Trends in Adsorption Characteristics of Benzene on Transition Metal

Sep 12, 2013 - Physics Department, University of Central Florida, Orlando, Florida 32816, .... based on the trends obtained in the adsorption energies...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/JPCC

Trends in Adsorption Characteristics of Benzene on Transition Metal Surfaces: Role of Surface Chemistry and van der Waals Interactions Handan Yildirim,*,†,§ Thomas Greber,‡ and Abdelkader Kara*,† †

Physics Department, University of Central Florida, Orlando, Florida 32816, United States Physik-Institut, Universität ZürichCH-8006 Zürich, Switzerland



ABSTRACT: The accurate description of interface characteristics between organic molecules and metal surfaces has long been debated in theoretical studies. A well-founded description of interface geometry and adsorption energy is highly desirable for these hybrid inorganic/organic interfaces. Using first principles calculations with the inclusion of five van der Waals functionals (vdW-DF family), benzene (C 6 H 6 ) adsorption on seven transition metal surfaces is studied to explore the performance of these vdW functionals under varying surface chemistry. Our results reveal that vdW interactions are crucial for an accurate description of bonding on transition metal substrates. We find that vdW interactions increase adsorption energy on coinage metal surfaces (Au, Ag, Cu) by about 0.7 eV, while they lead to even larger increases in the adsorption energies on the reactive transition metal surfaces (Pd, Pt, Rh, Ni). Our calculations also reveal that changes in adsorption energies stemming from vdW functionals show significant variation, and can be grouped. We find the adsorption energies and heights on the reactive transition metal surfaces obtained using vdW-DF and vdW-DF2 functionals to differ significantly from those of the opt-type functionals, revealing the intrinsic strong repulsion character at short ranges for the former functionals. A simple comparison between experimentally determined adsorption energies (averaged) and those of computed suggests that optPBE and optB88 functionals show systematically good agreement. The information acquired from our analysis on the performance of these functionals can be used as a basis for further refinement of these functionals for the adsorption on metal surfaces with varying chemistry.

I. INTRODUCTION The nature of organic molecule interaction with metal surfaces is of broad interest for both fundamental and applied research. For environmental and industrial needs, petroleum refining, and reforming processes, transformation of organic molecules into less hazardous components is crucial;1,2 hydrogenation and cracking of these molecules can be performed by transition metal catalysts.3 More recently, organic molecule interaction with metal surfaces has attracted much attention as they show potential applications in the design of devices based on “electroactive” organic molecules.4,5 Therefore, thorough understanding of this interaction is crucial, and it is the starting point for any quantitative insight into a catalytic process. Benzene is the smallest aromatic molecule, and it has often been used as a model system. Although basic interaction characteristics with metals are fairly well understood, such important details as adsorption site preferences, accurate adsorption energies, and geometries are often nontrivial to obtain from experimental studies. In particular, the interpretation of experimental results on adsorption characteristics on transition metals is challenging. Possible fragmentation of benzene6−10 may occur when it is strongly bound with substrate that may lead to notable uncertainties in adsorption energies.11−13 The experimental studies mostly report changes © 2013 American Chemical Society

in geometric and electronic structures, work functions, and vibrational properties of benzene upon adsorption. The available data on adsorption energies, however, is scarce, making the theoretical insights valuable; for a review see ref 14. From the available literature on benzene adsorption on coinage metals (Ag, Au, and Cu),15−20 a common understanding is that benzene weakly interacts with these surfaces. For close-packed (111) coinage metal surfaces, the adsorption is dominated by van der Waals (vdW) forces, and the nature of bonding is defined as weak physisorption. On the other hand, adsorption on the more reactive transition metals (for simplicity, these are labeled hereafter as transition metals) is significantly different, governed by both ionic and covalent bonding.1,2,21−27 These conclusions are obtained from the studies employing standard density functional theory (DFT), which is unable to describe accurately the long-range electron correlations (electron dispersion forces). The recent advances in the field made it possible to incorporate these dispersion forces into the description of bonding;28−34 see detailed discussions in these reviews, refs 35−38. Received: May 6, 2013 Revised: September 1, 2013 Published: September 12, 2013 20572

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

As there are several implementations of the dispersion forces to be tested for the case of adsorbed molecules on metal surfaces with different chemistries, we have focused in the present study on the vdW-DF family, which belongs to the third level in the stairway to heaven for vdWs. Five functionals have been chosen to explore their performances under varying surface chemistry, and they will be discussed hereafter. Four of these vdW functionals use the nonlocal implementation of vdW-DF,45 while the fifth uses vdW-DF250 nonlocal term. The main difference among these five functionals comes from the exchange term, i.e., the form of the enhancement factor and its parameters. Let us summarize briefly why there are several functionals for the same implementation shown in eq 2 (a detailed discussion can be found in ref 40). An evaluation of the performance of the original vdW-DF has pointed out the tendency of this approach to overestimate the long-range dispersion interactions.51 As a consequence, a new functional labeled vdWDF250 was developed at which two modifications have been introduced. The GGA functional, used in vdW-DF for the exchange term (revPBE), was replaced by a new functional (rPW86), and the nonlocal correlation term was also changed (called vdW2, with the original being vdW52). It turns out that both revPBE and rPW86 exchange functionals chosen for vdWDF and vdW-DF2, respectively, tend to poorly describe binding energies and distances, and this is due to the strong repulsive nature of these functionals at short ranges. Thus, new “less repulsive” functionals, within the vdW-DF family, have been proposed, and show substantial improvement in binding energies and distances.31,52−54 Among the proposed functionals we have chosen to include in our study “optB86b”, “optB88”, and “optPBE” functionals. The main focus of the present paper is to give insights into how the interaction of benzene varies when substrate characteristics are changed, and how vdW functionals considered in this study treat adsorption characteristics for chemically nonidentical substrates. To the best of our knowledge, such a detailed study is lacking for these vdW functionals. Here, the adsorption characteristics of benzene on close-packed (111) surfaces of Ag, Cu, Au, Pd, Pt, Rh, and Ni are studied using both the GGA functional of Perdew, Burke, and Ernzerhof (PBE)55,56 and those vdW functionals52 available in the Vienna Ab-initio Simulation Package (VASP).57−59 Note that these functionals introduce nonlocal correlations selfconsistently, which is needed for accurate inclusion of these effects.60 The calculations reveal that, for physisorbed systems, the inclusion of vdW interactions increases the adsorption energies regardless of the vdW functional used. For chemisorbed systems, on the other hand, much significant contribution to the adsorption energies is obtained using the three functionals (opt-types,31 less “repulsive” functionals within the vdW-DF scheme); however, this contribution is less significant for the two functionals (revPBE and rPW86, “strongly repulsive at short ranges” known as vdW-DF45 and vdW-DF2,50 respectively). For these two functionals, the adsorption energies on transition metal substrates are reduced in comparison with those obtained using GGA-PBE. This can be understood from the strong repulsive nature of these functionals at short range (see the discussion in refs 31, 52, 53, and 61). The results reveal the role of vdW interactions for accurate description of adsorption characteristics on metal substrates, even for strongly bound systems; however, the choice of which functional to use is still debatable as there is a

The reports have so far revealed that inclusion of vdW interactions in general provides better agreement with experimental data. Although several studies have been conducted for assessing the importance of vdW interactions on adsorption, most often, the conclusions are driven based on a few data sets, involving a limited number of surfaces, and explored only a few vdW functionals. Thus, a systematic study exploring the adsorption on both coinage and transition metal substrates is necessary to gain further understanding of the effect of vdW interactions on adsorption characteristics, while assessing the transferability of vdW functionals (within the same level of approximation made for obtaining dispersion forces) for varying substrate chemistry. A similar attempt has been made by a recent study focusing on the role of vdW interactions for benzene adsorption on selected coinage and transition metal substrates. However, the focus was not to assess the performance of the available vdW-DF family functionals; rather it was to provide a comparative view on assessing the performance of their own method.39 Many DFT-based techniques have been developed to account for vdW interactions, and here we briefly classify them by considering the level of approximations employed in each implementation of electron dispersion forces. We follow here what has been described in detail in the review article by Klimeš and Michaelides,40 in which they have introduced a “stairway to heaven” of vdW implementations. Starting with level zero at which no dispersion forces are taken into account, the authors introduced levels one and two (DFT-D2,41 DFTD3,42 vdW(TS),43 and BJ44 models), where a correction energy term is added to DFT energy in the following manner: Etot = E DFT + Edisp

