Introduction
Intermittent energy sources such as solar and wind require storage if they are to maximize their utilization while ensuring grid stability.[Reference Dunn, Kamath and Tarascon1, Reference Lewis2] The carbon dioxide reduction reaction (CO2RR) has the potential to store renewable energy in the form of chemical bonds through the conversion of CO2 and H2O to fuels or chemicals.[Reference Aresta3, Reference Whipple and Kenis4] This route could yield transportation fuels compatible with our current infrastructure with a CO2–neutral cycle or even offer an alternative method for chemical synthesis, sequestering CO2 in the process. Solar energy could be harnessed for converting CO2 directly via photoelectrochemistry or thermochemistry, and any form of renewable electricity could be used for electrochemical CO2 conversion. In any of these processes, catalysts for the CO2RR will be needed to maximize energetic efficiency to compete with other sources of fuel or energy storage. The reduction of CO2 in the presence of protons can yield a wide variety of products, and catalysts are also necessary to direct the reaction to desired products with high selectivity and current densities.[Reference Whipple and Kenis4, Reference Jitaru, Lowy, Toma, Toma and Oniciu5] Current heterogeneous catalysts can convert CO2 to formate or CO with >90% Faradaic efficiencies, or to further reduced species such as methane, ethylene, and methanol with moderate Faradaic efficiencies, but the catalysts typically require high overpotentials, leading to an energetically inefficient process.[Reference Hori, Wakebe, Tsukamoto and Koga6, Reference Kuhl, Cave, Abram and Jaramillo7] Strategies for developing heterogeneous catalysts with improved activity include nanostructuring,[Reference Kauffman, Alfonso, Matranga, Qian and Jin8, Reference Chen, Li and Kanan9] alloying,[Reference Watanabe, Shibata, Katoh, Sakata and Azuma10] using ionic liquids,[Reference Rosen, Salehi-Khojin, Thorson, Zhu, Whipple, Kenis and Masel11] and adding chemical modifiers to metal surfaces.[Reference Ogura, Endo and Nakayama12–Reference Zhang, Zhang, Lu, Wallace and Chen15]
This study initially explores the CO2RR on polycrystalline platinum (Pt) surfaces, and subsequently focuses on the effects of chemically modifying Pt surfaces by means of a thin polyaniline (PAni) film, which contains amine groups that are known to interact with CO2 and could potentially play a role in reaction activity and/or selectivity.[Reference Ogura, Endo and Nakayama12, Reference Aydin and Koleli14, Reference Rochelle16] More generally, this work explores the impact of functionalizing metal surfaces with organic groups, as suggested previously as a possible means to improve CO2RR catalysts.[Reference Peterson and Nørskov17] First, we give a detailed study of a polycrystalline Pt metal catalyst for the electrochemical CO2RR as a function of potential with sensitive gas and liquid product analysis. Then, a thin PAni film is electrodeposited on this Pt surface in an effort to tune the product distribution or lower the overpotential required for electrochemical CO2 reduction. PAni was chosen as the chemical modifier for several reasons: it contains amine groups that could interact with CO2 at the metal surface and possibly change the energetics of the reaction, it can be electrodeposited to form a conformal, porous film with tunable thickness, and it is highly stable in the explored reductive electrochemical conditions. The Pt foils performed similar to literature reports, with H2 as the dominant product with formate (<0.4%) as a minor product.[Reference Hori, Wakebe, Tsukamoto and Koga6] However, CO, methane, and methanol were also detected in small amounts (<0.2%), which were seen in two reports.[Reference Brisard, Camargo, Nart and Iwasita18, Reference Kuhl, Hatsukade, Cave, Abram, Kibsgaard and Jaramillo19] The PAni–Pt catalysts show noticeable increases in efficiency for formate (up to 1.5%) and CO (up to 0.8%) and a decrease in methanol production compared to the pure Pt metal catalyst. The reported enhancements in CO2RR activity from the use of a hybrid polymer–metal interface in the case of PAni–Pt suggest another class of surface structures to test for future CO2RR catalyst systems.
