Theoretical study of the oxidation of formic acid on a PtPd(111) surface

Abstract

By performing density functional theory calculations, the adsorption configurations of formic acid and possible reaction pathway for HCOOH oxidation on PtPd(111) surface are located. Results show that CO₂ is preferentially formed as the main product of the catalytic oxidation of formic acid. The formation of CO on the pure Pd surface could not possibly occur during formic acid decomposition on the PtPd(111) surface owing to the high reaction barrier. Therefore, no poisoning of catalyst would occur on the PtPd(111) surface. Our results indicate that the significantly increased catalytic activity of bimetallic PtPd catalyst towards HCOOH oxidation should be attributed to the reduction in poisoning by CO.

Introduction

The direct formic acid fuel cell (DFAFC) appears to be one of the promising clean energy sources due to its high energy conversion efficiency and low environmental pollution. ¹ Its direct oxidation releases two electrons per molecule ²

$$\mathrm{Anode}:\mathrm{HCOOH}\to \mathrm{C}{\mathrm{O}}_2+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}$$

(1)

$$\mathrm{Cathode}:1/2{\mathrm{O}}_2+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2\mathrm{O}$$

(2)

$$\mathrm{Overall}:\mathrm{HCOOH}+1/2{\mathrm{O}}_2\to \mathrm{C}{\mathrm{O}}_2+{\mathrm{H}}_2\mathrm{O}$$

(3)

It is generally accepted that the mechanism of the formic acid oxidation to CO₂ follows dual pathways,^3,4 that is, the dehydration pathway via adsorbed CO (HCOOH → CO + H₂O → CO₂ + 2 H⁺ + 2e⁻) or the dehydrogenation pathway without the participation of CO (HCOOH → CO₂ + 2 H⁺ + 2e⁻). In DFAFC, the sluggishness of formic acid oxidation on the anode catalyst is one of the primary bottlenecks. In general, Pd-based catalysts are commonly used in DFAFC. ⁵ However, the catalytic performance of pure Pd catalyst would be significantly diminished owing to the poisoning effect of CO generated from the HCOOH dehydration pathway. ³ Therefore, a highly selective and active catalyst towards the HCOOH dehydrogenation pathway ⁴ with formation of H₂ and CO₂ is desirable.

Recent studies showed that modification of the Pd surface with a secondary metal, such as Ni,^6,7 Au,^8,9 Co ¹⁰ and Ag,^11,12 is considered to be an efficient way to hinder the dehydration pathway, ¹³ by taking advantage of the so-called electronic ligand effect ^14–16 and the geometrical ensemble effect.^14,15,17,18 Among these Pd-based bimetallic catalysts, Pt-Pd nanostructures have attracted special attention due to their higher tolerance to CO poisoning and better catalytic activity compared with pure Pd catalysts. ¹⁹ Recently, the Pt-modified Pd catalyst has been prepared as the electrocatalyst for the HCOOH oxidation, such as Pt-decorated Pd/C nanoparticles, ²⁰ Pt sub-monolayers–decorated Pd black, ²¹ Pt-decorated Pd nanochain networks ²² and carbon-supported Pd-Pt alloy nanoparticles. ²³

Although much is known about the preparations of different morphological Pt-modified Pd catalysts, the underlying mechanism of their improved CO-poisoning tolerance still remains unclear at the molecular level. Herein, this study presents a theoretical study of formic acid oxidation on the bimetallic PtPd(111) surface. The (111) surface is considered as the most stable crystal plane of the fcc lattice and therefore is expected to contribute the most significant areas for the surface of bulk metals, nanoparticles and alloys and thus has been chosen for the present work. By performing the ultrasoft pseudopotential ²⁴ density functional theory (DFT) calculations, we have shown the mechanistic details of the elementary steps of HCOOH oxidation on a PtPd(111) surface, from which we hoped to elucidate the reason why Pt-Pd bimetallic catalysts possess enhanced catalytic activity and improve the CO-poisoning tolerance.

Computational details

The PtPd(111) surface was obtained based on the Pd(111) surface, where three Pd atoms in the top layer were substituted by three Pt atoms. In such a PtPd(111) surface model, the Pt atoms in the top layer were isolated by the Pd modifier and fixed to Pd(111) to follow the idea of maximizing the use of the efficiency of Pt materials. Our calculations were performed by using a triple-layer p(3 × 3) unit cell, where the atoms in the top layer were allowed to relax, whereas those in the two bottom layers were fixed. The vacuum region between slabs was 10 Å, which was sufficiently large to guarantee that the interactions between neighbouring cells in a direct normal to the surface were negligible.

