Operando Converting BiOCl into Bi2O2(CO3)xCly for Efficient Electrocatalytic Reduction of Carbon Dioxide to Formate

Highlights An operando synthetic approach was exemplified to enhance catalyst stability for efficient reduction of CO2 to formate. A highly stable Bi2O2(CO3)xCly electrocatalyst was synthesized by direct electrochemical conversion of BiOCl via a cathodic potential-promoted anion-exchange process under operando CO2RR conditions. The surface Cl− in Bi2O2(CO3)xCly changes the p-orbital electron states to enhance the stability and alters the CO2RR pathway to markedly reduce the energy barrier. Supplementary Information The online version contains supplementary material available at 10.1007/s40820-022-00862-0.


Introduction
The renewable electricity-powered electrocatalytic carbon dioxide reduction reaction (CO 2 RR) to produce chemicals/ fuels not only curbs greenhouse gas emissions but also reduces our reliance on the rapidly diminished petroleum resources [1]. In this regard, various C 1 (e.g., carbon monoxide, formate, methane and methanol), C 2 and C 2+ (e.g., ethylene, ethanol, acetylene, acetate, acetaldehyde, oxalic acid and n-propanol) CO 2 RR products have been obtained [2,3]. Among them, CO and HCOO − /HCOOH are the most energy-efficient CO 2 RR products as they can be formed by transferring two electrons to CO 2 . Comparing to CO, converting CO 2 to HCOO − /HCOOH is more desirable because HCOO − /HCOOH are more valuable commodity chemicals [4,5]. To date, the reported high-performance electrocatalysts for CO 2 reduction to HCOO − /HCOOH are almost exclusively made of p-block metals-based materials such as In, Pb, Sn, Sb and Bi [6,7].
Owning to their low toxicity and high selectivity toward HCOO − /HCOOH, Bi-based CO 2 RR electrocatalysts have attracted increasing attentions [8,9]. Various Bi-based CO 2 RR electrocatalysts such as metallic Bi o , oxides and subcarbonate (Table S1) have been employed to electrocatalytically convert CO 2 to HCOO − /HCOOH. As shown in Table S1, in general, the metallic Bi o -based ones perform better than other forms of bismuth-containing electrocatalysts. Nevertheless, the metallic Bi o -based electrocatalysts usually require high overpotentials, consequently the high cathodic potentials, to achieve their optimal performances [10,11], undesirable for energy efficiency. In addition, high cathodic potentials are favorable for the competing hydrogen evolution reaction (HER), which often leads to low Faradic efficiencies toward HCOO − /HCOOH (FE HCOO -/FE HCOOH ) [12]. The bismuth oxides-based electrocatalysts were also reported (Table S1). Noticeably, such electrocatalysts often encounter stability issues because the bismuth oxides in these electrocatalysts can be easily converted to metallic Bi o under CO 2 RR conditions [13]. For example, Deng et al. reported a Bi 2 O 3 electrocatalyst with the optimal performance at − 0.9 V (vs RHE) to achieve a FE HCOO -of 91% with a partial HCOO − current density (J HCOO -) of ~ 8 mA cm −2 [14]. However, the as-synthesized Bi 2 O 3 is found to be partially converted to metallic Bi o under the CO 2 RR conditions at − 0.9 V vs RHE. In fact, the reported bismuth oxides electrocatalysts require cathodic potentials ≥ − 0.9 (vs RHE) to concurrently achieve FE HCOO -> 90% with J HCOO -≥ 15 mA cm −2 [13,15,16]. Under such CO 2 RR conditions, the bismuth oxides in these electrocatalysts are either partially or completely converted to metallic Bi o . Other than metallic Bi o and bismuth oxides, Zhang's group reported the use of ultrathin bismuth subcarbonate (Bi 2 O 2 CO 3 ) nanosheets to catalyze CO 2 reduction to HCOO − [17]. Their Bi 2 O 2 CO 3 electrocatalyst exhibits a very low overpotential of 610 mV and can achieve a FE HCOO -of 85% with a J HCOOof ~ 11 mA cm −2 at − 0.7 V (vs HRE), however, partial conversion of Bi 2 O 2 CO 3 to the metallic Bi o occurs within 30 min under − 0.65 V (vs RHE).
As reviewed above, under the required cathodic potentials to concurrently achieve high FE HCOO -and J HCOO -, the reported bismuth oxide and subcarbonate electrocatalysts are unavoidably reduced to metallic Bi o , leading to the structural and compositional changes under operando CO 2 RR conditions. Critically, such operando structural transformation processes are progressive and potential-dependent, leading to great difficulties to confirm the actual active sites, hence the catalysis mechanisms. Parenthetically, the synthetic conditions of the reported bismuth oxide and subcarbonate electrocatalysts are vastly different to their electrocatalytic application conditions, which could be a cause of their structural transformation under the operando CO 2 RR conditions. If this is true, the severe operando stability issues might be effectively mitigated by employing identical synthesis and application conditions.
In this contribution, we report an approach to electrochemically convert bismuth oxychloride (BiOCl) into chloride-containing bismuth subcarbonate (Bi 2 O 2 (CO 3 ) x Cl y ) under operando CO 2 RR conditions (at − 0.8 V vs RHE in CO 2 -saturated 0.5 M KHCO 3 solution) and use it to exemplify that the operando synthesis can be an effective means to enhance the operando electrochemical stability of electrocatalysts. Systematic operando spectroscopic studies were conducted to depict the conversion mechanism and electrochemical stability. BiOCl is converted to Bi 2 O 2 (CO 3 ) x Cl y via the cathodic potential-promoted anion-exchange process.