Methods
Surface characterization
Catalyst surfaces were characterized by x-ray photoelectron spectroscopy (XPS, PHI Versaprobe) with an AlKα source and by scanning electron microscopy (SEM, FEI XL30 Sirion). XPS and top-down SEM were conducted on Pt foils and on PAni films on Pt foils. Cross-sectional SEM was conducted on a separate sample consisting of 200 nm Pt on 50 nm Ti, deposited on a quartz substrate designed for cleaving.
PAni electrodeposition
PAni was synthesized electrochemically with a Pt foil working electrode (0.1 mm, Alfa Aesar, 99.99% metals basis) and graphite foil (0.13 mm thick, Alfa Aesar, 99.8% metals basis) counter electrode in the electrolysis cell with 1.5 cm × 3 cm exposed in an N2-saturated solution of 0.05 m aniline, 0.5 m H2SO4, and 0.5 m Na2SO4. The Pt was stored in 40% nitric acid and rinsed with copious amounts of water and dried with an N2 gun before PAni deposition. Aniline (99.8%, Acros) was distilled over Zn and stored in an air-tight vessel before being used for electrochemical deposition. Cyclic voltammetry was used for the deposition, employing a range of −0.015 to 0.710 V versus Ag/AgCl (Fisher Accumet, 13-620-53). The oxidative potential was chosen to provide a slow, repeatable thin-film growth and to prevent oxidative degradation of the PAni. The limits were chosen to grow a film of ~10 nm based on an ellipsometric study of PAni growth.[Reference Cruz and Ticianelli20]
Electrolysis
A custom electrochemical cell was used, based on a design described in detail in previous work.[Reference Kuhl, Cave, Abram and Jaramillo7] The custom electrochemical cell was designed with a fairly large ratio of electrode area (4.5 cm2) to electrolyte volume (8 mL) to enable improved sensitivity for liquid product detection.[Reference Kuhl, Cave, Abram and Jaramillo7] The cell was made of Kel-F and fitted with Teflon-coated silicone o-rings (McMaster-Carr) and was used for all CO2 electrolysis experiments and PAni electrodeposition (Fig. 1). Pt foils (0.1 mm, Alfa Aesar, 99.99% metals basis) were used for both the working electrode and counter electrode. The reference electrode employed was Ag/AgCl (Fisher Accumet, 13-620-53). For the Pt foil experiments, the Pt was electropolished for 5 min holding at a potential of 3.5 V versus a Pt wire in 10% nitric acid and then rinsed thoroughly with Millipore water before electrolysis. PAni–Pt catalysts were rinsed thoroughly with Millipore water before electrolysis. The Pt foil was stored in 40% nitric acid when not in use. The electrolyte used for testing was freshly made solutions of 0.1 m KHCO3 that were purged with CO2 at 20 sccm after passing through a bubbler to hydrate the CO2 and minimize solvent evaporation during electrolysis. All experiments were performed using a Biologic VMP3 potentiostat, and all CO2 reduction experiments used a channel with electrochemical impedance spectroscopy (EIS) capabilities. Potentiostatic EIS was used for determining the uncompensated solution resistance (R u). The electrolysis was conducted while compensating for 85% of this value, and the data reported are after post-correction of the last 15% of R u, with further description in the Supporting Material. All electrolysis potentials plotted and discussed are reported versus Reversible Hydrogen Electrode (RHE) unless otherwise stated. The 13CO2 experiment was run after purging the 0.1 m KHCO3 for 20 min on both the WE and CE side to replace the electrolyte with KH13CO3.
Electrolysis cell schematic for CO2 electrolysis with high surface area to solution-volume ratio.
