Ir

HSa

b

a

ARRAA

KAEOF

1

uaip[iemiac

licy

g

h0

Catalysis Today 263 (2016) 11–15

Contents lists available at ScienceDirect

Catalysis Today

j our na l ho me page: www.elsev ier .com/ locate /ca t tod

mportance of Pd monomer pairs in enhancing the oxygen reductioneaction activity of the AuPd(1 0 0) surface: A first principles study

yung Chul Hama,b,∗, Gyeong S. Hwanga,∗∗, Jonghee Hanb, Sung Pil Yoonb,uk Woo Namb, Tae Hoon Limb

Department of Chemical Engineering, The University of Texas at Austin, Austin, Texas, 78712, USAFuel Cell Research Center, Korea Institute of Science and Technology (KIST), Seoul, Korea

r t i c l e i n f o

rticle history:eceived 2 May 2015eceived in revised form 15 July 2015ccepted 30 July 2015vailable online 21 August 2015

eywords:

a b s t r a c t

Based on density functional theory calculations, we present that pairs of 1st nearest Pd monomers play animportant role in significantly enhancing the oxygen reduction reaction (ORR) on the AuPd(1 0 0) surface.While the catalytic ORR activity tends to be sensitive to the surface atomic ordering, we find that the Pdmonomer pairs lead to a substantial reduction in the activation barrier for O/OH hydrogenation with nosignificant suppression of O–O bond scission, thereby considerably lowering the overall activation energyfor the ORR as compared to the case of pure Pd(1 0 0). On the other hand, an isolated Pd monomer tends

uPd(1 0 0)nsemblesRRirst-principles

to greatly suppress the O–O bond cleavage reaction, which in turn slows down the ORR kinetics. Unlikethe monodentate adsorption of O2 on an isolated Pd monomer, the pairing of Pd monomers allows O2

adsorption in a bidentate configuration and consequently facilitating O–O bond scission. However, thebarrier for OH hydrogenation at each Pd site shows no significant change between the isolated and pairedcases, while it is noticeably lower than the pure Pd case.

. Introduction

Bimetallic gold-palladium (AuPd) catalysts have been tested andsed for various chemical reactions such as production of vinylcetate monomer [1], low temperature oxidation of carbon monox-de [2–5], selective oxidation of formic acid [6–8], and selectiveroduction of hydrogen peroxide (H2O2) from molecular O2 and H26,9–11]. Recent studies suggest that the surface activity of AuPds governed by the various alloying effects such as the so-calledlectronic (ligand) effects [4,6,12–15] (electronic state change byetal–metal interactions) and ensemble (geometric) effects (mod-

fication of catalytic activity by the unique arrangement of surfacetoms) [16–20]. However, the underlying mechanism of AuPd alloyatalysis still remains unclear.

It has been theoretically and experimentally found that an iso-ated Pd monomer surrounded by Au atoms is a good site for

ncreasing the rate of O/OH hydrogenation reaction but signifi-antly slowing down the kinetics of O2 bond cleavage, therebyielding the significant enhancement of selective H2O2 formation

∗ Corresponding author. Tel.: +82 2 958 5889; fax: +82 2 958 5199.∗∗ Corresponding author. Tel.: +1 512 471 4847; fax: +1 512 471 7060.

E-mail addresses: hchahm@kist.re.kr, ham.hyungchul@gmail.com (H.C. Ham),shwang@che.utexas.edu (G.S. Hwang).

ttp://dx.doi.org/10.1016/j.cattod.2015.07.054920-5861/© 2015 Elsevier B.V. All rights reserved.

© 2015 Elsevier B.V. All rights reserved.

[21–25]. The reactivity of Pd monomer in the AuPd catalyst can bealso altered due to additional geometrical factors. For instance, ittends to be changed by the presence of low-coordination surfaceatoms and the strain imposed on the outer-layer atoms, in asso-ciation with the size and shape of nanoparticle catalysts and thelattice parameter mismatch between the substrate and the adlayer[13,26–28]. The different surface facets such as (1 1 1) and (1 0 0)[29] may also affect the reactivity of isolated Pd monomers, leadingprobably to the change in catalysis of the oxygen reduction reac-tion (ORR). For example, according to a recent density functionaltheory (DFT) calculation [23], ORR strongly depends on the sur-face arrangement of Au and Pd atoms Pd(1 1 1). In particular, on Pdmonomers surrounded by less active Au atoms, H2O2 is selectivelyproduced, while on Pd dimer, the reactivity to H2O is predicted tobe enhanced compared to pure Pd(1 1 1). But there has been littleinvestigation concerning the interplay between different facet andPd ensemble.

