Effect of the oxidation state and morphology of SnOx-based electrocatalysts on the CO2 reduction reaction

CO2 electrochemical reduction reaction (CO2RR) is an attractive strategy for closing the anthropogenic carbon cycle and storing intermittent renewable energy. Tin-based electrocatalysts exhibit remarkable properties for reducing CO2 into HCOOH. However, the effects of morphology and oxidation state of tin-based electrocatalysts on the performance of CO2 reduction have not been well-described. We evaluate the oxidation state and particle size of SnOx for CO2 reduction. SnOx was effective for converting CO2 into formic acid, reaching a maximum selectivity of 69%. The SnO exhibited high activity for CO2RR compared to SnO2 electrocatalysts. A pre-reduction step of a SnO2 electrocatalyst increased its CO2 reduction performance, confirming that Sn2+ is more active than Sn4+ sites. The microsized SnO2 is more effective for converting CO2 into formic acid than nanosized SnO2, likely due to the impurities of nanosized SnO2. We illuminated the role played by both SnOx particle size and oxidation state on CO2RR performance.


Introduction
Global energy consumption is highly dependent upon fossil fuels, and climate models have shown that CO 2 emissions are inducing climate change due to the greenhouse effect [1,2]. The conversion of CO 2 into valuable chemicals and fuels by means of electrochemical reduction could solve both the environmental and energy crises [2][3][4]. This process carries several advantages, such as low-temperature operation, that it can be run at ambient pressure, and the required energy input can be supplied from renewable energy sources (i.e., solar or wind), creating a netzero CO 2 emission condition in certain energy business cases [5][6][7]. Moreover, the performance and selectivity of such an electrochemical reaction can be tuned, and the scale-up of this process becomes simpler than other such as: photochemical and thermochemical process [8][9][10].
The CO 2 reduction reaction (CO 2 RR) is a multielectron process that may proceed via different reaction pathways, yielding diverse reduction products, such as CO and HCOOH (2 electrons), CH 3 OH (6 electrons), CH 4 (8 electrons), C 2 H 4 (12 electrons) depending on the electrocatalysts and experimental conditions [11]. Formic acid has been receiving significant attention as an CO 2 RR product due to its stability, remarkably high volumetric capacity, and its versatile potential use in various applications (e.g., direct formic acid fuel cells, and the leather, textile, food, and chemical industries) [12,13]. The economic viability of various chemicals from the CO 2 RR demonstrated that formic acid has a great business value, which is one of the most desired products [10]. However, the inertness of CO 2 due to its high chemical stability results in a process with high overpotential, sluggish kinetics, and broad distribution of products [3,14,15]. The physical-chemical properties of an electrocatalyst, such as morphology [16], chemical state [17][18][19], and surface features strongly depend on its CO 2 RR performance. Therefore, rationalizing the effect of each catalyst's properties on CO 2 reduction is especially important in the development Article of new kind of materials to overcome the major challenges in this field [20].
Among the materials for converting CO 2 into formic acid, tin (Sn)-based ones exhibit remarkable features, such as good selectivity, low costs, and nontoxicity [21][22][23][24]. However, bare Sn planar electrodes present a current density of − 5 mA cm −2 with 80% Faradaic efficiency and an overpotential of almost 0.90 V, which is too low for practical applications [25]. The role played by metal semiconducting oxide, oxidation state and morphology whether as catalysts for the formation of oxygenates, still remains unclear [18,26,27]. Some researchers have shown that a layer of metal oxide on the catalyst surface can decrease the reaction overpotential and increase the performance of the CO 2 RR [18,[26][27][28][29]. Additionally, Kanan et al. [28] demonstrated that the removal of the SnO x native layer from an Sn electrode results in near exclusive H 2 evolution activity. It was also demonstrated that CO 2 RR performance of SnO x -based electrocatalyst decreased after Sn 4+ reduction to Sn 0 , they verified the occurrence of three processes: CO 2 reduction into formic acid, hydrogen evolution reaction and the reduction of the SnO 2 catalyst which yields tin species of lower oxidation number (that is, Sn 0 and probably Sn 2+ species as well) [27]. Additionally, at moderately cathodic potentials, SnO 2 exhibited high selectivity for the production of formate, while at very negative potentials it was observed the reduction of oxide to Sn, and the efficiency of formate production was significantly decreased [27]. Therefore, Sn-based catalysts showed low stability during long periods of operation, as the reduction of SnO x species in Sn metallic decreases its activity [28,30]. However, it has also been found that the deposition of electrocatalysts on carbon paper can increase its performance and stability [31,32]. Nevertheless, SnO x -based nanoparticles deposited on carbon paper have thus far been little explored for CO 2 reduction applications.
