A coverage dependent study of the adsorption of pyridine on the (111) coinage metal surfaces

Journal Pre-proof

A Coverage Dependent Study of the Adsorption of Pyridine on the (111) Coinage Metal Surfaces Walter Malone , Abdelkader Kara PII: DOI: Reference:

S0039-6028(19)30508-4 https://doi.org/10.1016/j.susc.2019.121525 SUSC 121525

To appear in:

Surface Science

Received date: Revised date: Accepted date:

4 July 2019 30 October 2019 30 October 2019

Please cite this article as: Walter Malone , Abdelkader Kara , A Coverage Dependent Study of the Adsorption of Pyridine on the (111) Coinage Metal Surfaces, Surface Science (2019), doi: https://doi.org/10.1016/j.susc.2019.121525

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier B.V.

Highlights    

All functionals predict pyridine to adsorb with its plane perpendicular to the surface. SCAN+rVV10 predicts more charge transfer than to the pyridine molecule than any other functionals. All functionals disagree with experiment. The finial adsorption configuration fails to change with coverage.

A Coverage Dependent Study of the Adsorption of Pyridine on the (111) Coinage Metal Surfaces Walter Malone1 and Abdelkader Kara1,2* 1

Physics Department, University of Central Florida, Orlando, Florida 32816, United States

2

Renewable Energy and Chemical Transformations Cluster, University of Central Florida, Orlando, Florida 32816, United States

*

Corresponding author

Abstract Using density functional theory we explore the adsorption of pyridine (NC5H5) on Cu, Ag, and Au(111) surfaces. To study the effect of coverage we run calculations on a 3x4 (12 atom), 4x4 (16 atom), 5x4 (20 atom), and 4x6 (24 atom) surface. To examine the role of the van der Waals interaction we use the following vdW inclusive functionals: optB86-vdW, optB88-vdW, optPBE-vdW, revPBE-vdW, rPW86-vdW2, SCAN+rVV10 along with the PBE functional. Regardless of functional, metal, and coverage we find the most energetically stable adsorption site to be a vertical site with the pyridine’s N atom directly above a metal atom, which contradicts the available experimental literature. Introducing the vdW interaction lowers the energy gap between the most energetically favorable flat and vertical adsorption configurations. Moving to a higher coverage increases the overall interaction of the pyridine molecule with the surface. We also find that the SCAN+rVV10 functional performs similarly to the optB86b-vdW and optB88-vd functional. Along with the geometric properties of the substrate/adsorbate system, we report several electronic properties of the substrate/adsorbate system such as charge transfer from the surface to the molecule, change in the surface’s work function, and change in dband of the atoms in the first layer of the substrate. Keywords: pyridine, density functional theory, coinage metal surfaces, van der Waals interaction, SCAN+rVV10 1. Introduction Organic molecules have received widespread attention as they can be used to fabricate electronic devices [1,2]. The performance these devices are often dictated by the interaction of a layer of organic molecules with an inorganic substrate. The interaction of organic molecules with inorganic substrates is also important in the field of catalysis. Traditionally, transition metals have been used to catalyze ring cracking reactions that break apart potentially hazardous organic materials [3]. However, critically important to the description of the interface of organic

