Rational Design of Competitive Electrocatalysts for Hydrogen Fuel

Letter pubs.acs.org/JPCL

Rational Design of Competitive Electrocatalysts for Hydrogen Fuel Cells Sergey Stolbov* and Marisol Alcántara Ortigoza Department of Physics, University of Central Florida, Orlando, Florida 32816, United States S Supporting Information *

ABSTRACT: The large-scale application of one of the most promising clean and renewable sources of energy, hydrogen fuel cells, still awaits efficient and cost-effective electrocatalysts for the oxygen reduction reaction (ORR) occurring on the cathode. We demonstrate that truly rational design renders electrocatalysts possessing both qualities. By unifying the knowledge on surface morphology, composition, electronic structure, and reactivity, we solve that trimetallic sandwich-like structures are an excellent choice for optimization. Their constituting species are expected to couple synergistically yielding reaction-environment stability, cost-effectiveness, and tunable reactivity. This cooperativeaction concept enabled us to predict two advantageous ORR electrocatalysts: Pd/Fe/ W(110) and Au/Ru/W(110). Density functional theory calculations of the reaction freeenergy diagrams suggest that these materials are more active toward ORR than the so-far best Pt-based catalysts. Our designing concept advances also a general approach for engineering advanced materials. SECTION: Surfaces, Interfaces, Catalysis

D

conducted using the trial-and-error method. Some computational works have screened dozens of materials via combinatorial search (sometimes even without considerations of the surface stability in the reaction environment). Although not the most efficient way to discover optimum materials, applications of the latter approach have provided a deep insight into the microscopic mechanisms underlying ORR and inadvertently paved the basis for designing ORR catalysts. ORR is a complex kinetic phenomenon involving charge transfer between two electrodes with different Fermi levels. Nevertheless, it has been demonstrated that the thermodynamics of the reaction steps alone can determine the upper limit of the relative rate achieved by a catalyst toward ORR with respect to a reference material, Pt.9−11 In the approach proposed in ref 8, the activity of a surface toward ORR can be estimated through the maximal electrode potential U at which all reaction steps are still exothermic. (See the Supporting Information.) This gauge has shown to be greatly useful for evaluating the ORR activity on prospected surfaces3−6,9,11−17 and together with combinatorial screening proved successful in finding two active catalysts,11 yet with a high load of Pt. Turning to the predicting power of the current stage of knowledge, the reason for which new electrocatalysts can be designed rests on (1) U can be traced to free-energy diagram of the reaction; (2) the free energies of the reaction intermediates entering in this diagram9 can be accurately determined by first-

esigning the catalytic properties of metal surfaces is one of the ultimate goals of material science. Among its most challenging and, at the same time, promising aims is the efficient electrocatalysis of reactions in proton exchange membrane fuel cells (PEMFCs) because these devices are clean means for conversion of the chemical energy to electric power.1 PEMFCs, however, are unacceptably expensive because they rely on Pt-based electrocatalysts. Furthermore, even for these, the rate of the oxygen reduction reaction (ORR) on Pt cathodes1 is still lower than optimal. One major difficulty in developing new cost-effective materials that efficiently catalyze ORR is the limited set of suitable materials. The PEMFC cathode is exposed to an acidic solvent that, at the expected operating potential (+0.8 to +1.0 V), merely dissolves most transition metals.2 Only Pt, Ir, Au, and Pd are stable in such an environment, yet all four suffer from either of two serious disadvantages. The former two elements are extremely rare, whereas the latter two do not provide the desired activity toward ORR. Not surprisingly, the quest of efficient catalysts overcoming these drawbacks has explored several routes extensively. A significant progress in reducing the Pt load was achieved with Pt-based alloys, Pt3M, where M is another transition metal.3,4 Another leap toward this direction was attained by depositing only a Pt monolayer on a Pt-free substrate.5−7 Still, because the actual catalysts are constituted by small nanoparticles, surface Pt atoms represent a significant fraction of the material. Various Pt-free materials have also been explored as possible electrocatalysts for ORR (most notably Pd1−xMx alloys, where M = Fe, Co, Cu,8 but as yet they hardly match the performance of Pt. Part of the reason is that in most cases the search for advantageous catalysts is © 2012 American Chemical Society