This manuscript reports on a series of evaluations of the effect of tin oxidation states, as well as the particle size of SnO x deposited on carbon fiber on CO 2 RR activity and selectivity. Additionally, the electrocatalyst's stability was also assessed by means of physicochemical characterization of the electrodes before and after the CO 2 RR experiments. Finally, the influence of the reduction potential was evaluated in terms of CO 2 RR performance and the stability of SnO x -based materials.

Results and discussion
XRD patterns of the SnO x -based electrodes ( Fig. 1) were obtained to confirm the presence of SnO x material and crystalline phase on carbon paper electrode, for comparison purposes the XRD patterns of the pristine materials were also obtained ( Figure S1). All of the electrodes exhibited an identical crystalline phase before and after deposition onto carbon fiber, indicating that the method proposed was effective for impregnating the SnO x materials. All of the electrodes exhibited a broad peak at about 26° that could be related to the carbon fiber [13]. SnO (PDF2: 01-172-1012) exhibited small peaks due to the small amount in the carbon fiber (18 mg); however, its characteristic peaks related to the plane (101) could be observed. Microsized SnO 2 exhibited all peaks in a well-defined way; on the other hand, as expected, nanometric SnO 2 exhibited small and broad peaks due to its small particle size [33]. In addition, the main peak from SnO 2 , i.e., the (101) plane was overlapped by a carbon fiber peak of about 26º. The material's crystallite size was estimated from the XRD pattern using Scherrer equation (Table  SI). It can be seen that SnO and SnO 2 microsized exhibited a crystallite size value of ca. 43 and 45 nm, respectively. On the other hand, the SnO 2 nanosized exhibited a crystallite size of ca. 5 nm. The results are in agreement with the material's morphology. Therefore, the XRD results confirmed that the electrode fabrication method was efficient to obtain the SnO x electrodes without crystalline phase modification.
SEM images of the electrodes revealed that SnO x particles were successfully deposited on the carbon fiber, as can be seen in Figure S2. The formation of agglomerated spherical particles attached to the carbon fibers can also be observed, such a feature becomes deleterious in catalytical applications [34,35]. However, it can be seen that the carbon fiber was not entirely covered by the SnO x particles, and this feature could aid in CO 2 reduction processes using oxide-based materials. It was demonstrated that a complete recovery of carbon fiber or electrode by a semiconductor can be deleterious for electrocatalytic performance, as it increases the electrode resistance, thus hampering the charge transfer processes [13]. Nanosized SnO 2 electrode seems more homogeneous than the microsized SnO 2 and SnO, it is likely related to the lower particle size and higher specific surface area that will be easier spread on the carbon fiber. Additionally, the specific surface area of the pristine materials was determined Article applying the BET model to N 2 physisorption data. SnO 2 nanosized material exhibited a SSA value at least 30 times higher than the SnO x microsized materials, as can be seen in the Table  SI. This result is in agreement with the materials morphologies. Cyclic voltammetry was performed on the SnO x -based electrodes to identify the reduction potential of the CO 2 reduction reaction, the hydrogen evolution reaction, and for SnO x redox (Fig. 2). All of the electrodes exhibited two coinciding anodic peak in the potential region between 0.1 and − 0.1 V, and this was associated with the oxidation of Sn metallic to Sn 2+ and Sn +2 to Sn +4 , even in SnO 2 electrodes, which is likely due to in situ reduction of Sn 4+ to Sn 2+ and Sn metallic during cathodic sweeps. The enlarged cyclic voltammetry of microsized SnO 2 exhibited two anodic peaks at − 0.13 V and + 0.01 V ( Figure  S3a), while the microsized SnO exhibited two peaks at − 0.06 V and + 0.05 V ( Figure S3b). These peaks can be attributed to two different oxidation steps, the first one could be related to the oxidation of Sn to Sn 2+ , while the second one could be related to the oxidation of Sn 2+ to Sn 4+ [21,22,36,37].