(1)

where EDFT is calculated with the exchange-correlation (XC) functional at hand, and Edisp is an additional term to account for long-range interactions. The review article by Klimeš and Michaelides40 gives a detailed and very informative discussion of the implementations for levels one and two. At level three, which is targeted in this study, one finds the vdW-DF family functionals that were first introduced by Dion et al.,28,45 where the XC energy (Exc) now includes a density dependent dispersion term as the nonlocal correlation term, and it is given by: Exc = Ex GGA + Ec LDA + Ec nl

(2)

where the first energy in the right-hand side of eq 2 is the exchange term calculated using a generalized gradient approximation (GGA) type functional; the second and third energies are the local and nonlocal contributions, respectively, to the correlation term. The exchange term depends on the type of functional used, namely the form and parameters of the so-called enhancement factor F, which controls the effect of the density gradient on the local exchange energy. The third term can also be implemented in different ways as has been discussed in ref 40. Beyond the third level in the stairway to heaven for vdWs lies the fourth level containing approaches that are computationally demanding (such as RPA46 and post-HF methods (MP2,47 CC48,49)). These consider the dispersion beyond pairwise additive, thus accounting for collective excitations. These effects become important for adsorption where the bare interaction is screened. Note that the level four methods still require significant computational time. 20573

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

Table 1. Calculated Lattice Constants (in Å) Using vdW functionals and PBEa method

Ag

Au

Cu

Pt

Pd

Rh

Ni

PBE optB86b optB88 optPBE revPBE rPW86 expt

4.1542.2 4.1101.2 4.1472.1 4.1792.9 4.2584.8 4.3296.5 4.06365

4.1702.7 4.1401.9 4.1782.9 4.1973.3 4.2614.9 4.3527.2 4.06165

3.6351.1 3.5980.1 3.6260.9 3.6481.5 3.7023.0 3.7474.2 3.59565

3.9801.7 3.9601.2 3.9881.9 3.9992.2 4.0403.2 4.1175.2 3.91365

3.9602.2 3.9191.1 3.9511.9 3.9702.4 4.0233.8 4.0915.5 3.87665

3.8421.3 3.8200.7 3.8461.4 3.8581.7 3.8962.7 3.9574.3 3.79365

3.5250.5 3.489−0.5 3.5110.1 3.5290.6 3.5711.8 3.6102.6 3.50965

a

The experimental lattice constants with ZPEC are taken from ref 65. The percentage deviations of the calculated lattice constants from the experimental values are given as superscripts for different vdW functionals as well as PBE.

varying range of energies obtained when using different functionals. Overall, our results reveal the differences and the similarities obtained for benzene adsorption energies and heights, between different vdW-DF family functionals, and reveal the expected grouping of these functionals in two sets based on the trends obtained in the adsorption energies and geometries.

II. COMPUTATIONAL DETAILS The calculations are carried out within the DFT framework using the VASP code (version 5.2.12). For assessing the role of vdW interactions, and screening the transferability of the vdW functionals, the calculations are performed using optB88,31 optB86b,52 optPBE,31 revPBE,28 and rPW8650 functionals, and the comparisons are made with the GGA-PBE results. The interaction between the valence electrons and ionic cores is described by the projector augmented wave (PAW) method.62,63 The energy cutoff is set to 400 eV for the wave functions, and the Brillouin zone sampled with a 7 × 7 × 1 Monkhorst−Pack64 grid for the slab calculations for which a (4 × 4) super cell structure is used. The calculated lattice constants are summarized in Table 1 and show excellent agreement with the previously reported theoretical52 and experimental results (with the zero-point energy corrections (ZPEC)).65 The spin polarized calculations are performed for the adsorption on Ni(111). We use a (4 × 4) super cell structure with four layers and 19 Å vacuum separating the two surfaces. The convergence of the adsorption energies is tested with additional calculations using a six-layer slab for a few systems. Our results suggest a four-layer slab to be sufficient for reaching similar conclusions. Upon adsorption of benzene, the bottom two layers are fixed during the optimization with a force criterion of 0.01 eV/Å. The most stable adsorption site is frequently reported to be the bri30°, from theoretical and experimental studies.1,2,22,66−68 In this conformation, the benzene ring center is located on a bridge position above two surface atoms (see Figure 1a) with an angle of 30° between the C−C and metal−metal bonds with C−C bonds parallel to the [2̅11] direction. For all calculations, benzene is brought at a relatively high distance (∼3.5 Å) from the surface, and regardless of the surface and the vdW functionals, the equilibrium geometry of benzene is found to be the bri30° site. All results reported here are obtained for the bri30° adsorption configuration.

Figure 1. (a) Top view of the bri30° configuration of benzene on (111). (b) Equilibrium adsorption geometry on Pt(111). (c) Equilibrium adsorption geometry on Au(111). Light gray, blue, dark gray, red, and black spheres represent the first layer atoms, the second layer atoms, the third layer atoms, H atoms, and C atoms, respectively.

devoted to the adsorption characteristics of benzene on coinage metal surfaces and section III.2 is devoted to the adsorption on the transition metal surfaces; comparisons are made with the available theoretical (using different vdW functionals and GGA) and experimental results. In section III.3, we present the results on the changes in the electronic structures, and attempt to derive correlations between these changes and the adsorption properties. III.1. Benzene Adsorption Characteristics on Coinage Metal Surfaces. The common understanding reached from the available literature15−20 for benzene adsorption on coinage metals is that benzene weakly interacts with these surfaces. For the adsorption on close-packed (111) coinage metal surfaces, the adsorption is dominated by vdW interactions and classified as weak physisorption. In Table 2, we summarize our results of benzene adsorption energies on both coinage and transition metal substrates calculated using PBE and five vdW functionals, as well as the available experimental and theoretical results. The interaction of benzene with coinage metal substrates is much weaker than that on transition metal substrates as clearly reflected in the adsorption energies listed in Table 2. This can be understood as these metals have their d-bands well below the Fermi level and full, and weakly participating in the bonding. The adsorption energies calculated using PBE are small, and do not vary significantly among Cu(111), Ag(111), and Au(111). We observe a systematic increase in the adsorption energy with the inclusion of vdW interactions regardless of the functional employed. It is only the degree of enhancement that differs with the employed functional. On all the coinage substrates, the vdW interactions enhance the adsorption energy by about 500 meV using revPBE and rPW86 functionals and 700−800 meV when opt-type functionals are used. Benzene adsorption energies on all substrates calculated

III. RESULTS AND DISCUSSION Below, we will first discuss benzene adsorption energies and heights on coinage and transition metal surfaces calculated using PBE and five vdW-DF functionals. For the sake of clarity, we divide the discussion into three sections: section III.1 is 20574

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

Table 2. Adsorption Energies (in eV) Calculated Using PBE and vdW-DF Functionals along with Available Experimental and Theoretical Dataa Bz/Ag(111)

Bz/Au(111)

Bz/Cu(111)

PBE optB86b optB88 optPBE revPBE rPW86 expt

method

0.06 0.76 0.72 0.71 0.55 0.52 0.4273 0.66−0.8034 b

0.06 0.86 0.82 0.71 0.56 0.55 0.6474

existing theory

0.07−0.0881 c 0.5081 d 0.0982 g 0.8782 e 0.7582 f 0.7282 i 0.41−0.4970 q 0.3371

0.07−0.0981 c 0.6981 d 0.1582 g 0.8482 e 0.7482 f 0.76−1.3570 p 0.42−0.5570 q 0.4217 l 0.3171 m 0.6017 n 0.4617 o 0.7982 i

0.07 0.82 0.74 0.68 0.53 0.49 0.5875 0.6276 0.68−0.8134 b 0.06−0.0781 c 0.7081 d 0.0882 g 1.0782 e 0.8682 f 0.61−0.8670 p 0.45−0.5570 q

Bz/Pt(111)

Bz/Pd(111)

Bz/Rh(111)

Bz/Ni(111)