Received: November 23, 2011 Accepted: January 24, 2012 Published: January 24, 2012 463

dx.doi.org/10.1021/jz201551e | J. Phys. Chem.Lett. 2012, 3, 463−467

The Journal of Physical Chemistry Letters

Letter

conditions are met for the recently reported Pt-based catalysts, Pt/Ni/Pt.18 In this work, we shall thus explore sandwich-like structures with Pd and Au as the M1 element because, as previously mentioned, these elements are stable in the reaction environment and not as scarce as Pt and Ir. The issue to address is, of course, that of tuning their activity: Pd is too reactive for ORR, whereas Au is too noble. To adjust Pd activity, we turn to our recent work on electrocatalytic Pd−Co alloys.16 We have shown that the hybridization between the d states of the two metals reduces Pd reactivity. Specifically, whereas the Pd d-states do not overlap much energetically (and thus do not hybridize much) with Co minority-spin d-states, Co majority-spin d states, which are mostly located below the Fermi-level (EF), strongly hybridize with Pd d states. Consequently, the Pd d band overall recedes from EF, which decreases Pd reactivity. For the present sandwich-like structures, we thus choose another 3d magnetic metal, Fe, as the M2 element for M1Pd. Fe is likely to provide an effect similar to that of Co and perhaps even stronger because Fe spin polarization is larger than that of Co. Fe also satisfies that its cohesive energy is higher than that of Pd, which is essential to ensure the stability of the sandwich-like structure. Namely, such disparity anticipates that exposing a strained Felayer to the surface is not energetically favorable. At this point, though, it becomes critical that the selection of M3. M3 must bind strongly and stretch the M2 layer while keeping the M1 layer almost unstrained. We find W to be the best choice for M3. Indeed, the bond lengths in bulk Pd and W differ by only 0.4%. Bulk tungsten has the highest cohesive energy among all other elements, which should prevent any alloying of Fe or Pd with W at preparation and operation temperatures. In fact, it is known that a pseudomorphic Fe monolayer is stable on the W(110) surface at elevated temperatures.19 This study thus targets the properties of the Pd/Fe/W(110) structure. The (110) surface of bcc metals is known to have the lowest energy among other low-index surfaces, which suggests that the surface of W-based catalyst particles is mostly constituted by (110) facets. It is worth to remark that W has bcc structure. To our knowledge, this is the first investigation considering a bcc substrate for ORR catalysts. We now turn to adjust the activity of gold. Au is too noble to catalyze efficiently ORR. The d-electron band of pure Au is located much below EF, which determines its low reactivity. For enhancing it in the Au/M2/M3 structure, we choose Ru as the M2 element. Bulk Ru has a high density of d-electron states around EF, and its d band spreads to low energies where Au dband is located. (See ref 20, for example.) Therefore, it is plausible that the hybridization between Ru and Au d states in the Au/Ru/M3 structure will shift the Au LDOS toward EF and hence increase Au reactivity. Moreover, bulk Ru has a cohesive energy much larger than that of Au, which again ensures the structure’s stability. For the Au/Ru/M3 structure, we also choose W(110) as the substrate. Although the bond lengths in bulk Au are 4.5% larger than that in W, we shall find that the desired tuning is achieved. We thus choose Au/Ru/W(110) as the second system to study in this work. It is important to remind the reader that in the selected sandwich-like structures W is not exposed to the surface but Pd or Au, which have been proven to be stable in the reaction environment. Moreover, we will show below that both Pd and Au surface layers are more stable in the reaction environment than the pure Pd and Au. Next, we ascertain via first-principles DFT calculations (see details in the Supporting Information) whether the educated

