Density Functional Theory Studies of the Methanol ... - ACS Publications

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES

Article

Density Functional Theory Studies of the Methanol Decomposition Reaction on Graphene-Supported Pt Nanoclusters 13

Raymond J. Gasper, and Ashwin Ramasubramaniam J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.6b04557 • Publication Date (Web): 18 Jul 2016 Downloaded from http://pubs.acs.org on July 25, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Density Functional Theory Studies of the Methanol Decomposition Reaction on Graphene-Supported Pt13 Nanoclusters Raymond J. Gasper (1) and Ashwin Ramasubramaniam(2),* (1)

Department of Chemical Engineering,

University of Massachusetts, Amherst, MA 01003, USA. (2)

Department of Mechanical and Industrial Engineering,

University of Massachusetts, Amherst, MA 01003, USA. Abstract Defective graphene has been shown experimentally to be an excellent support for transitionmetal electrocatalysts in direct methanol fuel cells. Prior computational modeling has revealed that the improved catalytic activity of graphene-supported metal clusters is in part due to increased resistance to catalyst sintering and to CO poisoning, but the increased reaction rate for the methanol decomposition reaction (MDR) is not yet fully explained. Using density functional theory, we investigate the adsorption and reaction thermodynamics of MDR intermediates on defective graphene-supported Pt13 nanoclusters with realistic, low-symmetry morphologies. We find that the support-induced shifts in catalyst electronic structure correlate well with an overall change in adsorption behavior of MDR intermediates and that the reaction thermodynamics are modified in a way that suggests the potential of greater catalytic activity. We also show that adsorption energy predictors established for traditional heterogeneous catalysis studies of MDR on macroscopic crystalline facets are equally valid on catalyst nanoclusters (supported or otherwise) with irregular, low-symmetry surface morphologies. Our studies provide theoretical insights into experimental observations of enhanced catalytic activity of graphene-supported Pt

1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 42

nanoclusters for MDR and suggest promising avenues for further tuning of catalytic activity through engineering of catalyst–support interactions.

Keywords: methanol decomposition reaction, graphene, heterogeneous catalysis, density functional theory, d-band center * Corresponding author: [email protected]; Phone: +1-413-545-4866


Page 3 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Graphene has been demonstrated to be a very effective support for electrocatalysts for methanol and hydrogen fuel cells.1–7 In particular, for direct-methanol fuel cells, graphenesupported platinum clusters have been shown to have high catalytic activity for the methanol decomposition reaction (MDR), and increased resistance to both sintering and CO poisoning.1–7 The details of the binding of Pt nanoclusters to defective graphene supports have been studied extensively. Via computational and experimental work, Pt nanoclusters have been shown to adsorb strongly to point-defect sites in the graphene support.5,8,9 Upon binding at point defects, significant charge transfer takes place from the Pt nanoclusters to the graphene support, correspondingly modifying the electronic structure of the nanoclusters.10–14 This phenomenon suggests the use of predictors based upon electronic structure, in particular the d-band center model, which have been used to great effect to allow exploration of pure transition metal surfaces and surface alloys for use as catalysts.15–21 Other predictors for catalytic activity are based on scaling relationships, which correlate, for example, the adsorption of a molecule on different surfaces to the adsorption energy of a simpler probe molecule on the same surface. These have been demonstrated to be accurate in a wide variety of applications.16,17,22–24 In previous work, we showed that the d-band center can be used effectively as a predictor for CO adsorption on unsupported or graphene-supported Pt nanoclusters, but did not explore any of the MDR intermediates or provide any investigation of adsorption energy scaling relationships.25 The reduction of CO adsorption energies via support effects offers a possible explanation for the higher tolerance to CO poisoning that supported Pt nanoclusters exhibit, while the strong binding of Pt nanoclusters to defects explains resistance to sintering. However, neither of these observations is sufficient to explain fully the higher catalytic activity for the methanol 3 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 42

decomposition reaction (MDR) that is observed in experiments. Thus, the primary goal of this work is to present a detailed density functional theory (DFT) study of the adsorption of MDR intermediates on Pt13 nanoclusters with the goal of understanding the role of support effects on reaction thermodynamics. As Pt13 nanoclusters (with or without supports) have already been studied in detail, 9,11,25 they serve as a model reference system for further exploration of catalytic reactions. We also seek to further extend the use of the d-band center model and adsorption energy scaling relationships from their traditional applications for catalysis at macroscopic crystalline surfaces to more complex supported-cluster reaction networks, thereby expanding the generality of their potential use for rational catalyst design. Calculations of adsorption energies and reaction thermodynamics on clusters require overcoming difficulties that are not present on high-symmetry surfaces. Due to there being few (if any) symmetry-equivalent adsorption sites on low-symmetry clusters, some form of statistical sampling must be undertaken, which magnifies the effort required to understand a reaction on a single cluster. In addition, clusters adopt different geometries and electronic structures on different supports, requiring separate calculations even for simple quantities such as adsorption energies. The combination of these factors makes exploration of cluster-support systems for their catalytic potential extraordinarily time- and resource-consuming. Therefore, in order to achieve a broad and/or in-depth exploration of supported catalyst design, circumventing these extensive ab initio calculations via robust predictors is highly desirable. In this work, we demonstrate that the d-band center model can be extended for use not only as a predictor for CO adsorption, as we have done before,25 but more generally as a predictor for the qualitative adsorption behavior of the methanol decomposition reaction across nanoclusters with varying degrees of support interactions.


Page 5 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The MDR has been extensively investigated,24,26–33 in part due to its value for understanding and improving catalytic performance for direct methanol fuel cells. Pioneering work in the field was performed by Greeley and Mavrikakis, who investigated the reaction on Pt(111) in extensive detail.34,35 They showed via a combination of DFT and microkinetic modeling that the most likely pathway for MDR on a Pt(111) single-crystal surface under ultra-high vacuum is H3COH H2COH HCOH CO (henceforth, Path 1), with the first dehydrogenation step being ratedetermining. Two other viable, but less favorable, pathways were identified: (i) H3COH H3CO H2CO HCO CO (Path 2), with the first step also being the rate determining step, and (ii) H3COH H2COH H2CO HCO CO (Path 3), with the first and second steps having approximately the same barriers. In addition to the Pt(111) surface, the MDR pathway has also been fully or partially identified on stepped and defective platinum surfaces,20 B12N12 nanocages,36 the Cu(111) surface,37 the Cu(110) surface,27 the ZnO(101̅0) surface,38 among several others.30,31,33,39–42 Kandoi et al.

