increase in the number of collagen fibres showing discrete shrinkage of individual fibres in both (A1 + B1) and (A2 + B2) intervals. Tf values higher than 60 °C preclude the existence of gelatin and pregelatinized collagen fractions. On the other hand, the increase of (A2 + B2) interval might be attributed to collagen de-tanning and formation of multiple collagen–tannin fractions with distinct tanning degree and distinct thermal stability.
CaC1 and CaC2 leather samples behaves similarly to CC1 and CC2, respectively. SC1 leather sample behaves similarly to CC1, too, while the SC2’s behaviour is clearly different in that the (A2 + B2) interval fully disappears. A more rapid thermal destabilisation of the collagen–tannin matrix within sheep leather and a greater propensity to de-tanning during artificial ageing compared to calf leather has already been observed by some of us [25, 39].
Interestingly, for all leather samples, the more coherent collagen fibres (namely those showing shrinkage in the main shrinkage interval ΔC) shown a slight increase of their thermal stability (indicated by the Ts increase), while the length of ΔC interval increased. This behaviour is in contrast to what Larsen et al.  and Carsote and Badea  observed for most artificially aged and historical leathers, namely a parallel decrease of Ts value and ΔC length as deterioration increases. On the other hand, a parallel increase of Ts and ΔC interval was recently reported for vegetable tanned leather exposed to low-dose gamma irradiation and explained by the formation of molecular cross-links that draws the collagen molecules and fibrils closer together, thus increasing their structural order and thermal stability . In the case of buried leathers, we could infer a mechanical strengthening of collagen fibres through the cementation effect of soil minerals, especially silica, instead of a chemical strengthening (cross-linking), with similar effects on leather shrinkage temperature Ts and ΔC interval. Consequently, the extension of (A1 + B1) and (A2 + B2) intervals should also be partially ascribed to the slowdown of the contraction movements caused by “hardening and welding” of collagen fibres and silica minerals.
Finally, the ΔC disappearance, as for CaC4 sample, has been related to a major damage, i.e., the conversion of collagen ordered structures into coiled structures [38, 39].
Effects of leather burial on collagen thermal stability by DSC
The DSC curves for the new leather samples measured in open crucibles and gas flow (Fig. 2) displayed a broad endotherm ranging from room temperature to about 120 °C, associated with loss of moisture, followed by two small endotherms in the temperature intervals (135–150) °C and (220–240) °C, respectively (Table 2). These two small endotherms are in good agreement with the literature data [41,42,43] and our previous results [44, 45]. Budrugeac et al.  assigned these two endotherms to thermal denaturation of the crystalline collagen embedded in the amorphous matrix. Considering the very different thermal stability of the two collagen populations, we attribute the two endotherms to two collagen populations with distinct crystallinity and hydration degree according to the two-phase model of collagen in leather represented by a crystalline phase embedded into an amorphous matrix .
For all buried samples, these small endotherms disappeared after 1-year underground. A similar behaviour, previously observed for parchments exposed to the combined action of polluting gases (50 ppm NOx + SO2) and dry heat (100 °C), was attributed to the full denaturation of both the amorphous and crystalline collagen fractions . This indicates that he burial tests we performed had already a strong denaturing effect on collagen after the first year regardless of the type of tannin or animal species. The DSC results thus confirm that the increased thermal stability determined by MHT method is not real and can only be explained by the mechanical effect of fiber aggregation through soil minerals.
