The section is arranged in two segments as follows. First optical microscopy and SEM were used to identify image particles associated with the tone and brightness of selected regions, as well as tarnish features. SEM–EDX was used to establish element distribution. Second, SR-XRF imaging was used to further elucidate the element distribution followed by micro-XANES to study the detailed surface chemistry.
Morphology of daguerreotype surface from SEM
The daguerreotype (Fig. 1) revealed patches of white haze across the majority of the portrait, masking the entirety of the woman’s body. This type of haze, typical for tarnished daguerreotypes, has been previously attributed to AgCl [6], a conclusion that was confirmed with XANES and EDX analysis in this study. Under this photolytic tarnish, the characteristic image particles of each region can be seen with SEM. The particles are small and densely packed for the highlight region and unequally dispersed and shaped image agglomerates for the dark region (Fig. 2). The presence of these larger image particles confirms that the woman’s shoulder to be a dark region, despite the white haze that masks the image. While some smaller particles were observed in the shadow region alongside the image agglomerates, which are not expected in the shadow region, they may be the result of the photo-catalyzed reaction to Ag particles from the invasive AgCl that covers the surface. The majority of the particles within this region correspond in size to the reported values for image particles in mid-tone regions [4, p. 120].
The plate (Fig. 1) exhibits a variety of tarnish defects (Fig. 3) including flattened dark circles, white fogging across broad regions of the image, bubbles, and protruding crystalline particles. The presence of red and blue tones suggests the presence of Ag2O and Ag2S and/or silver gold sulfide compounds (such as AuAgS or AuAg3S2) on the surface, respectively [4, p. 215, 22]. This material, when not surrounding a specific defect spot, is concentrated on the exterior border of the plate, a region prone to tarnish from contact with the metal frame of the daguerreotype plate. The presence of bubbles, one of which was analyzed with an X-ray microprobe, will be discussed below (Fig. 6).
Figure 4 shows the SEM and EDX of a tarnished area on the plate, whose center presents a crystalline, petal-like structure. These petals, which fold outwards, are rich in K and Cl. Enrichment of C is seen at the base and P is present in hot spots at the edges of the petal features. Oxygen is observed to be associated with the P, suggesting the presence of a phosphate. The K on the surface may arise from the deterioration of the cover glass under which the daguerreotype was enclosed [13, 23]. Dendrite-like structures containing K and P have been previously observed with SEM as a result of leaching from a cover glass that was originally designed to protect the daguerreotype from corrosion [12]. Furthermore, dendrites have been previously observed in non-uniform corrosion on silver surfaces [24]. Another possible source for K is the production process. Humphrey discusses various uses of K within the 19th century daguerreotype process: such as the use of dilute potassium cyanide to increase the “brilliancy of the daguerreotype” after the gilding step [25]. The photosensitising step may also be the source for residual Cl [25]. The presence of Cl on the daguerreian surface has also been attributed to glass deterioration [25]. As the cover glass was unavailable for a comparative analysis, the origin of K remains unknown although it is not uncommon to have K and Cl in glass.
XRF and micro-XANES analysis
X-ray fluorescence images (collected at 3950 eV excitation energy, above the Ag L3-edge, 3351 eV) of Ag, Cl, and S (obtained with partial fluorescence yield of Ag Lα and S and Cl Kα X-ray emission) and accompanying XANES spectra of a tarnish spot on the plate can be seen in Fig. 5. The tonal variations in the Ag XRF map show the changing distribution of Ag on the surface; the dark strands to the left of the tarnish spot indicate a relatively greater concentration of Ag while the lowest concentration is observed at the center of the tarnish feature. The decrease in Ag intensity at the center of the tarnish spot correlates to an increase in Cl and S signal. The S XRF shows decreased concentrations along the exterior of the of the woman’s face.
The Ag L3-edge (Fig. 5), which arises from the excitation of a 2p3/2 electron to unoccupied bands above the Fermi level, displays no sharp peak at the edge (3351 eV) indicating a full Ag d band. The subsequent oscillations are indicative of a face centered cubic (fcc) structure of Ag metal [26, 27]. Spectra 1, 2, and 3 show little variation and most closely resemble the Ag standard. While the two outside spectra (Ag spectra 2 and 3) LCF results indicate a dominance of pure Ag, an AgAu alloy, and Ag2S. The presence of AgCl, AgBr, and AgI in all three regions is probable but not conclusive.