The micrometric SnO 2 exhibited the highest current density at − 1.0 V, followed by the SnO and nanometric SnO 2 electrode, respectively. It was unexpected, because the material with a lower particle size and higher specific surface area should exhibit the higher electrochemical active surface area and, consequently, a higher current density. The reasons of the low activity of nanosized SnO 2 was investigated by X-ray photoelectron spectroscopy ( Figure S4). The survey spectrum confirms the presence of Sn, O, C, and Cl. Note that, the presence of the elements Sn and O is in accordance with the composition and chemical state of the SnO 2 nanosized. The peaks related to carbon are from adventitious contamination commonly used as a charge reference for XPS spectra. The peak at 199.0 eV can be attributed to Cl 2p, which is an impurity derived from the precursor salt (SnCl 2 .2H 2 O) employed to synthesize the SnO 2 . Additionally, the FTIR spectrum ( Figure S5) demonstrated the presence of some impurities related to carbonic groups (-CH). Therefore, the low performance of nanosized SnO 2 can in principle be attributed to the presence of these impurities. On the other hand, the sample with the highest cathodic current does not necessarily achieve greater efficiency in CO 2 reduction, as the CO 2 RR competes with the H 2 evolution reaction and electrode redox reactions.
To confirm the CO 2 RR performance of the SnO x -based electrodes, electrolysis experiments were carried out under galvanostatic conditions (Fig. 3), applying − 50 mA (or − 6.25 mA/ cm 2 ) for 30 min. The microsized SnO 2 electrocatalyst required the lower potential to maintain a constant current (of ca. − 0.7 V, followed by SnO (of ca. − 1.0 V) and nanosized SnO 2 (of ca. − 1.4 V) in the electrocatalysts. Therefore, nanometric SnO 2 requires much more power consumption to keep the current constant, which is likely due to the presence of impurities on its surface as demonstrated by FTIR and XPS analyses ( Figure  S4 and S5).
The HPLC results showed that formic acid was the only product in the liquid phase ( Table 1). The SnO electrode   exhibited the highest performance in the HCOO − formation, followed by nanosized SnO 2 and microsized SnO 2 . The uncoated fibers did not exhibit any catalytic activity in this reaction, indicating that SnO x was the catalytically active site for CO 2 RR. The different performances achieved by the SnO x electrocatalysts indicate that the oxidation state of Sn, as well as the particle size, were both responsible for achieving significant CO 2 RR performance profiles. The results suggest that the lower performance of SnO 2 electrodes compared to those of SnO can be related to the Sn 4+ to Sn 2+ competition reaction, as observed by CV analysis (Fig. 2). Nevertheless, the nanosized SnO 2 requires a reduction potential of approximately − 1.4 V in order to keep the current constant, whereas the microsized SnO 2 requires a reduction potential of about − 0.8 V. Therefore, the faradaic efficiency for nanosized SnO 2 could be higher due to the increase in the reduction potential, and not only because of the particle size, since it is well known that reduction potential influences directly on CO 2 reduction activity and selectivity. The Latimer diagram of Sn species around pH 7 showed that Sn 4+ oxide should be reduced into Sn 2+ oxide in a reduction potential around − 0.17 V vs RHE, while the Sn 2+ oxide should be reduced into Sn metallic in a reduction potential around − 0.47 V vs RHE [27]. However, Dutta et al. [27] described the SnO x 's species stability under different electrochemical conditions, which was based on Pourbaix Diagram of Sn species and in operando Raman spectroscopy. They showed that the conversion of SnO 2 to metallic Sn requires more negative potentials than what could be predicted based on thermodynamic data, because the reduction of the SnO 2 is kinetically hindered. Therefore, both microsized SnO 2 and SnO are stable under galvanostatic conditions studied; however, the nanosized SnO 2 required a very high potential around − 1.4 V vs RHE to keep the − 50 mA. This catalyst should be partially converted into Sn 2+ oxide and Sn metallic, it explains why the nanosized SnO 2 exhibited a higher faradaic efficiency than the microsized material.