Product analysis
Gas products from the working electrode compartment were carried into a 1 mL sample loop of a gas chromatograph (GC, SRI 8610C in Multi-Gas #3 configuration) by the 20 sccm CO2 stream and analyzed after 5, 23, 41, and 59 min of the 1 h electrolysis. N2 was the carrier gas, a thermal conductivity detector was used for H2 detection and a flame ionization detector was used for hydrocarbon detection, utilizing a methanizer for CO detection. A standard 1000 ppm gas mixture, as well as H2 generation on Pt were used for creating calibration curves for CO, CH4, C2H4, C2H6, and H2. Liquid products were detected using 1H nuclear magnetic resonance (NMR) spectroscopy, with further details about the technique described in the Supporting Material.
Results and discussion
CO2RR on Pt
The electrolysis was carried out potentiostatically, and the resulting current density versus time for several potentials for Pt is shown in Fig. 2. The potentials were chosen to explore a range of current densities from ~0.5 to 15 mA/cm2, which is a relevant range for fundamental studies as well as for solar fuels applications.[Reference Chen, Jaramillo, Deutsch, Kleiman-Shwarsctein, Forman, Gaillard, Garland, Takanabe, Heske, Sunkara, McFarland, Domen, Miller, Turner and Dinh21] The overall current densities between the Pt and PAni–Pt samples were comparable; thus, only those of Pt are shown in Fig. 2 for clarity. Results for the product distribution for the Pt foil are shown in Fig. 3, both in terms of Faradaic efficiencies [Fig. 3(a)], as well as a Tafel plot [Fig. 3(b)]. Each point is the average of three separate experiments, and partial current densities with error bars are shown in Fig. S4. No metal contaminants on the Pt foil were detected via XPS either before or after electrolysis (Fig. S2). The Pt foil performed similar to reports in the literature, producing close to 100% H2 for the entire range of potentials explored with small amounts of formate (<0.5%) being produced at potentials cathodic of −0.67 V. However, small amounts of CO and methanol and trace amounts of methane are also detected at potentials cathodic of −0.75 V. One previous study also detected methanol and traces of methane on Pt in acidic solution using an in-line mass spectrometer.[Reference Brisard, Camargo, Nart and Iwasita18] However, our studies provide the first detailed reports of quantified production rates of these species on Pt.[Reference Kuhl, Hatsukade, Cave, Abram, Kibsgaard and Jaramillo19]
Current density versus time for 1 h potentiostatic electrolysis on Pt in 0.1 m KHCO3. Potentials are versus RHE.
(a) Faradaic efficiency and (b) Tafel plots for 1 h electrolysis of CO2 on Pt foils in 0.1 m KHCO3.
PAni film deposition and characterization
The Pt foil was chemically modified with PAni via an electrochemical deposition using oxidative polymerization, and a characteristic cyclic voltammogram for deposition is shown in Fig. 4. Cyclic voltammetry was chosen to allow the film to expand and contract during growth to achieve well adhered, porous films, and the anodic potential of 0.7 V versus Ag/AgCl was chosen to finely control deposition rate with cycle number while avoiding polymer degradation. A total of 150 cycles were used under these conditions to achieve a film of ~10 nm based on an ellipsometric study of PAni electrodeposition.[Reference Cruz and Ticianelli20] The thin and porous geometry was chosen to allow for diffusion of CO2 and products in and out of the film to reach the Pt surface. SEM micrographs in Figs. 5(a) and 5(b) were taken top-down of PAni films on Pt foils before electrolysis, and the cross-sectional micrographs of Figs. 5(c) and 5(d) were collected on thin films of Pt that were deposited on Ti on quartz using an e-beam evaporator after cleaving the quartz in half for cross-sectional views. The same PAni deposition technique was used on the Pt/Ti/quartz samples for cross-sectional SEM as for the Pt foils used for electrolysis experiments, and the deposition cyclic voltammograms (CVs) for each were nearly identical. The SEM micrographs indicate a PAni film of the order of ~10–20 nm and uniform coverage of PAni.
150-cycle CV electrochemical polymerization of aniline on a Pt foil at 50 mV/s.
SEM micrographs of top-down view of (a) Pt and (b) PAni–Pt foils and cross-sectional views of (c) Pt and (d) PAni–Pt using thin films of evaporated Pt on Ti/quartz.