molecules and inorganic substrates, along with a myriad of other systems, is the van der Waals (vdW) interaction [4-17]. Density functional theory (DFT) [18], a potent tool for studying systems on the molecular scale, fails to correctly model long range dispersion forces when using the standard generalized gradient approximation [19-21] (GGA) or the local density approximation (LDA). In simple terms, the LDA models systems at the atomic scale using the electron density while the GGA models systems at the atomic scale using the electron density and the gradient of the electron density. Several attempts have been made to include the vdW interaction within the GGA. These attempts have led to the creation of popular methods such as the DFT-D2 [22], DFT-D3 [23], vdW(TS) [24], and the Becke-Johnson (BJ) [25] approach. While these methods have had a large degree of success they all share a common weakness. They account for long range dispersion semi-empirically, requiring some predetermined input parameters. The vdW-DF [4,26] and vdW-DF2 [27] methods avoid semi-empiricism by adding a nonlocal correlation term to the exchange-correlation interaction. With this addition the total exchange-correlation energy reads: (1) Where the first term is a GGA exchange term, the second term is a LDA correlation term, and the last term is the nonlocal correlation term previously mentioned. By including this nonlocal term, vdW-DF and vdW-DF2 can seamlessly describe long range dispersion. Obviously though the accuracy of these methods depend on the choice of the exchange term that one pairs with the nonlocal correlation term. Several exchange terms are often paired with the vdW-DF correlation term leading to various different flavors of the method, notably optB86-vdW [4], optB88-vdW [26], optPBE-vdW [26], revPBE-vdW [28,29], and BEEF-vdW [30]. In contrast the vdW-DF2 correlation term is only commonly paired with one exchange partner leading to the rPW86vdW2 [27,31] method. In addition to the vdW-DF and vdW-DF2 methods, the relatively new Strongly Constrained Appropriately Normed (SCAN) [32] functional has shown a lot of promise. This functional, a meta-GGA functional [33-37] which uses the Laplacian of the electron density in addition to the electronic density and the gradient of electron density to model systems at the atomic scale, satisfies all 17 known exact constraints a semi-local functional should satisfy, making it strong candidate for further screening. However, SCAN too, like standard LDA and GGA, fails to appropriately account for long range dispersion. To include long range dispersion SCAN is paired with the rVV10 correlation term [38] creating the SCAN+rVV10 functional [39]. SCAN+rVV10’s novelty, its inclusion of the vdW interaction, and its satisfaction of all 17 known exact constraints a semi-local functional should satisfy mark it as an excellent choice to screen alongside other vdW inclusive methods.

Before going further we must mention that the Random-Phase Approximation (RPA) within the framework of the adiabatic-connection fluctuation dissipation theorem (ACFD) [40-42] provides an accurate way, possibly even more accurate than any method mentioned thus far, to calculate electron correlation energy. However, its computation cost is often prohibitive. For that reason we avoid using ACFD-RPA in this study. Given the wide variety of DFT methods that try to account for the vdW interactions and the importance of the vdW interactions when describing the interface between organic molecules and metallic substrates, additional studies are needed to both further screen the effectiveness of different methods and to study systems of technological interest. To screen SCAN+rVV10 and other vdW inclusive methods we use the model system of pyridine on transition metal surfaces. Pyridine (C5H5N) is one of the simplest aromatic compounds. Simple aromatic molecules such as pyridine, thiophene, and benzene are appealing for computational study for several reasons. They are the building blocks of larger organic molecules that can be used to fabricate organic electronics [43]. Their simplicity also allows the possibility of screening a wide variety of different methods. In this DFT investigation we study the adsorption of pyridine on Au, Ag, and Cu(111) surfaces using the following vdW inclusive methods: optB86-vdW, optB88-vdW, optPBE-vdW, revPBE-vdW , rPW86-vdW2, and SCAN+rVV10. For practical reasons we restrict the scope of our study to only consider the parameter free vdW-DF and vdW-DF2 methods along with the promising SCAN+rVV10 functional. We also restrict our study to explore pyridine adsorption on coinage metals. The coinage metals make for an attractive substrate choice as long range dispersion forces should play an especially important role in the bonding of pyridine with these less reactive surfaces. These systems have also been studied extensively in the literature [4462], providing plenty of results to compare with our own. Chief among those results is that pyridine is thought to undergo a coverage driven phase transition on these metal surfaces [44-49], presenting a unique opportunity to gauge the effectiveness of the vdW-DF, vdW-DF2, and SCAN+rVV10 methods. Taking advantage of this purported coverage-driven phase transition, we explore the adsorption of pyridine in a 3x4, 4x4, 5x4, and 4x6 unit cell using again, as mentioned, the vdW inclusive optB86-vdW, optB88-vdW, optPBE-vdW, revPBE-vdW , rPW86-vdW2, and SCAN+rVV10, functionals. In section 2, we offer the computational details of the study. In section 3, we examine the equilibrium adsorption geometries and energetics of the relaxed pyridine/substrate systems in the 3x4 unit cell. In section 4, we analyze the change in the electronic properties of the system including change in work function of the surface, modification of the partial density of states (total d states) of the surface, and charge transfer between the adsorbate and substrate in the 3x4 unit cell. In section 5, we examine the effects of coverage analyzing the results in the larger unit cells. Finally, in section 6, we present our conclusions.

