Selective recovery of precious metals through photocatalysis

Precious metals such as gold and platinum are valued materials for a variety of important applications, but their scarcity poses a risk of supply disruption. Recycling precious metals from waste provides a promising solution; however, conventional metallurgical methods bear high environmental costs and energy consumption. Here, we report a photocatalytic process that enables one to selectively retrieve seven precious metals—silver (Ag), gold (Au), palladium (Pd), platinum (Pt), rhodium (Rh), ruthenium (Ru) and iridium (Ir)—from waste circuit boards, ternary automotive catalysts and ores. The whole process does not involve strong acids or bases or toxic cyanide, but needs only light and photocatalysts such as titanium dioxide (TiO2). More than 99% of the targeted elements in the waste sources can be dissolved and the precious metals recovered after a simple reducing reaction that shows a high purity (≥98%). By demonstrating success at the kilogram scale and showing that the catalysts can be reused more than 100 times, we suggest that this approach might be industry compatible. This research opens up a new path in the development of sustainable technologies for recycling the Earth’s resources and contributing to a circular economy. Recovering precious resources from waste is essential to implement a circular economy, but the available methods carry environmental costs. In this Article, a greener photocatalytic process is shown to recover up to seven precious metals from waste successfully, offering the potential for wide application.

and imidazole) as an organic form of aqua regia to dissolve gold and palladium 23 . However, these methods prove effective for only one or two PMs, and the reagent design is not favourable in terms of operation and cost. Compared with the dissolution process, reduction of PM x+ to PM 0 requires materials that must be resistant to harsh media such as acidic solutions, which is challenging. A recent report showed that a porous porphyrin polymer can capture PM ions from the acidic exudate of e-waste 24 . Smith and colleagues used 1,3:2,4-dibenzylidenesorbitol-derived hydrogels to extract gold and silver ions 25 . Queen and colleagues prepared Fe-BTC/PpPDA composite material, which was proven to quickly extract trace amounts of gold ions from water mixtures 11 . However, the synthesis of the leaching material and the subsequent reaction processes are very complicated.
The photocatalytic reactions are known to produce a range of highly reactive free radicals that can react with other species of interest. Moreover, photocatalysis is characterized by low energy requirements, easy operation, high efficiency and zero hazardous emissions. In light of the redox potential (normal hydrogen electrode (NHE) and reversible hydrogen electrode (RHE)) of photogenerated holes (TiO 2 (2.91 V NHE )), it is sufficient for them to oxidize PM 0 into PM x+ (Rh (0.75 V RHE ) < Ir (0.9 V RHE ) < Pt (1.1 V RHE ) < Au (1.3 V RHE )) 26,27 . As a result, photocatalytic oxidation could be the technology of choice to tackle environmental and energy issues, although reports on the oxidation and dissolution of PMs using this green approach are lacking.
In this Article, we report the use of photocatalysis to dissolve and recover a series of PMs (Ag, Au, Pd, Pt, Ru, Rh and Ir) in the absence of strong acids, strong bases and toxic solvents. Interestingly, the dissolution is selective, enabling the separation of metals of interest from others. More importantly, this photocatalytic technology applies to different waste sources, including e-waste, automotive three-way catalysts and ores, and can function effectively on a large scale (kilograms). This research potentially offers a sustainable and affordable approach to the recovery of PMs and is expected to inspire more efforts into the advancement of technologies for resource recycling.

results
Here, photocatalysis is used to extract PMs from waste central processing unit (CPU) boards, three-way catalysts (TWC) and ore. We started by screening a list of solvents that are commonly used to dissolve PMs, confirming that acetonitrile (MeCN) is the best choice (Fig. 1a). Our results show that Au could be completely dissolved into MeCN, whereas other solvents show no uptake at all. Mixing MeCN with dichloromethane (DCM) could optimize the variety of metals to be leached. Remarkably, it was found that seven different PMs (Au, Ag, Pd, Pt, Ru, Rh and Ir) and three common participating non-precious elements in e-waste (Fe, Al and Cu) can be fully dissolved in the mixture solution ( Fig. 1b and Supplementary  Fig. 1). By contrast, only Au, Ag and Cu could be captured if MeCN is used as a single solvent (Fig. 1c).  after (e) the reaction. g,h, Photographs of retrieving gold from gold ore (28.8 g) before (g) and after (h) the reaction. j,k, Photographs of retrieving PMs from TWC (17.9 g) before (j) and after (k) the reaction. f,i,l, The amount of PMs obtained by photocatalysing an unbroken CPU board (f), gold ore (i) and TWC (l).