28

showed that the experimental performance of the

Pt(111) surface catalyst for the MDR can be predicted qualitatively by a combination of ab initio calculations and microkinetic modeling, along the way reaffirming the dominant pathway established by Greeley and Mavrikakis. In this work, we characterize the adsorption energies and reaction thermodynamics of the three pathways previously identified, considering especially the relationships between the most likely pathways. We believe that the level of understanding in the field, along with the previously identified gap in understanding regarding the effect of supports, makes this reaction-catalyst-support system ideally suited for exploration of predictors for supported catalysis.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 42

Our computational methodology is explained in Section II. Results and discussion on the effect of support binding on the adsorption and reaction thermodynamics of MDR intermediates are presented in Section III. Conclusions and closing remarks are presented in Section IV. 2. Methodology Spin-polarized DFT calculations were performed using the Vienna Ab Initio Simulation Package (VASP).43–46 The core and valence electrons were treated using the projector-augmented wave function method.47,48 The Perdew-Burke-Ernzerhof generalized-gradient approximation (PBE) was used to describe electron exchange and correlation.49 From convergence studies, we chose a kinetic energy cutoff of 400 eV and a first-order Methfessel-Paxton smearing50 of 0.05 eV. All atomic positions were relaxed with a force tolerance of 0.02 eV/Å. Initial low-energy Pt13 cluster configurations (supported or unsupported) were generated via simulated annealing as reported in Ref. 11. The supports considered were 8×8 single-layer graphene sheets that were either defect free or containing a point defect (vacancy or divacancy) at the cluster adsorption site. Graphene planes were separated in the normal direction by 18 Å of vacuum to prevent spurious interaction with periodic images. The Brillouin zone was sampled by a 2×2×1 Γ-centered k-point mesh, chosen from convergence studies. Unsupported (Isolated) Pt13 clusters were modeled in an 18 Å cube using a single Γ point for reciprocal space sampling. Adsorption energies and reaction energies were sampled on five random sites on each cluster. Sampling sites were chosen randomly, with repeated sites or sites hindered by graphene proximity (high steric hindrance) being discarded. Initial configurations for adsorbed molecules were set up to be as close to the Pt(111) adsorption configuration as possible but these eventually relaxed to various local minima as dictated by the local structure of the adsorption sites. Charge transfer analyses were performed using a Bader analysis.51,52 6 ACS Paragon Plus Environment

Page 7 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Reference Pt(111) surface calculations were performed on a three-layer, 3×3 surface cell with the bottom two layers frozen in their bulk FCC geometry. Slabs of greater thickness with additional free layers were tested and not found to influence the adsorption energies. The slabs were separated by 18 Å of vacuum normal to the surface to prevent spurious interactions. A Γcentered, 5×5×1 k-point mesh was used for Brillouin zone sampling. Dipole-moment corrections were applied normal to the surface. For adsorption energy (Eads) calculations, the following definitions are employed throughout this paper: Supported Clusters: ‫ܧ‬௔ௗ௦ = ‫ܧ‬௉௧భయ ା௦௨௣௣௢௥௧ା௔ௗ௦௢௥௕௔௧௘ − ‫ܧ‬௉௧భయ ା௦௨௣௣௢௥௧ − ‫ܧ‬௔ௗ௦௢௥௕௔௧௘ ,

(1)

Isolated Cluster:

‫ܧ‬௔ௗ௦ = ‫ܧ‬௉௧భయ ା௔ௗ௦௢௥௕௔௧௘ − ‫ܧ‬௉௧భయ − ‫ܧ‬௔ௗ௦௢௥௕௔௧௘ ,

(2)

Pt(111) Surface:

‫ܧ‬௔ௗ௦ = ‫ܧ‬௉௧ሺଵଵଵሻା௔ௗ௦௢௥௕௔௧௘ − ‫ܧ‬௉௧ሺଵଵଵሻ − ‫ܧ‬௔ௗ௦௢௥௕௔௧௘ .

(3)

In Eq. 1, ‫ܧ‬௉௧భయ ା௦௨௣௣௢௥௧ା௔ௗ௦௢௥௕௔௧௘ is the total energy of the relaxed Pt13 cluster, graphene support, and adsorbate molecule; ‫ܧ‬௉௧భయ ା௦௨௣௣௢௥௧ is the total energy of the Pt13 cluster and graphene support relaxed again after adsorbate removal; and ‫ܧ‬௔ௗ௦௢௥௕௔௧௘ is the reference energy of the adsorbate isolated in vacuum. In Eq. (2), for the isolated cluster, ‫ܧ‬௉௧భయ ା௔ௗ௦௢௥௕௔௧௘ is the total energy of the relaxed Pt13 cluster plus adsorbate, ‫ܧ‬௉௧భయ is the energy of the isolated Pt13 cluster relaxed after adsorbate removal, and ‫ܧ‬௔ௗ௦௢௥௕௔௧௘ is the reference energy of the isolated adsorbate; Eq. (3) is analogous to Eq. (2) with Pt(111) substituting for the Pt13 cluster. For all molecules considered, the reference state is the molecule in vacuum; for a hydrogen atom, however, the reference state is half the energy of the H2 molecule. Note that as defined, more negative adsorption energies correspond to a stronger bond between the adsorbate and the catalyst. The thermodynamics of the MDR were calculated using the following definitions of reaction energies: 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 42

Supported Clusters: ଵ

തതതതതതതത ߂‫ܧ‬௥௫௡ = ‫ܧ‬௉௧భయ ା௦௨௣௣௢௥௧ା௣௥௢ௗ௨௖௧ + ܰு ቂ ‫ܧ‬ுమ,ሺ೒ሻ + ‫ܧ‬ ௔ௗ௦,ு ቃ − ‫ܧ‬௉௧భయ ା௦௨௣௣௢௥௧ା௥௘௔௖௧௔௡௧ , ଶ

(4)

Isolated Cluster: ଵ തതതതതതതത ߂‫ܧ‬௥௫௡ = ‫ܧ‬௉௧భయ ା௣௥௢ௗ௨௖௧ + ܰு ቂଶ ‫ܧ‬ுమ,ሺ೒ሻ + ‫ܧ‬ ௔ௗ௦,ு ቃ − ‫ܧ‬௉௧భయ ା௥௘௔௖௧௔௡௧ ,

(5)

Pt(111) Surface: ଵ

߂‫ܧ‬௥௫௡ = ‫ܧ‬௉௧ሺଵଵଵሻା௣௥௢ௗ௨௖௧ + ܰு ቂ ‫ܧ‬ுమ,ሺ೒ሻ + ‫ܧ‬௔ௗ௦,ு ቃ − ‫ܧ‬௉௧ሺଵଵଵሻା௥௘௔௖௧௔௡௧ , ଶ