1.19 (0.91,1.09) 2.42 2.02 1.77 0.98 0.60 1.49−1.8377

1.19 2.36 2.01 1.75 1.02 0.77 1.3578 1.44−2.0479

1.54 2.81 2.40 2.21 1.33 0.90 −

0.98 2.19 1.79 1.48 0.69 0.27 0.7880

0.9022 h 0.8182 g 1.8082 e 1.9682 f 1.8439 i 0.85−1.2121,22,24,67,83,84

1.1922 h 1.1782 g 2.0182 e 2.1482 f 1.9139 i 1.1922,67,83

1.5322 h 1.4882 g 2.7982 e 2.5282 f 2.2739 i 1.5385

1.0022 h 0.9166 0.8781 j 1.4181 k

a The adsorption energy is defined as Eads = −(EBz/surf − Esurf − EBz), where the subscripts Bz/surf, surf, and Bz refer to the total energies of benzene on the surface, the clean surface, and isolated benzene systems, respectively. The numbers in parentheses for Pt are for (3 × 3) cells with four and six layers, respectively. bA recent interpretation of the temperature-programmed desorption (TPD) for benzene at the hcp30° site using the Redhead formula for the range of pre-exponential factors 1015−1018 s−1. cPBE and PW91. dPBE-XDM. ePBE+vdW. fPBE+vdWsurf. gPBE. hPW91. ioptB88vdW. jRPBE. kRPBE-XDM. lRPBE-vdW. mMP2. noptPBE-vdW. ovdW-DF1 and 2. pDFT-D: hybrid; Grimme, respectively. qvdW-DF.

using opt-type functionals are systematically larger than those calculated using revPBE and rPW86 functionals. These opttype functionals were introduced to address the problem of large binding distances of revPBE and rPW86 functionals by using less repulsive exchange functionals. It is clear from our results and from those reported in the literature that these new functionals improve the accuracy in adsorption energies for many systems.31,33,52,69 For weakly physisorbed systems, the vdW interactions become important, and hence several approaches including DFT-D,70 MP2,71 and vdW-DF72 have been employed to study benzene interactions with coinage metal substrates. The conclusions of these studies are the same: they all report a reduction for the molecule−surface height and an increase in the adsorption energy. Benzene adsorption heights (C−metal and H−metal) on both coinage and transition metal substrates summarized in Table 3 further reveal the role played by vdW interactions in the nature of the bonding. Our results suggest that adsorption heights on coinage metal surfaces are higher than those on transition metal surfaces. For coinage metal substrates, the adsorption heights calculated using PBE for Cu, Ag, and Au are high (ranging from 3.42 to 3.55 Å), suggesting that benzene is adsorbed at a high distance from these surfaces. The adsorption heights are systematically reduced with the inclusion of vdW interactions suggesting increased interaction with the substrate. As similar to the trends observed for the changes in adsorption energies upon inclusion of vdW interactions, the degree of reduction in the adsorption heights is also found to depend on the vdW functionals employed. While opt-type functionals substantially reduce the adsorption heightsthe lowest adsorption heights for all surfaces are obtained using the optB86b functional (i.e., reducing from 3.42 (PBE) to 2.83 Å

for Bz/Cu)much larger adsorption heights are predicted using vdW-DF and vdW-DF2 functionals, close to those obtained using PBE. The adsorption heights calculated with revPBE functional are the highest among the five functionals employedas high as those of PBEreflecting the strong repulsive behavior of this functional with which benzene is adsorbed at a high distance from the surfaces reflecting the weak interaction with the underlying substrate. As reported earlier, these functionals are known to cause too-large intermolecular binding distances, and most often lead to inaccurate binding energies.31,52,53,61 For all coinage metal substrates, we find a small or no deviation between the C−M and H−M heights, suggesting that benzene adsorbs flat on all these surfaces. This was predicted in the earlier experiments by Xi et al.76 for benzene on Cu(111) using high-resolution electron energy loss spectroscopy (HREELS) and near-edge X-ray absorption fine structure (NEXAFS) measurements, where they reported that benzene binds parallel to the Cu(111) surface. Again, by means of NEXAFS measurements, Yannoulis et al.86 reported a flat adsorption geometry for benzene on Ag(111). We now briefly review the earlier theoretical and experimental studies on benzene adsorption, and compare our calculated adsorption energies and heights with those reported. An earlier DFT study81 used different GGA functionals and evaluated the performance of the implementation of Becke and Johnson’s (BJ) exchange-hole dipole moment (XDM) method44 for modeling dispersive interactions within DFT, and reported benzene adsorption energies on Ag(111), Au(111), and Cu(111) at different adsorption sites. The adsorption energies calculated using PBE, at the bridge site, are reported to be 0.08, 0.07, and 0.06 eV, respectively. 20575

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

3.55 3.03 3.08 3.23 3.51 3.40

dH−M

3.7017 b 3.6071 c 3.4017 d 3.6017 e 3.1081 3.6282 f 3.2182 i 3.0582 j 3.8071 c 3.2339 k 3.1016 l

3.46 3.03 3.08 3.21 3.44 3.31 −

dC−M 3.46 3.05 3.09 3.21 3.44 3.31

dH−M

Bz/Au(111)

3.0081 h 3.7482 f 3.0482 i 2.7982 j 3.7072 2.9016 l

3.42 2.83 2.91 3.14 3.46 3.39 −

dC−M 3.42 2.85 2.92 3.15 3.46 3.39

dH−M

Bz/Cu(111)

2.1082 f 2.5482 f 2.2183 g 2.2222 g 2.1822 g 2.1182 i 2.0882 j 2.1239 k 2.1785

2.14 2.12 2.14 2.15 2.19 2.22 2.18−2.2587

dC−M

Bz/Pt(111) 2.57 2.55 2.57 2.58 2.61 2.63

dH−M

2.1282 f 2.4782 f 2.2183 g 2.2322 g 2.2022 g 2.1282 i 2.1082 j

2.15 2.13 2.16 2.17 2.24 2.31 −

dC−M 2.50 2.48 2.50 2.51 2.55 2.65

dH−M

Bz/Pd(111)

2.1482 f 2.5582 f 2.2022 g 2.1922 g 2.1482 i 2.1282 j

2.16 2.14 2.17 2.18 2.22 2.28 −

dC−M 2.56 2.54 2.56 2.57 2.60 2.63

dH−M

Bz/Rh(111) 2.07 2.01 2.08 2.09 2.16 2.23 1.9288 1.7989 1.9066 g 2.0522 g 2.1081 h

dC−M

2.42 2.39 2.42 2.43 2.48 2.52

dH−M

Bz/Ni(111)

a The distances are calculated from the average positions of the surface atoms. bRPBE-vdW. cMP2. doptPBE-vdW. evdW-DF. fPBE. gPW91. hRPBE-XDM. iPBE+vdW. jPBE+vdWsurf. koptB88-vdW. lDFTD. mDFT-D2 and D3.

3.1081 h 3.6982 f 3.1482 i 2.9682 j 3.7071 c 3.7072 2.75−2.9030 m 2.90−3.0030 m

existing theory

dC−M

3.55 3.02 3.08 3.23 3.51 3.40 −

PBE optB86b optB88 optPBE revPBE rPW86 expt

method

Bz/Ag(111)

Table 3. Benzene Adsorption Heights (in Å; C−Metal (dC−M) and H−Metal (dH−M) Distances) along with Available Experimental and Theoretical Dataa