2. Computational Details We perform all computations in the Vienna Ab initio simulation package (VASP) version 5.3.5 [53-65]. VASP runs calculations using the projector augmented wave (PAW) method [6667]. As previously stated, we utilize the following vdW inclusive methods: optB86b-vdW [4], optB88-vdW [26], optPBE-vdW [26], revPBE-vdW [28,29], rPW86-vdW2 [27,31], and SCAN+rVV10[39]. For reference we also run calculations using the Perdew–Burke–Ernzerhof (PBE) functional [68], a GGA type functional which fails to appropriately account for long range dispersion. To model our (111) coinage metal surfaces we use the slab method. Each slab consists of six layers with at least 25 Å of vacuum separating neighboring slabs. We construct slabs using the calculated theoretical lattice constant given in Table S1 in the supplementary information. In order to gauge the effects of coverage we explore several superstructures; one superstructure in which each layer contains 12 atoms (4x3), one superstructure in which each layer contains 16 atoms (4x4), one superstructure in which each layer contains 20 atoms (5x4), and one superstructure in which each layer contains 24 atoms (4x6). For the 4x3 and 4x4 unit cells we use a 6x6x1 Monkhorst-Pack grid to sample the Brillouin zone, and for the 5x4 and 4x6 unit cells we use a 4x4x1 Monkhorst-Pack grid to sample the Brillouin zone. Other than this difference in k-point sampling, all input parameters for the various unit cells are the same. We use a plane wave energy cutoff of 400 eV, a force criterion of 0.02 eV/Å, and we accomplish structural optimization using the conjugate gradient (CG) method [69,70]. We allow both the substrate and the pyridine molecule to relax separately before we place the pyridine molecule on the substrate. Once we place the pyridine molecule approximately 3Å above the substrate, we fix the bottom three layers of the substrate at their relaxed positions; their positions in the relaxed slab. Except for those bottom three layers of substrate atoms, we allow all other atoms to undergo structural relaxation. To find the equilibrium adsorption site in the 3x4 unit cell we screen 20 possible adsorption sites which we display in Fig. 1. For 8 of these adsorption sites, a)-h) in Fig. 1., the plane of the molecule is perpendicular to the surface. These are the so-called vertical configurations and are label with a ―v‖. The remaining 12 adsorption configurations, i)-t), position the molecule’s plane parallel to the surface, making these configurations the so-called flat configurations. Furthermore, we can subdivide adsorption sites based on what site the center of the molecule lies over on the surface whether it be a top site, a)-b) and s)-t), a bridge site, c)-d) and i)-l), a fcc site, e)-f) and m)-o), or a hcp site, g)-h) and p)-r). For the other, larger unit cells we only use two adsorption sites, v-a30 (Fig. 1. b) and bB1 (Fig. 1. k). These adsorption sites are the most stable vertical and flat adsorption sites, respectively, in the 3x4 unit cell. In addition to saving computational time, we use pyridine’s coverage-driven phase transition from a flat to a vertical adsorption configuration on these transition metal surfaces to justify our choice of studying only the most stable flat and vertical adsorption sites in the larger unit cells. If any changes in the adsorption properties of pyridine on the (111) coinage metal surfaces occur as a function of

coverage it should be their relative preference for a flat or vertical adsorption site. We also only utilize the SCAN+rVV10 and the optB88-vdW functional in the larger unit cells again to save computational time, and guided by the observation that all the vdW-DF and vdW-DF2 methods perform similarly in the 3x4 unit cell. Finally, we define adsorption energy as: Eads = - (Epyridine/metal – Epyridine – Emetal)