principles calculations; and (3) the key aspects involved in tuning these are, in principle, well-established. Namely, the free energy of the ORR intermediates is mostly determined by their binding energy to the catalytic surface.9,16 The optimization of the ORR rate can hence be achieved by tuning the binding energies of the ORR intermediates on the catalyst surface. The binding energies are well-known to depend on the local densities of electronic states (LDOS) of the surface atoms. Moreover, numerous investigations show the possibility of rationally modifying LDOS by alloying transition or noble metal elements. (See a review article, ref 17, and details in the Supporting Information.) Although the relation among alloy composition, LDOS distribution, binding energy of the ORR intermediates, and the ORR activity has been demonstrated in a number of studies,3,5−7,9,11−17 the patent long-chain correlation among them has not yet been exploited. Here we unify all available pieces of knowledge to advance a new concept for designing targeted materials. To tailor an efficient electrocatalyst, we first choose cost-effective elements arranged so that the designed structure surface can be tuned to optimal reactivity and reaction environment stability. We propose that such a system can be constituted of sandwichlike transition (noble)-metal superstructures M1/M2/M3, where M3 is the substrate material and M1 and M2 denote the surface and the sandwiched monatomic layers, respectively. (See Figure

Figure 1. Schematic illustration of the trimetallic sandwich-like structures for tuning catalytic properties. (See the text.)

1.) We shall see below that the components of such structures could work synergistically for the catalyst to meet all of the above requirements. ORR takes place on M1, which must sustain the reaction environment. The purpose of M2 and M3 is therefore manifold. M2 tunes the electronic structure of M1, so that the latter provides the optimal binding energies of the ORR intermediates and promotes M1’s stability at the surface and against the reaction environment. M3 provides a low-cost material substrate impenetrable by M2 at annealing temperatures, furnishes a suitable lattice parameter for the surface layer, and strains the M2 sandwiched layer through lattice mismatch. On account of this, two main competing scenarios determine the stability of M1/M2/M3 structures: (1) exposing M1 to the surface, which keeps fully coordinated strained-M2 and allows for M2−M3 bonds (or, if M1M3, maximizes the number of M2−M1) and (2) exposing strained-M2 to the surface, which keeps fully coordinated M1 and allows for M1-M3 bonds (or, if M1M3, maximizes the number of M1−M1 bonds). The M1/ M2/M3 structure will be stable if the energy of configuration (1) is lower than that of configuration (2). Sandwich-like structures will be, in general, favorable for a pair of elements in which M2−M3 bonds are stronger than M1−M3 bonds or M2− M2 bonds as well as for elements in which M2−M2 bonds are stronger/shorter than M1−M1 bonds. For example, these 464



Letter

It remains to evaluate to what extent the above-mentioned features of the “tuned” electronic structure impact the reactivity of Pd/Fe/W(110) and Au/Ru/W(110) in general. We therefore calculate the binding energy of atomic oxygen, EB(O), on these surfaces and compare these values to EB(O) on Pd (111) and Au(111), correspondingly. The atomic configuration for O adsorption on Pd/Fe/W(110) is illustrated in Figure 3 (left panel). We find that, for Pd/Fe/W(110),

guess on which this approach is grounded indeed renders these novel materials to be highly active and stable electrocatalysts for ORR. As expected and desired, the stability of the Pd-based structure with respect to Fe/Pd/W(110) is guaranteed by a 0.734 eV total-energy preference per surface atom toward the Pd/Fe/W(110) structure. Such a large difference ensures that the vibrational contribution to G of either system could not change qualitatively this result. We thus resolve that the Pd/Fe/ W(110) structure would be stable even at high temperatures. Likewise, we find for the Au-based structure that the total energy of Au/Ru/W(110) is lower than that of Ru/Au/W(110) by 0.503 eV per surface atom, which again suggests that the desired Au/Ru/W(110) structure is stable. We now turn to examine whether Pd/Fe/W(110) and Au/ Ru/W(110) are stable in the reaction environment. For this purpose, we calculate the dissolution potential (Udiss) of the corresponding M1 with respect to that in the M1(111) surface, following the technique developed in ref 21. We find that Udiss of Pd in Pd/Fe/W(110) is higher than that for Pd(111) by 0.323 V, that is, even higher than Udiss of Pt. Similarly, Udiss of Au in Au/Ru/W(110) is higher than that for Au(111) by 0.376 V. Note that the enhanced Udiss results from the relatively high formation energy of the M1 (Pd, Au) monolayer in the sandwich-like structure with respect to that in the M1(111) surfaces. Next, it is critical to confirm whether the chosen M2 elements do optimize the electronic structure of the M1 layers as it is expected. To this end, we calculate the LDOS of M1 and M2 atoms in both structures as well as LDOS of the surface atoms in Pd(111) and Au(111). Figure 2 shows that indeed the strong

Figure 3. Illustration of two adsorption configurations of ORR intermediates on Pd/Fe/W(110). Left panel: Atomic oxygen adsorption. The bright gray, orange, and red balls represent the Pd, Fe, and O atoms, respectively. The purple lines indicate the supercell boundaries. Note that O is shifted from the symmetric rhombic hollow site to make three equal bonds with Pd atoms. Right panel: coadsorption of OOH and water. Dark gray and small blue balls represent W and H atoms.

