2.1. Raman and Luminescence Spectroscopy Mapping for Monitoring the Modern Paper Natural Ageing
Before performing spectroscopic characterization of paper sample from ancient books, as the first step of our study, we exposed a piece of printer paper (2019) to natural light in the laboratory for about two years. Our aim was to follow the natural degradation evolution to define suitable contrast parameters for a proper Raman and luminescence spectral imaging of the ageing processes. A portion of paper was masked from light, as sketched in
Figure 1a. Raman and luminescence maps were acquired after 3, 7, 12, and 20 months of ambient light exposition. The optical microscope image of the region of the paper sample exposed for seven months, as shown in
Figure 1b, shows the browning of several cellulose fibers. The browning is not uniform, and it is particularly evident in correspondence of large fibers, likely due to inhomogeneity of the raw material. The corresponding topographic AFM images (
Figure 1c) of browned (A) and white (B) fibers show that the measured roughness is determined by the type and disposition of the basic components of paper and not by the change in color.
Figure 2 reports both Raman and photoluminescence spectra acquired from the paper sample before and after 7 and 20 months of exposition. The main peaks observed in Raman traces of
Figure 2a can be related to vibrational features of cellulose carbon groups, namely the stretching of C–O–C glycosidic bonds between cellulose monomers (1100 cm
−1), the internal vibrations of the C–H groups (1300–1470 cm
−1) and the CH
2 stretching peak, which is related to the crystalline degree of cellulose (2890 cm
−1). The fingerprint at 1602 cm
−1 showing the presence of lignin was not observed in this sample of laser printer paper. The exposition to ambient light causes intensity decrease and broadening of C–O–C and CH
2 peaks and the appearance of new bands, whose intensity increases with exposition time, related to the oxidized species attached to cellulose chain, namely the bands at 1580 cm
−1 (stretching of double bonds in C=C–O and O–C=O groups), 1640 cm
−1, 1740 cm
−1 (stretching of C=O in the carbonyl), 1850 cm
−1 (stretching of C=O in the carboxyl groups), and in the 2300–2800 cm
−1 region (overtones and combination of carboxylic group frequencies) [
5,
7,
8].
As paper ageing proceeds, the length of cellulose polymer chain decreases due to hydrolysis reactions. This process can be monitored by recording the intensity of the bands at 1100 and 1380 cm
−1. The intensity of the first band, ascribed to the stretching of the C–O–C glycosidic bond, is proportional to the number of linkages in the cellulose chain, while the intensity of the second band, related to the vibrations of C–H groups in the glucose monomer, is weakly sensitive to the cellulose polymerization degree. Thus, the intensity ratio
RH =
I1100/
I1380 calculated from the measured Raman spectra is proportional to the polymerization degree of cellulose in the paper sample [
4]. Indeed, the
RH marker decreases as the ageing/exposition time of the paper sample increases. Analogously, the index
CI =
I2890/I1380 of cellulose crystallinity describes the decrease in CH
2 peak intensity due to the reduction in content of crystalline form, consequent to the shortening of cellulose chain length in exposed paper.
The modification in cellulose chain induced by oxidation can be evaluated from the strength of new peaks appearing in the Raman spectra of exposed papers. The oxidation of hydroxyl groups in a cellulose chain in an ambient atmosphere leads to the formation of both C=C double bonds in glucose ring and of C=O bonds (carbonyl groups), which can be further oxidized to carboxylic groups, depending on the position of the C=O bonds in cellulose chain [
5,
8,
12]. The ratio
OI =
A1640–1850/
A1580 between the area of carbonyl bands in the range 1640–1850 cm
−1 (
A1640–1850), representing the final oxidation stage of cellulose, and that of the band at 1580 cm
−1 (
A1580), corresponding to the presence of intermediate species, gives information on how advanced the oxidation process is [
5,
10]. In contrast, the ratio
OT =
A1500–2800/
A700–3000 between the area of Raman bands of oxidized functional groups in the range 1500–2800 cm
−1 (
A1500–2800) and the area of whole spectrum (
A700–3000) measures the amount of oxidation products linked to the cellulose backbone.
