Electrochemical Recycling of Photovoltaic Modules to Recover Metals and Silicon Wafers

2.1 Materials and Chemicals

All chemicals were obtained by standard suppliers in analytical grade and used without further purification. For leaching experiments with copper and silver, metal foils with a single-side surface area of 1 cm² and a thickness of 0.2 mm for copper and 0.1 mm for silver have been used. In the case of tin, wires with a diameter of 0.8 mm and a length of 1 cm were used. Plating experiments were conducted with Ag₂SO₄ (VWR), CuSO₄ (Merck) and SnSO₄ (Acros Organics). PV modules were obtained from Phaesun (310164) and disassembled by removing the aluminum frame and scraping the backfoil. The EVA lamination foil was removed by swelling and softening it in toluene at 80 °C for several hours. Afterwards busbars can be retrieved from the mass.

2.2 Electrochemical Procedures

Cyclic voltammetry (CV) was performed using a glassy carbon working electrode, a platinum counter electrode and a Hg/HgSO₄ reference electrode (Sensortechnik Meinsberg Xylem Analytics).

Electrochemically assisted leaching experiments were conducted in an H-cell (Fig. 2b) with a volume of 100 mL per half-cell that has been described previously ²⁹. As anode, a boron-doped diamond electrode (CONDIAS) with a geometric surface area of 2.5 cm² was used. A glassy carbon plate (SIGRADUR, HTW Hochtemperatur-Werkstoffe) with a geometric surface area of 8 cm² was used as cathode. The current was controlled either by a power supply (CTL-3003, QUATPower) or a potentiostat (GAMRY-3000, GAMRY Instruments). The same setup was used for metal plating experiments except for potentiostatic experiments, in which case a Hg/HgSO₄ reference electrode (Sensortechnik Meinsberg Xylem Analytics) was placed in the cathode compartment. The half-cells were separated by a cation exchange membrane (F-930rfd, FUMATECH BWT) if not noted otherwise. Alternatively, porous PTFE membranes (Bohlender, BOHLN1690-64) with a pore diameter of 5 µm were used to separate the half-cells as a porous non-selective membrane material.

2.3 Analytical Procedures

The leaching progress was monitored by removing the metal from the solution, washing with deionized water, drying it in an air stream and weighing it. Concentrations of Ag, Cu, Pb and Sn were determined by periodically retrieving aliquots from the solution for analysis by atomic absorption spectroscopy (Varian AAS 240FS) at 328.1 nm for Ag, 324.8 nm for Cu, 235.5 nm for Sn and 217 nm for Pb.

3.1 Metal Leaching

To evaluate the potential of electrochemically assisted leaching of the relevant metals contained in PV modules the dissolution of pure Ag, Cu and Sn samples were investigated. First, Cu as a moderately noble metal was examined in an H-Cell setup comprising of a three-electrode setup with a membrane as separator. A metal foil with a surface area of 1 cm² was placed in the anode compartment and its mass weighed periodically during the leaching process with a constant current. As electrolyte solution, several acids were applied to identify an effective leaching medium, but only sulfuric acid (1 M) as well as hydrochloric acid (1 M) gave good leaching results (Fig. S1 and S2 in the Supporting Information, SI). In comparison, the fastest process is accomplished with HCl, however, this goes along with a notable formation and release of chlorine. This is on the one hand an environmental problem and on the other hand a loss of the leaching reagent that will quickly decrease the activity of the solution. Furthermore, the concentration of chloride is further decreased by the previously described formation of perchlorate ³⁰, ³¹. Therefore, HCl was disregarded and sulfuric acid was chosen as leaching reagent for further investigation.

The dissolution process is influenced mainly by the applied current. A higher current is leading to a faster increase of the peroxydisulfate concentration and thus a faster dissolution (Fig. 3a). Since the initial concentrations of peroxydisulfate are rather low, the dissolution proceeds at much higher rate at the end. The applied current determines the S₂O₈²⁻ concentration and thus the etching rate as shown in Fig. 3a. For comparison, etching times t_m were determined at a specific residual mass percentage m, comparable to a half-life period. For a current density of 100 mA cm⁻² and 5 M H₂SO₄ the dissolution of 50 % of a 1 × 1 × 0.2 cm³ copper foil requires a time t₅₀ of 3.1 h, while at 200 mA cm⁻² this value is reached in 1.8 h and at 300 mA cm⁻² only 1.3 h are necessary. The same trend is observed for the nearly completed dissolution at t₂₀. A further increase in current has a less significant effect as the maximum concentration of S₂O₈²⁻ is limited by decomposition according to Eq. 4 and (5). With higher current, this concentration is reached quicker and thus the dissolution rate increases. For copper, however, a higher concentration of H₂SO₄ does not improve the dissolution rate as depicted in Fig. 3b.

