Alteration markers in reference samples. Emission and excitation spectra collected on artificially aged reference samples show that different stages of CN degradation correlate with the detection of luminescent intermediates that display specific spectra, Fig. 4. An excitation spectrum mimics the UV-VIS absorption spectrum of the chromophore that emits, and these were acquired using spectrofluorimetry (spot analyses,10x2mm2). Originally, an unreacted CN film is characterized by a broad excitation spectrum centered at 320 nm, which during the first 20h of irradiation shifts to shorter wavelengths (290 nm) and only then it will evolve continuously to longer wavelengths, starting at 50h of irradiation with a maximum at circa 300 nm, Fig. 4 (left). The first shift to 290 nm can result from the degradation of chromophores formed in the dark. In the films that have suffered extensive degradation, a strong shift to higher wavelengths is observed in the excitation spectra; for example, at 130h of irradiation, several oxidized functions are identified by the presence of emission bands whose maxima are at 266, 325, 366 and 400 nm Fig. 4 (right). This agrees with the yellowing observed in the cellulose nitrate films. Based on the excitation spectra collected on the various samples, we selected the excitation wavelength of 290 nm to acquire maps of emission spectra with the DUV micro-imaging set-up of the DISCO beamline, Fig. 5. The overall signal collected at the microscale is consistent with the averaged signal collected using spectrofluorimetry, Supplementary Fig. S2. With an important advantage, the emission spectra obtained having a better resolution, it was possible to discriminate two bands in aged samples, in what was seen by spectrofluorimetry as an unresolved broad band, Supplementary Fig. S2. For this reason, we will only discuss the emission spectra acquired using synchrotron radiation. An important first result was to prove that degradation is homogeneous in the reference films, which was achieved by showing that the average spectrum obtained with POLYPHEME mapping is representative of the luminescence of each pixel, Fig. 5B. POLYPHEME shows that CN at t0 is characterized by a weak emission at 425 nm, this band increases during irradiation (Imax ca. 2600 cps). Also, a second band is formed and its intensity increases over time in comparison to the first band at 420 nm, being visible at 150h of irradiation with maximum at 510–520 nm. This is an important result that was tested to assess the conservation condition of CN by plotting this ratio I425nm/I510nm versus irradiation time. We observed a linear correlation between those two parameters showing that quantification of specific degradation state of CN can be assessed through the ratio between these two bands (I425nm/I510nm), Supplementary Fig. S3: the higher the intensity of the CN2 band at 510–520 nm over the first band CN1 (ca. 420 nm), the higher the extent of degradation when compared with the DS values obtained by infrared spectroscopy. In future work, more irradiation times will be measured as well as longer irradiation times to assess whether different mechanisms are at play.
When exciting at 290 nm, the spectrum of celluloid is dominated by the emission of camphor (Imax ca. 1900 cps). To avoid it and to collect only the CN emission, excitation was performed at 330 nm for celluloid references. As observed for CN, degraded celluloid is also characterized by a first band at 420 nm and a second one at 510–520 nm, Fig. 5C. However, the I425nm/I510nm ratios indicate that comparing to CN, celluloid has degraded more rapidly: celluloid value of 3.25 at 50h of irradiation is similar to CN at 100h of irradiation, 3.07.
In summary, the studies on reference samples identified relevant markers to probe alteration in CN films. To assess the relevance of those markers in naturally aged samples, we performed similar analytical procedures in the cinematographic films and celluloid artworks.
Cinematographic films dated from 1925 to 1960, naturally and artificially aged. The observation of the stratigraphy of the cinematographic films studied shows that they are composed of the celluloid support and the image layer, which is a proteinaceous matrix with dispersed silver particles43. On naturally aged DIF-50500 cross-section, the spatial distribution of luminescent species collected using POLYPHEME allows to distinguish three regions by their specific emission spectra, Fig. 6: i) in the proteinaceous material, the image layer was characterized by a band at 412 nm. This emission can be associated with collagen as it was made from animal sources, principally calfskin44; ii) in the celluloid support, the emission spectra presented a band at 425 nm and a shoulder at higher wavelengths (ca. 510 nm) that compare well with the 150 h irradiated celluloid reference (Fig. 5C) and, agree with the degree of substitution of 1.71 calculated for this film by infrared spectroscopy (indicative of a degraded CN)33, Table 1; iii) at the support interfaces, with air and with the image layer, the spectra presented two bands at 420 nm and 538 nm; this second band is more shifted and intense than the second band produced during degradation of celluloid references (centered at ca 510–520 nm), and can represent other types of degradation products. This emission at 538 nm can be related to a more extensive degradation or a different mechanism, Fig. 6. This degradation trend was confirmed in the cinematographic films S4, S5, S6 and 50509, where a more intense emission in the region between 500–550 nm at the interfaces was also observed, Supplementary Fig. S4.
DIF-50500 samples artificially aged (λ ≥ 280 nm, 40 ºC, 150h) were analyzed in situ since it was not possible to prepare a cross-section due to the film brittleness, Fig. 2. After 150h of irradiation, the emission spectrum acquired was very similar to that of the 150h artificial aged cellulose nitrate reference emission, Supplementary Fig. S5. This behavior comparable to cellulose nitrate is probably due to the low amount of camphor in DIF-50-500, ca. 6% w/w, estimated by infrared spectroscopy26, Supplementary Fig. S6. This further strengthens the results obtained in the reference samples, proving that the concentration of camphor influences the degradation rate of celluloid.