(6)

where NH is the number of hydrogen atoms being removed from the intermediate in a particular reaction step (NH=1 in all steps except for the semi-concerted removal of two H atoms from HCOH where NH =2), ‫ܧ‬ுమ,ሺ೒ሻ is the DFT energy of an H2 molecule, and തതതതതതതത ‫ܧ‬௔ௗ௦,ு is the average adsorption energy of hydrogen on the system being considered [Pt13, the exact value can be used on Pt(111)]. By way of benchmarking, this approach was compared with the more conventional approach wherein the products of a reaction step (i.e., the MDR intermediate and the leaving H atom) are co-adsorbed on the catalyst surface and the reaction thermodynamics calculated with respect to this configuration. Selected tests on Pt(111) and on the isolated Pt13 cluster (Fig. S5) reveal that both approaches provide consistent and reliable estimates of the reaction thermodynamics. On the Pt(111) surface both approaches yield nearly identical results; on the isolated Pt13 cluster, the results coincide to within the bounds of statistical error. Thus, employing the average value for hydrogen adsorption energy facilitates reaction thermodynamics analyses at half the computational cost of the co-adsorption approach while also minimizing the influence of variations in energy of the relaxed cluster and support, which causes excessive noise in the calculations without lending improved understanding. No entropic contributions or zeropoint energies were included in this study.


Page 9 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3. Results and Discussion 3.1 Supported and Unsupported Reference Pt13 Clusters The graphene-supported Pt13 clusters used for this study are shown in Fig. 1. Initial configurations for the systems used were taken from previous work.11 The systems chosen for this study are an isolated Pt13 cluster, a pristine-graphene supported Pt13 cluster, a Pt13 cluster

Figure 1: Relaxed Pt13 clusters that are a) isolated, b) on a pristine graphene support, c) bound at a vacancy defect in graphene, and d) bound at a divacancy defect in graphene. (C – brown spheres; Pt – gray spheres) supported on a graphene sheet with a single vacancy, and a Pt13 cluster supported on a graphene sheet with an unreconstructed divacancy. These systems were chosen due to previous work indicating that they provided a broad range of support interaction, leading to significant shifts in d-band center and CO adsorption behavior between the systems.25 The support size was increased relative to our previous work to an 8×8 graphene supercell to accommodate larger adsorbed molecules without spurious image interaction. As discussed in previous work, with the increasing size of point defects in the graphene support, both the binding strength and charge transfer between the Pt cluster and graphene– support increase.11 This charge transfer induces a change in the electronic structure of the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 42

clusters, manifested by a downshift in the d-band center value. The d-band centers of the clusters from Fig. 1 are listed in Table 1, and reaffirm these conclusions. The d-band centers of the unsupported cluster and the pristine-graphene-supported cluster (weak cluster–support interactions) are significantly higher than that of the Pt(111) surface; however, sufficiently strong support interaction in the divacancy case lowers the d-band center below the Pt(111) level. The d-band model then suggests that the adsorption energy of molecular adsorbates on these catalysts should be correlated consistently with respect to the d-band shift between the differently supported Pt13 nanoparticles. Indeed, for the case of CO adsorption on graphene-supported or unsupported Pt13 clusters, we have shown previously that trends in adsorbate binding (as well as oxidation barriers) can be rationalized in terms of the d-band shifts.25,53 Through this metric we will further unravel the relationship between support effects and adsorption for MDR intermediates below.

Table 1: Positions of d-band centers (εd) of Pt13 clusters and Pt(111) surface (relative to the Fermi level). System

εd (eV)

Pt(111)

-2.58

Isolated cluster

-2.25

Cluster on pristine graphene

-2.42

Cluster bound at vacancy in graphene

-2.49

Cluster bound at divacancy in graphene

-2.62


Page 11 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3.2 Adsorption of MDR Intermediates

Figure 2: Adsorption of H3CO at randomly chosen sites on a vacancy-supported Pt13 cluster. (C – brown spheres; Pt – gray spheres; O – red spheres; H – white spheres).

Unlike macroscopic crystal surfaces or faceted nanoparticles, it is impossible to identify a small set of symmetry-equivalent adsorption sites on our low-symmetry, graphene-supported Pt13 nanoclusters. Thus, in order to understand the adsorption behavior, five randomly chosen sites were sampled on each cluster−support system. Sample converged configurations of various adsorbates on the Pt13 cluster and graphene support are displayed in Figs. 2 and 3. Despite variations in local cluster geometry, the most favorable site for H3CO adsorption on the supported clusters is the same as on Pt(111), namely, at the on-top site. Such agreement is not universally true, however, and variations in adsorption configurations, likely dictated by variations in local cluster morphology, are seen for other adsorbate molecules. This is seen, for example, in Fig. 3 where atomic configurations of various adsorbates are displayed at a single 11 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 42

adsorption site on a Pt13 cluster: H3COH, H3CO, HCO, and H2COH all adopt the same adsorption configuration as on a Pt(111) surface (on-top site), but H2CO does not; instead, H2CO, which adopts a Pt-C-O-Pt bridge configuration on Pt(111), can adopt either this configuration, an on-top configuration, or a Pt-C-Pt bridge configuration on Pt13 clusters. As another example, CO, which adsorbs most strongly to the fcc site on the Pt(111) surface (as per DFT calculations; the disagreement with experiment is well known54,55), adopts on-top, bridge, and hollow low-energy adsorption configurations on Pt13 clusters.

Figure 3: Top row: selected MDR intermediates on Pt(111) surface. Bottom row: same selected MDR intermediates on a single adsorption site of divacancy graphene-supported Pt13. a) H3COH, b) H3CO, c) HCO, d) H2CO, e) H2COH. (C – brown spheres; Pt – gray spheres; O – red spheres; H – white spheres) To extend understanding of the MDR on nanoclusters, we sample all intermediates of the favorable pathways for the methanol dehydrogenation reaction under UHV conditions. To do this we consider the following adsorbate molecules: H3COH, H3CO, H2CO, HCO, H2COH, HCOH, CO, and H. This selection is based on the molecules that participate in the most likely UHV reaction pathways identified by Greeley & Mavrikakis,35 Paths 1, 2, and 3 in our current 12 ACS Paragon Plus Environment