The Journal of Physical Chemistry C Article

20576

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

PhD89and LEED88 experiments suggest the center of benzene ring to be located on the hcp 3-fold hollow site with its C−C bonds oriented parallel to the close-packed rows of the substrate. Additionally, both PhD89 and LEED88 results are consistent with bridge site adsorption, and particularly PhD results suggest this site for lower coverage (0.1 ML). It is also reported that benzene adsorbs molecularly at low temperatures (T < 150 K), and it strongly chemisorbs on Ni(111) as is deduced from TPD and ARUPS data.98,99 In our calculations for all transition metal surfaces, we find that H atoms in the benzene bend when it is adsorbed on these surfaces. From Table 3, such nonflat benzene adsorption geometry is evident from the differences between C−metal and H−metal distances. For the latter, we find larger distances to the metal surface, in line with some of the experimental evidence. In all our calculations, the optimized geometry of benzene adsorbed on transition metal surfaces suggests that the adsorption at the bri30° is the most stable site, again in agreement with some of the experimental data and most of the theoretical studies. From the calculated adsorption energies, which are summarized in Table 2, employing the PBE functional, we find that benzene adsorbs strongly on these surfaces with adsorption energies varying around and above 1 eV. Similarly to that observed for adsorption on coinage metal substrates, the inclusion of vdW interactions via opt-type functionals significantly increases adsorption energies. The effect of less “repulsive” behavior of the opt-type functionals is evident from the adsorption energy changes. The highest increase is obtained with the optB86b functional. Following that, optB88 and optPBE functionals increase the adsorption energies systematically for each surface reaching above 1 eV for some cases (see Table 2). This is an interesting observation as the vdW interactions are typically expected to play a minor role in chemisorbed systems.39 When we compare the adsorption energies calculated using revPBE and rPW86 functionals with those obtained using PBE, we find an interesting feature, however, which is somewhat expected. The results show that the adsorption energies are lowered from those of PBE on the order of 400−700 meV particularly when the rPW86 functional is used. Again this is the result of the strong repulsive behavior of revPBE and rPW86 functionals.31,52,53,61 A similar trend was reported recently for benzene on Pt(111).39 For benzene adsorption on transition metal surfaces, the most stable adsorption site is found to be the bri30° in our calculations using a (4 × 4) cell as well as in the other theoretical studies.39,66,82,83,85 Benzene adsorption on Pt(111) has been extensively studied by theory, and only a few of these studies utilize vdW effects.39,82 For benzene on Pt(111), the adsorption energies reported using GGA ranges between 0.85 and 1.21 eV21,22,24,67,83,84 in good agreement with our calculated values of 0.91−1.19 eV. A microcalorimetry experiment reports the adsorption energy (for 0.7 ML) to be 1.57−1.91 eV.77 On Pd(111), the benzene adsorption energy, again utilizing GGA, is reported to vary between 1.04 and 1.19 eV.22,67,83 For benzene adsorption on Rh, an earlier study using GGA reported the values of 1.5385 and 1.48 eV;82 both are in good agreement with our value of 1.54 eV. For benzene adsorption on Ni(111), utilizing GGA in the Perdew−Wang form, an earlier study reported the bridge site to be the most stable along with the adsorption energy of 0.91 eV.66 This is in good agreement with our value of 0.98 eV. The experimentally extracted adsorption energy based on desorption temperature is 0.78 eV.80 Theoretical studies which include the vdW

These are in excellent agreement with those obtained in our study. The benzene adsorption energies at the bridge site using PBE-XMD show a systematic increase. The adsorption energies reported using PBE-XMD are 0.70, 0.50, and 0.69 eV for Cu(111), Ag(111), and Au(111), respectively. A recent DFT calculation by Liu et al.82 reported benzene adsorption energies and heights at the hcp30° site. The adsorption energies, for the same site using (PBE), are larger for Au(111) than that of the previous study.81 Earlier calculations employing MP2, which is known to overestimate dispersion interactions,71,90 have reported lower adsorption energies of 0.35, 0.33, and 0.31 eV for Cu, Ag, and Au at the hcp site, respectively. Compared to the adsorption energy results, for the same adsorption site82 these energies are found to be even lower than those obtained by vdW-DF.17,70,72,91,92 Earlier calculations using the vdW-DF (revPBE) functional report adsorption energies of 0.45−0.55 eV for Cu, 0.41−0.49 eV for Ag, and 0.42−0.55 eV for Au(111), respectively.17,70,72,91,92 These adsorption energies are in general lower than the experimental data, resulting from the choice of the exchange functional for the original vdW-DF form. An earlier study employing DFT-D for benzene on Cu(111) and Au(111) reported adsorption energies of 0.61− 0.86 and 0.76−1.35 eV,70 respectively. The DFT-D2 and DFTD3 adsorption energies on Ag(111) are 1.20−1.29 and 0.95 eV,30 respectively. Another study employing the optB88-vdW functional reports 0.72 and 0.79 eV for Ag(111) and Au(111),39 respectively. The study using the methods of PBE +vdW (PBE+vdWsurf) reported adsorption energies of 1.07 (0.86) eV for Cu, 0.87 (0.75) eV for Ag, and 0.84 (0.74) for Au(111).82 The adsorption energies obtained using optPBE and optPB88 functionals in our study are similar to the experimental data, when taking into account the uncertainty of the experimentally determined adsorption energies (see Table 2). Compared to the adsorption energies82 calculated using PBE-vdWsurf, the adsorption energies calculated in our study using the opt-B88 functional show good agreement. III.2. Benzene Adsorption Characteristics on Transition Metal Surfaces. Benzene adsorption on transition metal surfaces is governed by both ionic and covalent bonding,1,2,21−27 and the nature of bonding on these surfaces is defined as chemisorption. Using NEXAFS measurements and X-ray photoelectron spectroscopy, Lee et al.79 reported that, at low coverage (0.16 ML), benzene binds strongly with the Pd(111) surface and lies flat, while above that coverage benzene is tilted with respect to the substrate. Although benzene adsorption on Pt(111) is an extensively studied system, the adsorption site remains as a debate among experimental studies. Using nuclear magnetic resonance, Tirendi et al.93 reported that benzene molecules are located at the atop site. On the other hand, Wander et al.,87 using diffuse LEED intensity analysis, reported that the bri30° site is the most stable. Weiss and Eigler94 using scanning tunneling microscopic (STM) image analysis suggested that benzene coexists in both hcp and fcc sites. The nature of strong binding of benzene with the Pt(111) surface is inferred from the STM images, as benzene molecules are found to be immobile on the surface.94,95 Both planar and buckling distortions are observed for benzene on Rh(111) using a low energy electron diffraction (LEED) intensity analysis.96 Benzene adsorption on Ni(111) has been studied in the past using several experimental techniques from angleresolved ultraviolet photoelectron spectroscopy (ARUPS)97 scanned energy mode photoelectron diffraction (PhD)89 and LEED88 to investigate the adsorption geometry. Both the 20577

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

Figure 2. (a) Benzene adsorption energies (Eads) calculated using different vdW functionals, and PBE on coinage substrates and transition metal substrates. (b) Benzene adsorption heights calculated using different vdW functionals, and PBE on coinage substrates and transition metal substrates,.

adsorption heights (averaged C−metal distances) on coinage and transition metal substrates with respect to the vdW functional used. Figure 2a reveals that benzene adsorption energies systematically increase on coinage metal surfaces with the inclusion of nonlocal correlations regardless of the vdW functional used, and the increase is in the order optB86b > optB88 > optPBE when using opt-type functionals. The vdWDF and vdW-DF2 functionals give the smallest increase in the adsorption energies. Similar to the trend obtained for adsorption energies, the adsorption heights on coinage metal substrates are significantly reduced when opt-type functionals are used, while the decrease is smaller for revPBE and rPW86 functionals (see Figure 2b). Figure 2a reveals some similarity between the trends obtained for adsorption energies as the adsorption energies on transition metal substrates also significantly increase when opt-type functionals are used. However, Figure 2a also indicates that revPBE and rPW86 functionals noticeably reduce adsorption energies on transition metal substrates, revealing the intrinsic problem of strong repulsion character of the exchange correlation functional used. Furthermore, the changes in the adsorption heights on transition metal substrates with the inclusion of nonlocal effects are small and comparable with those of PBE for opt-type functionals, while the adsorption heights increase from those of PBE when reVPBE and rPW86 functionals are used. This is the opposite of what is obtained for the adsorption on coinage metal substrates. In Figure 3, adsorption energies are plotted as a function of adsorption heights for coinage and transition metal substrates. As is evident from Figure 3, the adsorption heights on coinage metal substrates are the largest for PBE, then those for rPW86 (those of revPBE are similar to those of rPW86), and the smallest for opt-type functionals (for clarity, we show only results of optB86b; the results obtained using optB88 and optPBE are similar). The substantial reduction in adsorption heights obtained using opt-type functionals can be correlated with the increase in adsorption energies. For strongly bound systems, on the other hand, the adsorption is governed by both vdW interactions and covalent bonding. The analysis of the adsorption heights and energies on transition metal substrates shows that the adsorption heights calculated using opt-type functionals are similar to those found using PBE, however, the effect of nonlocal correlations manifests itself as the significant increase in adsorption energies (see Table 2 and Figure 2a). A similar observation was encountered recently for the adsorption