19th c./early 20t h c. celluloid everyday objects collection. Zinc oxide (ZnO) was the main pigment used to produce the ivory-like color and it was identified by its emission at 380 nm, in all the objects studied39,40,45−47. Zinc oxide is a semiconductor having a first narrow UV emission band at ca. 380 nm and one or more emission bands in the visible region that tend to be broad and centered at ca. 425 nm and 510 nm40,46. The latter can be attributed to defects or impurities that are related to levels located within the bandgap (trapped states). The ratio between the UV and visible bands can vary enormously in historical paints and may be used to distinguish between them40,46, Supplementary Fig. S7.
In the case of the celluloid objects of the Perlov collection, following excitation at 290 nm, the three bands described for ZnO were observed, Fig. 7 and Supplementary Figs. S7-S13. These bands were identified with maxima at ca 380 nm (ZnO1), at ca. 418 nm (ZnO2) and at ca. 507 nm (ZnO3), Supplementary Fig. S7. Every pixel of the POLYPHEME maps performed has a significant contribution of the 380 nm emission, even for those with higher intensity emissions above 400 nm, confirming the ubiquitous presence of this pigment in the samples. The ratios between the three bands varied from pixel to pixel, in the same sample and within samples, showing the heterogeneity of this ZnO-polymer systems, Fig. 7 and Supplementary Figs. S7-S14. However, for each of the objects studied, characteristic trends were observed, which will be discussed below.
The ZnO band at ca. 380 nm does not overlap with the CN emission, Fig. 5C. On the other hand, considering that these objects have an estimated camphor concentration between 20 − 15% w/w, Supplementary Fig. S11, the emission of this plasticizer at ca. 420 nm can contribute to an intensity increase in this region.
Overall, the four objects analyzed showed different emission profiles: in the 1901 postcard the emission at 510 nm (ZnO3) dominates over the other two bands; the 1899 calendar is characterized by a highly heterogeneous system with high-intensity emissions from the three bands; for the Holy Bible Pin and the American Flag Pin, the band at 380 nm displays the highest intensity, Fig. 7 and Supplementary Figs. S12-S14.
The American flag advertisement pin mappings were characterized by the strong emission of zinc oxide particles, Fig. 7. The heterogeneity of the system was observed on images collected with the full field endstation TELEMOS, by designating colors to each emission bandpass filter used: blue for the 352–388 nm range to image zinc oxide (ZnO); green for the 412–438 nm range to image ZnO trapped states, cellulose nitrate degradation, camphor, or other admixed materials; and red for the 535–607 nm range to image the ZnO trapped states and cellulose nitrate degradation, Fig. 7. In future work, we will investigate the origin of these differences by studying a set of aged and unaged reference samples. In this work, we will examine in more detail the American Flag Pin to better understand the information it is possible to extract based on the luminescent properties of ZnO, CN and camphor. The results correlated with POLYPHEME mapping in which we observed, spatially, spectral variations related to the main emission features, Fig. 7C. These emission features can be represented by three average spectra (calculated from 10 selected pixels) and are characterized by 1) 380 nm emission, accounting for 67% of the mapped area, 2) emissions between 400–450 nm (10%) and 3) ZnO trapped state emission above 500nm (23%), Fig. 7D, Supplementary Figs. S8 and S9. To better understand the causes of these variations, preliminary Raman and infrared analyzes were carried out, which will be further developed in future work. The main results are summarized in Supplementary Table S1. µRaman analysis of whitish particles in the American flag pin, studied by TELEMOS and depicted with a strong blue luminescence in Fig. 7, confirmed the presence of ZnO by its characteristic peaks at 330 and 475 cm 1 and identified a peak at 1054 cm− 1 indicative of cerussite (PbCO3), Supplementary Fig. S10 and table S1. This also agrees with the identification of Zn, Pb and Ca by EDXRF. Zinc stearate was also detected by Raman peaks at 1062, 1094, 1127, 1296 and 1445 cm− 1, its presence is confirmed by FTIR-ATR by comparison with a reference, Supplementary Figs. S10 and S11. It was also possible to identify peaks related to the polymer matrix, namely the nitrate groups (864, 1286, 1652 cm− 1) and camphor (650 cm− 1) main vibration. Zinc stearate and cerussite are compounds that can be responsible for the variations observed between the ZnO1 emission at 385 nm and the ZnO3 trapped state broad emission between 500 and 550 nm40,46,48,49, Fig. 7. Emissions between 400 and 450 can be due the ZnO defect-associated emissions, camphor, or localized cellulose nitrate degradation. Future work will investigate more in-depth the origin of these emissions and if they can be related to specific formulations of ZnO that could be used as a signature to a technological process (for colour production).
In summary, it was possible to visualize at the macro and sub microscale, the heterogeneity of these complex mixture of ZnO with celluloid. Three spectral regions of interest were identified: the ZnO1 (385nm) and ZnO3 (500-550nm) emissions whose intensity variations can be influenced by the presence of organic functionalized ZnO and other inorganic additives or impurities; and the 400–450 nm regions whose contributions can be due to ZnO crystal defects (ZnO2) or to the polymer associated emissions, either camphor or cellulose nitrate degradation. Comparing POLYPHEME results for the four objects analyzed it was possible to observe differences in their spectral signatures and spatial distribution at the submicrometric scale, Fig. 7 and Supplementary Fig. S12-S14. The different emission features of these two objects can be due to formulation differences, shown by Raman analysis, which identified massicot/litharge (PbO) and zinc stearate for the 1901 postcard and anatase (TiO2) and azurite for the holy bible pin, Supplementary table S1. These results will be further explored in future work, but already show the potentiality of this technique to distinguish color formulations that can also be linked with the three different ZnO manufacturing methods employed in the late 19th early 20th centuries40,50.