Stable, active CO2 reduction to formate via redox-modulated stabilization of active sites

Electrochemical reduction of CO2 (CO2R) to formic acid upgrades waste CO2; however, up to now, chemical and structural changes to the electrocatalyst have often led to the deterioration of performance over time. Here, we find that alloying p-block elements with differing electronegativities modulates the redox potential of active sites and stabilizes them throughout extended CO2R operation. Active Sn-Bi/SnO2 surfaces formed in situ on homogeneously alloyed Bi0.1Sn crystals stabilize the CO2R-to-formate pathway over 2400 h (100 days) of continuous operation at a current density of 100 mA cm−2. This performance is accompanied by a Faradaic efficiency of 95% and an overpotential of ~ −0.65 V. Operating experimental studies as well as computational investigations show that the stabilized active sites offer near-optimal binding energy to the key formate intermediate *OCHO. Using a cation-exchange membrane electrode assembly device, we demonstrate the stable production of concentrated HCOO– solution (3.4 molar, 15 wt%) over 100 h.

X-ray diffractometer using Cu Kα radiation at a scanning rate of 9°/min in the 2θ range 23 from 20° to 80°.
The ESCA is calculated by the following equation (2): 15 A ECSA = Specific capacitance 40 μFcm −2 cm ECSA −2 , The specific capacitance of the sample was obtained by CV. It was carried out at 16 different scan rates in the range of -0.077 VRHE to -0.177 VRHE. 40 mF cm -2 is the 17 specific capacitance of a flat surface for metallic and semiconducting materials with 1 18 cm 2 of the real surface area in the aqueous electrolyte 1 .

19
The gaseous products were quantified using gas chromatography (GC, PerkinElmer) 20 with a thermal conductivity detector (TCD) and a flame ionization detector (FID solution served as the sample used for NMR.

26
The Faradaic efficiency (FE) of the liquid product (HCOO -) were calculated using equation (3): where n is the transfer electron number, F is the Faraday efficiency constant (96485 C 2 mol -1 ), c is the mass concentration of the acid root generated by the reaction (in mg L -1 ), 3 V is the electrolyte solution volume (in L), M is the molar mass of formic acid (46.03 4 g mol -1 ), and Q is the total amount of charge consumed by the entire reaction as 5 monitored by the electrochemical workstation (in coulombs). The FEs of the gaseous 6 products were calculated using equation (4): where n is the transfer electron number, F is the Faraday efficiency constant (96485 C 8 mol -1 ), V is the generated volume of gaseous products (in L), and Q is the total amount 9 of charge consumed by the entire reaction as monitored by the electrochemical 10 workstation (in coulombs).

11
All measurements were conducted at room temperature under ambient pressure. The 12 half-cell energy conversion efficiency (CEE, also called cathodic energy efficiency) is 13 calculated using equation (5): where Eformate of -0.199 VRHE is the standard potential of the formate formation. FEformate 15 is the measured formate Faradaic efficiency. Ecathode is the applied potential vs. RHE.

16
To determine the Faraday efficiency of the liquid products, we quantified the liquid   (VASP) 3 . The projector-augmented wave 4, 5 was used to describe the ion-electron 5 interaction in the periodic boundary condition, and the generalized gradient 6 approximation and Perdew−Burke−Ernzerhof 6 was used. After benchmark calculations, 7 the cutoff energy was set to be 460 eV, and the k-points were set at 3 × 2 × 1, and 8 SIGMA was set to be 0.2 eV. The convergence criterion for the energy difference

23
The binding energy Ead* was estimated by using the following equations: Esurf and Emolecule represent the energy of the bound state and the isolated molecule, 27 respectively. 28 We take the following reaction (9) as an example to show the calculation of free energy changes by using equation (10), where e is the charge number, and U is the 1 applied voltage, for the calculation of free energy of H + and e -, the computational 2 hydrogen electrode (CHE) model 12 was used to calculate the free energy of proton and 3 electron: to the stainless-steel flow field plate by using a copper frame for homogeneously 5 distributing the electrical current. The anode (IrOx on Ti foam) and cathode were and Bi0.1Sn (346 uF cm -2 ) catalysts were larger than that of Sn catalysts (128 uF cm -2 ), indicating 5 that the ECSA of Bi0.1Sn is 2.7x larger than that of Sn. As control experiments, we studied the CO2R stability of Bi and Sn catalysts with the same carbon 1 NPs and graphite coatings under the same electrochemical conditions. SEM images reveal that small 2 Bi nanoparticles were formed all over the catalyst layer, likely due to the Bi surface reconstruction 3 during CO2R (Supplementary Fig. 24). The electrical resistance of Bi2O3/Bi nanoparticles was 4 increased, and the CEE was decreased. Small Bi2O3/Bi nanoparticles partially peeled away from the 5 PTFE, leaving cracks in the catalyst layer, which further increase electrical resistance. 6 Compared to Bi, Sn consistently produced in total ~20% H2 and CO (Supplementary Fig. 25). It 7 needed a large overpotential (Supplementary Fig. 23). The overall formate CEE is lower. Note that 8 (i) the pH near the surfaces is high compared to that in the electrolyte bulk at a high current density 9 during CO2R, and (ii) catalyst geometry could cause the non-uniform distribution of OHions near 10 the surfaces. We found there was increased H2 production with Sn catalysts during the 50-hour 11 stability test at pH 14 ( Supplementary Fig. 25). cell", the major advantage of the flow cell is to achieve a large CO2R current density. We therefore 4 compare the formate performance at a relatively high current density. In addition, the counter-5 electrode reaction in the full CO2R electrolysis is water oxidation, which is RHE dependent. We 6 therefore use the RHE scale to unite the potential on both sides. Error bars correspond to the standard 7 deviation of three independent measurements. Note that the amount of catalyst used for the electrochemical test is ~650 μg (area: 1 cm 2 , 1 thickness: 700 nm), the overall amount of electrolyte used for the stability test is over 1 kilogram (> 2 1 litre). We confirm catalyst loading by inductively coupled plasma atomic emission spectroscopy 3 (ICP-AES) analysis. We first completely dissolved different sizes of Bi0.1Sn (1 cm 2 , 4 cm 2 and 15 4 cm 2 ) in the HCl solutions. We then determined the amounts of dissolved Bi and Sn by ICP-AES 5 (Supplementary Figure 33). 6 We performed CO2R stability test with Bi0.1Sn over 100 hours. It indicates that the catalyst is 7 mostly stable at negative protecting potentials in the Pourbaix diagram. However, we need to be 8 careful with the unexpected corrosion that might randomly occur on the catalyst surfaces during the 9 long-term operation. Therefore, it is important to lower the electrolyte pH for achieving long 10 stability in combination with high selectivity and energy efficiency.