Achieving complete electrooxidation of ethanol by single atomic Rh decoration of Pt nanocubes.

SignificanceDirect ethanol fuel cells are attracting growing attention as portable power sources due to their advantages such as higher mass-energy density than hydrogen and less toxicity than methanol. However, it is challenging to achieve the complete electrooxidation to generate 12 electrons per ethanol, resulting in a low fuel utilization efficiency. This manuscript reports the complete ethanol electrooxidation by engineering efficient catalysts via single-atom modification. The combined electrochemical measurements, in situ characterization, and density functional theory calculations unravel synergistic effects of single Rh atoms and Pt nanocubes and identify reaction pathways leading to the selective C-C bond cleavage to oxidize ethanol to CO2. This study provides a unique single-atom approach to tune the activity and selectivity toward complicated electrocatalytic reactions.

Heterogeneous electrocatalysts have received ever-increasing attention owing to global issues such as climate change and energy supply. Since platinum group metals such as Pt, Pd, Ir, Rh, and Ru are particularly active in electrocatalysis, their development as catalysts has been fundamentally established by single crystals (1–6) and practically proceeded with alloying (7, 8), surface defect engineering (9, 10), and nanocrystal synthesis (7, 11). Specifically, single-atom catalysts (SACs) have been prepared by coordinating single metal atoms with carbon-based supports or by embedding them into the defects in metal oxides/metal-organic frameworks (MOFs) to prevent the clustering (12, 13). However, drawbacks remain that carbon-supported SACs are significantly deactivated during the irreversible oxidation of carbon (14), and some of the single atoms completely embedded in metal oxides/MOFs cannot be activated (13).

Meanwhile, single-atom alloy (SAA) catalysts have been developed in which the surface of metallic substrates is atomically alloyed by dispersing trace amounts of other metals. As an example, Rh–Pt SAA decorated on a PtBi surface promoted activity and stability toward the ethanol oxidation reaction (EOR) (15). In addition, density functional theory (DFT) calculations predicted that the Rh–Pt SAA boosted the C–C bond cleavage owing to the strain and ligand effects. However, such Rh–Pt SAA catalysts with bimetallic configurations have not achieved a high selectivity for the complete oxidation of ethanol to CO2, which is essential for ethanol fuel cells. In addition, the fundamental question has remained as to how an unalloyed, low-coordinated single metal atom decorated on a metallic substrate behaves as a catalyst.

Herein, we succeeded in the controlled synthesis of dispersing partially oxidized single Rh on the (100) surface of Pt nanocubes (RhatO-Pt NCs). As is known, the Pt(100) surface shows higher ethanol to CO2 conversion compared to Pt(111) and Pt(110), but it prefers partial oxidation to form acetic acid (CH3COOH) owing to the impeded kinetics of C–C bond cleavage (16). Therefore, EOR was tested for RhatO-Pt NCs to understand the unique role of RhatO in enhancing the oxidization of ethanol to CO2. Pt NCs decorated with partially oxidized Rh clusters (RhclO-Pt NCs), Pt NCs, and commercial Pt/C were also prepared for comparison. The catalyst prepared after loading RhatO-Pt NCs on a carbon support (RhatO-Pt NCs/C) demonstrated the highest EOR activity among all prepared electrocatalysts. At 0.75 V (versus reversible hydrogen electrode [RHE]), RhatO-Pt NCs/C showed 1.5-, 4.2-, and 11.4-fold higher current density than those of RhclO-Pt NCs/C, Pt NCs/C, and commercial Pt/C, respectively. Most importantly, the decoration of isolated RhatO rendered the RhatO-Pt NCs/C electrocatalyst capable to break the C–C bond of ethanol, resulting in >99.9% of CO2 selectivity in a record-low onset and wide potential region (0.35 to 0.75 V). In addition, in situ measurements, including infrared reflection absorption spectroscopy (IRRAS), X-ray absorption fine structure (XAFS), and DFT calculations, demonstrated that the single-atom Rh sites could facilitate the C–C bond cleavage and the removal of the *CO intermediate, promoting EOR performances. Our work demonstrated partially oxidized Rhat-O active sites as a unique SAC on shape-controlled nanocatalysts that can dramatically enhance the catalytic activity and selectivity compared with conventional nanocatalysts in complicated reactions such as EOR.