Page 13 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


nomenclature (see Sec. 1). In addition, monatomic carbon and oxygen are also studied to establish adsorption energy scaling relationships as was done previously for crystalline surfaces.16,17,22-24 Figure 4 displays the average adsorption energies of each MDR intermediate on all four Pt13 clusters (220 calculations in total) plotted as a function of the corresponding molecular adsorption energy on Pt(111). For each system (i.e., isolated cluster or cluster and support), a linear fit of all adsorption energies (not just average values) versus the respective Pt(111) values was performed, with best-fit parameters listed in Table 1. Despite statistical variation, the adsorption behavior for every system versus Pt(111) shows a very consistent trend: to a very good approximation, all linear fits essentially have the same slope as the dashed 1:1 line (Fig. 4) suggesting that the support induces rigid shifts in the adsorption energies of MDR intermediates on the Pt13 nanoclusters. These rigid shifts can be quantified by the y-intercepts of the best-fit lines and, as shown in the inset of Fig. 4, are inversely correlated to the calculated cluster d-band centers. In other words, as the cluster d-band center approaches that of Pt(111), so do the adsorption energies of MDR intermediates on the cluster (to within statistical variations). This is a noteworthy result as it suggests that by knowing (a) the adsorption energy of an intermediate on Pt(111), and (b) the d-band center for the cluster of interest (supported or unsupported), one may readily estimate the average adsorption energy for that intermediate on the cluster. Detailed statistical sampling over cluster adsorption sites is thus eliminated in favor of a simple scaling relationship. This finding could open up the potential use of support effects as a new control parameter for exploration when conducting high-throughput computational studies of supported catalysts. Systematic exploration of cluster size effects and their interplay with support effects will be reported elsewhere. Bifunctional support effects require special consideration but are not of relevance for our present study. (Consideration of more adsorbate



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 42

Figure 4: Adsorption energies of MDR intermediates on supported and unsupported Pt13 nanoclusters versus adsorption energies on Pt(111). Data points are averages of all sampled adsorption sites on that system and error bars are calculated from Student-T 95% confidence intervals. Lines represent linear fits of all (not just average) molecular adsorption energies for each Pt13 cluster. The dashed 1:1 line is plotted as a guide to the eye. The inset shows the shift (y-axis intercept) of the adsorption data fits for each system versus their respective d-band centers; the dashed line is a linear fit that serves as a guide to the eye. molecules that were not in favorable reaction pathways is provided in the Supporting Information, the general trends identified here are still applicable.)


Page 15 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 2: Linear best-fit parameters for adsorption energies (Eads) of MDR intermediates on Pt13 nanoclusters to their Pt(111) values (Fig. 4). System

Slope

y-Intercept (eV)

Isolated cluster

0.97

-0.55

Cluster on pristine graphene

1.01

-0.34

Cluster bound at vacancy in graphene

0.97

-0.32

Cluster bound at divacancy in graphene

0.94

-0.17

For completeness, the adsorption energies and trends in adsorption energy were checked against a multitude of other potential descriptors including charge transfer from the Pt13 cluster to the support, charge transfer from the adsorbate to the Pt13 cluster, the Fermi level shift of the supported Pt13 clusters versus the unsupported case and versus Pt(111), and the d-band center of the Pt13 clusters relative to the vacuum level; none of these descriptors proved useful. The excellence of the d-band center position (relative to the Fermi level) as a predictor of average adsorption behavior on catalyst nanoclusters appears to be unique.

3.3 Scaling Relationships on Supported nanoparticles Given the successful past use of adsorption energy scaling relationships for heterogeneous catalysis on macroscopic crystalline surfaces,16,22,24,56 it is natural to inquire whether such relations work on nanoclusters. To the best of our knowledge, beyond a recent series of studies of four-atom unsupported clusters,22,57,58 this question has received scant attention and certainly not for supported, low-symmetry nanoclusters. In Fig. 5, we display the adsorption energies of 15 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 42

the MDR intermediates on the unsupported and supported Pt13 nanoclusters as a function of the adsorption energy of either monatomic carbon or monatomic oxygen; the probe atom (C or O) was chosen depending upon whether the intermediate adsorbed most frequently to the cluster via a C-Pt or an O-Pt bond. In all cases, the most frequently observed cluster–adsorbate bond was the same as that on the Pt(111) surface. For H2CO, which was seen to bind via C, O, or both simultaneously, O was chosen as the probe molecule due to a superior agreement. As seen from Fig. 5, for the most part, the adsorption energy of an intermediate correlates very well with that of the relevant probe molecule leading to a set of robust scaling relationships. The correlation for HCO and H2COH appears somewhat less robust, but the outlying data for these two cases are contributed by the isolated Pt13 clusters. To the extent that the isolated Pt13 cluster merely sets the “zero” for probing support interactions and is not relevant as a catalyst, we deem these couple of outliers to be unimportant. The trend of adsorption energies for the C atom on the clusters may initially appear erratic, but actually is very similar to the results for carbon-binding molecules on the clusters. Within statistical error, adsorption energies for the isolated, pristine, and vacancy cases are fairly similar; adsorption in the divacancy case is significantly weaker though and in many cases the data even lie above the Pt(111) values. Just as for the d-band center model, the internal variation caused by the low-symmetry cluster sampling does not disrupt the general trends that make adsorption energy scaling relationships an effective predictor. As seen above, the scaling relationships are fairly robust for all adsorbates and cluster– support systems despite local variations in geometry and the corresponding variations in adsorption energies. By substituting the average C and O adsorption energies into the linear fit equations for the various adsorbates (see Fig. 5), one can “predict” the average adsorption energies of the MDR intermediates. The predicted adsorption energies for the MDR


Page 17 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5: Scaling relationships for adsorption energies of MDR intermediates versus adsorption energies of monatomic carbon or oxygen. Points are average adsorption energies for each system whereas linear fits are to all adsorption data. (The Pt(111) value for HCO overlaps with that for H2COH making the symbols indistinguishable.)

intermediates can then be compared to the DFT-calculated adsorption energies to assess the overall quality of the correlations. As seen from Fig. 6, the scaling relationships based on C and O adsorption energies (Fig. 5) function as excellent predictors across the entire spectrum of MDR intermediates and cluster–support systems. The deviation between values predicted by the 17 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 42

Figure 6: Adsorption energies of MDR intermediates on all systems predicted by fitted scaling relationships based on C and O adsorption energies versus their DFT calculated values. Error bars are Student-T 95% confidence intervals of the DFT-calculated adsorption energies. Equality line is plotted as a guide to the eye. scaling relationships and calculated from DFT is relatively small throughout the range of adsorption energies being considered. Indeed, the performance of these scaling relationships at predicting the average adsorption energies is surprisingly good given the variation in adsorption geometries and energies between systems and points to the general robustness of the chosen predictors, inspired from studies on crystalline facets, even for low-symmetry clusters.