The dissolution of silver is less facile, hence, in 3 M H₂SO₄ less than 10 % of a 0.1 mm thick silver foil are dissolved within 5 h. Thus, other acids such as hydrochloric acid, perchloric and methanesulfonic acid were tested, though, without success. Using HCl even leads to a mass increase by the formation of a AgCl layer. An increase in the H₂SO₄ concentration leads to a faster dissolution so that in 7 M H₂SO₄ 40 % of the silver is dissolved within 5.5 h. Higher H₂SO₄ concentrations did not result in significantly higher dissolution, e.g., in 10 M H₂SO₄ t₆₀ is 4.9 h. This can be attributed to the limited stability of peroxydisulfate in strongly acidic solution according to Eq. 4 and (5) as discussed above. Thus, maximum accessible concentrations are limited.

Therefore, 5 M H₂SO₄ has been chosen to investigate the dissolution of Ag by applying different current densities. As expected, the dissolution rate increases with the current (Fig. 4a). As seen for Cu, with increasing current the effect becomes less pronounced. Nevertheless, even at lower current densities an efficient leaching of silver is possible under these conditions.

The dissolution of elemental tin proceeds quickly due to the metal's unnoble nature. As for copper and silver, the rate is dependent on the applied current as shown in Fig. 5a. With increasing current density from 50 mA cm⁻² (125 mA) to 300 mA cm⁻² (750 mA) the time for a dissolution of 50 % t₅₀ of the material decreases from 2.1 h to 0.9 h in 1 M H₂SO₄. Again, as described for Cu, the higher the applied current, the lower the relative increase. For t₂₀ the same trend is observed, which confirms that a fast dissolution can be achieved under the investigated conditions.

Regarding the acid concentration, an opposite trend is observed compared to silver. For 1 M to 5 M H₂SO₄ no significant change of the dissolution rate was observed, however, in 7 M H₂SO₄ the dissolution slows down significantly. This counter-intuitive behavior can be rationalized by the reported low solubility of tin sulfate in concentrated sulfuric acid solution ³², ³³. Thus, only medium-concentrated H₂SO₄ is well-suited for tin leaching from PV modules.

The results from these experiments indicate that 5 M sulfuric acid is best suited to extract Ag, Cu and Sn from spent photovoltaic modules. Subsequently, the three metals were dissolved simultaneously, which resulted in a stepwise process depicted in Fig. 6. It can be seen that Sn is dissolved first, followed by Cu and finally the dissolution process of Ag begins. This behavior is easily rationalized by the degree of nobility of the metals.

3.2 Metal Recovery

The plating of dissolved metals was also investigated to demonstrate the viability of metal recovery. In order to electrochemically isolate the metals, their redox potentials need to be well-separated, thus, cyclic voltammograms (CV) of the relevant metal salts were recorded and are displayed in Fig. 7a.

The peak potentials for the reduction of silver, copper and tin are observed at 0.97, 0.35 and 0.11 V vs. standard hydrogen electrode (SHE), respectively. Therefore, the separation of these metals is feasible. A CV of a solution obtained from dissolving a PV busbar is depicted in Fig. 7b showing the copper reduction wave at about 0.4 V. It is overlayed at higher potentials by a strong reduction wave of peroxydisulfate, which is present in much higher concentration. Thus, the reduction wave of tin is superimposed as well as its oxidation wave, contrary to the copper oxidation, which can be clearly observed at 0.8 V.