(2)

Where Epyridine/metal is the energy of pyridine/metal system, E pyridine is the energy of the pyridine molecule in the gas phase, and Emetal is the energy of the clean substrate. With this definition the higher the adsorption energy the more stable the adsorption site. In the next section we will give a detailed analysis of both the adsorption energetics and geometry for each adsorption configuration/metal combination in the 3x4 unit cell.

Figure 1. Initial adsorption configurations for pyridine on Cu, Ag, and Au(111) surfaces. Going from left to right and top to bottom we have the following adsorption configurations: a) v-a, b) v-a30, c) v-b, d) v-b90, e) v-fcc, f) v-fcc30, g) v-hcp, h) v-hvp30, i) bA1, j) bA2, k)

bB1, l) bB2, m) fA, n) fB1, o) fB2, p) hA, q) hB1, r) hB2, s) t1, and t) t2. Orange atoms represent N atoms, red atoms represent C atoms, white atoms represent H atoms, silver atoms represent first layer substrate atoms, and teal atoms represent second layer substrate atoms.

3. Equilibrium Adsorption Geometry and Energy In the 3x4 unit cell, each functional unanimously predicts the most stable adsorption sites to be v-a and v-a30. Rotating the molecule by 30˚ from v-a to v-a30 has no effect on adsorption energy to 0.02 eV. See Fig. 2. for a complete illustration of the adsorption energies of pyridine in the 3x4 unit cell. In Fig. 2 the x-axis indicates the initial adsorption site. The y-axis indicates the adsorption energy of the final adsorption site, the configuration of pyridine after structural relaxation given the initial adsorption site. Unless a point in Fig. 2 sits in a shaded box the initial and final adsorption sites are roughly the same. Points located in the shaded area of Fig. 2 correspond to unstable adsorption sites; the initial and final adsorption configurations differ significantly. Interestingly only vertical adsorption sites are unstable. Specifically, v-b, v-fcc, vfcc30, v-hcp, and v-hcp30 could be unstable adsorption sites. The final adsorption configuration among these unstable adsorption sites is always a final adsorption configuration we call ―a‖. In this position the N atom sits directly above a metal atom just like the v-a and v-a30 configurations. However unlike the v-a or v-a30 sites, molecules in these positions are no longer strictly vertical and instead acquire a tilt away from the surface normal. Fig. 3 illustrates an example of the adsorption configuration ―a‖. Figures S1-S21 in the supplementary information provide an illustration of every individual final adsorption site. We will discuss the instabilities of the vertical adsorption sites later. First starting with the flat configurations, we notice that translating and rotating the molecule, when flat, along the surface modifies the adsorption energy little except perhaps in the cases of bB1, hB1, and fB1. Pyridine in these three adsorption sites often possesses larger adsorption energies than pyridine in the other initially flat adsorption sites. The reason for this is simple. When we place pyridine in either bB1, hB1, or fB1 its N atom is close enough to a metal atom to allow the molecule to begin to tilt away from the surface. More interestingly, only the SCAN+rVV10, optB6b-vdW, and optB88-vdW functionals appear to predict this type of behavior while the PBE, optPBE-vdW, revPBE-vdW, and rPW86-vdW2 functionals predict pyridine to remain relatively flat on the surface if initially flat on the surface. The most extreme example of pyridine picking up a tilt angle away from the surface occurs when the molecule sits initially in the hB1 site on Cu(111). Using SCAN+rVV10 pyridine in this configuration acquires a 40˚ tilt angle away from the surface plane during relaxation. Even though the molecule could tilt away from the surface when initially flat, the center of the molecule would always sit approximately over the initial adsorption sit.