interactions are scarce for benzene adsorption on these transition metal substrates. The common observation from these studies is that inclusion of vdW interactions significantly enhances the adsorption energies. Using the optB88-vdW functional, Liu et al.82 have reported benzene adsorption energies on the Pd(111), Pt(111), and Rh(111) surfaces to be 1.91, 1.84, and 2.27 eV, and our calculated adsorption energy values using this functional deviate by 100, 180, and 130 meV for Pd, Pt, and Rh, respectively. This difference may result from two effects: we use a (4 × 4) cell with four layers, as opposed to a (3 × 3) cell with six layers. Note that this particular study82 reports an increase of 220 meV in the adsorption energy for Pt(111) when a (4 × 4) cell is used. The adsorption energies reported recently using PBE+vdW (PBE+vdWsurf)82 are 2.01 eV (2.14 eV) for benzene on Pd(111), 1.80 eV (1.96 eV) on Pt(111), and 2.79 eV (2.52 eV) on Rh(111). Compared to the adsorption energies calculated using the vdW-DF family, we find that our optB88-vdW results are in good agreement with those of the PBE+vdWsurf.82 The analysis of the adsorption heights, which are summarized in Table 3, shows interesting trends among the vdW functionals used. The strong binding of benzene with these transition metal substrates is evident when adsorption heights are compared with those obtained on coinage substrates, for which the adsorption heights are in general above 3 Å even with the inclusion of vdW interactions. On the other hand, the adsorption heights on transition metal substrates are slightly above 2 Å even with PBE. Note that already low adsorption heights obtained using PBE on transition metal surfaces are lowered for some surfaces, or remain similar when we use opttype functionals. Although the adsorption heights do not change significantly when using these functionals, there are systematic increases in the adsorption energies of about 1 eV. The most significant trend is the increase obtained in adsorption heights from those of PBE when revPBE and rPW86 functionals are used (see Table 3). Note that this observation also correlates with the systematic reduction in adsorption energies obtained in comparison with those using PBE. Comparison between our calculated benzene adsorption heights with those reported in the literature using PBE, PBE +vdWsurf,82 and PBE+vdW82 show good agreement within 0.02−0.04 Å. In order to reveal visually the effects of both surface chemistry and nonlocal correlations on the adsorption energies and heights, we plot in Figure 2 the adsorption energies and 20578

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

observation underestimates the adsorption energy although the corresponding adsorption height is about 1.9 Å. For this exercise, we first determine the average experimental adsorption energies whenever upper and lower bound values are available from Table 2. Note here that, for benzene on Au(111), only one value was found. Thus, we have used the following averaged adsorption energies for benzene on Ag, Au, Cu, Pt, and Pd as 0.73, 0.64, 0.74, 1.66, and 1.70 eV, respectively. The percentage changes in the calculated adsorption energies from the average experimental values are calculated for each case reported in Table 2. Our analysis suggests that the best overall performance corresponds to the results obtained using the optPBE functional. For this functional, the percentage changes in the adsorption energies from those of the experimental values are 3% for Ag and Pd and 6, 8, and 11% for Pt, Cu, and Au, respectively. The results obtained using the optB88 functional show the next best variations of 0% for Cu and 1% for Ag, while we find a dramatic increase to 18% for Pd, 22% for Pt, and 28% for Au. The adsorption energies determined using the optB86b functional present a relatively good performance for Ag (4%) and Cu (11%), while large variations are obtained for Au (34%), Pd (39%), and Pt (46%). The percentage changes when using the revPBE functional, which gives systematically much lower adsorption energiesin particular on transition metal substrates as compared to opttype functionalsare 13, 25, 28, 40, and 41% for Au, Ag, Cu, Pd, and Pt, respectively. The largest percentage changes are found for the rPW86 functional as 14, 29, 34, 55, and 64% for Au, Ag, Cu, Pd, and Pt, respectively. In the light of this simple comparison, our observation suggests that optPBE and optB88 functionals show systematically good performances for predicting the adsorption energies. We should however stress that this is a rather simple analysis that begs for less uncertainty for the adsorption energies extracted from the experimental studies. III.3. Changes in Electronic Structure. We have examined the changes introduced to the electronic structure of metal substrates with varying vdW functionals. For this purpose, we analyzed the change in the position of surface d-band centers and d-band widths upon adsorption of benzene. Most often the changes in d-band features are used to explore the effect of adsorbed species on host electronic structure and/or to classify physisorbed and chemisorbed systems.91 The changes in the position of the surface d-band centers and d-band widths are calculated for all vdW functionals with respect to those obtained using PBE. The percentage changes in the position of the averaged (over all surface atoms) d-band center and d-band width as a function of the percentage change in the lattice constant obtained for each vdW functional are plotted in Figure 4. Figure 4a suggests that the positions of the surface d-band centers shift toward lower binding energies (toward the Fermi level) when revPBE and rPW86 functionals are used, while the positions of d-band centers shift toward higher binding energies for opt-type functionals as compared with those obtained using PBE. Figure 4a also reveals that the changes in the positions of the d-band centers are more significant for revPBE and rPW86 functionals than those obtained using opt-type functionals, and seems to correlate with differences in the associated lattice constants. Similar trends are obtained for the percentage changes in dband widths (see Figure 4b). We find the surface d-band widths to be narrower for revPBE and rPW86 functionals than those calculated using PBE, while they are broader for opt-type

Figure 3. Benzene adsorption energies as a function of adsorption heights on coinage and transition metal substrates calculated using PBE, rPW86, and optB86b functionals. The results obtained using revPBE, optB88, and optPBE show similar trends and are omitted for clarity (see Table 2 and Table 3 for comparison).

of olympicene radical on Cu(111)100 and the adsorption of water on metal surfaces.101 The lowest adsorption energies and the largest adsorption heights are encountered using the rPW86 (same trend for revPBE) functional as compared to both PBE and opt-type functionals. Figure 3 also reveals that these two functionals affect the adsorption characteristics differently for coinage and transition metal substrates. For simplicity, in Figure 3, we have presented only the results obtained using the two functionals from those of five vdW functionals as the others belong to the same group and show similar trends. Figure 3 reveals how the opt-type functionals treat the adsorption characteristics differently from those of revPBE and rPW86 functionals for coinage and transition metal surfaces. A direct comparison of the adsorption energies with the available experimental data is nontrivial if the adsorption energies are extracted using temperature programmed desorption (TPD), where desorption energies are determined at desorption temperatures. For molecules like benzene, particularly on strongly bound systems, fragmentation may even make it impossible to determine the desorption temperature. On the other hand, calorimetry experiments give true measures of adsorption energy, and the one example worth mentioning is benzene adsorption on Pt(111).77 In this experiment, the adsorption energy was extracted for different coverages, and an empirical formula was fitted to the following equation: E (in kJ/mol) = 197 − 48θ − 83θ2. The θ is coverage, with θ = 1 corresponding to saturation with an estimated error of 10% for the adsorption energy. From experimental observation, the adsorption energy can vary between 0.68 eV for full coverage and 2.04 eV for zero coverage, a change of 300%. Thus, the experimental adsorption energies (via TPD) should be taken as a lower bound, at best. In the calorimetry experiment,77 the saturation occurs when the C/Pt ratio is 6/7, while for our case the C/Pt ratio is 6/16. Using θ = 7/16, we find the lower and upper limits of the adsorption energy to be 1.49 and 1.83 eV, respectively. Although it is not straightforward to identify which functional gives the best performance based on scarce experimental data with high uncertainties in the derivation of adsorption energies, and scattered computed adsorption energies, we attempt hereafter a tentative quantitative comparison for which we exclude the adsorption energies on Ni(111). For this system, we suspect that the experimental 20579

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

we find a substantial increase on the order of 1 eV using opttype functionals. The significant increase in the adsorption energies also follows substantial reduction in the adsorption heights. Although the adsorption heights on transition metal surfaces are on the order of those of PBE, the effect of vdW interactions is revealed in the adsorption energies. On the other hand, our calculations reveal the strong repulsion character of revPBE and rPW86 functionals at short ranges leading to reduction in the adsorption energies on transition metals as compared to those of PBE. The observations based on the adsorption energy and height relations, and the changes in the surface electronic structure, indicate that these functionals can be grouped, and the nature of bonding, in particular, on transition metal substrates can vary dramatically depending on the functional used. A tentative quantitative comparison between the computed and the experimentally reported adsorption energies suggests that optPBE and optB88 functionals show a systematic good agreement with “averaged” experimental adsorption energies. This conclusion, however, should be taken with caution as the available experimental data, in general, may present large error bars due to the difficulties surrounding the experimental determination of adsorption energies for such organic systems as those presented here. Taking into account the abovementioned aspects, we conclude that, although opt-type functionals show promise, these pairwise additive methods neglect the many-body effects. It is clear that an accurate description of adsorption is still a challenging problem for dispersion-based DFT methods; nevertheless, our results provide a comparison for the performances of the most commonly used vdW functionals that is of broad interest for adsorption studies.