Synthesis of BiOCl-NSs
0.164 g of KCl and 0.868 g of Bi(NO 3 ) 3 ·5H 2 O were dissolved in 70 mL H 2 O and stirred for 1 h. The solution was transferred into a 100 mL Teflon-lined stainless-steel autoclave and kept at 120 °C for 24 h. The obtained BiOCl-NSs was adequately washed with deionized water and ethanol and dried at 60 °C for 6 h in vacuum oven.

Synthesis of Bi 2 O 2 (CO 3 ) x Cl y
Twenty milligrams of the as-synthesized BiOCl-NSs was mixed with 80 μL Nafion solution (5 wt%) and dispersed 0.92 mL isopropanol under sonication for 40 min to form the ink. 100 μL ink was then cast onto the pre-treated carbon fiber paper substrate with an exposed area of 1 × 1 cm 2 (2 mg cm −2 of BiOCl-NSs). The carbon fiber paper with loaded BiOCl-NSs was used as the working electrode and subjected to − 0.8 V (vs RHE) in CO 2 -saturated 0.5 M KHCO 3 solution for 2 h to electrochemically transform BiOCl-NSs to Bi 2 O 2 (CO 3 ) x Cl y .

Synthesis of Bi 2 O 2 CO 3
For comparative purpose, pure Bi 2 O 2 CO 3 was synthesized. Under constant stirring, 0.234 g of Bi(NO 3 ) 3 ·5H 2 O was dissolved into 10 mL H 2 O, followed by adding 1.502 g of CH 4 N 2 O and 10 mL of C 2 H 5 OH. The resultant solution was then placed in the oil bath under 90 °C for 4 h. The obtained pure Bi 2 O 2 CO 3 was adequately washed with deionized water and ethanol and dried in a vacuum oven of 60 °C for 6 h.