As shown in Supplementary Fig. 2, simply soaking the waste and TiO 2 powders in the mixed MeCN-DCM solvent and exposing them to light irradiation, Au present in CPU board (Fig. 1d) and gold ore (28.8 g; Fig. 1g), as well as several PMs contained in TWC (17.9 g)-including Pd, Pt and Rh (Fig. 1j), all of which can be leached-forms solutions with different colours (Fig. 1e,h,k). The reaction takes place under ambient temperature and pressure, and may need 1-24 h depending on the depth of leaching. After 24 h, around 5 mg and 1.2 mg Au from one piece of CPU board sample and 28.8 g ore, respectively, could be extracted (Fig. 1f,i). In the case of TWC, up to 0.7 mg Pt, 2.3 mg Pd and 0.5 mg Rh were dissolved into the solution (Fig. 1l). Unsurprisingly, crushing the bulk waste samples to scraps could increase the surface area of contact of the reactants and therefore accelerate the reaction and help to dissolve more metals (Supplementary Fig. 3). Some other elements such as Cu, Ni and Ag are present in the CPU board ( Supplementary  Fig. 4) and they can also be leached ( Supplementary Fig. 5). Note that the present reaction is mild compared with aqua regia, which reacts vigorously, producing toxic gases, such as NO and chlorine, and cracking the CPU board ( Supplementary Fig. 6).
Selective and scalable dissolution of PMs. The e-waste, TWC and ore usually contain a variety of metals. Ideally, a PM should be leached selectively in the presence of competing metal species. To evaluate the selectivity, a photocatalytic dissolution test was performed on TiO 2 samples loaded with four of the most widely used metals, including Cu, Ag, Au and Pt. As shown in Fig. 2a,b, Cu, Ag and Au were dissolved sequentially in MeCN over the course of the reaction before DCM was introduced to enable the dissolution of Pt, demonstrating that the present photocatalytic approach is selective. As a result, when CPU board was treated, the elements that were present, including Cu, Ag and Au, could be leached one by one with increasing irradiation time (Fig. 2c,d and Supplementary Table 2). The selectivity is rooted in the reactivity of the metals (Cu > Ag > Au), whereby the dissolution of Cu takes priority over Ag and Au. Otherwise, Ag and Au ions will be reduced by Cu metals. Furthermore, as the concentration of Cu is much higher than that of Ag and Au in the CPU board, the full dissolution of Cu took more time (8 h) compared with for Ag (4 h) and Au (another 4 h; Supplementary Table 3). We noted that the Ni on the CPU board cannot be adsorbed in MeCN ( Supplementary Fig. 7). This shows that base metals and PMs can be selectively dissolved and separated. We further prepared Fe-Au alloy to assess the selectivity of photocatalytic dissolution ( Supplementary Fig. 8a). It is obvious that only Au can be dissolved (100%) in MeCN without uptake of Fe ( Supplementary  Fig. 8b). In the case of e-waste, which contains Fe, Ni, Cu, Ag, Au and Pd, similarly selective recovery of Ag, Au and Pd can be realized by adjusting the solvent and reaction time (Fig. 2e,f). First, e-waste is soaked in dilute hydrochloric acid to remove Fe and Ni, and Cu, Ag and Au are then dissolved in MeCN by photocatalysis. Finally, Pd is dissolved in mixed MeCN and DCM.
The photocatalytic dissolution process is not only simple to be implemented but also scalable. We have further designed a reactor that is suitable for dissolution on a larger scale ( Supplementary  Fig. 9). With this, we are able to treat 1.137 kg of CPU board and 1.169 kg of gold ore (Fig. 2g-l). As the photocatalytic reaction proceeds, the gold on the CPU board and the ore continues to be dissolved. In 24 h, 18.3 mg and 26.1 mg of Au were dissolved from the CPU board and the ore, respectively. After 48 h, 32.5 mg and 65.9 mg of Au were captured, respectively.