CO2RR on PAni–Pt
PAni films were electrodeposited on Pt foil, rinsed with Millipore water, and dried with an N2 gun before each electrolysis experiment. One-hour long CO2 electrolysis experiments were carried out on the PAni–Pt electrodes at the same potentials as for the Pt foil, with liquid and gaseous products detected by NMR and GC, respectively. The Faradaic efficiencies and partial current densities for each product are plotted in Fig. 6, and partial current densities with error bars are plotted in Fig. S4. In several respects, the Pt and PAni–Pt samples behaved similarly; they both exhibited similar total current densities as a function of potential as well as similar partial current densities for H2. Furthermore, both electrodes produced the same overall set of products: H2, CO, formate, methanol, and methane. Recently, we reported on the ability of Pt to electrochemically reduce CO2 to methane and methanol, albeit in small quantities.[Reference Kuhl, Hatsukade, Cave, Abram, Kibsgaard and Jaramillo19] Both methane and methanol were also observed on PAni–Pt, and in similarly small amounts near the detection limits of the experimental system. One final similarity worthy to note is that the overpotential needed to produce carbon products remained the same with or without PAni at the Pt surface, with an observed onset potential for CO at −0.57 V versus RHE for each catalyst system.
(a) Faradaic efficiency and (b) Tafel plots for 1 h electrolysis in 0.1 m KHCO3 on PAni–Pt.
Two major differences set apart PAni–Pt versus pure Pt as observed in Fig. 7: (1) the overall Faradaic efficiency for carbon-containing products, and (2) the quantitative distribution among carbon products. Not surprisingly, H2 is the dominant electrolysis product on both systems; however, Fig. 7 shows that compared with the bare Pt surface, PAni–Pt exhibits an increased fraction of carbon-containing products versus H2 by a factor of 2–4, depending on the applied potential. Controlling the selectivity of the CO2RR is an important goal in this field, including the ability to steer the reaction chemistry towards the reduction of CO2 and away from the evolution of H2, which can typically be produced more efficiently in the absence of CO2.[Reference Leiva, Santos and Iwasita22] The bulk of this increase in carbon-containing products arises from increased formate production at each potential, with a maximum of a fivefold increase in Faradaic efficiency from 0.3% to 1.5% observed at a potential of −0.82 V versus RHE. This value is significantly higher than what has been reported previously for formate production on Pt foils (e.g., 0.1% at 5 mA/cm2 total current).[Reference Hori, Wakebe, Tsukamoto and Koga6]Figure 7 also shows that increased CO production on PAni–Pt, particularly at the higher overpotentials, also contributes to the greater selectivity for CO2 reduction products versus H2.
Comparison of CO2RR Faradaic efficiencies on Pt and PAni–Pt.
Experiments clearly revealed an increase in carbon-containing products (particularly formate) on the PAni–Pt system compared with pure Pt; however, additional experiments were necessary to confirm whether the additional organic products were derived from PAni degradation or from enhanced CO2RR catalysis. Although PAni is known to degrade at oxidative potentials above approximately +0.95 V versus RHE,[Reference Cui, Su and Lee23] one reason PAni was chosen for this study was its expected aqueous electrochemical stability within the range of negative potentials of interest for the CO2RR. Two experiments were conducted to explicitly confirm the origins of the additional carbon products: 1H NMR using 13CO2 and XPS before and after electrolysis. Electrolysis at −0.82 V versus RHE was performed using 13CO2 to examine if the increased production of HCOO− arose from the 13C of CO2 or the 12C of PAni. If the 1H atoms of formate are bound to 12C, then a singlet peak would be observed at 8.32 ppm; however, if the 1H atoms of formate are instead bound to 13C, then 1H–13C spin–spin coupling would occur and beget a doublet in the NMR spectrum. The 1H NMR spectrum of the electrolyte in Fig. 8 shows a strong doublet indeed, with ~95% of the HCOO− coming from H13COO−. This indicates that the vast majority (if not, all) of the increased production of formate must have come from CO2RR catalysis and not from PAni degradation. Further evidence comes from the XPS spectra measured on a PAni–Pt electrode before and after electrolysis; the spectra (Fig. S3) show that the PAni is nearly identical in terms of chemical composition and chemical states, thus PAni degradation appears unlikely. Metal contaminants were not observed, either. Thus, the observed ~5% of 12C formate may have come from incomplete equilibration between the H12CO3− electrolyte and the flowing 13CO2, and/or the existence of 12CO2 impurities in the 13CO2 tank or gas manifold. In any event, the small ~5% magnitude of 12C formate observed in the isotope labeling experiments does not affect the primary conclusion of this study: the addition of PAni to a Pt surface perturbs the interfacial chemistry to increase CO2RR activity, particularly to formate.