In this paper, we investigate the arrangement of Pd atomsand its effect on ORR on the AuPd(1 0 0) surface, particularly therole of pairs of 1st nearest Pd, using periodic density functionaltheory (DFT) calculations. Here, we only consider small-sized Pd

ensembles including monomer, monomer pair, and dimer whichare experimentally found to be energetically favorable whenthe surface coverage of Pd is sufficiently low [30,31]. We firstexamine the relative stabilities among those small Pd ensembles

12 H.C. Ham et al. / Catalysis Today 263 (2016) 11–15

F ed Pda preseo ion of

cadmsm

2

otS(dvteoutacrbe

rwssc(rslig

ig. 1. Tilted side view of the model PdAu surfaces considered in this work, an isolat trimer (T), a tetramer (TE) and a Pd(1 0 0) slab (P). The green, gold, and gray balls ref the color information in this figure legend, the reader is referred to the web vers

onsidered and also calculate and compare the reaction energeticsnd barriers for O-O bond scission and O/OH hydrogenation on theifferent Pd sites. Next, we analyze the surface electronic structureodifications upon O2 adsorption to better understand the rea-

ons underlying the activity enhancement of pairs of 1st nearest Pdonomers toward O-O scission.

. Computational methods

The calculations reported herein were performed on the basisf spin polarized DFT within the generalized gradient approxima-ion (GGA-PW91) [32], as implemented in the Vienna Ab-initioimulation Package (VASP) [33]. The projector augmented wavePAW) method [34] with a plane-wave basis set was employed toescribe the interaction between core and valence electrons. Thealence configurations employed to construct the ionic pseudopo-entials are: 5d10 6s1 for Au, 4d9 5s1 for Pd, and 2s2 2p4 for O. Annergy cut-off of 350 eV was applied for the planewave expansionf the electronic eigenfunctions. For Brillouin zone sampling, wesed a (3 × 3 × 1) Monkhorst-Pack mesh of k points to determinehe optimal geometries and total energies of systems examined,nd increased the k-point mesh size up to (8 × 8 × 1) to re-evaluateorresponding electronic structures. Reaction pathways and bar-iers were determined using the climbing-image nudged elasticand method (c-NEBM) [35] with eight intermediate images forach elementary step.

For a model surface, we used a supercell slab that consists of aectangular 4 × 4 surface unit cell with four atomic layers each ofhich contains 16 atoms. For each AuPd surface model, the topmost

urface layer that is overlaid on a three-layer Pd (1 0 0) slab containselected Pd ensembles including an isolated Pd monomer (indi-ated by 1M), a pair of 1st nearest monomers (1st 2MP), a dimerD), a trimer (T), and a tetramer (TE) (see Fig. 1). A slab is sepa-ated from its periodic images in the vertical direction by a vacuum

pace corresponding to seven atomic layers. While the bottom twoayers of the four-layer slab were fixed at corresponding bulk pos-tions, the upper two layers were fully relaxed using the conjugateradient method until residual forces on all the constituent atoms

monomer (indicated by 1M), a pair of 1st nearest monomers (1st 2MP), a dimer (D),nt surface Pd, surface Au, and subsurface Pd atoms, respectively. (For interpretationthe article.)

become smaller than 5 × 10−2 eV/Å. The lattice constant for bulkPd is predicted to be 3.95 A, which is virtually identical to previousDFT-GGA calculations and also in good agreement with the exper-imental value of 3.89 A. The binding energy (Eb) of O2 is calculatedby Eb = E(M) + E(O2) − E(O2/M), where E(O2/M), E(M), and E(O2) rep-resent the total energies of the O2/slab system, the bare slab, andan isolated triplet O2 molecule in the gas state, respectively.

3. Results and discussion

3.1. Relative stability of small Pd ensembles

Before looking at the reactivity of various Pd ensembles towardORR, we first examined the relative stability of Pd ensembles on the(1 0 0) surface. Table 1 summarizes the predicted formation ener-gies per Pd atom for the 1M, 1st 2MP, D, T, and TE ensembles,given by Ef = [EPdAu − EAu + NPd × (EAu-bulk − EPd-bulk)]/NPd, whereEPdAu, EAu, EAu-bulk, and EPd-bulk represent the total energies ofPdAu/Pd(1 0 0) [Pd ensembles are overlaid on a three-layer Pd(1 0 0) slab], Au/Pd(1 0 0) [Au monolayer are overlaid on a three-layer Pd (1 0 0) slab], bulk Au (per atom), and bulk Pd (per atom),respectively, and NPd indicates the number of Pd atoms on a givenPdAu surface. Here, the smaller the formation energy, the higherthe stability of Pd ensembles.