To further understand the effects of particle size and reduction potential, we evaluated CO 2 RR performance in potentiostatic conditions, i.e., by applying − 0.8 V over 3 h for nanosized and microsized SnO 2 . It was observed that the microsized SnO 2 produced 789 ppm of formic acid, with a faradaic efficiency of 56%, whereas the nanosized SnO 2 achieved 21 ppm and an FE of 22%. Furthermore, microsized SnO 2 exhibited superior activity and selectivity for formic acid production. Again, this could reflect the impurities on the surface of nanosized SnO 2 electrocatalysts. This finding confirms the last experiment observation (galvanostatic), the reduction potential difference was the key to boosting the performance of the nanosized SnO 2 .
In this step, the effect of the oxidation state of tin on the CO 2 RR performance was evaluated. To attain this objective, we carried out a pre-reduction step in the SnO 2 electrodes by applying − 50 mA for 30 min under Ar purge in the same electrochemical cell. Then, the CO 2 RR performance was evaluated under potentiostatic conditions, applying − 1.0 V for 2.5 h (Fig. 4). The obtained results were then compared with the performance of SnO 2 electrodes without any prereduction treatment ( Table 2). It can be observed that the performance of SnO 2 electrocatalysts was enhanced following the pre-reduction step, and therefore this finding confirmed that the reduction of Sn 4+ can compete with the CO 2 reduction and decrease the FE of formic acid formation. Additionally, we can confirm that Sn 4+ is not the active site of the CO 2 reduction reaction. This corresponds to previously published data, for example, Lee et al. [38] showed that, under neutral conditions, metallic Sn is most likely the active site of CO 2 RR. On the other hand, some authors have found that the Sn electrode possesses an oxide layer and shows excellent catalytic activity for CO 2 reduction, and that the Sn electrode removed its oxide layer, resulting in poor catalytic activity for CO 2 reduction, but with the hydrogen evolution reaction accelerating [23]. The results reported by Baruch et al. suggest that the active species for catalysis is an Sn 2+ species rather than an Sn 4+ one, as the first species can react with CO 2 to form the surfacebound carbonate, which is the intermediate products in the reduction of CO 2 to formate (Scheme S1). The first step of the reduction of CO 2 is preceded by a two-electron reduction of   [22]. It was demonstrated that CO 2 reduction overpotential is decreased by lowering the free energy of formation of carbonate [39]. These finding were confirmed by electrochemical experiments and DFT calculations [40]. Figure 4 shows a microsized SnO 2 electrocatalyst with a higher current density, production, and FE for formic acid formation than nanometric SnO 2 . The CO 2 RR performance was evaluated at four different reduction potentials (− 0.2, − 0.6, − 0.8 and − 1.0 V), the polarization curves were obtained by average the current over a time period of 9000 s to each point, as is displayed in Fig. 5. It can be observed that when the reduction potential increased, the FE for formic acid formation was also enhanced. The increase in the reduction potential not only enhanced the current density/amount of formic acid formed but also increased the selectivity of the microsized SnO 2 electrocatalyst. This behavior is due to the H 2 evolution reaction, which exhibited a low reduction potential compared to that of the CO 2 RR to formic acid [25].
The stability of the SnO x -based electrodes was evaluated by means of two cycles of reuse under galvanostatic conditions, applying − 50 mA during 30 min (Table 1). It was verified for all SnO x -based electrodes that, after the first cycle, the CO 2 RR performance was increased. It can be observed that the increase in performance of the SnO 2 samples after the first cycle was more pronounced than for the SnO sample, likely due to the competition of the Sn 4+ /Sn 2+ reduction with the CO 2 RR in the first cycle, this reaction may lead to a loss of the selectivity for formate production. This result corresponds to those observed in the pre-reduction experiment, i.e., after partial reduction of the SnO 2 electrodes, the catalytic activity was increased. To confirm the presence of the SnO 2 sample in the carbon fiber surface, even after two cycles of the CO 2 RR experiment, an XRD and SEM analysis were performed on the micrometric SnO 2 (Fig. 6). The former showed that the crystalline SnO 2 phase did not change during the CO 2 RR experiment; however, it was clear that the SnO 2 peaks decreased after the CO 2 RR experiment. This peak decrease can be related to the Sn 4+ reduction and/or its leaching into the solution. The SEM images showed that the morphology of microsized SnO 2 did not change during the CO 2 RR; however, it could be verified that the SnO 2 particles became less agglomerated and more uniformly dispersed on the carbon fibers, which could relate to the surface reconstruction during CO 2 RR operation, as observed in other studies [41][42][43]. Therefore, the enhanced CO 2 RR performance after the first cycle could also be related to the better distribution of the SnO 2 particles across the carbon fiber surface.