Moving on, we notice some other trends among flat and vertical configurations. For PBE the vertical configurations always possess larger adsorption energies than that of the flat configurations. For the vdW inclusive functionals we see that, with the exception of v-a, v-a30, or initial adsorptions sites that move to an ―a‖ site, the vertical adsorption sites tend to possess adsorption energies similar to or lower than that of the flat adsorption sites. Overall including the vdW interactions lowers the energy difference between the flat configurations and both the v-a and v-a30 configurations. This suggests that the vdW interaction could play an important role in keeping pyridine parallel to the surface.

Figure 2. Adsorption Energies for pyridine adsorbed on a) Cu(111), b) Ag(111), and c) Au(111) in the 3x4 unit cell. Points in the shaded box correspond to unstable adsorption configurations where the final adsorption configuration differed from the initial adsorption configuration. The final structures after optimization are discussed in the main text and displayed in Supporting Information, Figures S1-S2.

Figure 3. The final adsorption configuration “a” where the pyridine molecule would move off an initially vertical adsorption site such that its N atom could bond with a metal atom. During relaxation pyridine in this configuration would pick up a tilt angle away from the surface normal. Orange atoms represent N atoms, red atoms represent C atoms, white atoms represent H atoms, silver atoms represent first layer substrate atoms, and teal atoms represent second layer substrate atoms. As mentioned, some configurations move during relaxation making those configurations unstable. First let us point out that for PBE and revPBE-vdW we observe no instability among the initial adsorption sites over all three coinage metal surfaces. In other words the final adsorption configuration was always the same as the initial adsorption configuration. Using rPW86-vdW2 we calculate v-hcp30 and v-fcc30 to be unstable sites on Ag(111). The optPBEvdW functional predicts v-fcc30 and v-hcp30 to be unstable on Cu(111) and Ag(111). The optB86b-vdW and optB88-vdW functionals predict v-b, v-fcc, v-fcc30, v-hcp, and v-hcp30 to be unstable on Cu(111), v-fcc30 and v-hcp30 to be unstable on Ag(111), and v-hcp, v-hcp30, v-fcc, and v-fcc30 to be unstable on Au(111). Finally, SCAN+rVV10 predicts v-b, v-fcc, v-fcc30, vhcp, and v-hcp30 to be unstable on Cu(111), and v-fcc30 and v-hcp30 to be unstable on Ag(111) and Au(111). For these ―a‖ configurations the pyridine molecule could acquire a tilt angle up to

35˚ away from the surface normal. Moreover, these new ―a‖ configurations gave adsorptions energies close to the strictly vertical v-a and v-a30 configurations (see Fig. 2), demonstrating two important things. First the bonding of the pyridine’s N atom with the metal atom appears to be paramount to the stability of the final stable adsorption site. Second, the pyrdine molecule can be rotated by a noticable amount, sometimes up to 35˚, away from the surface normal without a large change in adsorption energy so long as the N atom remains closely bonded to a metal atom. Once we establish the configuration with the highest binding energy for each surface/functional combination, we delve deeper into the geometric properties of the adsorbate/surface system. As we stated earlier, v-a and v-a30 often tie for highest binding energy configuration or are within 0.02 eV of one another. As expected, when we run the geometrical analysis both v-a and v-a30 possess nearly identical geometrical properties, making distinguishing them inconsequential. However, for transparency we list in Table 1, along with adsorption energies, which configuration, v-a or v-a30, yields the highest adsorption energy, and several geometric properties of the highest adsorption energy configuration. Specifically we list the distance between the pyridine’s N atom and the closest metal atom (N-Metal), N-C bond lengths (N-C), buckling of the first layer of the substrate which we define as the maximum zcoordinate minus the minimum z-coordinate of the metal atoms that compose the first layer of the substrate, and the tilt angle (Θ) of the molecule which we define as the angle the plane of the molecule makes with the surface. As we can ascertain from Table 1 and Fig. 2, the adsorption energies increase, omitting SCAN+rVV10, as PBE < revPBE-vdW < rPW86-vdW2 < optPBEvdW < optB88-vdW < optB6b-vdW. In our own previous studies of small organic molecules on coinage metal surfaces we observe the exact same trend [17,71,72]. Moreover, the adsorption energy trend PBE < optPBE-vdW < optBb86b-vdW agrees well with a previous benchmarking study of small organic molecules on Pt(111) [73]. Moving further, the SCAN+rVV10 functional yields adsorption energies on the larger end of the functional spectrum. SCAN+rVV10 calculates an adsorption energy between optB86b-vdW and optB88-vdW for pyridine on Cu(111), between optB88-vdW and optPBE-vdW for pyridine on Ag(111), and as high as optB86b-vdW for pyridine on Au(111). Finally we must mention including the vdW interaction enhances the adsorption energy. This enhancement is maximum on Au(111) where we calculate up to a 540 meV increase in adsorption energy going from PBE to our vdW inclusive functionals. This again suggests the vdW interaction plays an important role in the bonding of pyridine to the (111) coinage metal surfaces.