The calculation results suggest that 1M, D, T, and TE in orderof decreasing stability; that is, Pd atoms would have a tendencyto remain isolated, rather than form aggregates in the AuPd (1 0 0)surface, consistent with recent theoretical and experimental results[30,36,37]. In addition, our calculations show that the Ef of 1st 2MPis comparable to that of 1M, implying that 1st 2MP can exist stablyfor surface catalytic reactions. Hence, we only considered the 1M,1st 2MP, and D sites for the study of O-O bond scission and O/OHhydrogenation, with comparison to the pure Pd(1 0 0) case.

3.2. O-O bond scission

In this study, we examined the three elementary steps(the so-called dissociative mechanism) for the ORR in acidic

H.C. Ham et al. / Catalysis Today 263 (2016) 11–15 13

Table 1Calculated formation energies (in eV) of various Pd ensembles on AuPd(1 0 0) surface. The green and yellow represent the Pd and Au atoms, respectively.

1M 1st 2MP D T TE

0.42 0.43

c[gioi[

ateas

sb(ptcgOa

ia1bteSh

isbPOitmOuale

soopb

Fig. 2. Predicted molecular configurations of the initial, transition and final statesin the O-O bond scission and O/OH hydrogenation reactions at the 1st 2MP site. Red,yellow, green, and white balls indicate O, Au, Pd, and H atoms, respectively. (Forinterpretation of the color information in this figure legend, the reader is referredto the web version of the article.)

0.39 0.39 0.42

ondition: (A) O2 adsorption [O2(g) → O2], (B) O2 scissionO2 → O + O], (C) O hydrogenation [O + H → OH], and (D) OH hydro-enation [OH + H → H2O(g)]. Although the ORR is complicated andts detailed mechanism still remains under debate, the comparisonsf these scission and hydrogenation reactions would give importantnsight into the activity of different ensemble sites toward the ORR23,38,39].

Table 2 summarizes the predicted total energy changes (�E) andctivation barriers (Ea) for O-O bond scission and O/OH hydrogena-ion reactions. The results clearly show that 1st 2MP significantlynhances the ORR as compared to pure Pd(1 0 0), while for the 1Mnd D cases the opposite is true. Next, we will discuss each reactiontep in detail.

The O2 binding energies (Eb, with respect to the triplet groundtate of gas phase O2) of small Pd ensembles are predicted toe much smaller as compared to the pure Pd(1 0 0) case, i.e., P1.73 eV) > D (1.10 eV) > 1st 2MP (0.60 eV) > 1M (0.48 eV). It is worthointing out that 1st 2MP has a noticeably higher Eb (by 0.12 eV)han 1M. According to our Bader charge analysis, the amount ofharge transferred to the adsorbed O2 from the surface is muchreater in 1st 2MP (0.63e−) than in 1M (0.32e−); consequently, the-O bond is more elongated (1.39 A in 1st 2MP and 1.29 A in 1M)s the transferred charge fills the antibonding O2 2p states.

The O2 scission reaction is predicted to be exothermic by 0.52 eVn 1st 2MP, whereas endothermic by 0.02 eV in 1M. Our calculationslso predict a substantially lower barrier for O-O bond cleavage inst 2MP (0.48 eV) than the 1M case (1.04 eV). Note that the largearrier for the O2 scission reaction in 1M causes an increase inhe formation of H2O2 rather than H2O. According to the reactionnergetics/barriers in the associative mechanism (see Table S1 inupporting Information), the selectivity for H2O2 formation is theighest in 1M.

The enhanced kinetics of O-O bond cleavage on 1st 2MP over 1Ms related to the unique arrangement of Pd monomers on the (1 0 0)urface. As illustrated in Fig. 2, O2 on 1st 2MP can be stabilized in theridge-hollow-bridge (b-h-b) configuration by being bonded to twod atoms, indicating that both Pd atoms in 1st 2MP are involved in2 adsorption. This is possible because the Pd-Pd distance of 3.95 A

n 1st 2MP is sufficiently short that O2 adsorbed can be in a biden-ate configuration. Note that the distance between two nearest Pd

onomers in the (1 1 1) surface is much longer (4.83 A) such that2 prefers to adsorb onto a single Pd atom in a monodentate config-ration with Eb of 0.25 eV. Moreover, the O atoms in the transitionnd final states on 1st 2MP can be significantly stabilized by beinginked to two Pd atoms, leading to a reduction in the activationnergy barrier compared to the 1M case (Figs. 2 and 3).