Electrochemical Measurements
The electrochemical measurements were performed using a Nafion 115 proton exchange membrane separated twocompartment electrochemical cell consisting of a threeelectrode system controlled by an electrochemical station (CHI 660E). For CO 2 RR, the Bi 2 O 2 (CO 3 ) x Cl y working electrode (1 × 1 cm 2 ) was fabricated by operando electrochemical transformation of the immobilized BiOCl-NSs on carbon fiber paper, while the Bi 2 O 2 CO 3 working electrode was prepared by immobilizing 2 mg cm −2 of Bi 2 O 2 CO 3 on carbon fiber paper (1 × 1 cm 2 ). For all electrochemical measurements, an Ag/AgCl (3.5 M KCl) reference electrode, a Pt mesh counter electrode and CO 2 -saturated 0.5 M KHCO 3 electrolyte (pH of 7.2) were employed. During CO 2 RR, the electrolyte in the cathode compartment was constantly stirred at a rate of 800 rpm and bubbled with CO 2 at a flow rate of 5 mL min −1 controlled by a universal flow meter (Alicat Scientific, LK2). All reported potentials were converted to the reversible hydrogen electrode (RHE) in accordance with E RHE = E Ag/AgCl + 0.059 × pH + 0.205. The gas chromatography (GC, RAMIN, GC2060) equipped with a flame ionization detector (FID) and a thermal conductivity detector (TCD) was used to qualitatively and quantitatively determine the gaseous products (e.g., H 2 and CO or other gaseous hydrocarbons). The CO and H 2 Faradaic efficiency were calculated as below: (1) FE CO = 2x CO pGF∕IRT where x CO and x H2 (vol%) are the volume fractions of CO and H 2 in the exhaust gas, I (A) is the steady-state current, G = 5 mL min −1 is the CO 2 flow rate, p = 1.013 × 10 5 Pa, T = 273.15 K, F = 96,485 C mol −1 , R = 8.3145 J mol −1 K −1 .
1 H nuclear magnetic resonance ( 1 H-NMR) was used to qualitatively and quantitatively determine the liquid phase products, including HCOOH. After reaction, 0.5 mL electrolyte from the cathode compartment was mixed with 0.1 mL D 2 O containing 3-(trimethylsilyl)propanoic acid (TMSP) as the internal standard and subjected to NMR analysis. The Faradaic efficiency for formation of HCOO − was calculated as below:

Characterizations
The morphologies and structures of the samples were characterized by SEM (JEOL JSM-7100) and TEM (Tecnai F20, 200 kV). The STEM images were recorded using a probe corrected JEOL JEM-ARM200F instrument with at an acceleration voltage of 200 kV. AFM measurements were performed using a Bruker Dimension Icon system. XRD patterns were collected from a Bruker D8 diffractometer. The operando XRD patterns were recorded using a Bruker D8 diffractometer and a home-made three-electrode electrochemical cell. Raman spectra were taken by a REN-ISHAW mVia Raman Microscope using a 532 nm excitation laser. The operando Raman studies were performed on a RENISHAW mVia Raman Microscope equipped with a microscopic lens immersed under the electrolyte to capture Raman signals and a home-made three-electrode electrochemical cell consisting of a BiOCl-NSs or Bi 2 O 2 CO 3 working electrode, an Ag/AgCl (3.5 M KCl) reference electrode and a Pt mesh counter electrode. The working electrodes were prepared by immobilizing BiOCl-NSs or Bi 2 O 2 CO 3 on a commercial Si substrate (1 × 1 cm 2 ). XPS spectra were recorded by Kratos Axis ULTRA using the C1 s at 284.8 eV as the internal standard. C and O K-edge XAS measurements were performed at the Soft X-ray spectroscopy beamline at Australian Synchrotron Facility, Australia's Nuclear Science and Technology Organisation (Clayton, Victoria, Australia). Bi L 3 -edge XAS measurements were performed at the 10-ID-B beamline of the Advanced Photon Source (APS), Argonne National Laboratory (ANL). Data reduction, processing and subsequent modeling were performed using the Demeter XAS software package [18]. Modeling of the EXAFS data of Bi 2 O 2 (CO 3 ) x Cl y was performed using Bi-O, Bi-C and Bi-Bi backscattering paths from the crystal structure of Bi 2 O 2 CO 3 [19], while the Bi-Cl contributions were generated from the optimized structure generated from the DFT calculations. All EXAFS fitting was performed using an S 0 2 value of 0.868, which were obtained by modeling the EXAFS of a reference Bi foil (L 3 -edge at 13,419 eV). To minimize error in CN and NND values, Debye-Waller factors were optimized in initial rounds of EXAFS fitting and then held constant.