Reaction kinetics. Several parameters affect the dissolution rate of the PM-leaching process, such as solvent ratio, PM content and rotation rate. An examination of Pt dissolution provided information regarding the kinetics of the present photocatalytic process, which is a fluid-solid heterogeneous reaction and is commonly described by the shrinking-nucleus model 28,29 : where x is the leaching fraction of Pt, k c represents the apparent rate constants (h −1 ) for diffusion through the product layer chemical reaction and t is the reaction time. The dissolution fractions of Pt (x) were calculated according to the following equation: where M 0 is the mass of Pt in the Pt/C and M t is the mass of Pt in the leached solution. At a MeCN:DCM ratio of 3:1, most Pt nanoparticles (NPs) could be dissolved ( Supplementary Fig. 10a). Following equation (1), the k c could be determined and we found that the k c could be enhanced by 1.8× and 6.4× for the ratios 1:1 and 1:3, respectively ( Supplementary Fig. 10d). The leaching curve of Pt under different stirring rates is shown in Supplementary Fig. 10b,e. The results show that, when the rotating speed exceeded 600 rotations per min, the leaching rate was not affected. The larger the platinum content in the sample, the longer the leaching time that was needed, but it did not affect the leaching rate ( Supplementary Fig. 10c,f).

Recovery of PMs.
The dissolution solutions contain immobilized PM x+ ions that must be further separated and reduced to retrieve PMs. Before doing so, the solution was first filtered and evaporated to recycle the solvent ( Fig. 3a-d). The obtained products were then calcined in a tube furnace at 500 °C for 2 h under a H 2 atmosphere, yielding different NPs ( Supplementary Fig. 11b). The corresponding powder X-ray diffraction (XRD) patterns show the characteristic peaks for Ag, Au, Pt and Pd metals ( Fig. 3e-h). Scanning electron microscopy (SEM) images indicate that the PMs are NPs (Fig. 3i-l). Elemental analysis then confirmed the chemical composition of each sample and that the purity is higher than 98% (Fig. 3m,n).

Quantum yield.
Taking the amount of consumed photon energy and dissolved and recycled PMs into account, we determined the quantum yield of photodissolution reactions in MeCN and mixed MeCN-DCM solvents ( Supplementary Fig. 12). The maximum value of the incident photon-to-electron conversion efficiency (IPCE) spectra appears at ~360 nm. The apparent quantum efficiency (AQE) of the Au-MeCN system reached 0.053% at 360 nm under the optimized dissolution conditions. The AQE of the Au-MeCN-DCM system reached 0.056% at 360 nm under the optimized dissolution conditions.

Discussion
The remarkably high uptake and selectivity for PMs prompted us to understand the mechanism of the photocatalytic dissolution and reduction process. A control experiment was performed using a commercial 5% Pt/C catalyst. As shown in Fig. 4a and Supplementary Fig. 13, Pt NPs are completely dissolved after 4 h of illumination with ultraviolet light (Fig. 4b). There is no redox peak before the reaction starts ( Supplementary Fig. 14). As the reaction continues, the corresponding peak to Pt ions grows gradually, confirming that Pt is progressively dissolved into the solution and consistent with X-ray photoelectron spectroscopy (XPS) spectra in which the binding energies of Pt 0 4f drops substantially 30 (Fig. 4c).
To reveal the structure of the solid product after solvent evaporation ( Supplementary Fig. 15a), the infrared spectrum was collected, showing a different feature that was not observed for the sample in solution. As shown in Fig. 4d, the new infrared absorption peak appeared in the region of 3,344-2,913 cm −1 , which is a typical N-H stretching vibration peak 31 . Moreover, the original   (Fig. 4f). The N and Cl elements also exhibit corresponding peaks of N-H and Pt-Cl ( Supplementary Fig. 17). Apparently, the obtained solid is (NH 4 ) 2 PtCl 6 , as further evidenced by the similar colour with a commercial sample ( Supplementary Fig. 15b-c). As a result, the N-H peak in the infrared spectrum should come from the amino vibration. Furthermore, the control experiments indicate that the presence of oxygen, ultraviolet light and photocatalyst are all inevitable for Pt NP dissolution (Fig. 4g). According to the dissolution efficiency of capturing electrons (the superoxide radicals (O 2

•−
) formed by the combination with oxygen) and holes (Fig. 4h), the photogenerated electrons and holes are the main active charge carriers. Furthermore, electron spin resonance spectroscopy (ESR) was used to detect the species that were generated during the reaction. Compared with a blanket test in which no free radical could be observed ( Supplementary Fig. 18a), the ESR spectra in Supplementary Fig. 18b evidence the formation of O 2 •− and methyl radicals (CH 2 R • ) under irradiation. Furthermore, the absence of hydrogen peroxide (H 2 O 2 ) was confirmed using the iodometric method, excluding the possibility that the superoxide radical could turn to H 2 O 2 ( Supplementary Fig. 19). Nor do gas chromatography-mass spectrometry spectra suggest that any of the by-products are in the gas phase ( Supplementary Fig. 20). The experiments of discoloured silica gel were used to prove that there was water in the solution after the reaction, and the water content in the dissolution reaction was quantitatively detected using Karl Fischer analysis. ( Supplementary  Fig. 21).