The enhanced O2 activation on 1st 2MP can also be demon-trated by the analysis of the local density of states (LDOS) projected

n the 2p orbital of O2 adsorbed on a single Pd monomer (1M), a pairf 1st nearest monomers (1st 2MP), a dimer (D), a trimer (T) and aure Pd(1 0 0) (P) (Here, the LDOS of a tetramer (TE) is not shownecause it has a similar LDOS to a trimer case). As displayed in Fig. 4,

Fig. 3. Predicted potential energy diagram on various Pd ensembles forO2(g) + 2H2(g) → 2H2O(g) reaction.

14 H.C. Ham et al. / Catalysis Today 263 (2016) 11–15

Table 2Calculated total energy changes (�E) and activation barriers (Ea in parenthesis) for O-O bond scission and O/OH hydrogenation reactions on various Pd ensembles withrespect to fully separated co-adsorbed species. The symbol “P” indicates the pure Pd(1 0 0) slab. The All energy values are given in eV.

1M 1st 2MP D P Pt(1 1 1)

(A) O2(g) → O2 −0.48 (–) −0.60 (–) −1.10 (–) −1.73 (–) −0.83 (–)(B) O2 → O + O 0.02 (1.04) −0.52 (0.48)

(C) O + H → OH −1.40 (0.34) −0.92 (0.48)

(D) OH + H → H2O(g) −0.67 (0.68) −0.54 (0.63)

Fig. 4. Local density of states projected onto the 2p orbital of O2 adsorbed on a singleP(l

fcalapficaOtitmt

3

n1ftihlimb

tPsitc

gD

d monomer (1M), a pair of 1st nearest monomers (1st 2MP), a dimer (D), a trimerT) and a pure Pd(1 0 0) (P). The 2p orbital of gas state O2 is also presented. The Fermievel (Ef) is marked by the vertical line.

or the Pd monomer pairs (1st 2MP), trimer (T) and pure Pd(1 0 0)ases, the spin-up and spin-down states are nearly degenerate with

magnetic moment of � ≈ 0.02, indicating the formation of peroxo-ike O2 by completely filling the antibonding 2p orbital (�g*) ofdsorbed O2 by the charge transfer from the surface. Here, O2 isreferentially adsorbed in the bridge-hollow-bridge (b-h-b) con-guration. On the contrary, for the single Pd monomer and dimerases, we predict the significant magnetic moment of � ≈ 0.99 (1M)nd � ≈ 0.50 (D), implying the superoxo-like character of the bound2 by partially filling antibonding 2p orbital (�g*) by less charge

ransfer from surface than the 1st 2MP, T and P cases. Here, O2s preferentially adsorbed in the top-bridge-top (t-b-t) configura-ion. This demonstrates that both Pd atoms in a pair of 1st nearest

onomers (1st 2MP) can play an key role in significantly enhancinghe O2 activation on AuPd(1 0 0) surface.

.3. O/OH hydrogenation

Looking at O/OH hydrogenation, our calculations predict veryoticeable enhancement in the kinetics on 1M (0.34/0.68 eV) andst 2MP (0.48/0.63 eV) when compared to the pure Pd(1 0 0) sur-ace (0.54/0.95 eV), while the O/OH hydrogenation barriers tendo increase on D (0.71/1.03 eV). This suggests that Pd monomersn both 1M and 1st 2MP can contribute to facilitating the O/OHydrogenation reactions; that is, the monomer pair can behave

ike an isolated monomer for O/OH hydrogenation. It would be alsonteresting to note that the rate-limiting step on each ensemble

ay change considering the significant differences in the activationarriers for the O2 scission and OH hydrogenation reactions.

First, we find that for the Pd(1 0 0) surface the OH hydrogena-ion reaction is rate-limiting with a barrier of 0.95 eV, while for thet(1 1 1) surface, the O hydrogenation reaction is rate-determiningtep with a barrier of 1.06 eV, indicating the improved ORR kineticsn Pd(1 0 0) over Pt(1 1 1). This agrees well with a recent experimen-al result where the ORR activity in Pd(1 0 0) is higher than Pt(1 1 1)

ase [40].