Results and Discussion

The RhatO-Pt NCs/C catalysts were prepared from a two-step solution method, with Pt NCs/C being prepared first as the substrate for further Rh decoration. During the synthesis of Pt NCs, CO gas was used directly instead of the common capping agent such as metal carbonyls to avoid the pollution of trace amount of metal (8). The transmission electron microscopy (TEM) image in showed the as-prepared Pt NCs with an average edge length of 10.8 ± 1.0 nm. These NCs can be uniformly dispersed on carbon to form the final electrocatlysts (Pt NCs/C, ). The atomic level decoration of Rh on Pt NCs/C was achieved by adding a trace amount of Rh precursors [Rh(III) acetylacetonate or Rh(acac)3] in the organic mixture containing Pt NCs/C (see Materials and Methods for details). In order to thermally decompose the metal–ligand bonds in Rh(acac)3, an annealing temperature of 250 °C was applied. In addition, this annealing condition was sufficient to prevent the clustering of Rh atoms because of the high surface free energy and the Rh–Rh interatomic bond energy (93 kJ ⋅ mol−1) (17, 18). Therefore, we expect that the thermally decomposed Rh atoms favor a single-atom distribution rather than cluster or particle growth on the surface of Pt NCs. By adding twice the amount of Rh precursor using a similar procedure, RhclO-Pt NCs were obtained. After the Rh decoration, the cubic morphologies were maintained with average edge lengths of 10.4 ± 1.2 and 10.9 ± 1.3 nm for RhatO-Pt NCs/C and RhclO-Pt NCs/C, respectively (). The same d-spacings of the Pt(200) plane (0.196 nm) were observed for RhatO-Pt NCs/C and RhclO-Pt NCs/C, suggesting the nonalloy formation during the decoration process (Fig. 1 ). The high-angle annular dark-field–scanning TEM (STEM) with energy dispersive X-ray spectroscopy (EDS) elemental mapping images of RhatO-Pt NCs/C and RhclO-Pt NCs/C represent the weak Rh signals in the outer layer of Pt NCs (white dots region in Fig. 1 ), suggesting that Rh was distributed on the surface of Pt NCs without the formation of the Rh shell. The O signals are evenly distributed over the single cubes, revealing the presence of oxygen in both samples. To further examine the decorated structure of RhclO-Pt NCs/C, STEM–EDS line scans were conducted at five different locations on a single RhclO-Pt NC (). As marked by the black arrows in , some EDS line profiles showed Rh signals at the edges, while others did not, indicating the presence of Rh clusters on Pt NC but not a complete coating of Rh shell. X-ray diffraction (XRD) patterns did not show any shift in the Pt peak position, suggesting that Pt and Rh did not form alloys in the Rh-modified Pt NCs/C samples (). The chemical states of Pt and Rh were identified by X‐ray photoelectron spectroscopy (XPS). Both Pt and Rh were in the oxidized form as indicated by the presence of Pt(II) in the deconvoluted 4f peaks and positively shifted Rh 3p peaks in both samples (). The quantitative Pt-to-Rh ratios of samples were determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES), suggesting stoichiometric composition of Rh1Pt19.8 and Rh1Pt11.4 for RhatO-Pt NCs/C and RhclO-Pt NCs/C, respectively (). Also, based on the ICP-AES results and the particle size of Pt NCs (), about 33% of the Pt surface was decorated with Rh atoms in RhatO-Pt NCs/C, but the Rhcl coverage was not quantified because of the clustering of Rh atoms on the Pt surface in the RhclO-Pt NCs/C sample.

Fig. 1.

High-angle annular dark-field STEM images and EDS mapping image of (A) RhatO-Pt NCs and (B) RhclO-Pt NCs. The k3-weighted Rh K-edge EXAFS spectra of (C) RhatO-Pt NCs/C and (D) RhclO-Pt NCs/C. Schematic models of (E) RhatO and (F) RhclO decorated Pt surface.