Effects of leather burial on collagen susceptibility to thermo-oxidative deterioration by TG/DTG
It was reported that the non-isothermal deterioration of leather occurs through three processes accompanied by mass loss: (i) water loss at T < 150 °C, (ii) pyrolytic thermo-oxidation and (iii) thermal decomposition of the material [17, 23]. These processes are evidenced in the DTG curves by the peaks I, II and III (Fig. 3). The rate of the thermo-oxidation process corresponds to the maximum of the thermo-oxidation peak, namely the peak II (Fig. 3). For various collagen-based materials it was shown that the rate d%Δm/dt of the pyrolytic thermo-oxidation process calculated at 310 °C well correlates with the collagen matrix cross-linking degree. Accordingly, Budrugeac et al.  used this parameter as an indicator of the damage level in historical collagen-based materials relating the high rates of the thermo-oxidation process to not-damaged/well conserved leather, and the lower rates to various damage levels. In fact, the lower the rate the higher the deterioration degree. The rates d%Δm/dt of the pyrolytic thermo-oxidation process for the investigated leather samples reported in Table 2 show a progressive decrease of the oxidation rate as burial time increases. This trend of the thermo-oxidation susceptibility clearly indicates an increase of the damage level with burial time.
Effects of leather burial on molecular alteration of collagen–tannin matrix by FTIR-ATR
FTIR spectroscopic analysis is widely used to characterise the molecular changes in the collagen secondary structure [46,47,48,49,50,51]. Due to its non-invasivity and non-destructivity, FTIR-ATR technique has proved suitable for studying the alterations in the secondary structure of collagen within historical leathers [20, 52] as well as for characterizing their vegetable tannin composition [53, 54]. The following aspects were considered when studying the FTIR-ATR spectra of the investigated leather samples: (i) the effects of burial tests on the complex spectral bands of leather characterised by the overlapping of collagen (amide I and amide II) and vegetable tannin bands; (ii) the identification of other compounds on the leather surface, due to either the fabrication process or burial test.
In Fig. 4, the FTIR-ATR spectra of the buried leather samples, CaC1, CaC2 and CaC4, are illustrated and compared with that of the new leather CaC0. For all leather samples, the main infrared absorption bands of collagen were cleary detected and attributed: (i) the amide A (AA) at ~ 3300 cm−1 related to stretching vibrations of the amide N–H bonds; (ii) the amide B (AB) at ~ 3080 cm−1 attributed to the Fermi resonance overtone of the amide II vibration; (iii) the amide I (AI) at ~ 1633 cm−1 arising mainly from the C=O stretching vibration; (iv) the amide II (AII) at ~ 1538 cm−1 attributed to NH in plane bend and CN stretching vibration and (v) amide III (AIII) at ~ 1235 cm−1 corresponding to the NH bending, CN stretching vibration and small contributions from both CO in plane bending and CC stretching vibration. The presence of lipids typical bands at ~ 2920 and 2850 cm−1 (νasCH2 and νsCH2) and at ~ 1735 (νasC=O) could be ascribed to the fatliquoring products used in the leather finishing process to impart softness and flexibility to leather. Worthy of note, these bands show a significant intensity reduction in the CaC4 spectrum indicating the loss of fatliquoring oils which are water-soluble and susceptible to oxidation. Oils leaching makes leather more vulnerable to hydrolysis and to losing its collagenous ordered structure. This finding well correlates with the lack of the main shrinking interval in case of CaC4 sample. Soiling effects are indicated by the aluminosilicates specific bands at ~ 1030 cm−1 (νSi–O), 525 (δAl–O–Si) and 465 cm−1 (δSi–O–Si), as well as by the calcium carbonate main absorption bands at 1405 cm−1 (νasCO3) and 874 cm−1 (δCO3).
To discern the peak positions in the overlapping IR bands of collagen and tannins, the second derivatives of FTIR-ATR spectra were calculated in the spectral range of amide I and amide II (1700–1450 cm−1) for all the investigated samples (Fig. 5a–d). The overlapping spectral bands were assigned to the following components: amide I components centered near 1660 cm−1 and 1622–1628 cm−1, respectively; amide II band observed near 1550 cm−1; vegetable tannins bands near 1602–1605 cm−1 (visible as a shoulder of the second component of amide I) and 1502–1515 cm−1 (this band completely overlap the second component of amide II). Previously, amide I band at about 1658 cm−1 has been reported for native collagen, while the bands at 1651 cm−1 and 1633 cm−1 were correlated with denatured collagen and gelatin, respectively . Similarly, the amide II components at lower wavelengths have been associated to collagen disordered structures (i.e., the band near 1530 cm−1) [48, 52] and gelatin (i.e., the band near 1518 cm−1) [38, 48].