Figure 4. Percentage changes as a function of the lattice constants [% alatt change = %(alattvdW − alattPBE)/alattPBE)] calculated using the five vdW functionals from those of PBE for (a) the positions of the d-band centers [% Ed change = %(EdvdW − EdPBE)/|EdPBE|] and (b) the d-band widths [% W change = %(WdvdW − WdPBE)/WdPBE]. Black squares are for the opt-type functionals, and red circles are for vdW-DF functionals. The solid lines represent the PBE results, which are used as the reference.

functionals. These observations can be tied to the associated lattice constants obtained using different functionals. A close look at Table 1 shows that vdW-DF and vdW-DF2 functionals have the largest lattice constants, which deviate by 5% from those calculated using PBE, while those calculated using opttype functionals are smaller than (using optB86b) or very close (using optB88 and optPBE) to those of PBE. It is expected that for the vdW functionals, which give larger lattice constants than PBE, the overlap of wave functions is reduced, leading to narrowing of d-band widths, while the opposite trend is expected for opt-type functionals as shown in Figure 4b. With this analysis, we show that the trends observed for changes in d-band features employing different vdW functionals correlate with the associated lattice constants, and hence can be grouped as similar to the trends in the changes obtained for the adsorption energies and heights. However, note that it is not trivial to use the changes in d-band features to further comment on the strength of benzene adsorption on these metal substrates.



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Present Address §

H.Y.: School of Chemical Engineering, Purdue University, West Lafayette, IN, 47907, USA. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS A.K. acknowledges support from the U.S. Department of Energy Basic Energy Science under Contract No. DE-FG0211ER16243. This research used resources of 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-FG02-11ER16243. A.K. thanks the University of Zurich for hospitality.

IV. CONCLUSIONS In summary, benzene adsorption characteristics on coinage and transition metal surfaces are studied to explore the role of nonlocal correlations and to compare the performances of the vdW-DF family functionals. For the adsorption on coinage substrates we find a systematic increase in the adsorption energies with the inclusion of vdW interactions. Such an increase can reach above 0.7 eV, with the highest increase obtained using opt-type functionals, while the contribution to the adsorption energies using revPBE and rPW86 functionals is lower. We note that vdW interactions are found to be even more significant for adsorption on transition metal surfaces, and



REFERENCES

(1) Sayes, M.; Reyniers, M.-F.; Marin, G. B.; Neurock, M. Density Functional Study of Benzene Adsorption on Pt(111). J. Phys. Chem. B 2002, 106, 7489. (2) Sayes, M.; Reyniers, M.-F.; Neurock, M.; Marin, G. B. Density Functional Theory Analysis of Benzene on Pt(111): Addition and Removal of the First Two H-Atoms. J. Phys. Chem. B 2003, 107, 3844. (3) Somorjai, G. A. Introduction to Surface Chemistry and Catalysis; Wiley: New York, 1994. (4) Tour, J. M.; Jones, L.; Pearson, D. L.; Lamba, J. J. S.; Burgin, T. P.; Whitesides, G. M.; Allara, D. L.; Parikh, A. N.; Atre, S. V. Self-

20580

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

Assembled Monolayers and Multilayers of Conjugated. Thiols, α,ωDithiols, and Thioacetyl-Containing Adorbates. Understanding Attachments between Potential Molecular Wires and Gold Surfaces. J. Am. Chem. Soc. 1995, 117, 9529. (5) Witte, G.; Wöll, C. J. Growth of aromatic molecules on solid substrates for applications in organic electronics. J. Mater. Res. 2004, 19, 1889. (6) Zaera, F. An Organometallic Guide to the Chemistry of Hydrocarbon Moieties on Transition Metal Surfaces. Chem. Rev. 1995, 95, 2651. (7) Wolkow, R. A. Controlled Molecular Adsorption on Silicon: Laying a Foundation for Molecular Devices. Annu. Rev. Phys. Chem. 1999, 50, 413. (8) Barlow, S. M.; Raval, R. Complex organic molecules at metal surfaces: bonding, organisation and chirality. Surf. Sci. Rep. 2003, 50, 201. (9) Held, G. Adsorption and dissociation of benzene on bimetallic surfacesthe influence of surface geometry and electronic structure. J. Phys.: Condens. Matter 2003, 15, R1501. (10) Filler, M. A.; Bent, S. F. The surface as molecular reagent: organic chemistry at the semiconductor interface. Prog. Surf. Sci. 2003, 73, 1. (11) Lei, R. Z.; Gellman, A. J.; Koel, B. E. Desorption Energies of Linear and Cyclic Alkanes on Surfaces: Anomalous Scaling with Length. Surf. Sci. 2004, 554, 125. (12) Tait, S. L.; Dohnalek, Z.; Campbell, C. T.; Kay, B. D. n-Alkanes on MgO(100) II: Chain length Dependence of Kinetic Desorption Parameters for Small n-Alkanes. J. Chem. Phys. 2005, 122, 164708. (13) Campbell, C. T.; Sellers, J. R. V. The entropies of adsorbed molecules. J. Am. Chem. Soc. 2012, 134, 18109. (14) Jenkins, S. J. Aromatic adsorption on metals via first-principles density functional theory. Proc. R. Soc. A 2009, 465, 2949. (15) Vanin, M.; Mortensen, J. J.; Kelkkanen, A. K.; Garcia-Lastra, J. M.; Thygesen, K. S.; Jacobsen, K. W. Graphene on metals: A van der Waals density functional study. Phys. Rev. B 2010, 81, 081408. (16) Toyoda, K.; Hamada, I.; Lee, K.; Yanagisawa, S.; Morikawa, Y. Density functional theoretical study of pentacene/noble metal interfaces with van der Waals corrections: Vacuum level shifts and electronic structures. J. Chem. Phys. 2010, 132, 134703. (17) Wellendorff, J.; Kelkkanen, A.; Mortensen, J. J.; Lundqvist, B. I.; Bligaard, T. RPBE-vdW Description of Benzene Adsorption on Au(111). Top. Catal. 2010, 53, 378. (18) Abad, E.; Dappe, Y. J.; Martínez, J. I.; Flores, F.; Ortega, J. C6H6/Au(111): Interface dipoles, band alignment, charging energy, and van der Waals interaction. J. Chem. Phys. 2011, 134, 044701. (19) Bilic, A.; Reimers, J. R.; Hush, N. S.; Hoft, R. C.; Ford, M. J. Adsorption of benzene on copper, silver, and gold surfaces. J. Chem. Theory Comput. 2006, 2, 1093. (20) Medeiros, P. V. C.; Gueorguiev, G. K.; Stafström, S. Benzene, coronene, and circumcoronene adsorbed on gold, and a gold cluster adsorbed on graphene: Structural and electronic properties. Phys. Rev. B 2012, 85, 205423. (21) Morin, C.; Simon, D.; Sautet, P. Density-Functional Study of the Adsorption and Vibration Spectra of Benzene Molecules on Pt(111). J. Phys. Chem. B 2003, 107, 2995. (22) Morin, C.; Simon, D.; Sautet, P. Chemisorption of benzene on Pt(111), Pd(111) and Rh(111) metal surfaces: a structural and vibrational comparison from first principles. J. Phys. Chem. B 2004, 108, 5653. (23) Orita, H.; Itoh, N. Simulation of phenol formation from benzene with a Pd membrane reactor: ab initio periodic density functional study. Appl. Catal., A: Gen. 2004, 258, 17. (24) Gao, W.; Zheng, T.; Jiang, Q. Dehydrogenation of benzene on Pt(111) surface. J. Chem. Phys. 2008, 129, 164705. (25) Da Silva, J. L. F.; Stampfl, C.; Scheffler, M. Adsorption of Xe Atoms on Metal Surfaces: New Insights from First-Principles Calculations. Phys. Rev. Lett. 2003, 90, 066104.