On the basis of these results, we were able to decode the reaction process (Fig. 4i, proposed schematic mechanism). After ultraviolet light excites TiO 2 to generate electrons and holes (stage 1), the photogenerated electrons react with O 2 molecules to form O 2 •− , while holes react with MeCN in mixed solvents to deprotonate CHCN • radicals (stage 2) 35,36 . Meanwhile, DCM is decomposed into oxidizing CH 2 Cl • (stage 3). These active species work synergistically to oxidize PM 0 to PM x+ (stage 4) form (NH 4 ) x PMCl y solids (stage 5). This mechanism applies to the cases of Cu and Au when (NH 4 ) x CuCl y ·2H 2 O (Supplementary Fig. 22a) and (NH 4 ) x AuCl y (Supplementary Fig. 22b) are obtained. After dissolution, both Cu and Au are oxidized to ions ( Supplementary  Fig. 22c,d). Here, it is oxygen rather than PM ions that deprives electrons of the ability to generate active species in a mixed solvent of MeCN and DCM. This is because oxygen is generally more easily adsorbed onto the surface of the semiconductor to occupy the sites at which PM ions are reduced 37  there are differences in the activation of oxygen molecules in different solvents 38 .
It is worth mentioning that the choice of supports for Pt NPs could be extended to SiO 2 , Al 2 O 3 and a molecular sieve ( Supplementary  Fig. 23). The photocatalytic approach can be extended to other photocatalysts, such as ZnO (under ultraviolet-light irradiation) or cadmium sulfide, which uses visible light without compromising the performance too much (Supplementary Fig. 24). Moreover, the importance of the cyano group and chloric substituent are notable for the selection of dissolving solvents. Organics containing -C ≡ N and -Cl functional groups can promote the dissolution reaction, while organics containing only -NH x functional groups cannot initiate the dissolution of Pt (Supplementary Fig. 25). The aqueous solution of ammonium chloride cannot dissolve Pt through photocatalytic technology, and the inorganic chloride is also ineffective for the dissolution of Pt (Supplementary Fig. 26).
For a technoeconomic analysis of our recovery method, a tabulation of the costs of all of the reagents is provided in Supplementary  Table 4. Here it must be pointed out that the dissolution solvent used in this research can be reused for more than 45 times ( Supplementary Fig. 27). The photocatalyst can be recycled at least 100 times (Supplementary Fig. 28a) with neither morphology nor structure changed after reaction ( Supplementary Figs. 28b and 28c), as well as leaving no Ti in the solution (Supplementary Fig. 28d). Without taking the recyclability into account, we estimate that the cost of photocatalytic dissolution is US$0.016 for each gram of Pt. By contrast, it will take US$0.069 to recover the same amount of Pt if aqua regia is used.   In summary, taking advantage of photocatalytic oxidation, we show a more sustainable process that features highly selective dissolution of PMs from different waste sources under ambient conditions. A high leaching efficiency of 99% is achieved for e-waste, TWC and ore, and the recycled metals show high purity. The method is simple, more environmentally friendly, scalable and is suitable for a wide range of PMs. The subsequent optimization of this work can focus on how to use water instead of organic solvents. The present solvent selection is acetonitrile and DCM, although it is a new method to avoid strong acids and cyanide. To further enhance its sustainability, water as a solvent is the best choice.
Synthesis of Pt NPs. Pt NPs were prepared using a typical reduction method 41 . H 2 PtCl 6 ·6H 2 O aqueous solution (2 ml of 10 mg ml −1 ) was dispersed in 10 ml of deionized water, and 1.8 g of cetrimonium bromide was dissolved therein. After thoroughly stirring and dissolving at 50 °C, the temperature was raised to 60 °C. Then, 6.5 ml of 0.01 M NaBH was quickly added to the stirred solution and the colour change of the solution was then observed. When the solution turned brown, it was reacted for another 5 min. The resulting solution was the Pt NPs and was stored at room temperature.