Table 1. Adsorption energy (Eads), N-Metal distances (N-Metal), N-C bond lengths (N-C), buckling of the first layer of the substrate, and tilt angle of the molecule (Θ) for pyridine adsorbed on Cu, Ag, and Au(111) in the 3x4 unit cell. Unit Cell

Funtional/Adsorption Site

Eads(eV)

Ag(111)

PBE/v-a30 optB86-vdW/v-a30 optB88-vdW/v-a30 optPBE-vdW/v-a revPBE-vdW/v-a rPW86-vdW2/v-a30 SCAN+rVV10/v-a30 PBE/v-a optB86b-vdW/v-a30 optB88-vdW/v-a30 optPBE-vdW/v-a30 revPBE-vdW/v-a30 rPW86-vdW2/v-a SCAN+rVV10/v-a30 PBE/v-a optB86b-vdW/v-a30 optB88-vdW/v-a30 optPBE-vdW/v-a revPBE-vdW/v-a rPW86-vdW2/v-a SCAN+rVV10/v-a30

0.28 0.76 0.73 0.67 0.51 0.54 0.72 0.43 0.95 0.90 0.81 0.61 0.63 0.94 0.34 0.88 0.85 0.76 0.57 0.62 0.88

Cu(111)

Au(111)

NMetal(Å) 2.54 2.43 2.43 2.51 2.65 2.59 2.42 2.15 2.10 2.13 2.17 2.24 2.23 2.08 2.43 2.35 2.37 2.45 2.65 2.56 2.34

N-C(Å) 1.35 1.35 1.34 1.35 1.35 1.35 1.34 1.35 1.35 1.35 1.35 1.36 1.35 1.34 1.35 1.35 1.34 1.35 1.35 1.35 1.34

Buckling (Å) 0.20 0.13 0.14 0.16 0.17 0.16 0.16 0.32 0.25 0.28 0.29 0.30 0.34 0.26 0.18 0.11 0.11 0.13 0.13 0.13 0.13

Θ(o) 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90

Despite the notable enhancement in adsorption energy, the vdW inclusive functionals give similar N-Metal distances to that of PBE with PBE’s N-Metal distance lying right in between optB88-vdW and optPBE-vdW’s N-Metal distance for pyridine on Cu(111) and Au(111), and right in between optPBE-vdW and rPW86-vdW2’s N-Metal distance for pyridine on Ag(111). The revPBE-vdW and rPW86-vdW2 functionals consistently give the largest N-Metal distances. This again agrees well with our previous studies [17,71,72] and agrees well with a previous study that determine that the exchange part of these two functionals is too repulsive [4], leading to an underbinding characterized by large adsorption heights. Moving on to the buckling of the first layer of the substrate, on Cu(111) PBE’s results sit between revPBE-vdW and rPW86-vdW2’s results. On Ag(111) and Au(111) PBE actually predicts the most buckling of the surface. The SCAN+rVV10 functional, in contrast, predicts a rather intermediate buckling of the substrate for each metal. For tilt angle and N-C bond lengths