DFT Calculations
All computation studies were performed using density functional theory (DFT) implemented in the Vienna Ab-initio Simulation Package (VASP) code in this study [20,21]. For the effects of electron-electron exchange and correlation, the Perdew-Burke-Ernzerhof (PBE) functional at the generalized gradient approximation (GGA) level was employed [22]. The projected augmented wave (PAW) potentials were used throughout for ion-electron interactions [23], with the 5d 10 6s 2 6p 3 , 2s 2 2p 2 , 2s 2 2p 4 , 3s 2 3p 5 and 1s 1 treated as valence electrons of Bi, C, O, Cl and H, respectively. The plane-wave cutoff of 520 eV was set for all the computations. The (1 × 2) clean {001} faceted Bi 2 O 2 CO 3 was modeled by a 14-atomic layer slab separated by a vacuum layer of 20 Å in this study. When geometries of all structures were optimized, top seven layers of the surfaces including adsorbate were relaxed, while the bottom seven layers were fixed. The gamma-centered Monkhorst-Pack k-point meshes with a reciprocal space resolution of 2π × 0.04 Å −1 were utilized for structural optimization. For the calculations on CO 2 and formic acid molecules, a (20 × 20 × 20) Å 3 unit cell and a Γ-only k-point grid were used. All atoms were allowed to relax until the Hellmann-Feynman forces were smaller than 0.01 eV Å −1 , and the convergence criterion for the electronic self-consistent loop was set to 10 -5 eV. The adsorption energy of each adsorbate [ΔE (eV/n)] was calculated as follows: where E ad , E surf and E ad/surf are the energies of an adsorbate, the clean {001} facet and the surface with adsorbates, respectively. And n is the number of adsorbates on the surface. Based on computational hydrogen electrode (CHE) model [24,25], each electrochemical reaction step can be regarded as a simultaneous transfer of the proton-electron pair as a function of the applied potential. The reaction mechanism of CO 2 reduction should consist of the following elementary reactions: or where * means the corresponding surface and adsorbed states. The free energy for all intermediate states and nonadsorbed gas-phase molecule is calculated as: where the E elec is the electronic energy obtained from DFT calculation; E ZPE is the zero-point vibrational energy estimated by harmonic approximation; ∫C p dT is the enthalpic correction and TS is the entropy. Here, reported values of E ZPE , ∫C p dT and TS are adopted [24]. The solvation effect has been considered for *COOH by stabilizing 0.25 eV [24].