The Cl XRF image (Fig. 5) does not follow the elemental distribution found in Ag and S. The greatest intensity of Cl was found at the center of the blemish with a lower, essentially uniform distribution throughout the rest of the image, which corresponds to the haze that is observed in the optical images. The Cl K-edge, which represents the excitation of an electron from a 1 s orbital to a previously unoccupied 3p orbital, taken outside of the blemish spot matches that of the AgCl standard and EDX was used to detect the presence of Ag and Cl. Silver chloride is commonly detected as a white haze that evenly covers the daguerreian image [6]. The majority of the signal within the blemish originates from NaCl, an observation confirmed by the micro-XANES standard. This is corroborated with the LCF analysis (Additional file 2: Tables). In the surrounding region, HgCl2, NaCl, KCl, and AgCl, were noted to be present through LCF analysis. The presence of KCl and NaCl at the center of the tarnish spot could be a result of the migration of Na+ and K+ to from the cover glass surface through an ion-exchange reaction by hydrogen ions (Eq. 1), resulting in Na+ and K+ precipitation onto the daguerreian surface [13, 14, 28].
$$ - {\text{SiO}} - {\text{Na}}^{ + } /{\text{K}}^{ + } + \, \left( {{\text{H}}^{ + } + {\text{ OH}}^{ - } } \right){} \rightarrow- {\text{SiOH}}^{ - } + {\text{ Na}}^{ + } /{\text{K}}^{ + } + {\text{ OH}}^{ - } . $$
(1)
Given the contained environment in which the daguerreotype would have been held, there was most likely an active and ongoing exchange between the daguerreotype surface and the cover glass. Although environmental contamination should also be considered [4, p. 175], the effects of such contaminants are expected to be minimal in this case, on the basis of the typical indoor concentrations for S and Cl as well as the daguerreotype being contained within an enclosed environment. As this daguerreotype was removed from its case for an unknown period of time (prior to being acquired by the NGC), exposure to the environment could play a larger role in the degradation of the surface. Any contamination from indoor air would depend both on S and Cl concentrations in the air and the time of exposure. Long exposure to low concentrations of S and Cl might be expected to generate the effects observed in this report. Examples of typical indoor (i.e. museums) concentrations include H2S (0.3 ppb), COS (0.6 ppb), SO2 (30 ppb), HCl (0.4 ppb), and Cl2 (not detected) [29,30,31,32] with the deposition rate of indoor atmospheric particles to be 4 ng/m3 for Cl− and 720 ng/m3 for \( {\text{SO}}_{4}^{ 2- } \) [33]. These values should be considered as representative concentrations and will vary depending on the environment.
Only one location for the S K-edge XANES is reported (Fig. 5), as the counts were too low to collect a spectrum in other regions of the map. The primary inflection point for the absorption spectrum collected at the blemish center occurs at 2482.7 eV with a post edge absorption feature at 2499.2 eV [34]. Compared to standards with LCF analysis, the principal contribution is solely from \( {\text{SO}}_{4}^{ 2- } \), with the characteristic resonance at ~ 2487 eV. The presence of residual sulfate ions is common on the surface of any S containing compounds of lower oxidation states. It may also be the result of cover glass deterioration as sulfate complexes have been established as a common degradation product [12].
Bubbles were observed at multiple locations on the daguerreotype; one example is shown in Fig. 6. This location shows a cluster of bubbles that are enclosed by a darkened circular band. Small image particles are observed above the circle, suggesting the presence of a precipitate. Although conservators have informally noted bubbles on the daguerreian surface and have attributed their presence as a product of “weeping glass” [4, p. 175, 12, 35], their presence has not been discussed in depth in the literature. While there is no direct notion of cover glass corrosion leading to these particular surface features on the daguerreotype surface, sticky droplets on the inner surface of daguerreotype cover glasses are often observed [4, p. 175, 12].
The bubble is primarily comprised of Cl and S, with hotspots of S observed outside of the bubble, where the precipitate products are observed in the optical image (Fig. 6). The S XRF also shows a concentrated band that inversely correlates to a band in the Ag XRF map.
The Ag L3-edge XANES of this bubble shows similar results to those in Fig. 6, where the majority of the signal is impacted by the Ag substrate. There is no sharp peak at the rising edge, indicating that the Ag d band is full. Metals with partially filled d bands always exhibit an intense resonance, historically called the whiteline [36]. While the white line is absent in the two Ag L3-edge spectra, the absorption bands are weaker in comparison to the Ag standard; this is a result of s–d hybridization. As the area under the curve for the experimental spectra (regions 1 and 2) is decreased when compared to the Ag standard, the presence of an AgAu alloy is confirmed, a result of the gilding process. While the Ag absorption spectra primarily provide information on the AgAu alloy, collecting S and Cl K-edge XANES does provide information regarding the tarnish composition.