EB(O) = 4.069 eV, which is 0.463 eV lower than that on Pd(111). For Au/Ru/W(110), EB(O) = 3.763 eV, which is 0.340 eV higher than that on Au(111). These calculations are perfectly consistent with the expectations brought on by the LDOS: the shift of the Pd d band from EF to lower energies in Pd/Fe/W(110) reduces its reactivity, whereas the increase in the LDOS around EF of Au enhances its reactivity. The obtained values of EB(O) for Pd/Fe/W(110) and Au/ Ru/W(110) fall in the region between EB(O) of Pt(111) and Au(111,) already suggesting that the reactivity of our sandwichlike structures is close to optimal. (See the Supporting Information.)9,10,22 Ultimately and most importantly, the specific activity of these materials toward ORR is evaluated via calculations of the free energies of the ORR intermediates coadsorbed with water, as described in ref 9 and as implemented in ref 11. We have done so for both sandwichlike structures under consideration as well as for Pt(111), for reference. The most favorable OOH and H2O coadsorption configuration is shown in the right panel of Figure 3, as an example. In general, a high coverage of O or OH may reconstruct the surface and impair its reactivity. For example, DFT calculations have shown that for the Pt/Cu/Pt surface alloy the second layer Cu atoms can be pulled up on the top in the presence of an adsorbate.23 This is partly because Pt atoms in the third layer can make bonds with Pt surface atoms that are stronger than those they can make with Cu atoms (and partly because the O binds stronger to Cu than to Pt). However, while thermodynamic considerations guarantee that the system will restructure and overcome high interatomic exchange activation barriers; this may happen within a time period that may extend to infinite. In fact the experimental evidence in ref 23 shows that the Cu−Pt exchange becomes significant only at ∼460 K,

Figure 2. Local densities of the d-electron states calculated for M1 and M2 atoms in the sandwich-like structures and for the corresponding metal M1(111) surfaces. Left panel: Pd/Fe/W(110). Right panel: Au/ Ru/W(110).

hybridization of Pd d states with the low-lying spin-up d states of Fe (positive LDOS) increases Pd LDOS at low energies. As a result, Pd d band in Pd/Fe/W(110) is shifted considerably away from EF toward low energies as compared with the d band of Pd(111). The opposite occurs to Au d band because of the high density of states of Ru around EF. The hybridization between Ru and Au d states enhances Au LDOS in the vicinity of EF as compared with that of Au(111) but to a lesser extent, partially because of Au shrunken bonds. To test this assumption, we calculated LDOS for Au/Ru/W with artificially extended in-plane periodicity by 4.5%. We find that, while such an expansion does not affect LDOS around the Fermi level, it does narrow the Au d band and shifts the d-band center toward the Fermi level. 465



Letter

Figure 4. ORR free-energy diagrams calculated for Pd/Fe/W(110) and Au/Ru/W(110) compared with that for Pt(111) (dashed black lines) for an electrode potential U = 0.85 V.