All DFT calculations were performed using the CASTEP code.^25,26 The exchange-correlation effects were described with the spin polarization generalized gradient approximation (GGA) using the Perdew–Burke–Ernzerhof (PBE) functional, ²⁷ which has often been used to investigate small molecule oxidation on noble metal surfaces and the calculated results are reliable. ²⁸ Our calculated results show that the lattice constant is 3.92 Å for the bulk Pd, which is in good agreement with the experimental value of 3.89 Å. ²⁹ A mesh of 2 × 2 × 1 Monkhorst–Pack special k-point grid was used for the integrations of the Brillouin zone. The electron wave functions were expanded by a plane-wave basis set with a cut-off energy of 400 eV. The criteria for maximum force convergence and energy were set to 0.05 eV Å⁻¹ and 2.0 × 10⁻⁵ eV atom⁻¹. The transition states (TSs) were determined with the linear and quadratic synchronous transit (LST/QST) method. ²⁵

The adsorption energies for HCOOH were defined as E_ad = E_surf + E_adsorbate − E_total, where E_surf, E_adsorbate and E_totel refer to the energies of the clean surface, free formic acid molecule and the adsorbed system, respectively. The activation barrier energy (ΔE_a) of each elementary step was computed by ΔE_a = E_TS − E_R, where E_TS and E_R are the energies of the transition state and reactant, respectively.

Results and discussion

Formic acid adsorption on PtPd(111) surface

The adsorption of formic acid on heterogeneous catalysts is the first and governing step in the process of oxidation. It should be noted that formic acid has two isomers: transoid-HCOOH and cisoid-HCOOH. Initially, our results showed that the transoid-HCOOH is more stable than the cisoid one in the gas phase by14.5 kJ mol⁻¹, which is consistent with the experimental finding of 16.4 kJ mol⁻¹. ³⁰ We then examined the adsorption of the two isomers of HCOOH on the PtPd(111) surface, including top, bridge, fcc and hcp sites. Figure 1 shows four adsorption configurations, labelled as 1a, 1b, 1c and 1d, respectively. It was found that HCOOH always binds to the bimetallic surface in an upright fashion through its different atoms in all three adsorption structures. The most stable configurations is 1a, where the carbonyl oxygen attaches to a top Pd atom and the hydroxyl group points asymmetrically to an adjacent top Pt and Pd atoms, which agrees well with the results of HCOOH adsorbed on Pt(111) and Pd(111) surfaces.^3,31 The Pd-O distance is 2.251 Å, and the Pt-H and Pd-H distances are 2.286 and 2.595 Å, respectively. The calculated adsorption energy of 1a is 42.5 kJ mol⁻¹. 1b is around 4.8 kJ mol⁻¹ less stable than 1a; however, HCOOH in 1b coordinates its carbonyl oxygen to a top Pt atom with Pt-O distance of 2.326 Å and the hydroxyl group points asymmetrically to two neighbouring Pd atoms. 1c shows an adsorption geometry in which the carbonyl O binds strongly and the carbon H binds weakly with the surface with Pd-O, Pd-H and Pt-H distances of 2.339, 2.859 and 2.942 Å, respectively. With the adsorption energy of 19.3 kJ mol⁻¹, this state has about half the value of E_ad as configuration 1a. For cisoid-HCOOH in structure 1d, HCOOH goes far away from the surface with Pt-O and Pd-O distances of 2.362 and 3.229 Å, respectively. The interaction between the molecule and surface is also weaker than that of transoid-HCOOH (17.6 kJ mol⁻¹). As a result, the most stable configuration of the optimized transoid-HCOOH is used as the initial state of HCOOH oxidation on the PtPd(111) surface.

Figure 1.

Optimized adsorption configurations of HCOOH on PtPd(111) surface with crucial structural parameters (in Å).

From our calculations, the above three adsorption structures of HCOOH on the PtPd(111) surface are similar to those on the Pd(111) surface shown in our previous study. ³ However, the adsorption energies of HCOOH seem to decrease compared to those of their counterparts (42.5, 37.7 and 19.3 kJ mol⁻¹ on the PtPd(111) surface and 59.8, 36.7 and 40.5 kJ mol⁻¹ on the Pd(111) surface). Generally, the d-band centre shift reported by Norskov and colleagues^32,33 can be used as the simplest descriptor to measure the variation of the interaction of surface atoms with adsorbates: the d-band centre moving up towards the Fermi level can strengthen the binding energies of adsorbates on catalyst surfaces. In this study, the d-band centre for the Pt/Pd(111) surface was calculated to up-shift to −0.19 eV from −1.86 eV for the Pd(111) surface. However, the trend of the calculated adsorption energies shows deviations from the d-band model. Such unexpected results are due to the modification of the local d-band by the presence of the adsorbate, which is similar to that found in previous researches.^34,35

Formic acid oxidation on PtPd(111) surface

The oxidation of formic acid to CO₂ would produce hydrogen atoms and electrons. The atomic charge population of H⁺ and Au surface was first calculated to understand the electronic structure of adsorbed H⁺ on Au surface. When H⁺ is adsorbed on an Au surface, the electron transfers from Au surface to H⁺. From our calculation, the atomic charge population of adsorbed H⁺ is close to that of H. Moreover, the metal acts as an electron sink and the transferred electron from Au to H is used as an electric power during the DFAFC operation condition. As a result, we assume the neutral model for the slab. We then present the possible pathways for formic acid oxidation. The optimized structures of the reaction intermediates and products located on the path are summarized in Figure 2, while the calculated energy profiles are shown in Figure 3, where the sum of the energies of the isolated reactants (HCOOH and the clean PtPd(111) surface) is taken as zero energy.