Page 19 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3.4 Reaction Thermodynamics The thermodynamics of the MDR have been explored quite thoroughly on a variety of surfaces and through a variety of methods, as discussed in Sec. 1. Here we perform an initial investigation of the MDR thermodynamics on low-symmetry supported Pt13 nanoclusters. On Pt(111) Greeley and Mavrikakis identified the pathway H3COH H2COH HCOH CO (Path 1) as the most dominant one for the MDR.35 Two additional pathways, H3COH H3CO H2CO HCO CO (Path 2) and H3COH H2COH H2CO HCO CO (Path 3), were identified as possible but relatively noncompetitive as, in comparison to the other two pathways, the rate limiting step of Path 1 has both a significantly lower barrier and higher reverse barrier, yielding an effective rate that is orders of magnitude higher than Paths 2 and 3. COH was shown to be present as a spectator species, with high barriers of formation and decomposition, but also high adsorption energy, leading to a minor poisoning effect on the catalyst surface. Since we used a different exchange-correlation functional (and a different DFT program) than Greeley and Mavrikakis, we revisited their proposed MDR pathways on Pt(111) and obtained essentially the same conclusions. The calculated reaction thermodynamics (Eqs. 46) for the Pt(111) surface as well as for a Pt13 cluster bound at a divacancy in graphene are displayed in Fig. 7. (Results for other cluster–support systems are displayed in Figure S4.) While we have not calculated actual reaction barriers in this work, we can nevertheless compare

the

reaction

thermodynamics

between

the

Pt(111)

surface

and

the

supported/unsupported Pt13 clusters, and arrive at a preliminary set of conclusions. As predictors for reaction barriers, based on the Brønsted-Evans-Polyani (BEP) principle, have been shown to be quite robust for Pt(111)35 among other catalyst surfaces,17,40,59 and even for four-atom metal clusters,57,58 we hypothesize the existence of analogous relationships on our graphene-supported



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 42

Figure 7: Schematic and thermodynamics of the MDR on Pt(111) and on a Pt13 cluster bound at a divacancy in graphene. Brown circles represent carbon atoms, red circles represent oxygen atoms, grey circles represent hydrogen atoms, the grey bar represents the catalyst surface. All molecules/atoms involved are considered adsorbed until after the decomposition pathway is completed. The adsorbed hydrogens are shown stacked to the right for visual clarity. Blue arrows represent negative reaction energies, while red arrows represent positive reaction energies. The size of each arrow is in exact proportion to the magnitude of the reaction energy, which is indicated beside the arrow. Relaxed structures of all reaction intermediates on a single site for the divacancy-bound cluster can be found in Figure S4.

Pt13 clusters. (Rigorous calculations of BEP relationships for our Pt13 systems will be undertaken in future work.) The focus of our comparison for now is between the Pt(111) surface and the divacancy-bound cluster (Fig. 7), as the d-band center calculation and adsorption behavior show that the divacancy defect has the strongest effect on the supported Pt13 cluster. The thermodynamics of Path 1 are relatively unchanged between the Pt(111) surface and the 20 ACS Paragon Plus Environment

Page 21 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


divacancy-bound Pt13 cluster. The statistical nature of our analysis does mean that some reactive sites on the cluster will exhibit reaction energies lower than average, indicating the possibility of outlier sites that contribute significant fractions of the total activity, analogous to the way that step-edges and defects have been shown to contribute significant activity and dominate the overall reaction on larger, faceted nanoparticles.60,61 The other two pathways (Paths 2 & 3), however, present interesting differences from the Pt(111) surface. Path 2 (H3COH H3CO H2CO HCO CO) on Pt(111) is unfavorable due to higher reaction energy and barrier for the first reaction step (H3COH H3CO); however, on the divacancy-bound Pt13 cluster, this first step is less uphill. Because of these interactions we believe that the second pathway could be more active on the graphene-divacancy-bound Pt13 cluster than on the Pt(111) surface. Path 3 (H3COH H2COH H2CO HCO CO) is less active on Pt(111), once again because of the unfavorable and high barrier for the H2COH H2CO step. Again, for the divacancy-bound Pt13 cluster case this step is significantly less unfavorable, with average reaction energy of essentially zero, and thus Path 3 could become much more active on the graphene-divacancybound Pt13 cluster than on the Pt(111) surface. Overall, the interaction between the three MDR pathways appears to be significantly changed on the divacancy-bound Pt13 cluster with respect to the Pt(111) surface: instead of having one dominant pathway contributing most of the catalytic activity as on the Pt(111) surface, the reaction thermodynamics suggest a mildly dominant pathway (that is possibly more active on the divacancy-bound Pt13 cluster than on the Pt(111) surface), with two competing pathways also generating significant activity. As all reaction energies are statistical distributions rather than precise values on the low-symmetry supported Pt13 clusters, the possibility of contributed activity from extraordinarily active outlier sites on the cluster is also a matter of future interest.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 42

The thermodynamics and adsorption of COH are also of some interest in the graphenedivacancy-bound Pt13 case. On Pt(111), COH has a very high adsorption energy, a high barrier of formation, and a very high barrier for decomposition into CO and H. It was found that this combination led to COH being a minor poison to reaction, occupying roughly 10% of available surface sites.35 However, on the divacancy-bound Pt13 cluster, the energetics are very different. COH still has a high adsorption energy, but it is about 1 eV lower than the Pt(111) case (see Table S3 and Fig. S1). The reaction energy for forming COH from HCOH is much higher on the divacancy-bound Pt13 cluster than on Pt(111), becoming thermodynamically even more unfavorable. This higher reaction energy then suggests the potential of an even higher barrier of formation, and means that the backwards reaction becomes thermodynamically favorable. Finally, the reaction energy for decomposition of COH into CO and H is dramatically more favorable on the divacancy-bound Pt13 cluster than on the Pt(111) surface. Considering all of these factors, it seems plausible that less COH would be formed on the divacancy-bound Pt13 cluster to begin with, and any COH that is formed could be hydrogenated, dehydrogenated, or desorbed more easily than on the Pt(111) surface, having the net effect of increasing the availability of reaction sites. The other graphene-supported Pt13 clusters also exhibit interesting reaction thermodynamics. More information on their specific details can be found in Figure S3 and Table S5, but a few salient points will be discussed here. First, similarly to the graphene-divacancy-bound Pt13 case, it seems likely that on all the Pt13 clusters COH either would not be formed in appreciable amounts or would be able to hydrogenate/dehydrogenate to HCOH/CO, and not behave as a minor poison/spectator. Second, in general, most reaction steps have more negative reaction energies on all Pt13 clusters relative to Pt(111). Furthermore, the reaction energies are also


Page 23 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


correlated with the strength of adsorption of the reactants and products on the supported Pt13 clusters: weaker/stronger adsorption leads to smaller/larger reaction energies. To the extent that the adsorption energies of intermediates are in turn correlated with the cluster d-band centers, which are sensitive to support defects, the support effect is then manifested on the reaction thermodynamics through this sequence of correlations. This suggests the possibility of different rate-limiting steps for certain pathways on the clusters, and for overall lower barriers assuming that BEP relationships hold for this novel system. Finally, some of the reaction steps have fairly large uncertainties, which suggests the possibility that the non-uniformity of the clusters could offer a non-zero concentration of sites with much lower barriers, therefore increasing the effective reaction rate of the system in a way that the average values would not indicate; however, it is impossible currently to make any conclusions based on this and more in-depth analysis is surely required.