Synthesis and Characterization of Bi 2 O 2 (CO 3 ) x Cl y
In this work, Bi 2 O 2 (CO 3 ) x Cl y was synthesized by direct electrochemical conversion of the pre-synthesized BiOCl under operando CO 2 RR conditions (Fig. 1a). The BiOCl nanosheets (BiOCl-NSs) were firstly synthesized as the precursor via a one-pot hydrothermal method [26]. The X-ray diffraction (XRD) pattern of the as-synthesized BiOCl-NSs (Fig. S1) can be indexed to the tetragonal BiOCl (PDF No. 06-0249). The Raman spectrum (Fig. S2) displays two strong peaks centered at 143 and 199 cm −1 , assignable to A 1 1g (external) and A 2 1g (internal) Bi-Cl vibration modes, respectively, while the weak peak at 400 cm −1 can be attributed to B 1 g mode [27]. The atomic force microscopy (AFM) and field-emission scanning electron microscopy (FE-SEM) images (Fig. S3a, b) disclose that the obtained BiOCl-NSs are octagonal shaped with sizes between 600 and 800 nm and thicknesses of ~ 150 nm. The high-resolution inverse fast Fourier transformation transmission electron microscopy (IFFT-TEM) image perpendicular to the nanosheet plane ( Fig. S3c) reveals lattice spacings of 2.75 and 2.75 Å with an interplanar angle of 90°, corresponding to the (110) and (1͞ 10) facets of BiOCl. The selected area electron diffraction (SAED) pattern (Fig. S3d) coincides to the diffraction pattern of the single crystal BiOCl (k±l0, k = l = n) from [001] zone axis. The aberration-corrected high-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM) image and the corresponding energy-dispersive X-ray spectroscopy (EDX) elemental mapping images confirm the homogeneous distribution of Bi, O and Cl throughout the entire BiOCl-NSs (Fig. S3e).
The as-synthesized BiOCl-NSs were then immobilized onto a conductive carbon paper substrate (1.0 × 1.0 cm 2 ) with a loading density of 2.0 mg cm −2 (Fig. S4) and subject to − 0.8 V (vs RHE) for 2 h in CO 2 -saturated 0.5 M KHCO 3 solution to electrochemically convert the loaded BiOCl-NSs into Bi 2 O 2 (CO 3 ) x Cl y . The XRD pattern (Fig. 1b) of the resultant Bi 2 O 2 (CO 3 ) x Cl y can be assigned to the tetragonal phased Bi 2 O 2 CO 3 (PDF No. 41-1488). The Raman spectrum (Fig. 1c) displays two strong peaks at 163 and 1068 cm −1 , attributing to the external vibration of Bi 2 O 2 CO 3 crystal and the ν 1 mode of the intercalated CO 3 2− between the (BiO) 2 2+ planes [28,29]. Raman peak at 182 cm −1 could be assigned to the A 1g mode of the intercalated Cl − in the interlayer [29,30]. The FE-SEM and AFM images (Figs. 1d and S5) unveil that Bi 2 O 2 (CO 3 ) x Cl y possesses a sheeted structure with lateral sizes of 600-800 nm and thicknesses of 130-140 nm.
The TEM image (Fig. 1e) shows that Bi 2 O 2 (CO 3 ) x Cl y is formed by multiple thin-layer structures with "doughnutslike" shape, resulting from the substitution of chloride by carbonate. The SAED pattern normal to the nanosheets (inset of Fig. 1e) manifests the reflections of Bi 2 O 2 CO 3 (k00 and 0l0, k = l = n) with [001] zone axis. The high-resolution TEM image (HRTEM, Fig. 1f) displays a lattice spacing of 0.273 nm, corresponding to Bi 2 O 2 CO 3 (110) plane, which is also confirmed by the high-resolution IFFT-HRTEM image (Fig. 1g). The HAADF-STEM image and the corresponding EDX elemental mapping (Fig. 1h)  The X-ray photoelectron spectroscopy (XPS) analysis was then carried out. The high-resolution XPS Bi 4f spectra (Fig. 2a) confirm the presence of Bi 3+ and Bi-O  [31] in BiOCl. The Bi 3+ peaks of Bi 2 O 2 (CO 3 ) x Cl y show a negative shift of 0.26 eV, consistent with that of reported Bi 2 O 2 CO 3 [32]. Figure 2b shows the high-resolution XPS O 1s spectra of BiOCl and Bi 2 O 2 (CO 3 ) x Cl y . The former could be deconvoluted into the binding energy peaks assignable to the Bi-O lattice O (530.8 eV), the surface adsorbed hydroxyl (~ 531.9 eV) and O species in Nafion (536.3, 533.3 and 531.9 eV) [33,34], while the deconvoluted binding energy peaks at 530.2 and 531.0 eV from the later are ascribed to the Bi-O lattice O and C=O, respectively [35,36]. The lattice O peaks in Bi 2 O 2 (CO 3 ) x Cl y shifted to lower energies due to the substitution of chloride by carbonate. The two binding energy peaks at 199.2 and 200.8 eV assignable to Cl 2p 3/2 and Cl 2p 1/2 can be deconvoluted from the high-resolution XPS Cl 2p spectra of both BiOCl and Bi 2 O 2 (CO 3 ) x Cl y (Fig. 2c), indicating the presence of the lattice Cl − [37]. Notably, a Bi/Cl atomic ratio of 61.5:1 is determined from the XPS Cl 2p spectrum of Bi 2 O 2 (CO 3 ) x Cl y (Fig. S7), confirming the presence of chemically bonded Cl on the surface of Bi 2 O 2 (CO 3 ) x Cl y .