Next, looking at Pd ensembles cases, we find that the OH hydro-enation reaction is rate-limiting on 1st 2MP (Ea = 0.63 eV), and

(1.03 eV), whereas the O2 scission reaction is on 1M (1.04 eV).

−0.54 (0.89) −0.89 (0.15) −1.77 (0.56)−0.83 (0.71) −0.55 (0.54) 0.00 (1.06)

0.05 (1.03) 0.14 (0.95) −0.30 (0.34)

Considering also the activation barriers, we could expect that thekinetics of ORR can be significantly enhanced on 1st 2MP comparedto pure Pd(1 0 0), while the opposite is true for the 1M and D cases.

The kinetics O/OH hydrogenation can be correlated to the rel-ative binding strengths of OH/O/H2O. Our calculations show nosignificant variation in the binding energy of OH at the bridgesites of 1M (Eb = 2.41 eV) and 1st 2MP (2.40 eV) and H2O at thetop sites 1M (0.31 eV) and 1st 2MP (0.32 eV). For O adsorption,the hollow site in 1st 2MP (Eb = 3.99 eV) tends to be energeticallymore favorable than that in 1M (Eb = 3.66 eV). As such, the pre-dicted exothermicties for the hydrogenation reactions are smallerin 1st 2MP than 1M (see Table 2). Overall, our study clearly high-lights that the unique arrangement of Pd monomers on the (1 0 0)surface, especially 1st 2MP, can contribute to significantly enhanc-ing the ORR by facilitating O-O bond scission without substantialsuppression of O/OH hydrogenation.

4. Conclusion

We present the impact of pairs of 1st nearest Pd monomers onthe catalytic ORR activity of the AuPd(1 0 0) surface based on DFTcalculations. Our study suggests that the ORR strongly depends onthe distribution of Pd atoms in the AuPd(1 0 0) surface. Especially,the pair of 1st nearest Pd monomers can greatly enhance the kinet-ics of ORR compared to the pure Pd(1 0 0) surface by substantiallyreducing the kinetics of the rate-determining OH hydrogenationreaction without the significant suppression of the O-O bond scis-sion. However, for the single Pd monomer case, the barrier for O-Obond scission is largely increased, which results in the reduction ofORR kinetics. This activity enhancement of the pair of 1st nearestPd monomers over a single Pd monomer toward ORR is due to thelarge stabilization of O2/O by being linked to two Pd atoms in thecourse of O-O bond breaking reaction and no significant variationof OH binding energies in O/OH hydrogenation reaction. Here, OHis adsorbed at the bridge site by being bonded with one Pd and oneAu atom rather than at the hollow site by being connected with twoPd atoms. This study also highlights that, when designing the AuPdnanocatalysts, the preferential exposure of (1 0 0) facet in surfacecan be one of key requirements to enhancing the ORR.

Acknowledgement

This work was supported by the KIST institutional programof Korea Institute of Science and Technology (2E25412) and theR.A. Welch Foundation (F-1535). The authors also thank the TexasAdvanced Computing Center for use of their computing resources.

Appendix A. Supplementary data

Supplementary material related to this article can be found,in the online version, at http://dx.doi.org/10.1016/j.cattod.2015.07.054

References

[1] S. Venkatachalam, T. Jacob, Phys. Chem. Chem. Phys. 11 (2009) 3263–3270.[2] J.A. Boscoboinik, C. Plaisance, M. Neurock, W.T. Tysoe, Phys. Rev. B 77 (2008).

sis To

[

[[[

[[

[[

[

[[

[

[[

[

[

[[[[[

[[[[[[[[38] H.C. Ham, D. Manogaran, K.H. Lee, K. Kwon, S.A. Jin, D.J. You, C. Pak, G.S.

Hwang, J. Chem. Phys. 139 (2013).

H.C. Ham et al. / Cataly

[3] H.C. Ham, J.A. Stephens, G.S. Hwang, J. Han, S.W. Nam, T.H. Lim, J. Phys. Chem.Lett. 3 (2012) 566–570.

[4] V. Pallassana, M. Neurock, L.B. Hansen, B. Hammer, J.K. Norskov, Phys. Rev. B60 (1999) 6146–6154.

[5] J. Zhang, H.M. Jin, M.B. Sullivan, F. Chiang, H. Lim, P. Wu, Phys. Chem. Chem.Phys. 11 (2009) 1441–1446.

[6] N.N. Edwin, J.K. Edwards, A.F. Carley, J.A. Lopez-Sanchez, J.A. Moulijn, A.A.Herzing, C.J. Kiely, G.J. Hutchings, Green Chem. 10 (2008) 1162–1169.