4. Conclusions We investigated the adsorption of MDR intermediates and examined the thermodynamics of several MDR pathways on Pt13 nanoparticles with varying amounts of interaction with a graphene support. The average adsorption energies of the intermediates of the MDR experienced a rigid shift relative to Pt(111) as a result of support interaction, and this shift could be correlated well with the d-band center of the system, which is in turn influenced by the strength of support interaction. In addition, scaling relationships and d-band center correlations established on highsymmetry crystal surfaces appear to be successful for predicting the average adsorption behavior of molecules on supported Pt13 clusters. A comparison of the MDR thermodynamics between the Pt(111) and divacancy graphene-supported Pt13 suggest shifts in the interactions between the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 42

favorable reaction pathways, and may indicate overall higher activity for graphene-supported Pt13 clusters compared to the Pt(111) surface. Overall, our studies indicate that defective graphene-supported Pt13 is a promising MDR catalyst and possesses qualitatively different adsorption and reaction thermodynamics from the Pt(111) surface, which may further explain the high activity seen in defective graphene-supported Pt nanoparticles. The findings in this work suggest new questions that should be addressed next to move results towards the goal of rational design of MDR catalysts. Specifically, it remains to be investigated whether the reaction barriers on the supported clusters are different than those on single-crystal surfaces, and if those differences match expectations from experimental work. The validity of BEP-like relationships between the reaction barriers and reaction thermodynamics for these supported-cluster systems remains a significant open question. Confirmation of BEP-like relationships would greatly facilitate computationally inexpensive calculation of the reaction barriers for the full reaction network on the low-symmetry supported Pt13 clusters, which could enable further investigation into the interplay of multiple support effects under realistic conditions. These and other questions will be addressed in future work.

Supporting Information Additional figures and data for adsorption of MDR intermediates; reaction thermodynamics on unsupported and all graphene-supported Pt13 clusters

Acknowledgements We gratefully acknowledge research funding from the U.S. Department of Energy under Award Number DE-SC0010610. Computational resources were provided by the Massachusetts Green


Page 25 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


High Performance Computing Center, and also by the National Energy Research Scientific Computing Center, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

References (1)

Antolini, E.; Lopes, T.; Gonzalez, E. R. An Overview of Platinum-Based Catalysts as Methanol-Resistant Oxygen Reduction Materials for Direct Methanol Fuel Cells. J. Alloys Compd. 2008, 461 (1-2), 253–262.

(2)

Antolini, E. Graphene as a New Carbon Support for Low-Temperature Fuel Cell Catalysts. Appl. Catal. B Environ. 2012, 123-124, 52–68.

(3)

Dong, L.; Gari, R. R. S.; Li, Z.; Craig, M. M.; Hou, S. Graphene-Supported Platinum and Platinum-Ruthenium Nanoparticles with High Electrocatalytic Activity for Methanol and Ethanol Oxidation. Carbon 2010, 48 (3), 781–787.

(4)

Yoo, E.; Okata, T.; Akita, T.; Kohyama, M. Enhanced Electrocatalytic Activity of Pt Subnanoclusters on Graphene Nanosheet Surface. Nano Lett. 2009, 9 (6), 2255–2259.

(5)

Kou, R.; Shao, Y.; Wang, D.; Engelhard, M. H.; Kwak, J. H.; Wang, J.; Viswanathan, V. V.; Wang, C.; Lin, Y.; Wang, Y.; et al. Enhanced Activity and Stability of Pt Catalysts on Functionalized Graphene Sheets for Electrocatalytic Oxygen Reduction. Electrochem. commun. 2009, 11 (5), 954–957.

(6)

Li, Y.; Gao, W.; Ci, L.; Wang, C.; Ajayan, P. M. Catalytic Performance of Pt Nanoparticles on Reduced Graphene Oxide for Methanol Electro-Oxidation. Carbon 25 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 42

2010, 48 (4), 1124–1130. (7)

Xin, Y.; Liu, J. G.; Zhou, Y.; Liu, W.; Gao, J.; Xie, Y.; Yin, Y.; Zou, Z. Preparation and Characterization of Pt Supported on Graphene with Enhanced Electrocatalytic Activity in Fuel Cell. J. Power Sources 2011, 196 (3), 1012–1018.

(8)

Vedala, H.; Sorescu, D. C.; Kotchey, G. P.; Star, A. Chemical Sensitivity of Graphene Edges Decorated with Metal Nanoparticles. Nano Lett. 2011, 11 (6), 2342–2347.

(9)

Okamoto, Y. Density-Functional Calculations of Icosahedral M13 (M = Pt and Au) Clusters on Graphene Sheets and Flakes. Chem. Phys. Lett. 2006, 420 (4-6), 382–386.

(10)

Chi, D. H.; Cuong, N. T.; Tuan, N. A.; Kim, Y. T.; Bao, H. T.; Mitani, T.; Ozaki, T.; Nagao, H. Electronic Structures of Pt Clusters Adsorbed on (5, 5) Single Wall Carbon Nanotube. Chem. Phys. Lett. 2006, 432 (1-3), 213–217.

(11)

Fampiou, I.; Ramasubramaniam, A. Binding of Pt Nanoclusters to Point Defects in Graphene: Adsorption, Morphology, and Electronic Structure. J. Phys. Chem. C 2012, 116 (11), 6543–6555.

(12)

Ashrafian, S.; Jahanshahi, M.; Ganji, M. D.; Agheb, R. Greatly Enhanced Adsorption of Platinum on Periodic Graphene Nanobuds: A First-Principles Study. Appl. Surf. Sci. 2015, 351, 1105–1115.

(13)

Zheng, H.; Mukherjee, S.; Gangopadhyay, K.; Gangopadhyay, S. Ultrafine Pt Nanoparticle

Induced

Doping/strain

of

Single

Layer

Graphene:

Experimental

Corroboration between Conduction and Raman Characteristics. J. Mater. Sci. Mater. Electron. 2015, 26 (7), 4746–4753. 26 ACS Paragon Plus Environment

Page 27 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(14)

Mahmoodinia, M.; Ebadi, M.; Åstrand, P.-O.; Chen, D.; Cheng, H.-Y.; Zhu, Y.-A. Structural and Electronic Properties of the Ptn-PAH Complex (n = 1, 2) from Density Functional Calculations. Phys. Chem. Chem. Phys. 2014, 16 (34), 18586–18595.