Figure 2b reports the photoluminescence spectra of the same paper samples of
Figure 2a, exhibiting two broad bands centered at 580 and 650 nm. By increasing the exposition time, the total area of luminescence spectrum increases, while the intensity ratio I
R between the first and second peak decreases. Due to the similarity in shape between the 580 nm peak and the one measured for the wood cellulose, with two peaks (one centered at 488 nm and the other at 580 nm), recent works correlate the presence of the long cellulose chains in paper to lower wavelength peak in the explored spectral region [
13,
14]. Following this hypothesis, fiber fragmentation promoted by hydrolysis causes an intensity decrease in the first peak. The increase in luminescence intensity of the second peak is ascribed to the formation of compounds originating from the degradation of cellulose, hemicellulose, and lignin as simple sugars, cellulose oligomers, and phenolic products.
Figure 3a shows the white light optical image of non-exposed paper sample. The area was scanned in Raman/photoluminescence spectral imaging mode, and 2D spectral arrays of (30 × 30) complete Raman/luminescence spectra were recorded. From these 2D spectral arrays, five distinct maps for cellulose characterization were extracted, using as contrast parameters the ratio
IR =
I580/
I645 between the intensities of the luminescence peaks (
Figure 3b) and the
CI,
RH,
OI, and
OT markers calculated from Raman spectra (
Figure 3c).
These marker maps are referred to the same area of crossing fibers shown in the white light image of optical microscope (
Figure 3a). The used color code displays the increasing value in the sequence blue–red–yellow. Larger values of
CI and R
H markers, measured along the cellulose fiber wall, are correlated with smaller values of
OI and
OT markers. This finding gives rise to a bimodal distribution of
RH and
OI values, extracted from the map of
Figure 3c, as shown in
Figure 4. By increasing the exposition time, the
RH (
OI) distribution peak decreases (increases) as expected, and the shape of distributions changes, becoming unimodal.
Figure 5 reports the mean values of markers, averaged over the distribution, as a function of exposition time. The decrease in crystalline index (
CI) is accompanied by a decrease in both polymerization degree (
RH) and 580 nm luminescence peak intensity (
IR), related to the long cellulose fiber.
IR and
RH markers show an initial rapid decrease followed by a slower variation, whereas the first stage rapid increase in
OT marker, proportional to the content of oxidized groups in the cellulose backbone, saturates after 12 months of exposure (
Supplementary Material S2). In contrast, the continuous increase in
OI marker indicates a progressive advancement of oxidation state with the production of C=O double bonds.
This behaviour agrees with literature data, which report that cellulose degradation can be regarded as taking place in three or two steps [
3,
5,
15]. In the first rapid stage, the weak or acid sensitive links are firstly hydrolysed, whereas in the second one, only the amorphous part of cellulose is randomly hydrolysed, as shown by the bimodal distribution of ageing marker values. At this stage, the ageing reaction decelerates, and the crystalline part of cellulose is attacked by hydrolysis/oxidation processes. Thereon, the ageing proceeds slowly and homogeneously, and the ageing marker value distributions are unimodal. It is worth noticing that, opposed to what happens for the
OT index, the
OI marker increase does not saturate, suggesting a continuous transformation of the intermediate products of oxidation into carbonyl groups.
2.2. Raman and Luminescence Spectroscopy Mapping of Modern and Ancient Paper
Figure 6 shows HR-SEM images of two paper samples of XIX (
Figure 6a,b) and XXI (
Figure 6c,d) centuries. Long cellulose fibers are clearly visible in the pictures related to the XIX century paper sample, whereas in those of modern paper, the fibers are mixed with the presence of granular material. This finding demonstrated the change in paper composition over the centuries. In fact, modern paper is composed of only up to 50% short cellulosic fibers, the remaining part being hemicellulose and lignin. On the other hand, ancient paper was manufactured from long cellulose fibers in larger percentages with respect to the hemicellulose/lignin content [
7,
9].