(26) Chen, D. L.; Al-Saidi, W. A.; Johnson, J. K. Noble gases on metal surfaces: Insights on adsorption site preference. Phys. Rev. B 2011, 84, 241405. (27) Silvestrelli, P. L.; Ambrosetti, A.; Grubisic, S.; Ancilotto, F. Adsorption of rare-gas atoms on Cu(111) and Pb(111) surfaces by van der Waals corrected density functional theory. Phys. Rev. B 2012, 85, 165405. (28) Dion, M.; Rydberg, H.; Schröder, E.; Langreth, D. C.; Lundqvist, B. I. Van der Waals Density Functional for General Geometries. Phys. Rev. Lett. 2004, 92, 246401. (29) Björkman, T.; Gulans, A.; Krasheninnikov, A. V.; Nieminen, R. M. Are we van der Waals ready? J. Phys. Condens. Matter 2012, 24, 424218. (30) Furche, F. Developing the random phase approximation into a practical post-Kohn−Sham correlation model. J. Chem. Phys. 2008, 129, 114105. (31) Klimeš, J.; Bowler, D. R.; Michaelides, A. Chemical accuracy for the van der Waals density functional. J. Phys.: Condens. Matter 2010, 22, 022201. (32) Tkatchenko, A.; Romaner, L.; Hofmann, O. T.; Zojer, E.; Ambrosch-Draxl, C.; Scheffler, M. Van der Waals Interactions Between Organic Adsorbates and at Organic/Inorganic Interfaces. MRS Bull. 2010, 35, 435. (33) Carrasco, J.; Santra, B.; Klimeš, J.; Michaelides, A. To Wet or Not to Wet? Dispersion Forces Tip the Balance for Water Ice on Metals. Phys. Rev. Lett. 2011, 106, 026101. (34) Ruiz, V. G.; Liu, W.; Zojer, E.; Scheffler, M.; Tkatchenko, A. Density-Functional Theory with Screened van der Waals Interactions for the Modeling of Hybrid Inorganic-Organic Systems. Phys. Rev. Lett. 2012, 108, 146103. (35) Grimme, S. Density functional theory with London dispersion corrections. Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2011, 1, 211. (36) Burke, K. Perspective on density functional theory. J. Chem. Phys. 2012, 136, 150901. (37) Sherrill, C. D. Frontiers in Electronic Structure Theory. J. Chem. Phys. 2010, 132, 110902. (38) Riley, K. E.; Pitoňaḱ , M.; Jurečka, P.; Hobza, P. Stabilization and Structure Calculations for Noncovalent Interactions in Extended Molecular Systems Based on Wave Function and Density Functional Theories. Chem. Rev. 2010, 110, 5023. (39) Liu, W.; Carrasco, J.; Santra, B.; Michaelides, A.; Scheffler, M.; Tkatchenko, A. Benzene adsorbed on metals: Concerted effect of covalency and van der Waals bonding. Phys. Rev. B 2012, 86, 245405. (40) Klimeš, J.; Michaelides, A. Perspective: Advances and challenges in treating van der Waals dispersion forces in density functional theory. J. Chem. Phys. 2012, 137, 120901. (41) Grimme, S. Semiempirical GGA-type density functional constructed with a long-range dispersion correction. J. Comput. Chem. 2006, 27, 1787. (42) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010, 132, 154104. (43) Tkatchenko, A.; Scheffler, M. Accurate Molecular Van Der Waals Interactions from Ground-State Electron Density and FreeAtom Reference Data. Phys. Rev. Lett. 2009, 102, 073005. (44) Becke, A. D.; Johnson, E. R. A density-functional model of the dispersion interaction. J. Chem. Phys. 2005, 123, 154101. (45) Puzder, A.; Dion, M.; Langreth, D. C. Binding energies in benzene dimers: Nonlocal density functional calculations. J. Chem. Phys. 2006, 124, 164105. (46) Furche, F. Molecular tests of the random phase approximation to the exchange-correlation energy functional. Phys. Rev. B 2001, 64, 195120. (47) Jansen, G.; Liu, R.-F.; Á ngyán, J. G. On the equivalence of ringcoupled cluster and adiabatic connection fluctuation-dissipation theorem random phase approximation correlation energy expressions. J. Chem. Phys. 2010, 133, 154106. 20581

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

(48) Purvis, D. G.; Bartlett, R. J. A full coupled-cluster singles and doubles model: The inclusion of disconnected triples. J. Chem. Phys. 1982, 76, 1910. (49) Voorhis, T. V.; Head-Gordon, M. Two-body coupled cluster expansions. J. Chem. Phys. 2001, 115, 5033. (50) Lee, K.; Murray, D. E.; Kong, L.; Lundqvist, B. I.; Langreth, D. C. Higher-accuracy van der Waals density functional. Phys. Rev. B 2010, 82, 081101. (51) Vydrov, O. A.; Voorhis, T. V. Nonlocal van der Waals Density Functional Made Simple. Phys. Rev. Lett. 2009, 103, 063004. (52) Klimeš, J.; Bowler, D. R.; Michaelides, A. A. Van der Waals density functionals applied to solids. Phys. Rev. B 2011, 83, 195131. (53) Cooper, V. R. Van der Waals density functional: An appropriate exchange functional. Phys. Rev. B 2010, 81, 161104. (54) Wellendorff, J.; Lundgaard, K. T.; Møgelhøj, A.; Petzold, V.; Landis, D. D.; Nørskov, J. K.; Bligaard, T.; Jacobsen, K. W. Density functionals for surface science: Exchange-correlation model development with Bayesian error estimation. Phys. Rev. B 2012, 85, 235149. (55) Perdew, J. P.; Ernzerhof, M.; Burke, K. Rationale for mixing exact exchange with density functional approximations. J. Chem. Phys. 1996, 105, 9982. (56) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865. (57) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 1993, 47, 558. (58) Kresse, G.; 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. (59) Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 1996, 54, 11169. (60) Thonhauser, T.; Cooper, V. R.; Li, S.; Puzder, A.; Hyldgaard, P.; Langreth, D. C. Van der Waals density functional: Self-consistent potential and the nature of the van der Waals bond. Phys. Rev. B 2007, 76, 125112. (61) Wellendorff, J.; Bligaard, T. On the Importance of GradientCorrected Correlation for van der Waals Density Functionals. Top. Catal. 2011, 54, 1143. (62) Kresse, G.; Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 1999, 59, 1758. (63) Blöchl, P. Projector augmented-wave method. Phys. Rev. B 1994, 50, 17953. (64) Monkhorst, H. J.; Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 1976, 13, 5188. (65) Hao, P.; Fang, Y.; Sun, J.; Csonka, G. I.; Philipsen, P. H. T.; Perdew, J. P. Lattice constants from semilocal density functionals with zero-point phonon correction. Phys. Rev. B 2012, 85, 014111. (66) Yamagishi, S.; Jenkins, S. J.; King, D. A. Symmetry and site selectivity in molecular chemisorption: Benzene on Ni{111}. J. Chem. Phys. 2001, 114, 5765. (67) Mittendorfer, F.; Thomazeau, C.; Raybaud, P.; Toulhoat, H. Adsorption of Unsaturated Hydrocarbons on Pd(111) and Pt(111): A DFT Study. J. Phys. Chem. B 2003, 107, 12287. (68) Gao, W.; Zheng, W.; Jiang, Q. Dehydrogenation of benzene on Pt(111) surface. J. Chem. Phys. 2008, 129, 164705. (69) Mittendorfer, F.; Garhofer, A.; Redinger, J.; Klimeš, J.; Harl, J.; Kresse, G. Graphene on Ni(111): Strong interaction and weak adsorption. Phys. Rev. B 2011, 84, 201401. (70) Tonigold, K.; Groß, A. Adsorption of small aromatic molecules on the (111) surfaces of noble metals: A density functional theory study with semiempirical corrections for dispersion effects. J. Chem. Phys. 2010, 132, 224701. (71) Caputo, R.; Prascher, B. P.; Staemmler, V.; Bagus, P. S.; Wöll, C. Adsorption of Benzene on Coinage Metals: A Theoretical Analysis Using Wavefunction-Based Methods. J. Phys. Chem. A 2007, 111, 12778. (72) Berland, K.; Einstein, T. L.; Hyldgaard, P. Rings sliding on a honeycomb network: Adsorption contours, interactions, and assembly of benzene on Cu(111). Phys. Rev. B 2009, 80, 155431.