(15)

Greeley, J.; Jaramillo, T. F.; Bonde, J.; Chorkendorff, I. B.; Nørskov, J. K. Computational High-Throughput Screening of Electrocatalytic Materials for Hydrogen Evolution. Nat. Mater. 2006, 5 (11), 909–913.

(16)

Fernández, E. M.; Moses, P. G.; Toftelund, A.; Hansen, H. a; Martínez, J. I.; AbildPedersen, F.; Kleis, J.; Hinnemann, B.; Rossmeisl, J.; Bligaard, T.; et al. Scaling Relationships for Adsorption Energies on Transition Metal Oxide, Sulfide, and Nitride Surfaces. Angew. Chem. Int. Ed. Engl. 2008, 47 (25), 4683–4686.

(17)

Lee, K.; Lee, E.; Song, C.; Janik, M. J. Density Functional Theory Study of Propane Steam Reforming on Rh–Ni Bimetallic Surface: Sulfur Tolerance and scaling/Brønsted– Evans–Polanyi Relations. J. Catal. 2014, 309, 248–259.

(18)

Yoo, J. S.; Abild-Pedersen, F.; Nørskov, J. K.; Studt, F. Theoretical Analysis of Transition-Metal Catalysts for Formic Acid Decomposition. ACS Catal. 2014, 4 (4), 1226–1233.

(19)

Bandarenka, A. S.; Koper, M. T. M. Structural and Electronic Effects in Heterogeneous Electrocatalysis: Toward a Rational Design of Electrocatalysts. J. Catal. 2013, 308, 11– 24.

(20)

Niu, C.; Jiao, J.; Xing, B.; Wang, G.; Bu, X. Reaction Mechanism of Methanol Decomposition on Pt-Based Model Catalysts : A Theoretical Study. J. Comput. Chem. 27 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 42

2010, 31 (10), 2023–2037. (21)

Mavrikakis, M.; Hammer, B.; Nørskov, J. Effect of Strain on the Reactivity of Metal Surfaces. Phys. Rev. Lett. 1998, 81 (13), 2819–2822.

(22)

Mehmood, F.; Rankin, R. B.; Greeley, J.; Curtiss, L. A. Trends in Methanol Decomposition on Transition Metal Alloy Clusters from Scaling and Brønsted-EvansPolanyi Relationships. Phys. Chem. Chem. Phys. 2012, 14, 8644–8652.

(23)

Studt, F.; Abild-Pedersen, F.; Wu, Q.; Jensen, A. D.; Temel, B.; Grunwaldt, J. D.; Nørskov, J. K. CO Hydrogenation to Methanol on Cu-Ni Catalysts: Theory and Experiment. J. Catal. 2012, 293, 51–60.

(24)

Ferrin, P.; Nilekar, A. U.; Greeley, J.; Mavrikakis, M.; Rossmeisl, J. Reactivity Descriptors for Direct Methanol Fuel Cell Anode Catalysts. Surf. Sci. 2008, 602 (21), 3424–3431.

(25)

Fampiou, I.; Ramasubramaniam, A. CO Adsorption on Defective Graphene-Supported Pt13 Nanoclusters. J. Phys. Chem. C 2013, 117 (39), 19927–19933.

(26)

Wang, J.-J.; Liu, Y.-T.; Chen, I.-L.; Yang, Y.-W.; Yeh, T.-K.; Lee, C. H.; Hu, C.-C.; Wen, T.-C.; Chen, T.-Y.; Lin, T.-L. Near-Monolayer Platinum Shell on Core–Shell Nanocatalysts for High-Performance Direct Methanol Fuel Cell. J. Phys. Chem. C 2014, 118 (5), 2253–2262.

(27)

Mei, D.; Xu, L.; Henkelman, G. Potential Energy Surface of Methanol Decomposition on Cu(110). J. Phys. Chem. C 2009, 113 (11), 4522–4537.

(28)

Kandoi, S.; Greeley, J.; Sanchez-Castillo, M.; Evans, S. T.; Gokhale, A.; Dumesic, J.; 28 ACS Paragon Plus Environment

Page 29 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Mavrikakis, M. Prediction of Experimental Methanol Decomposition Rates on Platinum from First Principles. Top. Catal. 2006, 37 (1), 17–28. (29)

Zhang, C. J.; Hu, P. A First Principles Study of Methanol Decomposition on Pd(111): Mechanisms for O–H Bond Scission and C–O Bond Scission. J. Chem. Phys. 2001, 115 (15), 7182.

(30)

Zuo, Z.J.; Wang, L.; Han, P.D.; Huang, W. Insight into the Size Effect on Methanol Decomposition over Cu-Based Catalysts Based on Density Functional Theory. Comput. Theor. Chem. 2014, 1033, 14–22.

(31)

Zhou, Y.H.; Lv, P.H.; Wang, G.C. DFT Studies of Methanol Decomposition on Ni(100) Surface: Compared with Ni(111) Surface. J. Mol. Catal. A Chem. 2006, 258 (1-2), 203– 215.

(32)

Guo, W.; Tian, W. Q.; Lian, X.; Liu, F.; Zhou, M.; Xiao, P.; Zhang, Y. A Comparison of the Dominant Pathways for the Methanol Dehydrogenation to CO on Pt7 and Pt7−xNix (x=1, 2, 3) Bimetallic Clusters: A DFT Study. Comput. Theor. Chem. 2014, 1032, 73–83.

(33)

Yuan, D.; Gong, X.; Wu, R. Decomposition Pathways of Methanol on the PtAu(111) Bimetallic Surface: A First-Principles Study. J. Chem. Phys. 2008, 128 (6), 064706.

(34)

Greeley, J.; Mavrikakis, M. A First-Principles Study of Methanol Decomposition on Pt(111). J. Am. Chem. Soc. 2002, 124 (24), 7193–7201.

(35)

Greeley, J.; Mavrikakis, M. Competitive Paths for Methanol Decomposition on Pt(111). J. Am. Chem. Soc. 2004, 126 (12), 3910–3919.

(36)

Esrafili, M. D.; Nurazar, R. A Density Functional Theory Study on the Adsorption and 29 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 42

Decomposition of Methanol on B12N12 Fullerene-like Nanocage. Superlattices Microstruct. 2014, 67, 54–60. (37)

Greeley, J.; Mavrikakis, M. Methanol Decomposition on Cu(111): A DFT Study. J. Catal. 2002, 208 (2), 291–300.

(38)

Vo, C. T.; Huynh, L. K.; Hung, J. Y.; Jiang, J. C. Methanol Adsorption and Decomposition on ZnO( 1010 ) Surface: A Density Functional Theory Study. Appl. Surf. Sci. 2013, 280, 219–224.