To further explore the coordination structure and the nature of Rh decoration in both samples, XAFS measurements were carried out. The X-ray absorption near edge structure (XANES) and Fourier-transformed extended XAFS (EXAFS) profiles of Pt () indicated the absence of Pt–Rh alloying signal in both samples (19). Different from the XPS results, the nearly identical XANES spectra of samples and the Pt foil revealed the metallic status of Pt in the bulk. Therefore, it could be concluded that only surface Pt was oxidized to a certain extent since XPS is a surface-sensitive technique. Unlike in the Pt L3-edge, the XANES profiles of Rh K-edge () represented the average Rh oxidation state of samples in the order of Rh foil < RhclO < RhatO < Rh2O3. EXAFS profiles at the Rh K-edge displayed distinguishable features from those of the Rh foil. The strong peak near ∼1.5 Å and the absence of the Rh–Pt peak in the EXAFS profile (Fig. 1 and ) were attributed to the presence of oxidized Rh atoms, indicating that Rh atoms were preferentially located on the aforementioned oxygen atoms on the Pt NC surfaces. In addition, there was no Rh–Rh peak for Rhat-Pt NCs/C, confirming the success of our synthetic approach for atomically dispersed Rh on the surface of Pt NCs. While RhatO-Pt NCs/C exclusively revealed the Rh–O–Pt bonding with its coordination number of 3.2(6) (), RhclO-Pt NCs/C displayed both Rh–O–Pt and metallic Rh–Rh bonding with their coordination numbers being 1.7(1) and 1.7(3), respectively (). The presence of the latter bonding could be verified by the peak near ∼2.6 Å (Fig. 1). These Rh coordination data are in good agreement with the Rh oxidation trend obtained by XANES. Based on the XAFS observations, expected surface models were illustrated for RhatO-Pt NC and RhclO-Pt NC (Fig. 1 ), showing that the RhatO-Pt NCs can provide high accessibility in an open structure.

The cyclic voltammetry (CV) test also revealed different surface properties of the prepared electrocatalysts (Fig. 2). The Pt(100) surface of Pt NCs/C was identified by the typical H adsorption/desorption features (0.08 to 0.45 V versus RHE) of single crystal Pt(100) shown in its CV curves (20). The distinct surface elemental distributions of RhatO-Pt NCs/C and RhclO-Pt NCs/C were represented by the reduction peaks of M–O (M = Pt or Rh). The cathodic CV curve of RhatO-Pt NCs/C showed the reduction of Pt-O (0.75 to 1.0 V) and Rh-O (0.45 to 0.65 V) peaks, confirming that both Pt and Rh were present on the surface. For RhclO-Pt NCs/C, the absence of an obvious reduction peak of Pt–O implied an Rh-dominated surface because of the decoration of Rh clusters.

Fig. 2.

(A) CV curves of the prepared electrocatalysts in Ar-saturated 0.1 M HClO4 solution. (B) Anodic polarization EOR curves for electrocatalysts in Ar-saturated 0.1 M HClO4 + 0.2 M ethanol solution with a scan rate of 50 mV ⋅ s−1. (C) Specific activity measured at different potentials of electrocatalysts. (D) The anodic CV curves of RhatO-Pt NCs/C and Pt/C before and after the stability test. The current densities are normalized to the electrochemical surface area.

The preferentially exposed Pt(100) facets of Pt NCs/C are active toward EOR, presented by the negatively shifted onset potential of 200 mV in comparison with Pt/C (Fig. 2 and ). However, the enhancement in terms of current densities was limited, with the current density of Pt NCs being only 2.7-fold higher than that of commercial Pt/C at 0.75 V. After Rh decoration, not only the EOR overpotentials further reduced but also the current densities showed significant enhancement. For example, compared to Pt/C, RhatO-Pt NCs/C showed more negative shifts of the EOR onset and main peak potentials by 350 and 85 mV, respectively (Fig. 2). Furthermore, the EOR current densities of RhatO-Pt NCs/C and RhclO-Pt NCs/C showed 11.4- and 7.7-fold enhancement over Pt/C at 0.75 V, respectively (Fig. 2). The chronoamperometric test also showed the same EOR activity trend among the electrocatalysts: Pt/C < Pt NCs/C < RhclO-Pt NCs/C < RhatO-Pt NCs/C at 0.65 V (). As for the decoration effect, RhatO is better for EOR than RhclO. To explore the reasons, Rh nanocubes/C (Rh NCs/C) and Rh NCs with decorated Pt clusters (Ptcl-Rh NCs/C) were prepared. Rh NCs/C was relatively inactive for EOR, but the EOR activity of PtclO-Rh NCs/C was enhanced by 83% after the deposition of Pt clusters, indicating that the EOR activity is mainly from Pt rather than Rh (). Then the activity enhancement from the RhatO and RhclO could be attributed to the Rh–O bonds, which might facilitate the formation of adsorbed OH species to remove the strongly adsorbed *CO during the EOR. Moreover, the RhatO single-atomic dispersion renders the corresponding catalyst the highest utilization of Rh–O active sites, making full use of Rh atoms. The stability of RhatO-Pt NCs/C was examined by a prolonged chronoamperometric test at 0.55 V for every 10,000 s followed by the CV cleaning scans (). After CV cleaning to remove the poisoned intermediates, the current density of RhatO-Pt NCs/C could be recovered during each chronoamperometric test. After 30,000 s, the maximum current density of RhatO-Pt NCs/C from CV curves only decreased by 8% (Fig. 2), showing a superior stability to Pt/C (decreased by 35%).