we note very homogenous results with a tilt angle of 90˚ for every functional/metal combination, and a maximum spread of 0.02 Å for N-C bond lengths. All this data suggests while the vdW interaction makes flat adsorption sites more stable, it has little affect on the observed most energetically favorable adsorption site. All these functionals predict that pyridine will adsorb perpendicular to the (111) coinage metal surface with the N atom bonding to a metal atom with similar N-Metal distances, similar buckling of the first layer of substrate, nearly identical N-C bond lengths, and identical tilt angles. If we had to make a judgment about SCAN+rVV10, since we are especially interested in that functional, we could say that the SCAN+rVV10 functional performs similarly to optB88-vdW and optB86-vdW for these pyridine/coinage metal systems. The adsorption energies of all three functionals are similar, they all predict similar unstable adsorption sites, they all predict similar, within 0.05 Å, N-Metal distances, and they all predict a similar buckling of the substrate, within 0.03 Å. Briefly touching on the results by metal, Fig. 4 a) plots the adsorption energy of pyridine in highest adsorption energy configuration on Cu, Ag, and Au(111). From Fig. 4 a) we see that each functional predicts that pyridine interacts the most strongly with Cu(111) followed by Au(111) followed by Ag(111). The N-Metal distances mirror this trend if we assume a larger adsorption energy should lead to a smaller N-Metal distance. Each functional predicts pyridine to have the shortest N-Metal distances on Cu(111) followed by Au(111) followed by Ag(111).

Figure 4. a) Adsorption energies and b) N-Metal distances for pyridine adsorbed on Cu, Ag, and Au(111) in 3x4 unit cell in the highest adsorption energy configuration. While all the functionals predict pyridine to interact the most strongly with Cu(111) and the least strongly with Ag(111) and all the functionals predict a similar final adsorption site, we still have not yet touched on whether the functionals give results in line with the available experimental literature. Comparing with the literature we note that v-a30 or v-a is largely not the experimentally observed adsorption site on any of the coinage metal surfaces at low coverage.

Table 2 compares the available experimental literature with our own work. In the following paragraphs we will take the time to go over the data presented in Table 2 in detail. Starting with Cu(111), Davies and Shukla using X-ray photoelectron spectroscopy (XPS) and vibrational electron-energy-loss spectroscopy (VEELS) conclude that pyridine adsorbs flat on Cu(111) [57]. They use two exposures: 0.01 L and 0.02 L. At 0.02 L they estimate their coverage to be 2.8x1014 molecules/cm2. Moreover, our coverage ranges from 1.37-1.53 x 1014 molecules/cm2 depending on what functional and lattice constant we use, indicating our coverage, in any case, to be lower than that of Davies and Shukla. Given that at a higher coverage Davies and Shukla observe pyridine adsorbing with its ring parallel to the Cu(111) surface it would make sense that we should observe this phenomenon too. Moreover, Zhong et al. using thermal desorption spectroscopic (TDS) and two-photon photoemission (2PPE) spectroscopy too proposes that pyridine adsorbs flat at low coverage, below 0.7 ML [47]. We though unambiguously observe a vertical configuration.

Table 2: A comparison between the available experimental literature and the results presented in this report for the final adsorption configuration of pyridine on Cu, Ag, and Au(111). Θ corresponds the tilt angle of the molecule away from the surface with Θ = 90˚ corresponding to a vertical configuration. Φ corresponds to the tilt of the molecule about its C2ν symmetry axis. Surface

Literature Methods

Cu(111) XPS, VEELS [57] TDS, 2PPE[47] Ag(111) NEXAFS [44] UPS,EELS, TDS [45]

ARUP[52]

Au(111) SEIRAS, STM[48]

Exposure/Coverage (molecules/cm2) 0.01L 0.02L/(2.8x1014) 0.7 ML 4.0L to 5.5L Less than 1.0 ML Phase transition around: 0.5 L / (3x1014)