(39)

Miller, A. V.; Kaichev, V. V.; Prosvirin, I. P.; Bukhtiyarov, V. I. Mechanistic Study of Methanol Decomposition and Oxidation on Pt(111). J. Phys. Chem. C 2013, 117 (16), 8189–8197.

(40)

Wang, G. C.; Zhou, Y. H.; Morikawa, Y.; Nakamura, J.; Cai, Z. S.; Zhao, X. Z. Kinetic Mechanism of Methanol Decomposition on Ni(111) Surface: A Theoretical Study. J. Phys. Chem. B 2005, 109 (25), 12431–12442.

(41)

Tenney, S. A.; Shah, S. I.; Yan, H.; Cagg, B. A.; Levine, M. S.; Rahman, T. S.; Chen, D. A. Methanol Reaction on Pt−Au Clusters on TiO2(110): Methoxy-Induced Diffusion of Pt. J. Phys. Chem. C 2013, 117 (51), 26998–27006.

(42)

Zuo, Z. J.; Wang, L.; Han, P. D.; Huang, W. Insights into the Reaction Mechanisms of Methanol Decomposition, Methanol Oxidation and Steam Reforming of Methanol on Cu(111): A Density Functional Theory Study. Int. J. Hydrogen Energy 2014, 39 (4), 1664–1679.

(43)

G. Kresse; Hafner, J. Ab Initio Molecular Dynamics for Liquid Metals. Phys. Rev. B 1993, 30 ACS Paragon Plus Environment

Page 31 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


47, 558. (44)

G. Kresse; Hafner, J. Ab Initio Molecular-Dynamics Simulation of the Liquid-MetalAmorphous-Semiconductor Transition in Germanium. Phys. Rev. B 1994, 49 (20), 14251.

(45)

G. Kresse; Furthmüller, J. Efficient Iterative Schemes for Ab Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Rev. B 1996, 54 (16), 11169–11186.

(46)

G. Kresse; Furthmüller, J. Efficiency of Ab-Initio Total Energy Calculations for Metals and Semiconductors Using a Plane-Wave Basis Set. Comput. Mater. Sci. 1996, 6 (15), 1550.

(47)

Blochl, P. E. Projector Augmented-Wave Method. Phys. Rev. B 1994, 50, 17953.

(48)

G. Kresse. From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method. Phys. Rev. B 1999, 59, 1758.

(49)

Perdew, J. P.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Atoms, Molecules, Solids, and Surfaces: Applications of the Generalized Gradient Approximation for Exchange and Correlation. Phys. Rev. B 1992, 46 (11), 6671.

(50)

Methfessel, M.; Paxton, A. T. High-Precision Sampling for Brillouin-Zone Integration in Metals. Phys. Rev. B 1989, 40 (6), 3616–3621.

(51)

Tang, W.; Sanville, E.; Henkelman, G. A Grid-Based Bader Analysis Algorithm without Lattice Bias. J. Phys. Condens. Matter 2009, 21 (8), 084204.

(52)

Bader, R. F. W. A Quantum Theory of Molecular Structure and Its Applications. Chem. Rev. 1991, 91 (5), 893–928. 31 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(53)

Page 32 of 42

Fampiou, I.; Ramasubramaniam, A. The Influence of Support Effects on CO Oxidation Kinetics on Graphene-Supported Pt13 Nanoclusters. J. Phys. Chem. C 2015, 119 (16), 8703–8710.

(54)

Feibelman, P. J.; Hammer, B.; Nørskov, J. K.; Wagner, F.; Scheffler, M.; Stumpf, R.; Watwe, R.; Dumesic, J. The CO/Pt(111) Puzzle. J. Phys. Chem. B 2001, 105 (18), 4018– 4025.

(55)

Wang, Y.; De Gironcoli, S.; Hush, N. S.; Reimers, J. R. Successful a Priori Modeling of CO Adsorption on Pt(111) Using Periodic Hybrid Density Functional Theory. J. Am. Chem. Soc. 2007, 129 (34), 10402–10407.

(56)

Abild-Pedersen, F.; Greeley, J.; Studt, F.; Rossmeisl, J.; Munter, T.; Moses, P.; Skúlason, E.; Bligaard, T.; Nørskov, J. Scaling Properties of Adsorption Energies for HydrogenContaining Molecules on Transition-Metal Surfaces. Phys. Rev. Lett. 2007, 99 (1), 016105.

(57)

Mehmood, F.; Greeley, J.; Curtiss, L. A. Density Functional Studies of Methanol Decomposition on Subnanometer Pd Clusters. J. Phys. Chem. C 2009, 113 (52), 21789– 21796.

(58)

Mehmood, F.; Greeley, J.; Zapol, P.; Curtiss, L. A. Comparative Density Functional Study of Methanol Decomposition on Cu4 and Co4 Clusters. J. Phys. Chem. B 2010, 14458– 14466.

(59)

Tripković, V.; Skúlason, E.; Siahrostami, S.; Nørskov, J. K.; Rossmeisl, J. The Oxygen Reduction Reaction Mechanism on Pt(111) from Density Functional Theory Calculations. 32 ACS Paragon Plus Environment

Page 33 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Electrochim. Acta 2010, 55 (27), 7975–7981. (60)

Nørskov, J. K.; Bligaard, T.; Rossmeisl, J.; Christensen, C. H. Towards the Computational Design of Solid Catalysts. Nat. Chem. 2009, 1 (1), 37–46.

(61)

Honkala, K.; Hellman, A.; Remediakis, I. N.; Logadottir, A.; Carlsson, A.; Dahl, S.; Christensen, C. H.; Nørskov, J. K. Ammonia Synthesis from First-Principles Calculations. Science 2005, 307 (5709), 555–558.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

H3COH

H3CO + H

CO + 4H

Page 34 of 42

H2CO + 2H

HCO + 3H

CO(g) + 2H2(g)


Page 35 of 42 a 1 2 3 4 5 6

The Journal of Physical Chemistry c d b

ACS Paragon Plus Environment

a 1 2 3 4 5 6 7 8 c 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

The Journal of Physical Chemistry Page 36 of 42

b

d

e


Page 37 of 42

1 2 3 4a 5 6 7 8 9 10 11


b

c

d


e


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37


Page 38 of 42

Page 39 of 42

The Journal of Physical Chemistry Text

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46


Page 40 of 42

Page 41 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29


Pt(111) -0.34

-0.20

-0.71

-0.41

Pt13 at graphene divacancy

0.48

0.45

-1.12

0.23

-0.33

-0.37

-0.22

-0.71

0.20

-1.06

-1.20


0.04

-0.53

-1.12

-0.64

-0.62

The Journal of Physical Chemistry Page 42 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14


Density Functional Theory Studies of the Methanol ... - ACS Publications

Recommend Documents