*2.3. Chemical Cleaning and Electrocleaning*

The chemical cleaning cell was clamped onto the daguerreotype surface, as shown in Figure S1c. Each of the three wells had a different solution pipetted into them and were left for 1 h. After that, the solutions were removed, and each well was gently rinsed with Type 1 water. The daguerreotype was then rinsed and patted dry with Kimwipes. Each site was then inspected at Surface Science Western with a VHX-6000 optical microscope (Keyence), as well as VP-SEM (Hitachi SU3900) and EDX (Oxford Instruments Ultim Max 65).

Electrocleaning and measurements were performed with a Solartron 1287 potentiostat on the daguerreotype plate. A three-electrode setup was used, as shown in Figure S1a,b, where the daguerreotype was the working electrode, Ag/AgCl was the reference electrode, and platinum foil was the counter electrode. The solution cell was clamped onto the daguerreotype, then filled with one of the several electrolytes being studied, and a 5 min open circuit potential (OCP) measurement was performed. The OCP measurement determines the resting potential of the working electrode. This was followed by an electrolytic cleaning step using the Wei method [11] or the Barger method [12], described as follows.

In the Wei method, we applied a constant cathodic polarization (constant negative potential) for 90 s, as seen in the potential versus time graph (Figure 1, top-left). In the corresponding current versus time graph (Figure 1, top-right), a typical current response is shown; the current started at a rather negative value, while reducible species were abundant on the surface, and then gradually approached zero, suggesting that the oxidized surface species were becoming depleted as the cleaning procedure progressed. *Heritage* **2021**, *4* FOR PEER REVIEW 5

**Figure 1.** (**Top**): the potential versus time, and current versus time profiles for the Wei method; only the first 30 s were plotted in a 90-s test. (**Bottom**): the potential versus time, and current versus time profiles for the Barger method; only the first 20 s were plotted in an 80-s test, including the 10 s of the constant negative potential applied at the end of each test. **Figure 1.** (**Top**): the potential versus time, and current versus time profiles for the Wei method; only the first 30 s were plotted in a 90-s test. (**Bottom**): the potential versus time, and current versus time profiles for the Barger method; only the first 20 s were plotted in an 80-s test, including the 10 s of the constant negative potential applied at the end of each test.

*2.4. SEM and EDX Characterization*  The morphology and elemental distribution of the plate before and after the cleaning were examined with SEM and EDX, respectively [14]. A Hitachi SU3900 Large Chamber Variable Pressure SEM combined with an Oxford ULTIM MAX 65 SDD X-ray analyzer was used. High resolution (up to 100 k X magnification) [15] FE-SEM imaging was performed using a Hitachi SU8230 Regulus Ultra High-Resolution Field Emission SEM. Selected areas on the daguerreotype were imaged using FESEM (image resolution of 0.6 nm at 15 kV acceleration or 0.8 nm at 1 kV acceleration). In the Barger method, we applied a modified version in which the applied potential was alternated between anodic and cathodic polarizations in 2-s intervals for 80 s, followed by a 10-s cathodic cleaning step, also seen in Figure 1, bottom-left, in the potential versus time graph. The corresponding current versus time graph (Figure 1, bottom-right) does not show the same approach to zero current seen in the Wei method, for several reasons. First, the electrode was not given enough time during any of the cathodic stages to achieve a steady state. Second, the anodic phase preceding each cathodic phase of the oscillation generated more oxidized species [7] for reduction during the subsequent cathodic phase,

XRF images of the plate were recorded at the microprobe station at CLS@APS at the ID beamline of sector 20 of the Advanced Photon Source at Argonne National Laboratory [16]. The ID line was equipped with a Si(111) double crystal monochromator and a KB mirror capable of focussing the X-ray beam down to 5 micrometres routinely. We used an excitation energy of 13 keV; this energy is just above the Hg L3 (12,284 eV) and Au L3 edge (11,919 eV), producing Hg Lα1,2 (9989 eV and 9898 eV) and Au Lα1,2 (9713 eV and 9628 eV) X-ray fluorescence lines, as well as other lines of interest, e.g., Kα of first row transition elements [17]. The incident beam was tuned to a spot size of ~ 50 μm to optimize the data acquisition efficiency. The experiment was conducted when APS was running in a top-up mode, 24 bunches with a total current of 100 mA. This mode, together with the incident focussed beam (Io) being monitored with an ionization chamber, ensured the beam stability and proper normalization, which is essential as it normally takes several hours to scan the entire plate. In this run, the photon flux was approximately 1012 photons per second over a spot size of ~50 μm with a step size of 50 μm. The illumination time was 50 ms per pixel, number of pixels of the map was 881 X 1061, and the total scan time was 13 h 24 min and 23 s. It should be noted that synchrotron XRF imaging has been widely used and the

The daguerreotype plate was mounted on a three-axis platform and the scanning was done by moving the plate across the beam pixel by pixel. The XRF image was obtained by

scope of its application can be found in a recent contribution [18].

and finally, the current during the anodic phase should never be eliminated because it could correspond to the oxidation of the Ag that makes up the bulk of the daguerreotype. Following the application of one of these cleaning profiles, the electrolyte solution was immediately removed, and the daguerreotype surface was gently flushed with Type 1 water. Each site was then analyzed optically, as well as by VP-SEM and EDX.