The advantage of RhatO decoration is more prominent for improving the ethanol fuel utilization by completely oxidizing ethanol to CO2. From in situ IRRAS measurements, four important intermediates and products were identified, CO (linearly adsorbed COL, 2,050 cm−1), CO2 (asymmetric stretching, 2,341 cm−1), CH3CHO (933 cm−1 as a representative peak), and CH3COOH (1,280 cm−1 as a representative peak) (Fig. 3 and ) (21). The onset potential for CO2 formation is an indicator of the ability of the electrocatalyst to break the C–C bond and the subsequent oxidation of adsorbed CO. The potential for the initial generation of CO2 over different electrocatalysts followed the order: RhatO-Pt NCs/C (0.35 V) < RhclO-Pt NCs/C (0.45 V) < Pt NCs/C (0.55 V) < commercial Pt/C (0.65 V). The CO2 generation potential of RhatO-Pt NCs/C was the closest to the thermodynamic potential for the oxidation of ethanol to CO2 (0.143 V) among the Pt-based electrocatalysts in the literature (), demonstrating its high activity to break the C–C bond. Besides, the peak located at 1,705 to 1,706 cm−1 represents the C = O stretching of C2 products (CH3CHO and CH3COOH). As shown in Fig. 3, even at 1.05 V, the IRRAS spectra of RhatO-Pt NCs/C still showed the negligible peak of 1,705 to 1,706 cm−1, suggesting that only a trace amount of C2 products were produced during the EOR. In addition, liquid products at the completion of electrolysis at 0.85 VRHE and 1.05 VRHE for 1 h were quantified by using high-performance liquid chromatography (HPLC) (see details in Materials and Methods). The Faradaic efficiency toward CH3COOH was presented in , showing a consistent trend with that from the IRRAS measurements.

Fig. 3.

Recorded in situ IRRAS spectra during CV test from 0.25 to 1.05 V on (A) RhatO-Pt NCs/C, (B) RhclO-Pt NCs/C, (C) Pt NCs/C, and (D) commercial Pt/C in 0.1 M HClO4 + 0.2 M ethanol solution. The red dot regions highlighted the projection files of the peak at 1,705 to 1,706 cm−1. (E) The calculated CO2 selectivity of all samples from 0.25 to 1.05 V.

The distribution of different EOR products were quantitatively determined from 0.35 to 1.05 V via the peak integration method developed by Weaver et al. (22) (Fig. 3 and ). Consistent with previous reports, the main product for Pt NCs/C and Pt/C was CH3COOH, and Pt NCs/C presented a higher CO2 selectivity than Pt/C (17.0 versus 2.4% at 1.05 V) (23, 24). It was noted that the CO2 selectivity of Pt NCs/C was much higher than the previous study using W(CO)6 to synthesize Pt NCs, confirming the importance of a clean Pt surface without trace amounts of W metal in EOR (7, 8). The Rh decoration remarkably improved the CO2 selectivity. The RhatO-Pt NCs/C catalyst exhibited >99.9% CO2 selectivity from 0.35 to 0.75 V, demonstrating its strong ability to break the C–C bond of ethanol in a wide potential range. We believe this is the first case that can achieve complete oxidation of ethanol to CO2 in such a wide potential range. The CO2 selectivity of RhatO-Pt NCs/C is still very high after 0.75 V, such as 32% of CO2 selectivity at 1.05 V. Consistent with the trend in EOR activity, the RhatO decoration is better than the RhclO on Pt NCs for the C–C bond scission in ethanol.