The recovery of the metals was demonstrated by subjecting solutions containing the dissolved metals to a cathodically polarized electrode while monitoring the metal salt concentration. The corresponding concentrations for Cu are displayed in Fig. 8a and show initially a linear increase that approaches 100 % after 2 h. This means a coulombic efficiency of close 100 % is achieved as a charge of 1800 C is transferred, which equals the reduction of 9.3 mM Cu²⁺. The remaining copper is plated more slowly, as at lower concentration the process becomes limited by mass transport. From 1 M to 5 M H₂SO₄ no significant difference was observed. However, when S₂O₈²⁻ is present at 0.1 M concentration in the Cu²⁺ solution the plating was slowed down significantly. This is caused by the reduction of S₂O₈²⁻ in a competing reaction to metal plating, as well as the reoxidation of plated Cu by S₂O₈²⁻. Thus, very high concentrations of S₂O₈²⁻ should be avoided in the cathode half-cell. If the Cu²⁺ solution is not placed directly in the cathode compartment, but the ions rather have to migrate through the dividing proton exchange membrane the plating rate is decreased substantially (Fig. 8b). This is a result of the low permeability for metal ions of the utilized proton exchange membranes and the competing migration of protons, which are present at a 10 to 50 times higher concentration. The use of a PTFE membrane with 5 µm pores results in a faster migration of Cu ions that is, nevertheless, much slower compared to the process without membrane.

The recovery of Ag was investigated utilizing lower concentrations of 0.01 M Ag⁺ as it is less abundant in photovoltaic waste. As displayed in Fig. 9a, the plating takes place quickly as most of the solutions silver content is recovered within 30 min. However, this denotes a coulombic efficiency of only 20 %, which is caused by the employed low concentration. As expected, the presence of S₂O₈²⁻ does not influence the recovery rate since the potential of the Ag/Ag⁺ couple is far from the reduction potential observed for S₂O₈²⁻.

The recovery of tin from a 0.01 M solution in 5 M H₂SO₄ was also investigated and is comparable to the recovery of Ag. This means the current efficiency is higher as two electrons are required per reduced Sn²⁺ ion. In the presence of S₂O₈²⁻ no recovery of tin was achieved, even if enough charge was passed to the solution to completely reduce S₂O₈²⁻. The reason for this might be an oxidation of Sn(II) to Sn(IV), which can precipitate as a dispersed complex form of SnO₂. This is confirmed by the absence of a Sn(II)/Sn reduction wave in CV measurements (Fig. S9 in the SI).

The electrochemical separation of the three metals was demonstrated by applying a potential to a solution 0.1 M in Cu and 0.01 M in both Sn and Ag that was stepwise decreased starting at 0.8 V for the recovery of Ag followed by 0.2 V for Cu and finally 0.0 V vs. SHE for Sn. The concentration profiles for this process are represented in Fig. 10a and show that the plating occurs stepwise and the metals can be separated with the proposed method.

3.3 Metal Extraction and Recovery from PV Waste

The composition of the investigated photovoltaic busbars was determined by fully dissolving them in HNO₃. The anticipated main component is copper with 80 wt % that is coated with solder consisting of Sn and Pb in equal amounts of 10 % of the total mass. Such a busbar was dissolved in 5 M H₂SO₄ while applying a current of 250 mA (100 mA cm⁻²). The course of the dissolution is displayed in Fig. 10b and a similar trend as for the pure metals is observed. A decrease in the rate is observed at 1.5 h, which occurs when most of the solder coating is dissolved and the dissolution of copper commences, which is slower, as described above. A small amount of about 5 mg of a gray residue remained that was identified as lead. Due to the extremely low solubility of PbSO₄ in sulfuric acid no dissolution was achieved. Nevertheless, the powder can be easily separated to be recovered and to avoid contamination of the environment.

Subsequently, the obtained solution was placed in the cathodic compartment of the cell, where a potential was applied and stepwise decreased to separate the metals. Following this procedure, over 99 % of the copper could be recovered. For tin, no recovery was possible due to the presence of S₂O₈²⁻ as described above. The silver coating of PV Si wafers was also leached. From 7.1 g of wafer fragments 112 mg of Ag were leached, which is slightly more than the previously reported Ag content of 0.7–1.4 wt % on wafers ³⁴, ³⁵. The obtained solution had a concentration of 10 mmol L⁻¹ and was subjected to plating conditions with a current of 250 mA (100 mA cm⁻²). By this, 59 % of the silver content was recovered after 2.5 h of electrolysis.

After 24 h the recovery rate increased to 88 %. The slow recovery is a result of the relatively low silver content compared to the high amount of generated S₂O₈²⁻ present in the solution that causes a competitive leaching reaction as well as competitive H₂ evolution. This problem is overcome by leaching larger amounts of PV waste to obtain higher silver concentration to make its recovery more feasible.