Conclusions
A simple, easy, and scalable method for impregnating SnO x in carbon fibers was proposed in this study. All SnO x electrodes were capable of converting CO 2 into formic acid, with a maximum FE of almost 70% for the SnO 2 catalyst. On the other hand, carbon fiber without tin oxide was not effective for the CO 2 reduction reaction. It was demonstrated that both the particle size and oxidation state played an important role in the CO 2 RR performance. The SnO electrocatalyst exhibited high activity for CO 2 RR compared to SnO 2 electrocatalysts, likely due to the requirement of reducing Sn 4+ . Nanosized SnO 2 exhibited a low CO 2 reduction performance due to the impurities on this material surface, as demonstrated by FTIR and XPS analysis. The pre-reduction step for both SnO 2 electrodes increased their CO 2 RR performance, indicating that Sn 2+ is the active site under potentiostatic conditions, and the microsized SnO 2 electrode showed greater efficiency in the generation of formic acid. The CO 2 reduction mechanism on the SnO x surface was proposed based on our results and in the literature. The increase in the reduction potential increased the CO 2 RR conversion of formic acid selectivity of the micrometric SnO 2 electrocatalyst. The results showed that the oxide remained on the surface, and therefore it constitutes a stable surface for reactions.

Methodology Preparation and characterization of SnO x -based electrodes
A general procedure was used to deposit the SnO x catalyst ink on the carbon fibers. To obtain the catalyst ink, 18 mg of SnO x with Nafion (5% w/w) was dispersed in 2 mL of isopropanol, using sonication. The catalyst/Nafion ratio was 90:10 (wt:wt), respectively. The mixture was dried in a fume hood and then deposited to the carbon fiber electrode (area of 8 cm 2 ) by means of brushing. The electrodes were then cured at 80 °C for 1 h. Three different SnO x -based materials were evaluated: (i) microsized SnO 2 (99.9%, Sigma-Aldrich); (ii) microsized SnO (99.99%, Sigma-Aldrich); and (iii) nanosized SnO 2 that was prepared by means of hydrothermal treatment, as described elsewhere [33]. In detail, 0.564 g of SnCl 2 .2H 2 O (Sigma-Aldrich) was dissolved in 100 mL of ethanol anhydrous (99.5% purity, Dinâmica) and then under vigorous stirring, 22.5 mL of distilled water was added dropwise at room temperature for 12 h. After that, the suspension was cleaned by dialysis in 5 L of distilled water, often renewed, until removing all chloride ions, which was confirmed by tests with AgNO 3 solution (0.1 mol L −1 ). Then, the solution inside the dialysis bag was dried at 70 °C for 12 h, ground with an agate mortar. After that, the as-prepared sample was treated hydrothermally up to 200 °C for 4 h. The concentrations of the SnO x investigated (2.2 mg cm −2 ) were chosen on the basis of previous reports [26,44,45]. The samples were characterized by X-ray diffraction (XRD) at 2θ = 10° to 70°, using a monochromatized X-ray beam from nickel-filtered Cu kα radiation (λ = 0.15406 nm, 30 kV, 30 mA-Rigaku Multiflex-Ultima IV). The morphological properties of the samples were characterized using a scanning electron microscope (SEM, JEOL JSM 670F) operating at 5 kV (secondary electron detector) at different magnifications. A Fourier Transform Infrared spectrometer (FTIR) (Bruker VER-TEX 70) was used to investigate surface changes using nanosized SnO 2 -containing KBr disks with 64 scans and 4 cm −1 resolution in the 4000-400 cm −1 . X-ray photoelectron spectroscopy (XPS) analyses were performed on a Scienta Omicron, model ESCA + spectrometer using monochromatic AlKα (1486.6 eV) radiation. Peak decomposition was performed using a Gaussian-Lorentzian line shape with a Shirley nonlinear sigmoidtype baseline. The binding energies were corrected for charging effects by assigning a value of 284.8 eV to the adventitious C 1s line. The data were analyzed using CasaXPS software (Casa Software Ltd., U.K.). N 2 adsorption-desorption isotherms were recorded on a Micromeritics ASAP 2020 analyzer at 77 K. Samples were previously degassed at 80 °C under vacuum until a degassing pressure, 10 μmHg. The Brunauer-Emmett-Teller (BET) method was used to calculate the SSA (S BET ).