Specifically, the role of Rh in breaking the C–C bond was also evidenced by the in situ IRRAS spectra of Rh NCs/C and Ptcl-Rh NCs/C (). For Rh NCs/C, despite its low activity, a CO2 peak was generated at the beginning of the EOR (0.15 V) without the detection of adsorbed CO. While for Ptcl-Rh NCs/C, the CO2 peak was visible from 0.55 V, and the CO peak was formed from 0.35 V. Thus, the roles of Rh and Pt in EOR appear to be different, with Rh being the active component to break the C–C bond in ethanol and Pt as the active center for the overall EOR activity. Therefore, Rh–O–Pt can combine the activity of Rh and Pt and additionally assist the formation of surface OH to remove the poisoning *CO as confirmed in the CO stripping analysis (). The oxidation of CO began in the order of RhatO-Pt NCs/C, RhclO-Pt NCs/C, Pt NCs/C, and commercial Pt/C, presenting the higher capability of RhatO-Pt NCs/C to oxidized *CO. These results suggest the unique role of the single RhatO decoration on a shape-controlled electrocatalyst to effectively break the C–C bond and the subsequent oxidation of *CO.

Based on the electrochemical and in situ IRRAS results, the unprecedented EOR selectivity of RhatO-Pt NCs/C suggests a unique environment of RhatO on the Pt(100) surfaces as a new active site. In order to verify the catalysis behavior of RhatO-Pt NCs/C, in situ XAFS analysis was conducted at the Rh K-edge using the chronoamperometric mode of EOR at 0.75 V. When comparing the XANES profiles, the oxidation state of Rh was apparently reduced during the chronoamperometric measurements (Fig. 4). This result reveals that the reaction intermediates such as *CO and *CH3CH2CO donate their electrons to RhatO to form a sigma (σ) bond. The σ bond formation on the RhatO site was further confirmed in the EXAFS profiles by the onset of a new intense peak in the range of 1∼1.5 Å (Fig. 4). At the same time, the peak for the Rh–O bond shifted toward higher R values (2.00∼2.20 Å), indicating an extended Rh–O bond because of the partial reduction of Rh.

Fig. 4.

(A) XANES and (B) EXAFS spectra for the Rh K-edge of Rhat-O Pt NCs/C before, after, and during EOR chronoamperometric test in 0.1 M HClO4 + 1.0 M ethanol solution at 0.75 V. As shown in the linear combination fitting results of XANES in , the average oxidation state of Rh before and during EOR is approximately +1.7 and +1.0, respectively.

In general, such σ bond formation between Rh and intermediates is not surprising as observed in transition metal carbonyl complexes (25–27). In addition, a similar reaction configuration was observed that an intermediate Rh–C bond was formed in RhCo-MCM-41 during the reaction between CO2 and ethane to yield C3 products (28). However, in the present work, EXAFS signals surprisingly recovered back to the pristine profile upon the open circuit voltage condition after the chronoamperometric measurement, indicating the robust structural feature of the RhatO site on the Pt NCs over the prolonged EOR period. Especially, the removal of the intermediate Rh–C bond led to the shrinkage of the Rh–O bond to the original value (∼1.5 Å), which indicates that the RhatO site can adopt the reversible configuration of Rh–O and C–Rh–O bonds under pristine and reaction conditions, respectively, without significant energy penalties. Adzic et al. (21) previously investigated an efficient catalyst composed of Pt–Rh–SnO2 supported by carbon black (Pt–Rh–SnOx/C), which showed a beneficial role of SnO2 in strongly adsorbing water and interacting with Pt and Rh to form M–OH as well as the Pt–Rh alloy effect in promoting C–C bond cleavage. Our work provides a strategy by utilizing single-atom Rh to promote the complete EOR with optimized utilization of Rh without requiring the SnOx component.

In situ IRRAS and XAFS results confirmed that the partially oxidized RhatO sites serve as a robust catalytic active site for EOR and play an important role in the complete oxidation of ethanol. To gain further insight into the role of RhatO in EOR, DFT calculations were performed to calculate the free energy change (ΔG) of C–C bond cleavage of CH3CH2OH along various possible pathways and the activation energy for the removal of *CO. The DFT-calculated binding energies (BEs) of intermediates in their most stable configurations show that RhatO/Pt(100) enhances the adsorption of CH3CH2OH, which should lead to a facilitated rate of CH3CH2OH conversion (). However, most other intermediates bind weakly on RhatO/Pt(100) compared to Pt(100), leading to a higher tolerance of RhatO single sites toward these species. The DFT obtained results on RhatO/Pt(100) in Fig. 5 show that the C–C bond scission occurs when the *CH3CO intermediate is formed via three sequential dehydrogenation steps: *CH3CH2OH ⟶ *CH3CHOH + 1/2H2(g), *CH3CHOH ⟶ *CH3COH + 1/2H2(g), and *CH3COH ⟶ *CH3CO + 1/2H2(g). This pathway on RhatO/Pt(100) is similar to the DFT-predicted pathways on Rh surfaces by Hu et al. (29). However, the C–C bond cleavage differs from the pathway reported on Pt/Rh/SnO2 in which the C–C bond cleavage is found to be facilitated via the formation of the *CH2CH2O intermediate (21). A similar reaction pathway for the C–C bond cleavage is predicted on Pt(100) (). Fig. 5 also shows that the potential limiting step: *CH3CH2OH ⟶ *CH3CHOH + 1/2H2(g) along the most favorable pathways of C–C bond cleavage on RhatO/Pt(100) has a ΔG of 0.27 eV, which is comparable to a value of 0.14 eV predicted on Pt(100) ().