■

suggesting that even for the relatively weak Cu−Pt bonds the exchange activation barriers are high for the temperatures of interest here. Nevertheless, to guarantee the stability of the M2 layer (formed by Fe and Ru atoms, elements with high O affinity), both systems considered in the present work are such that M3 atoms make stronger bonds with M2 than with M1, which is expected to largely impede any exchange process. Furthermore, a fuel cell works in the kinetic regime generating voltage only if the O* + H+ + e− → HO* and HO* + H+ + e− → H2O steps of ORR have a high rate. The electrode potential thus, in principle, cannot reach the value at which the O or OH adsorption is stable. Therefore, one should not expect any restructuring in the systems under the working conditions of PEMFC. Coincidently, our calculations show that both systems under consideration have almost the same onset potential (the maximal potential at which all reaction steps are still exothermic): we find these potentials to be 0.844 and 0.851 V for Pd/Fe/W and Au/Ru/W, respectively. We thus plot in Figure 4 the reaction free energy diagrams for U = 0.85 V for both systems. Neglecting this minor difference, all reaction steps on Pd/Fe/W(110) and Au/Ru/W(110) are shown to be exothermic, whereas the last two steps for Pt(111) are endothermic with significant (0.15 to 0.2 eV) thermodynamic barriers. In other words, these diagrams indicate that the proposed materials are more active toward ORR than Pt(111). It is important to note that in the case of Pd/Fe/W(110) the *O and *OH free energies are equal, suggesting that the O* + (H+ + e−) → HO* is the rate-limiting step for ORR on this surface, suggesting that this surface is still more reactive than the optimal one. Au/Ru/W(110) is expected to be less reactive than an optimal surface because the equal free energy obtained for (O2, 2H2) and *OOH states show that the O2* + (H+ + e−) → *OOH is the rate-limiting step for ORR on this system. (See the Supporting Information.) The main result obtained here, however, is that according to our calculations both proposed sandwich-like systems are more efficient catalysts for ORR than Pt. The success of this approach points to its potential and generality in tailoring the reactivity of advanced materials.

■

REFERENCES

(1) Handbook of Fuel Cells: Fundamentals, Technology, and Applications; Vielstich, W., Lamm, A., Gasteiger, H. A., Eds.; Wiley: Hoboken, NJ, 2003. (2) CRC Handbook of Chemistry and Physics; Lide, D. R., Ed.; CRC Press: New York, 1996. (3) Stamenkovic, V. R.; Mun, B. S.; Arenz, M.; Mayrhofer, K. J. J.; Lucas, C. A.; Wang, G.; Ross, P. N.; Markovic, N. M. Trends in Electrocatalysis on Extended and Nanoscale Pt-Bimetallic Alloy Surfaces. Nat. Mater. 2007, 6, 241−247 andreferences therein.. (4) Lai, F.-J.; Chou, H.-L.; Sarma, L. S.; Wang, D.-Y.; Lin, Y.-C.; Lee, J.-F.; Hwang, B.-J.; Chen, C.-C. Tunable Properties of PtxFe1−x Electrocatalysts and their Catalytic Activity Towards the Oxygen Reduction Reaction. Nanoscale 2010, 2, 573−581. (5) Zhang, J.; Vukmirovic, M. B.; Xu, Y.; Mavrikakis, M.; Adzic, R. R. Controlling the Catalytic Activity of Platinum-Monolayer Electrocatalysts for Oxygen Reduction with Different Substrates. Angew. Chem., Int. Ed. 2005, 44, 2132−2135. (6) Adzic, R. R.; Zhang, J.; Sasaki, K.; Vukmirovic, M. B.; Shao, M.; Wang, J. X.; Nilekar, A. U.; Mavrikakis, M.; Valerio, J. A.; Uribe, F. Platinum Monolayer Fuel Cell Electrocatalysts. Top. Catal. 2007, 46, 249−262. (7) Zhou, W.-P.; Yang, X.; Vukmirovic, M. B.; Koel, B. E.; Jiao, J.; Peng, G.; Mavrikakis, M.; Adzic, R. R. Improving Electrocatalysts for O2 Reduction by Fine-Tuning the Pt-Support Interaction: Pt Monolayer on the Surfaces of a Pd3Fe(111) Single-Crystal Alloy. J. Am. Chem. Soc. 2009, 131, 12755−12762. (8) Savadogo, O.; Lee, K.; Oishi, K.; Mitsushima, S.; Kamiya, N.; Ota, K.-I. New Palladium Alloys Catalyst for the Oxygen Reduction Reaction in an Acid Medium. Electrochem. Commun. 2004, 6, 105− 109. (9) Nørskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.; Kitchin, J. R.; Bligaard, T.; Jónsson, H. Origin of the Overpotential for Oxygen Reduction at a Fuel-Cell Cathode. J. Phys. Chem. B 2004, 108, 17886−17892. (10) Taylor, C. D.; Wasileski, S. A.; Filhol, J.-S.; Neurock, M. First Principles Reaction Modeling of the Electrochemical Interface: Consideration and Calculation of a Tunable Surface Potential from Atomic and Electronic Structure. Phys. Rev. B 2006, 73, 165402. (11) Greeley, J.; Stephens, I. E. L.; Bondarenko, A. S.; Johansson, T. P.; Hansen, H. A.; Jaramillo, T. F.; Rossmeisl, J.; Chorkendorff, I.; Nørskov, J. K. Alloys of Platinum and Early Transition Metals as Oxygen Reduction Electrocatalysts. Nat. Chem. 2009, 1, 552−556. (12) Greeley, J.; Nørskov, J. K. Combinatorial Density Functional Theory-Based Screening of Surface Alloys for the Oxygen Reduction Reaction. J. Phys. Chem. C 2009, 113, 4932−4939. (13) Zhang, J.; Vukmirovic, M. B.; Xu, Y.; Mavrikakis, M.; Adzic, R. R. Controlling the Catalytic Activity of Platinum-Monolayer Electrocatalysts for Oxygen Reduction with Different Substrates. Angew. Chem., Int. Ed. 2005, 44, 2132−2135.