[7] D.A. Bulushev, S. Beloshapkin, J.R.H. Ross, Catal. Today 154 (2010) 7–12.[8] H. Ohtani, M.A. Vanhove, G.A. Somorjai, Surf. Sci. 187 (1987) 372–386.[9] B.E. Solsona, J.K. Edwards, P. Landon, A.F. Carley, A. Herzing, C.J. Kiely, G.J.

Hutchings, Chem. Mater. 18 (2006) 2689–2695.10] J.K. Edwards, A.F. Carley, A.A. Herzing, C.J. Kiely, G.J. Hutchings, Faraday

Discuss. 138 (2008) 225–239.11] C. Samanta, Appl. Catal. A-Gen. 350 (2008) 133–149.12] J. Kitchin, J. Nørskov, M. Barteau, J. Chen, Phys. Rev. Lett. 93 (2004).13] H.C. Ham, G.S. Hwang, J. Han, S.W. Nam, T.H. Lim, J. Phys. Chem. C 114 (2010)

14922.14] R. Hoffmann, Rev. Modern Phys. 60 (1988) 601–628.15] J.A. Stephens, H.C. Ham, G.S. Hwang, J. Phys. Chem. C 114 (2010)

21516–21523.16] J.P. Perdew, Y. Wang, Phys. Rev. B 45 (1992) 13244–13249.17] P.B. Balbuena, S.R. Calvo, E.J. Lamas, P.F. Salazar, J.M. Seminario, J. Phys. Chem.

B 110 (2006) 17452–17459.

18] J.A. Rodriguez, P. Liu, J. Hrbek, J. Evans, M. Perez, Angew. Chem. Int. Ed. 46

(2007) 1329–1332.19] P. Liu, J.K. Norskov, Phys. Chem. Chem. Phys. 3 (2001) 3814–3818.20] F. Maroun, F. Ozanam, O.M. Magnussen, R.J. Behm, Science 293 (2001)

1811–1814.

[[

day 263 (2016) 11–15 15

21] F. Pittaway, L.O. Paz-Borbon, R.L. Johnston, H. Arslan, R. Ferrando, C. Mottet, G.Barcaro, A. Fortunelli, J. Phys. Chem. C 113 (2009) 9141–9152.

22] A. van de Walle, G. Ceder, J. Phase Equilibria 23 (2002) 348–359.23] H.C. Ham, J.A. Stephens, G.S. Hwang, J. Han, S.W. Nam, T.H. Lim, Catal. Today

165 (2011) 138–144.24] L.K. Ouyang, G.J. Da, P.F. Tian, T.Y. Chen, G.D. Liang, J. Xu, Y.F. Han, J. Catal. 311

(2014) 129–136.25] H.C. Ham, G.S. Hwang, J. Han, S.W. Nam, T.H. Lim, J. Phys. Chem. C 113 (2009)

12943–12945.26] P. Vassilev, M.T.M. Koper, J. Phys. Chem. C 111 (2007) 2607–2613.27] J.J. Burton, E.S. Machlin, Phys. Rev. Lett. 37 (1976) 1433–1436.28] B. Hammer, Y. Morikawa, J.K. Norskov, Phys. Rev. Lett. 76 (1996) 2141–2144.29] N.M. Markovic, P.N. Ross, Surf. Sci. Rep. 45 (2002) 121–229.30] P. Han, S. Axnanda, I. Lyubinetsky, D.W. Goodman, J. Am. Chem. Soc. 129

(2007) 14355–14361.31] F. Gao, Y.L. Wang, D.W. Goodman, J. Am. Chem. Soc. 131 (2009) 5734–5735.32] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.33] S. Stolbov, M.A. Ortigoza, T.S. Rahman, J. Phys.-Cond. Matter 21 (2009).34] P.E. Blochl, Phys. Rev. B 50 (1994) 17953–17979.35] E.D. German, M. Sheintuch, J. Phys. Chem. C 112 (2008) 14377–14384.36] Z.J. Li, F. Gao, W.T. Tysoe, J. Phys. Chem. C 114 (2010) 16909–16916.37] J.A. Stephens, G.S. Hwang, J. Phys. Chem. C 115 (2011) 21205–21210.

39] J. Greeley, J.K. Norskov, J. Phys. Chem. C 113 (2009) 4932–4939.40] S. Kondo, M. Nakamura, N. Maki, N. Hoshi, J. Phys. Chem. C 113 (2009)

12625–12628.