Fig. 5.

DFT-calculated free energy profile of C–C bond cleavage of CH3CH2OH on (A) RhatO/Pt(100) and (B) Rhat/Pt(100). As shown in the Bader charge analysis in , the average oxidation state of Rh in A and B is approximately +1.2 and +0.1, respectively.

Furthermore, CO binding is found to be significantly weakened on the RhatO/Pt(100) site compared to the sites on Pt(100) (). Additional transition state calculations were performed to compute the activation energy (Ea) of OH-assisted *CO oxidation to *CO2 (i.e., *HOCO formation: *CO + *OH ⟶ *HOCO followed by *HOCO dissociation: *HOCO +* ⟶ *CO2 + *H). The DFT-calculated Ea values of *HOCO formation and its dissociation were 0.72 and 0.24 eV, respectively, on RhatO/Pt(100), which were smaller than the corresponding values on Pt(100) of 0.92 and 0.94 eV, respectively. Therefore, consistent with the experimental observation, the DFT calculations predict the thermodynamically favorable C–C bond cleavage of *CH3CO to form *CH3 + *CO intermediates on RhatO/Pt(100) at moderate applied potentials as well as the facile removal of *CO, enhancing the complete oxidation of ethanol to CO2.

Because in situ XANES results (Fig. 4) suggested that Rh should be in the partially oxidized state during EOR, additional DFT calculations were performed on RhatO/Pt(100) and Rhat/Pt(100) surfaces with different Bader charges ( and Fig. 5). As shown in the linear combination fitting results of XANES in , the average oxidation state of Rh is reduced from +1.7 in the pristine sample to +1.0 during EOR, indicating that a partially oxidized Rh state is responsible for the high EOR activity. DFT results (Fig. 5) reveal similar C–C bond cleavage pathways over RhatO/Pt(100) and Rhat/Pt(100) surfaces and all the steps are predicted to be downhill in energy at moderate potential. Furthermore, the comparison of DFT results over surfaces with different Bader charges () illustrates that the C–C bond scission pathway, based on the difference in ΔG between the C–C bond scission of *CH3CO and its oxidation to acetic acid, becomes more favored as Rh is less oxidized, consistent with the experimental observation of C–C bond scission being preferred over acetic acid production.

Conclusions

In summary, we demonstrated that the decoration of single atomic RhatO on the metallic Pt(100) surface notably enhances the C–C cleavage and *CO oxidation to achieve completely oxidizing ethanol to CO2. The RhatO-Pt NCs/C presented 11.4-fold current density of Pt/C at 0.75 V and >99.9% of CO2 selectivity in a wide potential range of 0.35 to 0.75 V. Our studies revealed that the single-atomic RhatO sites could exploit particular advantages through the reversible configuration of Rh–O and C–Rh–O bonds during EOR: boosting the C–C bond cleavage to facilitate the conversion of ethanol and enhancing the formation of adsorbed -OH to remove the poisoning *CO. This work provided not only a fundamental understanding of the catalytic behavior of partially oxidized single Rh atoms on the Pt substrates to achieve the complete oxidation of ethanol to CO2 but also a unique single-atom approach using low-coordination active sites to tune the activity and selectivity toward other electrocatalytic reactions.

Materials and Methods

Materials.

Platinum(II) acetylacetonate [Pt(acac)2, Pt 48.0%] was obtained from Alfa Aesar. Rhodium(III) acetylacetonate [Rh(acac)3, 97%], oleic acid (OAc, 90%), oleylamine (OAm, 70%), benzyl alcohol (≥99%), and benzyl ether (98%) were obtained from Sigma-Aldrich. All chemicals were used as received without further treatment.