Synthesis of Au NPs. Au NPs were synthesized according to the method developed by Frens 42 . In brief, 1 ml of 10 mM chloroauric acid was added to 100 ml of deionized water, and the solution was boiled. Next, 5.0 ml of 1 M trisodium citrate was added to the solution to obtain AuNPs. The solution was refluxed until a colour change from dark blue to red.
Synthesis of Ag NPs. Ag NPs were prepared using a typical reduction method 43 . A solution of 5 × 10 −3 M AgNO 3 (100 ml) was added portionwise to 300 ml of vigorously stirred ice-cold 2 × 10 −3 M NaBH 4 . A solution of 1% poly(vinyl) alcohol (50 ml) was added during the reduction. The mixture was then boiled for around 1 h to decompose any excess of NaBH 4 . The final volume was adjusted to 500 ml.
Synthesis of Pt/TiO 2 , Ag/TiO 2 , Au/TiO 2 , 5% Ag/C and 5% Au/C. A certain amount of TiO 2 (Degussa P25) was placed in a solution of the prepared Pt NPs. After stirring for 2 h, the colour of the powder changed from white to light brown. The Pt/TiO 2 sample in the reactor was collected by centrifugation and washed several times with water and ethanol, and dried overnight at 80 °C. Samples of 1% Pt/SiO 2 , 1% Pt/Al 2 O 3 and 1% Pt/4 Å molecular sieves were synthesized using the same method, whereby TiO 2 was replaced by SiO 2 , Al 2 O 3 and 4 Å molecular sieves. Samples of 1% Ag/TiO 2 and 1% Au/TiO 2 were synthesized using the same method, whereby Pt NPs were replaced by Ag NPs or Au NPs. Samples of 5% Ag/C and 5% Au/C were synthesized using the same method, whereby TiO 2 was replaced by C and Pt NPs were replaced by Ag NPs or Au NPs.

Synthesis of Cu/TiO 2 .
Cu/TiO 2 was prepared by a simple precipitation method 44 . In brief, 0.5 g of TiO 2 (Degussa P25) was dispersed in NaOH aqueous solution, and then a certain volume of CuCl 2 solution was added dropwise under stirring. After further stirring for 6 h, the precipitates were washed with deionized water until pH 7, and then dried at 80 °C for 12 h. Finally, Cu/TiO 2 was calcined at 400 °C for 2 h in nitrogen. Samples of 1% Cu/SiO 2 were synthesized using the same method, in which TiO 2 was replaced by SiO 2 .
Photocatalytic dissolution of PMs from CPU, ore and TWC. At 20 °C, an amount of CPU, gold ore or TWC was placed into a mixed solution of 50 ml MeCN and DCM (3:1). Next, 50 mg TiO 2 was added directly into the solution. For the dissolution of 1 kg of CPUs and ore, 500 ml of a mixture of MeCN and DCM (3:1) and 500 mg of TiO 2 was added. Ultraviolet-light irradiation (100 W 365 nm LED lamp, Beijing Perfectlight) was then applied to the mixed solution for 50 h. The dissolution was calculated by measuring the concentration of PM in the solution using inductively coupled plasma emission spectrometry (ICP, Jarrel-Asm Atom Scan 2000).

Selective dissolution of PMs from CPU and metal catalyst.
At 20 °C, a piece of CPU was placed into 50 ml acetonitrile solution. Next, 50 mg TiO 2 was directly added into the solution. Then ultraviolet-light irradiation was applied to the mixed solution for a certain time. An amount of 1% Cu/TiO 2 (50 mg), 1% Ag/ TiO 2 (50 mg), 1% Au/TiO 2 (50 mg) and 1% Pt/TiO 2 (50 mg) was placed into a 15 ml acetonitrile solution. Next, ultraviolet-light irradiation was applied to the mixed solution for a certain time. First, copper, silver and gold were completely dissolved after 6 h. Then, the undissolved residual solid was centrifuged and placed into a mixed solvent of acetonitrile and DCM, which can dissolve platinum.