The Cl K-edge XANES (Fig. 6), with a threshold located at 2822 eV, was collected at a hotpot found within the bubble. While AgCl is often the primary Cl tarnish feature noted in literature [6], at this location it only accounts for 2.4 wt. % of the signal. The major contributor to the Cl K-edge spectrum in Fig. 6 is from KCl, which was estimated to be 97.6 wt. %, of the experimental spectrum, by LCF analysis. Similar to the tarnish feature discussed in Fig. 5, this may be another example of K+ ion migration from the corroding cover class glass that is then deposited on the daguerreian surface [37].
The S K-edge is comprised of two major composite peaks at approximately 2473 and 2483 eV (Fig. 6), which arises from a 1s→ 3p transition. The location of this first transition doublet suggests that the \( {\text{S p}}\pi * \) orbital is directly involved in the covalent Ag–S interaction [38]. This metal-sulfur bond is seen as the double feature of the first peak [39]. Differing from those shown in Figs. 5 and 7 (see below), the feature at 2473 eV shows a single peak. The second peak at about 2482 eV corresponds to the oxidized form of sulfur (\( {\text{SO}}_{4}^{ 2- } \), characteristic σ* multiple scattering in a td environment) [40]. The relative weight percent determined from LCF analysis of the experimental spectrum was estimated to be 6.7 and 49.9 from \( {\text{SO}}_{4}^{ 2- } \) (6 +) and S−2 (2-), respectively. The LCF analysis revealed approximately 43.4 wt. % to be from CuS. The presence of Cu on the surface may also be a result of Cu diffusing through grain boundaries and/or holes in the Ag, as Marquis have previously noted Cu on the Ag surface [41]. Crystalline Cu salts have been reported on the surface of daguerreotypes [25, 41] and the accretion of Cu salts, such as basic sodium copper carbonate (Na3[Cu2(CO3)3(OH)]∙4H2O), has been noted by Barger and White [4, p. 167] as a result of cover glass deterioration. Other studies have also reported that glass corrosion can produce copper oxides and copper sulfides [19, 24, 41, 42]. The LCF analysis should be taken as semi-quantitative at best, since non-uniform distribution of substances and varying penetration depth of the photon will contribute to the uncertainty.
The tarnish spot located on the right side of the daguerreotype (Fig. 7) shows a central, flat blemish encircled by a green ring, followed by a yellow halo. The entire feature is surrounded by blue tinted ring (optical image supplied in Fig. 3; top center). A precipitate is observed on the surrounding ring. The Ag XRF map reveals an even distribution of Ag across the image except for a sharp decrease in concentration at the center of the tarnish feature as well as a slight decrease in intensity where the blue band encircles the blemish. The Cl XRF image shows the center of the blemish has the greatest amount of Cl with a slight increase in concentration of Cl along the exterior. Similar concentrations of S are observed in the outer-most ring of the feature that correlates with the visibly darker degradation products seen in the optical image, attributed to S accumulation [5].
Similar to the Ag L3-edge XANES collected at the other tarnish locations (Figs. 5, 6), the absorption regions 1 and 2 indicate the dominance of the Ag substrate. The decreased area under the curve, when compared to the Ag standard, indicates a small concentration of Au alloyed with Ag since the appearance of the Ag XANES is clearly AgAu alloy-like [43]. Both AgCl and AgI were detected with LCF analysis outside of the tarnish location.
The location of the Cl K-edge jumps (Fig. 7) in both spectra and the subsequent oscillations, match those of the AgCl standard. Like the other examined locations, NaCl and KCl were also detected, pointing to cover glass deterioration as a source for the degradation spot. While LCF analysis also noted the presence of HgCl2, which may be due to residual halide that was trapped in the substrate binding to Hg during plate “fixing”, the reported amounts do not provide substantial evidence.
Only one S K-edge was collected due to low counts of S on the perimeter of the blemish. The threshold resonance of the S K-edge is due to the transition of a 1s electron to p-like states in the conduction band. The XANES intensity maximum of the disulfide is almost a factor of 3 less than that of \( {\text{SO}}_{4}^{ 2- } \); this is not unexpected since p densities of states are depleted in the highly oxidized S [44]. Linear combination fit results indicate the experimental spectrum to be comprised of 20.2 wt. % HgSO4, 4.3 wt. % Na2SO4, 58.8 wt. % Ag2S, and 16.8 wt. % thiosulfate. The presence of multiple forms of sulfur points to a complicated degradation process that occurs within the daguerreian system.