Synthesis of Pt NCs.

In a standard synthesis, Pt(acac)2 (0.102 mmol), OAc (2.0 mL), OAm (4.0 mL), and benzyl ether (14.0 mL) were added in a glass vial with a magnetic stirring bar. The mixture was heated to 130 °C under an Ar atmosphere. When the temperature reached 130 °C, CO gas was bubbled into the mixture with a flow rate of 30 mL ⋅ min−1, and Ar purging was stopped at the same time. The mixture was then heated to 210 °C at a heating rate of 8 °C min−1 under CO gas and held at 210 °C for 40 min without CO gas bubbling. The resulting suspension was cooled down to room temperature naturally, and the Pt NCs were precipitated out by sequential addition of toluene (10 mL) and ethanol (15 mL). The supernatant was discarded by centrifugation at 3,000 rpm for 5 min. The resulting Pt NCs were dispersed in toluene for further treatment.

Preparation of Pt NCs/C.

A suspension of Pt NCs was added into a toluene solution containing 40 mg Vulcan XC-72R carbon and kept under ultrasonic wave agitation for 30 min. The resulting Pt NCs/C catalyst was centrifuged three times with toluene at 3,000 rpm for 5 min and then dried under Ar protection at room temperature.

Synthesis of Rh-Decorated Pt NCs/C.

As-prepared Pt NCs/C (20 mg), OAm (5 mL), and benzyl ether (5 mL) were added in a glass vial with a magnetic bar with a magnetic stirring bar. The mixture was heated to 250 °C under an Ar atmosphere. Rh(acac)3 dissolved in benzyl ether (1 mL) was injected in a reaction mixture. The amount of Rh(acac)3 was 1 mg (0.002 mmol) for a single RhatO-Pt NCs and was 2 mg (0.005 mmol) for RhclO-Pt NCs. Then, the reaction mixture was maintained at 250 °C for 1 h. The resulting product was cooled down to room temperature naturally, and the Rh-decorated Pt NCs/C catalysts were centrifuged three times with toluene at 3,000 rpm for 5 min and then dried under Ar protection at room temperature. The surface coverage of Pt NCs/C by Rh atoms for RhatO-Pt NCs/C was calculated using the following equation:

Morphological, Structural, and Elemental Characterization.

TEM images were obtained using an H-7100 microscope (Hitachi) operated at an acceleration voltage of 120 kV. High-resolution TEM and EDS studies were carried out in a JEM-2100F (JEOL) and Titan G2 ChemiSTEM Cs Probe (FEI) operated at an acceleration voltage of 200 kV. The metal contents in catalysts were determined using ICP optical emission spectroscopy (PerkinElmer, Optima 7300DV) and ICP mass spectrometry (PerkinElmer, NexION 300×). XRD patterns were obtained with a D2 phaser X-ray diffractometer (Bruker). XPS was carried out using a spectrometer (Thermo Fisher Scientific) with Al Kα X-ray (1,486.6 eV) as the light source. All XPS spectra were aligned using the C1s peak at 284.8 eV as reference.

Electrochemical Measurements.