Several topographic AFM images were acquired on ancient and modern papers by scanning areas of different size, as shown in
Figure 7a,b. The AFM images show that the cellulose fibers of both samples are in a good conservation state without visible breakages, at least in the analyzed regions, and demonstrate the presence of disordered material in between the fibers. For each area, the height histogram was drawn, and the corresponding local roughness was calculated as standard deviation from average value. The obtained values are reported in the plot of
Figure 7c. By increasing the AFM image area, the roughness of ancient paper increases and saturates at 1.7 μm, whereas modern paper roughness progressively increases with any saturation observed.
This behavior can be ascribed to the more homogeneous composition of ancient paper with respect to modern paper, as evidenced in the HR-SEM images. The literature data report that paper ageing causes a lack of fiber surface compactness [
12]; instead, in this experiment, the roughness of modern paper is larger than that of ancient paper due to its inhomogeneous composition and its short cellulosic fibers content.
Figure 8a reports representative Raman spectra from XIX century paper compared with those of modern paper. The measured Raman profiles of ancient paper are quite similar to those collected from exposed paper, with oxidized group bands above 1500 cm
−1, whereas in the Raman spectra of modern paper, a lignin peak at 1600 cm
−1 is present. The use of wood pulp rather than rags causes an increase in lignin content in paper that can be eliminated only with secondary long chemical processes using toxic substances. Today, it is preferred not to eliminate lignin, leading to the production of low-cost, low-quality paper [
16]. Therefore, the analyzed modern paper samples were manufactured from short cellulose fibers as the main component (see
Figure 6b), but also with lignin and hemicellulose as secondary components. This difference in manufacturing is reflected in the value of I
R ratio from the luminescence spectra reported in
Figure 8b, which is smaller in modern than in ancient paper. Therefore, this parameter cannot be used to estimate the degradation degree of paper in a direct and simple way. However, the wavelength distance between the two luminescence peaks clearly shortens with increasing paper age. In fact, a shift of wavelength peak (w
p) of smaller energy band toward the 580 nm peak is observed.
We acquired Raman and luminescence spectra from an area of 60 μm × 60 μm for XIX and XXI century paper samples. The distribution of marker values calculated from Raman spectra as defined in the previous paragraph are reported in
Figure 9a–c, together with the distribution of w
p obtained from the measured luminescence spectra (
Figure 9d).
Modern paper displays a long tail of RH values larger than those of ancient paper, which experienced 124 years of hydrolysis attack. The OI distribution of ancient paper peaked at about 0.65 and 0.85, whereas that of modern paper peaked at 0.3, with a shoulder at 0.7. The OT values peak is at about the same position for both paper samples, with a larger width for modern paper. As expected from the above discussion, in the ancient paper, oxidation proceeds homogeneously and is more advanced, with a larger content of carbonyl groups, whereas the increase in hemicellulose and lignin contents, which are easier to oxidize, accelerates the degradation rate of modern paper.
As shown in
Figure 9d, the w
p distribution in ancient paper samples shows two peaks, one centered at about the same value of modern paper and the other one at 640 nm. The first peak may be due to the permanence of long cellulose fiber in the ancient paper sample, whereas the shifted peak is associated with changes occurred in paper structure related to the more rapid denaturation of shorter chains and to other bio-polymeric compounds present in the paper [
13,
17].
Non-printed areas of books of XIX, XX, and XXI centuries were analyzed. For each paper sample, several regions were mapped (
Supplementary Material S3), obtaining the distributions of different marker values.
Figure 10 plots the evolution of
RH,
OI, and
OT markers with ageing time. The reported values are the mean values calculated over the distributions, and the error bars are the calculated standard deviations. The observed difference in the paper manufacturing accounts for the large
OT values of modern paper with respect to those of XIX century paper, in which degradation is slow. The
OI values increase monotonically with ageing time, showing several fluctuations that can be explained with the different storage condition experienced by the books from which the sample paper is taken. In this respect, the
RH marker shows a steep decrease with ageing time, although the initial values of XXI century paper samples are scattered.