Selective dissolution of PMs from e-waste powder. First, we immersed e-waste powder (1 g) in 50 ml diluted hydrochloric acid (HCl, 1 M) and reacted for 5 h to remove some non-PMs (such as Fe and Ni). Second, the undissolved sample was separated and washed, and then put into the MeCN solution (50 ml). Next, 100 mg TiO 2 was added directly into the solution. Ultraviolet-light irradiation was then applied to the mixed solution for a certain time. Finally, the undissolved sample was separated and washed, and then put into the mixed solution of MeCN and DCM (3:1, 50 ml). Next, 100 mg TiO 2 was added into the above solution, and then ultraviolet-light irradiation was applied to the mixed solution for a certain period of time. The dissolution was calculated by measuring the concentration of PM in the solution using ICP (Jarrel-Asm Atom Scan 2000).
Dissolution of platinum NPs. At 20 °C, the PM-dissolution properties of the system were tested by photocatalytic induction of a platinum catalyst (5% Pt/C). Typically, 50 mg of 5% Pt/C catalyst and 50 mg TiO 2 were dispersed into a mixed solution of 20 ml MeCN and DCM (3:1) and irradiated with a 100 W 365 nm LED lamp (15 mW m −2 ). Control experiments were conducted under the conditions of darkness, nitrogen atmosphere or SiO 2 as a catalyst. The degree of dissolution was calculated by measuring the concentration of platinum in the solution during the dissolution process using ICP (Jarrel-Asm Atom Scan 2000).
Calculating the dissolution rate. Owing to the volatilization of chlorine in the aqua regia, we put the reacted sample in the aqua regia into a fume hood and reacted for 5 h such that the remaining PMs on the sample were completely leached. Then, the photocatalytic dissolution efficiency was calculated according to the remaining metal content, and the dissolution rate was calculated according to the equation below: Dissolution rate (%) = The amount of PM leaching by photocatalytic dissolution ( mg l −1 ) The amount of PM leaching by aqua regia (mg l −1 ) × 100 (3) As silver and rhodium cannot be dissolved in aqua regia, silver and rhodium were dissolved in concentrated nitric acid and hydrofluoric acid, respectively.
PM ion reduction process. The leachate was recovered after distillation at a low temperature (60 °C), and the solid product was calcined and reduced under an H 2 /Ar atmosphere at 500 °C.
Calculating AQE. The AQE at each centred wavelength of the monochromatic light was calculated from the ratio of the number of PM dissolution to those of the irradiated photons using the following expression: Characterization. The dissolution was calculated by measuring the concentration of PM in the solution using ICP (Jarrel-Asm Atom Scan 2000). The catalyst samples were characterized using scanning transmission electron microscopy (STEM) (high-angle annular dark-field STEM, FEI G2 Tecnai F20, 200 kV and JEOL ARM200F with a probe corrector), XPS (PerkinElmer PHI 5000C, Al Kα), UV-VIS diffuse reflectance spectra (Shimadzu, UV2600). The ESR spectra of the transition metal in the solution were recorded using the JEOL JES-FA200 model spectrometer to estimate the valence composition of platinum. The solution samples after reaction were characterized using Fourier transform infrared spectrometry (Nicolet IS 10). Powder samples (obtained by evaporating the reacted solution under low temperature) were characterized using field emission SEM (HITACHI S4800, operated at 50 kV), Fourier transform infrared spectrometry (Nicolet Nexus 470), XPS (PerkinElmer PHI 5000C, Al Kα) and XRD (Rigaku D/MAX-2000). All of the binding energies were calibrated using the contaminant carbon (C1s = 284.8 eV) as a reference. Gas compositions were measured using GC-MS (GCMS-QP2010, Shimadzu). The GC system was equipped with a SH-Rxi-5Sil-fused silica capillary column. Quantitative detection of water content in the dissolution reaction was performed using Karl Fischer analysis (870 KF Titrino plus). Photoelectrochemical measurements were performed in a conventional three-electrode, single-compartment quartz cell on an electrochemical station (CHI 660D). A bias voltage of 0.5 V was used to drive the transfer of photogenerated electrons from the working electrode to the platinum electrode. A LiCl acetonitrile or mixed solution (0.50 M) was used as the electrolyte.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings of the study are available within the paper and its Supplementary  Last updated by author(s): Feb 4, 2021 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.
n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability The data that support the findings of this study are available from the corresponding author upon reasonable request.