(73) Zhou, X.-L.; Castro, M. E.; White, J. M. Interactions of UV photons and low energy electrons with chemisorbed benzene on Ag(111). Surf. Sci. 1990, 238, 215. (74) Syomin, D.; Kim, J.; Koel, B. E.; Ellison, G. B. Identification of Adsorbed Phenyl (C6H5) Groups on Metal Surfaces: Electron-Induced Dissociation of Benzene on Au(111). J. Phys. Chem. B 2001, 105, 8387. (75) Koschel, H.; Held, G.; Steinruck, H. P. The orientation of benzene on bimetallic surfaces. Surf. Rev. Lett. 1999, 6, 893. (76) Xi, M.; Yang, M. X.; Jo, S. K.; Bent, B. E.; Stevens, P. Benzene adsorption on Cu(111): Formation of a stable bilayer. J. Chem. Phys. 1994, 101, 9122. (77) Ihm, H.; Ajo, H. M.; Gottfried, J. M.; Bera, P.; Campbell, C. T. Calorimetric Measurement of the Heat of Adsorption of Benzene on Pt(111). J. Phys. Chem. B 2004, 108, 14627. (78) Tysoe, W. T.; Ormerod, R. M.; Lambert, R. M.; Zgrablich, G.; Ramirez-Cuesta, A. Overlayer structure and kinetic behavior of benzene on palladium(111). J. Phys. Chem. 1993, 97, 3365. (79) Lee, A. F.; Wilson, K.; Lambert, R. M.; Goldoni, A.; Baraldi, A.; Paolucci, G. On the Coverage-Dependent Adsorption Geometry of Benzene Adsorbed on Pd{111}: A Study by Fast XPS and NEXAFS. J. Phys. Chem. B 2000, 104, 11729. (80) Myers, A. K.; Schoofs, G. R.; Benziger, J. B. Comparison of benzene adsorption on nickel(111) and nickel(100). J. Phys. Chem. 1987, 91, 2230. (81) Chwee, T. S.; Sullivan, M. B. Adsorption studies of C6H6 on Cu (111), Ag (111), and Au (111) within dispersion corrected density functional theory. J. Chem. Phys. 2012, 137, 134703. (82) Liu, W.; Ruiz, V. G.; Zhang, G.-X.; Santra, B.; Ren, X.; Scheffler, M.; Tkatchenko, A. Structure and energetics of benzene adsorbed on transition-metal surfaces: density-functional theory with van der Waals interactions including collective substrate response. New J. Phys. 2013, 15, 053046. (83) Morin, C.; Simon, D.; Sautet, P. Intermediates in the hydrogenation of benzene to cyclohexene on Pt(111) and Pd(111): A comparison from DFT calculations. Surf. Sci. 2006, 600, 1339. (84) Saeys, M.; Reyniers, M.; Marin, G.; Neurock, M. Density Functional Study of Benzene Adsorption on Pt(111). J. Phys. Chem. B 2002, 106, 7489. (85) Morin, C.; Simon, D.; Sautet, P. Trends in the Chemisorption of Aromatic Molecules on a Pt(111) Surface: Benzene, Naphthalene, and Anthracene from First Principles Calculations. J. Phys. Chem. B 2004, 108, 12084. (86) Yannoulis, P.; Dudde, R.; Frank, K. H.; Koch, E. E. Orientation of aromatic hydrocarbons on metal surfaces as determined by NEXAFS. Surf. Sci. 1987, 189−190, 519. (87) Wander, A.; Held, G.; Hwang, R. Q.; Blackman, G. S.; Xu, M. L.; de Andres, P.; Van Hove, M. A.; Somorjai, G. A. A diffuse LEED study of the adsorption structure of disordered benzene on Pt(111). Surf. Sci. 1991, 249, 21. (88) Held, G.; Bessent, M. P.; Titmuss, S.; King, D. A. Realistic molecular distortions and strong substrate buckling induced by the chemisorption of benzene on Ni{111}. J. Chem. Phys. 1996, 105, 11305. (89) Schaff, O.; Fernandez, V.; Hofmann, Ph.; Schindler, K.-M.; Theobald, A.; Fritzsche, V.; Bradshaw, A. M.; Davis, R.; Woodruff, D. P. Coverage-dependent changes in the adsorption of benzene on Ni{111}. Surf. Sci. 1996, 348, 89. (90) Bagus, P. S.; Hermann, K.; Wöll, C. The interaction of C6H6 and C6H12 with noble metal surfaces: Electronic level alignment and the origin of the interface dipole. J. Chem. Phys. 2005, 123, 184109. (91) Kelkkanen, A.; Lundqvist, B.; Nørskov, J. Van der Waals effect in weak adsorption affecting trends in adsorption, reactivity, and the view of substrate nobility. Phys. Rev. B 2011, 83, 113401. (92) Toyoda, K.; Nakano, Y.; Hamada, I.; Lee, K.; Yanagisawa, S.; Morikawa, Y. First-principles study of benzene on noble metal surfaces: Adsorption states and vacuum level shifts. Surf. Sci. 2009, 603, 2912. 20582

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583

The Journal of Physical Chemistry C

Article

(93) Tirendi, C.; Mills, G.; Dybowski, C.; Neue, G. Platinum-proton coupling in the NMR spectrum of benzene on alumina-supported platinum catalyst. J. Phys. Chem. 1992, 96, 5045. (94) Weiss, P.; Eigler, D. Site dependence of the apparent shape of a molecule in scanning tunneling micoscope images: Benzene on Pt{111}. Phys. Rev. Lett. 1993, 71, 3139. (95) Stranick, S.; Kamna, M.; Weiss, P. Atomic-Scale Dynamics of a Two-Dimensional Gas-Solid Interface. Science 1994, 266, 99. (96) Van Hove, M.; Lin, R.; Somorjai, G. Efficient Scheme for Calculation of Low-Energy Electron-Diffraction Intensities in the Presence of Large Superlattices, with Application to the Structural Analysis of Benzene Adsorbed on Rh(111). Phys. Rev. Lett. 1983, 51, 778. (97) Huber, W.; Zebisch, P.; Bornemann, T.; Steinruck, H.-P. Lateral interactions and azimuthal orientation of pure and co-adsorbed benzene on Ni(111). Surf. Sci. 1991, 258, 16. (98) Steinruck, H.-P.; Huber, W.; Pache, T.; Menzel, D. The adsorption of benzene mono- and multilayers on Ni(111) studied by TPD and LEED. Surf. Sci. 1989, 218, 293. (99) Huber, W.; Steinruck, H.-P.; Pache, T.; Menzel, D. The electronic structure and molecular symmetry of pure benzene and benzene coadsorbed with CO on Ni(111). Surf. Sci. 1989, 217, 103. (100) Yildirim, H.; Kara, A. Effect of van der Waals Interactions on the Adsorption of Olympicene Radical on Cu(111): Characteristics of Weak Physisorption versus Strong Chemisorption. J. Phys. Chem. C 2013, 117, 2893. (101) Carrasco, J.; Klimeš, J.; Michaelides, A. The role of van der Waals forces in water adsorption on metals. J. Chem. Phys. 2013, 138, 024708.

20583

dx.doi.org/10.1021/jp404487z | J. Phys. Chem. C 2013, 117, 20572−20583