As a result of the burial tests, the intensity of both amide I components decreased, suggesting a loosening of the collagen–tannin interactions (Figs. 5a–d). This destabilization process led to the dissappearance of the shoulder at 1602–1605 cm−1 after the second year, most likely due to a de-tanning process. The loosening and then breaking of collagen–tannin interactions during burial is also suported by the changes in the amide II band, namely the shift of the amide II component at 1550 cm−1 (α-helix) towards lower wavenumbers (1546–1538 cm−1) accompanied by a progressive depletion of the complex band at 1502–1515 cm−1 [38, 52, 55]. In case of CH leather samples, the intensity of the 1660 cm−1 component (α-helix) decreased while the intensity of the 1628 cm−1 component (gelatine) increased [37, 38, 40, 55]. This behaviour, explained by the unfolding of the native triple helix structure and conversion to gelatin, confirms the gelatinisation of CH leather revealed by the dramatic decrease of Tf evidenced through the MHT method. A similar spectral behaviour was previosly explained by Stani et al.  through the formation of a larger number of water-mediated hydrogen bonds in denatured collagen. They assigned the component near 1630 cm−1 mainly to the carbonyl stretching of hydroxyproline and of the others amino acids involved in water mediated hydrogen bonds in the native state. Recently, some of us related the concomitant increase in intensity of amide I component at 1632 cm−1 and depletion of amide II component near 1549 cm−1 to collagen gelatinisation as a result of parchment dry heating .
Effects of burial tests on leather surface morphology by SEM
Scanning electron microscopy (SEM) is an invasive but non-destructive high-resolution technique used for imaging a wide range of materials including the collagenous materials. Depending on magnification, changes in the surface microstructure of leather, i.e. the morphology of fibre network, fibres and microfibrils, can be observed as a result of either natural or induced deterioration [56,57,58].
The collagen structural damage documented by SEM imaging of both new and buried samples are illustrated in Figs. 6, 7 and 8. The newly manufactured leather showed ordered fibrillar networking with the fibrils grouped in bundles (Fig. 6a) and their typical periodic structure cleary visible at higher magnification (Fig. 7a). The buried samples displayed discontinuities and loosening of the fibre matrix cohesion, progresive shrinkage and extended melted/gelatinised areas (Fig. 6b, c). The distorted fibrilar structure and massive presence of amourphous and gelatinised structures are already evident at higher magnification after 1-year burial period (Fig. 7b). Bozec and Odlyha  related the loss of the repeted periodicity of collagen fibril ultrastructure to collagen entirely structurally denaturated. This is in accordance with the DSC results presented above.
In addition, a microbial attack with the leaching of mineral component on the leather surface (most problably calcium carbonate, as indicated by FTIR-ATR results) was observed (Fig. 8). The chains of bacterial spores and filaments resemble the typical structure of some Actinomycetes [58, 60], aerobic spore forming gram-positive bacteria, which are the most abundent organism that form thread-like filaments in the soil and play a major rôle in the cycling of organic matter .
pH variation as a result of the burial tests
The pH value of the new vegetable tanned leathers was in the range 3.5–4.6, depending on the tannin type and manufacturing process. The pH values of the buried leather increased to 6.5–7.3, which is actually the pH value of the soil. In fact, the soil acts as a buffer, so the pH value of the archaeological leather does not depend on the type of damage, but on the type of soil in which it was buried. In the burial experiment we performed, the pH increase facilitated de-tanning of collagen–tannin matrix, thus promoting the subsequent deterioration of collagen until complete denaturation and gelatinization . In addition, at pH > 6, the leaching out of vegetable tannins is favoured. They are easily washed out by rainwater thus speeding up de-tanning process.