Figure 2.

Optimized structures of intermediates and transition states. Bond lengths are in angstroms.

Figure 3.

Potential energy profiles for HCOOH oxidation.

The stable configuration 1a is employed as the initial state of HCOOH decomposition on the PtPd(111) surface. According to numerous experimental and theoretical investigations, the formate^12,13,36,37 species is believed to be a viable intermediate in DFAFCs. This observation prompted the suggestion that HCOOH decomposition can start from its O–H bond cleavage via structure 1a on the PtPd(111) surface. Along the reaction coordinate, the hydroxyl of formic acid has been significantly stretched via TS_1a-2a, resulting in the HCOO intermediate (structure 2a), which is in a bidentate form via binding to Pd and Pt atoms by its two oxygen. Note that the HCOO adsorption structure is consistent with experimental observation ¹³ and similar to the adsorption on Pt and Pd surfaces.^28,38 This process needs to overcome a barrier of 78.2 kJ mol⁻¹. From structure 2a, forming CO₂ directly in the gas phase is difficult due to the C–H bond oriented along the surface. However, forming CO₂ and H is much easier if HCOO can reorient itself to be bound through one of the oxygen atoms with a monodentate structure via the transition state TS_2a-3a. The barrier associated with this transition is 53.0 kJ mol⁻¹. The following step (the C–H bond scission) is exothermic by 121.5 kJ mol⁻¹ and involves an activation barrier of 50.2 kJ mol⁻¹. The calculated result indicates this process to be both kinetically and thermodynamically favourable. Alternatively, if 3a directly decomposes to 5a (CO and H₂O) instead, then the step requires overcoming an extremely high barrier of 202.7 kJ mol⁻¹, and this can be ruled out as a possible reaction process on the PtPd(111) surface.

To better understand the increased catalytic activity of bimetal PtPd catalysts, the activation energy barriers are compared with our previous results ³ for the monometallic Pd(111) surface. On one hand, for the dehydrogenation process of HCOOH to CO₂ on Pd(111), the transformation of HCOO from the bidentate structure to the monodentate structure is the rate-determining step, with a barrier of 83.9 kJ mol⁻¹. However, the barrier of this step on the PtPd(111) surface is 53.0 kJ mol⁻¹, which is not notably different from that on the Pd(111) surface, indicating that the PtPd(111) surface can perform similar activity for the catalytic dehydrogenation of HCOOH to CO₂. On the other hand, we find that the barrier in the rate-determining step is 202.7 kJ mol⁻¹ for the decomposition of HCOOH to CO on PtPd(111), whereas the corresponding result is 55.0 kJ mol⁻¹ on Pd(111), implying that the CO pathway on the PtPd(111) surface is more difficult than that on a pure Pd(111) surface, which is in good agreement with the observed CO antipoisoning ability of PtPd bimetal catalysts for HCOOH oxidation. Given the present results, we propose that the enhanced catalytic activity of bimetal PtPd catalysts towards HCOOH oxidation should be attributed to their suppression to the dehydration pathway.

It is important to note that the reduced CO-poisoning effect of PtPd bimetallic is influenced not only by the relative rate of CO formation but also by the binding ability of CO on the surface. The stronger the binding of CO, the easier the poisoning of the catalyst. In order to understand the enhanced CO tolerant of PtPd catalysts, we calculated the adsorption energies of CO on PtPd(111) and Pd(111) surfaces, respectively. Our results showed that the adsorption energy of CO on the PtPd(111) surface is noticeably smaller than that on the pure Pd(111) surface (37.6 vs 377.3 kJ mol⁻¹), implying that CO desorption from PtPd(111) is easier than that from Pd(111), also supporting the enhanced activity of PtPd bimetallic catalysts towards formic acid oxidation.

Conclusion

Formic acid oxidation on PtPd(111) surface has been investigated by DFT to provide a clue for the increased catalytic activity of PtPd bimetal catalysts. Comparing the activation energy barriers with Pd(111), our calculated results illustrate that PtPd(111) can perform as well as pure Pd for the catalytic dehydrogenation of HCOOH to CO₂. On the contrary, HCOOH oxidation to CO on PtPd(111) becomes more difficult than that on the pure Pd(111) surface. Thus, the enhanced catalytic activity of bimetallic PtPd(111) surface towards HCOOH oxidation is owing to its suppression of the dehydration pathway.