Evaluation of CO 2 reduction performance
The CO 2 reduction reaction experiments were studied in a two-chamber electrochemical cell, separated by a proton exchange membrane (Nafion 117) in an aqueous KHCO 3 (0.5 mol L −1 ) solution. The electrolyte was saturated with CO 2 (pH 6.9) for 30 min prior to each measurement. The SnO x -coated carbon fiber electrode was used as the working electrode, and a Pt mesh and Ag/AgCl electrode were used as the counter and reference electrodes, respectively. All of the potentials presented herein were converted to the RHE reference scale using the following equation: E (vs RHE) = E (vs Ag/ AgCl) + 0.210 V + 0.0591*pH. The electrochemical measurements were carried out using a Potentiostat/Galvanostat (Model 273, Princeton Applied Research).
These cyclic voltammetric measurements were performed from 1.1 to − 1.0 V. The measurements were conducted in order to verify the regions where the reactions of interest occurred. The performance of the SnO x -based catalysts was evaluated under galvanostatic conditions, applying a current of − 0.05 A for 30 min. The stability of the catalysts was evaluated by two reuse cycles of galvanostatic experiments. The species produced by the reduction of CO 2 in the liquid phase were analyzed by high-performance liquid chromatography (HPLC) in order to calculate the FE (%) of the process. The sample was collected in the cathode compartment of the electrochemical cell following the CO 2 reduction reaction and injected into the chromatograph (HPLC-LC-20AD, Shimadzu) with an Aminex HPX-87H column (300 × 7.8 mm) capable of analyzing carboxylic acids and alcohols. A diluted solution of H 2 SO 4 (3.3 mmol L −1 ) was used as the mobile phase with a flow rate of 0.6 mL min −1 . A sample volume of 20 μL was injected into the column loop. The column and detectors were kept at 40 °C. The chromatograph was equipped with a differential refractive index detector (RID-10A) that is suitable for the detection of alcohols, and a UV-Vis detector (SPD-20A, deuterium lamp, λ = 210 nm), which is suitable for the analysis of carboxylic acids. Data obtained by HPLC were used to calculate the Faradaic efficiency (FE) of the CO 2 electrochemical reduction in relation to formation of formate (HCOO − ). The Faradaic efficiency (FE) of the CO 2 electrochemical reduction of products was calculated as where n represents the number of electrons transferred from CO 2 for the production of one molecule of formate, F is the Faraday constant, n HCOO-is the number of moles of the formate, and Q is the total charge. In this case, n = 2; F is Faraday's constant (96 485 C mol −1 of electrons); n HCOO-is calculated by HPLC measurements. To evaluate the effect of Sn oxidation state, it was performed a pre-reduction step in the same electrochemical cell in both SnO 2 nanosized and microsized, i.e., a current of − 0.05 A was applied for 30 min under an Ar purge in order to reduce the species of Sn 4+ ; then, an experiment was performed under potentiostatic conditions, with − 0.8 V applied for 2.5 h. The performance of both electrodes was compared with and without this pre-treatment, to investigate the Sn site active. Additionally, the effect of the reduction potential (− 0.2 to − 1.0 V) on the FE of the microsized SnO 2 was also analyzed after the same pre-reduction step.
as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.