The electrocatalyst ink was prepared by dispersing 2.0 mg electrocatalysts in the mixed 2 mL Milli-Q water and isopropanol solution (Milli-Q water: isopropanol = 4: 1) and 8 µl Nafion (5%). A total of 10 µl electrocatalyst ink was used for the electrochemical test. Glassy carbon, graphite rod, and Ag/AgCl (3 M ⋅ Cl−) were used as the working, counter, and reference electrodes, respectively. All the potentials were calibrated to an RHE. To obtain the stable electrochemical result, 20 fast CV cycles were scanned first in Ar-saturated 0.1 M HClO4 solution at 100 mV ⋅ s−1 in the range of 0.05 to 1.20 V (versus RHE). The CV curves were then recorded in 0.1 M HClO4 solution with a scanning rate of 50 mV ⋅ s−1. The EOR activities were measured in Ar-saturated 0.1 M HClO4 and 0.2 M C2H5OH solution. The EOR CV curves were obtained with a scanning rate of 50 mV ⋅ s−1. The chronoamperometric curves were measured at 0.65 V. For CO stripping tests, CO was adsorpted on the electrocatalyst surface by holding the potential at 0.05 V for 10 min in CO-saturated 0.1 M HClO4 solution. The CO stripping curves were obtained after purging Ar for 30 min with a scanning rate of 50 mV ⋅ s−1. The stability test was performed by a prolong chronoamperometric test at 0.55 V for every 10,000 s followed by the CV cleaning. Three cycles of 10,000 s were examined. The liquid products were analyzed by using HPLC (Agilent, 1260 Infinity-II) equipped with Hi-Plex H columns. Upon the completion of electrolysis at 0.85 and 1.05 VRHE for 1 h, 1.5 mL electrolyte was collected for HPLC analysis. The faradaic efficiency of acetic acid was calculated by the ratio of the number of electrons consumed toward CH3COOH to the number of electrons passing through the working electrode. The IRRAS tests were performed on the Nicolet IS50 spectrometer equipped with a mercuric cadmium telluride detector. The homemade electrochemical cell was used, including a ZnSe hemisphere window, an Au working electrode, an Ag/AgCl reference electrode, and a Pt wire counter electrode (30). The IRRAS spectra were obtained in the 0.1 M HClO4 and 0.2 M C2H5OH solution from 0.05 to 1.05 V. The 127 interferograms were used in the test with the 4 cm−1 of resolution. The obtained spectra were postprocessed by subtracting the first reference spectrum. The selectivity for each of the EOR products were calculated by the following equation:in which is the integrated band intensities, is the effective absorption coefficient taken from the work of Weaver and coworkers (31, 32), and is the transferred electron numbers from ethanol to the respective product.

XAFS Analysis.

The XAFS analysis was conducted on the 7-BM (QAS) beamline at National Synchrotron Light Source-II in Brookhaven National Laboratory (BNL). The obtained spectra were processed using the IFFEFFIT package (33, 34). For the EXAFS analysis, the original EXAFS pattern [χ(k)] was weighted with k3 in order to intensify the high-k oscillation regime and was Fourier transformed using a Hanning window. The amplitude reduction factor (SO2) was attained from the corresponding foil. All of the EXAFS fittings were done in the R space. All fitting parameters and results are tabulated in .

In situ XAFS analysis was conducted on the 7D (XAFS) beamline at Pohang Light Source in the Pohang Accelerating Laboratory. The XAFS signal was recorded with fluorescent mode by using a seven-channel germanium (Ge) detector mainly because of the diluted amount of Rh in the samples. The laboratory-made acryl kit was used as described in our previous work (35). The areal loading of the working electrode was c.a. 3 mg ⋅ cm−2. During the measurement, Ar gas was continuously bubbled into the electrolyte. The working electrode potential was controlled by an RHE (Hydroflex, EDAQ), and graphite paper (highly oriented pyrolytic graphite) was used as a counter electrode. The other conditions were similar to the electrochemical operation except for the ethanol concentration (1 M) in order to compensate the higher mass loading of the working electrode.

Computational Methods.

Spin-polarized DFT (36, 37) calculations were performed at a generalized gradient approximation (38) level using the plane wave Vienna Ab-Initio Simulation Package code (39, 40). The core electrons were described using the projector augmented wave (41) potentials using PW91 functionals (42). A kinetic energy cutoff of 400 eV and 3 × 3 × 1 k-point mesh were used in all structure optimization calculations.

The Pt(100) surface was modeled using a four-layer 3 × 3 surface slab. Rhat-Pt NCs in our DFT calculations were represented by an Rh atom adsorbed at the most favorable site on ML O-covered Pt(100). Rh single-atom adsorbed on Pt(100) namely Rhat-Pt(100) was modeled to represent the Rh single-atom catalyst without surface O on the Pt surface. A vacuum layer of ∼15 Å thick was added in the slab cell along the direction perpendicular to the surface to minimize the artificial interactions between the surface and its periodic images. Atoms in the bottom two layers were fixed, while all other atoms were allowed to relax during geometry optimization until the Hellmann–Feynman force on each ion was smaller than 0.02 eV ⋅ Å−1.

The BE of adsorbate was calculated as the following:in which E(slab + adsorbate), E(slab), and E(adsorbate) are the total energies of the slab with adsorbate, clean slab, and adsorbate species in the gas phase, respectively.

The free energy change (ΔG) along various reaction pathways was calculated using the approach developed by Nørskov and coworkers (43). In this approach, the free energy of adsorbed species at applied potential (U) = 0 V is calculated as the following:in which EDFT is the DFT calculated total energy of adsorbed species, ZPE is the zero-point energy, and TS is the entropic corrections ().

Activation energy calculations were performed using the climbing image nudged elastic band method (44).