ASSOCIATED CONTENT

S Supporting Information *

Basis for rational design; computational details. This material is available free of charge via the Internet at http://pubs.acs.org. 466



Letter

(14) Abild-Pedersen, F.; Greeley, J.; Studt, F.; Rossmeisl, J.; Munter, T. R.; Moses, P. G.; Skúlason, E.; Bligaard, T.; Nørskov, J. K. Scaling Properties of Adsorption Energies for Hydrogen-Containing Molecules on Transition-Metal Surfaces. Phys. Rev. Lett. 2006, 99, 016105. (15) Stamenkovic, V.; Mun, B.; Mayrhofer, K. J. J.; Ross, P. N.; Markovic, N. M.; Rossmeisl, J.; Greeley, J.; Nørskov, J. K. Changing the Activity of Electrocatalysts for Oxygen Reduction by Tuning the Surface Electronic Structure. Angew. Chem., Int. Ed. 2006, 45, 2897− 2991. (16) Zuluaga, S.; Stolbov, S. Factors Controlling the Energetics of the Oxygen Reduction Reaction on the Pd-Co Electro-catalysts: Insight from First Principles. J. Chem. Phys. 2011, 135, 134702. (17) Nørskov, J. K.; Bligaard, T.; Rossmeisl, J.; Christensen, C. H. Towards the Computational Design of Solid Catalysts. Nature Chem. 2009, 1, 37−45. (18) Wadayama, T.; Todoroki, N.; Yamada, Y.; Sugawara, T.; Miyamoto, K.; Iijama, Y. Oxygen Reduction Reaction Activities of Ni/ Pt(111) Model Catalysts Fabricated by Molecular Beam Epitaxy. Electrochem. Commun. 2010, 12, 1112−1115. (19) Przybylski, M.; Kaufmann, I.; Gradmann, U. Mossbauer Analysis of Ultrathin Ferromagnetic Fe(110) Films on W(110) Coated by Ag. Phys. Rev. B 1989, 40, 8631−8640. (20) Stolbov, S. First-Principles Study of Formation of Se Submonolayer Structures on Ru Surfaces. Phys. Rev. B 2010, 82, 155463. (21) Greeley, J.; Nørskov, J. K. Electrochemical Dissolution of Surface Alloys in Acids: Thermodynamic Trends from First-principles Calculations. Electrochim. Acta 2007, 52, 5829−5835. (22) Okamoto, Y.; Sugino, O. Hyper-Volcano Surface for Oxygen Reduction Reactions over Noble Metals. J. Phys. Chem. C 2010, 114, 4473−4478. (23) Andersson, K. J.; Calle-Vallejo, F.; Rossmeisl, J.; Chorkendorff, I. Adsorption-Driven Surface Segregation of the Less Reactive Alloy Component. J. Am. Chem. Soc. 2009, 131, 2404−2407.

467


Rational Design of Competitive Electrocatalysts for Hydrogen Fuel

Recommend Documents