The X-ray absorption spectroscopy (XAS) measurements were then conducted to probe the electronic structure and local atomic environments. The O K-edge near-edge X-ray absorption fine structure (NEXAFS) spectra of BiOCl, Bi 2 O 2 (CO 3 ) x Cl y and reference samples are shown in Fig. 2d. The observed binding energy peaks at 532.0 and 537.0 eV from Bi 2 O 2 (CO 3 ) x Cl y are assignable to the hybridization of O 2p with Bi 6 s orbitals [38,39], while the binding energy peak at 534 eV corresponds to the π * C=O transition, indicating the presence of lattice carbonyl oxygen species [40]. The displayed binding energy peaks at 539.4 and 543.2 eV in the spectrum of Bi 2 O 2 (CO 3 ) x Cl y are ascribed to the non-equivalent σ * C-O bonds in the carboxylic group originated from the adsorbed carbonate [41]. Based on the C K-edge NEXAFS spectra of Bi 2 O 2 (CO 3 ) x Cl y and reference samples (Fig. 2e), the binding energy peak at 289.5 eV can be attributed to the σ * states of C-O [41], while the peaks at 297.2 and 299.8 eV are assignable to the σ * C = O resonances associated with the presence of carbonate species [42]. It is to note that the O K-edge and C K-edge NEX-AFS spectra obtained from Bi 2 O 2 CO 3 and Bi 2 O 2 (CO 3 ) x Cl y exhibit very similar characteristics, implying that the crystal structure of Bi 2 O 2 CO 3 in Bi 2 O 2 (CO 3 ) x Cl y is not noticeably altered by the presence of Cl − . According to the Bi L 3 -edge X-ray absorption near-edge structure (XANES) spectra (Fig. 2f), the same valence states of Bi 3+ exist in BiOCl, Bi 2 O 2 CO 3 and Bi 2 O 2 (CO 3 ) x Cl y . Figure 2g shows the k 2weighted Fourier transformed Bi L 3 -edge extended X-ray absorption fine structure (k 2 -weighted FT-EXAFS) spectra of Bi 2 O 2 CO 3 (Fig. 3a) are almost identical to that of the as-synthesized BiOCl (Fig. S1). The initial conversion of BiOCl to Bi 2 O 2 (CO 3 ) x Cl y occurs at E App = − 0.3 V as  To further elaborate the electrochemical conversion pathway, the operando potential-dependent Raman spectra of BiOCl immobilized on the carbon fiber paper were recorded under different cathodic potentials in CO 2 -saturated 0.5 M KHCO 3 solution (Fig. 3c). Under the OCP and E App ≤ − 0.3 V conditions, the recorded spectra are almost identical to that of the as-synthesized BiOCl-NSs (Fig. S2). With E App = − 0.4 V, the characteristic peaks of BiOCl at 143 ( A 1 1g ) and 199 cm −1 ( A 2 1g ) are markedly reduced and disappeared, respectively, which are accompanied by the appearance of a new peak at 182 cm −1 that might be assigned to the A 1g mode of the intercalated Cl − in the interlayer, although this could be complicated by the reduced crystal symmetry due to the disorder or free rotation of CO 3 2− in the interlayer [29,30]. These observed changes in the Raman spectrum signify the initial conversion of BiOCl to is rapidly decreased, while the peaks at 163 and 1068 cm −1 are evolved and intensified, which are likely due to the changes in the local atomic symmetry rather than the crystal cell parameters because of the unchanged operando XRD patterns within the same potential range (Fig. 3a)

DFT Calculations
It is known that electrocatalytic CO 2 RR to HCOOH has normally proceeded via a proton-coupled electron transfer (PCET) step to form * COOH or * OCHO intermediates and followed by another PCET step to generate HCOOH [47]. It  x Cl y , corresponding to a LUMO potential of − 2.3 eV. It is known that for semiconductor electrodes, the reduction reaction takes place via the injection of electrons into LUMO. Therefore, the LUMO potential corresponds to the minimum required cathodic potential for electron injection. As illustrated in Fig. S28

Conclusions
In summary, we reported an approach to electrochemically convert bismuth oxychloride (BiOCl) into chloridecontaining bismuth subcarbonate (Bi 2 O 2 (CO 3 ) x Cl y ) under operando CO 2 RR conditions. We demonstrated that the operando synthesis is an effective strategy to enhance the electrochemical stability of bismuth-based electrocatalysts. The exemplified approach in this work could be widely applicable to enhance the electrochemical stabilities of other electrocatalysts for other reactions.