1H NMR spectra showing peak splitting of formate during labeled 13CO2 reduction at −0.82 V on PAni–Pt.
Although there is a measurable increase in CO2RR activity on the PAni–Pt catalyst system versus pure Pt, the origins of this increase are not clear. In fact, it is not even clear as to how bare metal electrodes catalyze the electrochemical reduction of CO2 to formate or CO; each process requires two electrons to be transferred, and several mechanisms have been proposed for each pathway.[Reference Schouten24, Reference Zhang, Kang and Meyer25] For the PAni–Pt systems involved in this study, the XPS and SEM indicate a PAni film of the order of 10–20 nm on the Pt surface. It is unlikely that electron transfer takes place on the PAni itself as the PAni is expected to be in a non-conducting state, leucoemeraldine, the favored state under reaction conditions, at negative potentials and a pH of 6.8.[Reference MacDiarmid, Chiang and Richter26] Thus electron transfer likely takes place at the Pt electrode, with the PAni film affecting the reaction by means of chemical interactions with: (1) CO2 near the electrochemical interface, (2) adsorbed intermediates on the Pt surface, and/or (3) Pt surface sites.
Further studies are necessary to examine the reaction mechanism in greater detail, both with and without PAni thin films. Electronic and vibrational spectroscopies directed at determining bond distances and/or vibrational frequencies could help elucidate interactions between PAni and the reaction environment. For instance, probing the interaction between PAni and CO2, e.g., due either to an N–C interaction similar to CO2 capture systems[Reference Rochelle16] or to some of the proposed mechanisms for pyrdinium-based CO2 catalysis[Reference Morris, McGibbon and Bocarsly27, Reference Cole, Lakkaraju, Rampulla, Morris, Abelev and Bocarsly28] where hydrogen bonding with CO2 intermediates bound to the Pt surface[Reference Vogiatzis, Mavrandonakis, Klopper and Froudakis29] could play a role. Adsorbate–adsorbate interactions between CO2RR intermediates and PAni sites would be particularly interesting to explore. A greater understanding of these interactions could lead to further improvements in catalyst systems where chemical modifiers at the interface, such as PAni in this particular study, could have a strong effect in steering the reaction chemistry toward desired directions.
Conclusions
Pt surfaces were explored for catalyzing the CO2RR in aqueous, room temperature 0.1 m KCHO3 electrolytes with detailed product analysis during 1 h potentiostatic electrolysis experiments, providing new insights into CO2RR catalysis on this surface. Electrodepositing a thin, electrodeposited PAni film onto a Pt foil significantly modified the reaction selectivity, including a fivefold increase in formate production. Further studies to elucidate the cooperative nature between Pt and PAni could lead to new designs for improved catalyst systems.
Supplementary materials
For supplementary material for this article, please visit http://dx.doi.org/10.1557/mrc.2015.30
Acknowledgments
The authors acknowledge support from the Global Climate Energy Project at Stanford University as well as the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-SC0008685. D. N. A. acknowledges fellowship support from Chevron through the Stanford Graduate Fellowship (SGF) Program. E. R. C. and K. P. K. acknowledge fellowship support from the National Science Foundation Graduate Research Fellowship Program (NSF GRFP). The authors also acknowledge Ms. Michelle Vezie for technical assistance on this project.