1.5 L to12L

In Solution

This Work Adsorption Coverage Configuration (molecules/cm2) flat 1.37-1.53x1014 < 0.08 ML flat 1.37-1.53x1014 < 0.08 ML Θ = 40˚to 70˚ 1.03-1.17x1014 < 0.08 ML Below phase 1.03-1.17x1014 transition: < 0.08 ML Θ = 5˚ Above phase transition: Θ = 55˚, Φ = 30˚ Vertical 1.03-1.17x1014 0 √3x√3/30 < 0.08 ML LEED Structure Flat (low 1.02-1.12x1014 coverage & < 0.08 ML negative potentials) vertical (high coverage & positive potentials

Adsorption Configuration vertical (v-a30/v-a) vertical (v-a30/v-a) vertical (v-a30/v-a) vertical (v-a30/v-a)

vertical (v-a30/v-a)

vertical (v-a30/v-a)

On Ag(111) Bader et al. using near-edge x-ray-absorption fine-structure measurements (NEXAFS) at 100K purported to observed a phase transition from an inclined phase with a tilt angle of 40˚ to a more inclined phase with a tilt angle of approximately 70˚. This transition they claim occurs around 4.0 L to 5.5 L, which they estimate to still be less than a monolayer [44]. Our results disagree; we always calculate a strictly vertical configuration, with a tilt angle of 90˚, to be the equilibrium adsorption site. A tilt angle of 70˚ does correspond well to the unstable ―a‖ configurations which are close in adsorption energy to v-a30. Moving on, Demuth et al. found, using uv-photoemission spectroscopy (UPS), electron energy loss spectroscopy (EELS), and TDS at 140 K, a coverage driven phase transition from a nearly flat configuration with a tilt

angle of approximately 5˚ to a more inclined configuration with a tilt angle of 55˚ rotated a further 30˚ about the molecular C2ν symmetry axis at an exposure of 0.5 L or an estimated coverage of around 3*1014 molecules/cm2. As we can see Demuth et al. agrees with Bader et al. on the presence of a coverage driven phase transition of pyridine on Ag(111). However both papers disagree about the exposure needed to achieve such a phase transition and the tilt angles involved. Again, our strictly vertical configurations find little agreement with Demuth et al.’s results. Moreover, Demuth et al. is gracious enough to estimate a coverage at which the phase transition occurs, 3*1014 molecules/cm2. Our coverage ranges from 1.03-1.17 *1014 molecules/cm2 depending on what functional and lattice constant we use. According to Demuth et al. we should be observing a flat equilibrium configuration, which we clearly do not. Furthermore, we must note that Otto et al. found, employing UPS, evidence for the phase transition mentioned by Demuth et al. [49]. Finally, we must mention that Drudde et al. claims pyridine adsorbs on Ag(111) in a vertical configuration from exposures ranging from 1.5 L to 12 L. They find no indication of a coverage drive phase transition and observe a √3x√3/30˚ structure using low energy electron diffraction (LEED) [52]. These results agree well with our results, namely that pyridine adsorbs in a vertical configuration. However, their LEED pattern suggests a much higher coverage than what we are modeling. Moreover, given the numerous papers that claim to observe this phase transition on Ag(111) [44,45,49] and the other studies that claim this phase transition exists on other coinage metals such as Cu(111) [47], we cannot rule out the possibility that perhaps Drudde et al. was looking at the wrong coverage. Demuth purported to observe this phase transition at 0.5 L [45] while Drudde et al. only went as low 1.5 L [52]. Moving forward, on Au(111) one study finds, using scanning tunneling microscopy (STM) and surface-enhanced infrared absorption spectroscopy (SEIRAS), pyridine to adsorb flat on Au(111) in solution at low coverage and negative potentials, and vertical on Au(111) in solution at high coverage and positive potentials [48]. Besides that study, we regrettably could only find theoretical studies to compare with. We largely find good agreement with these studies. Tonigold and Groß taking a semiempirical C6R-6 approach implemented in VASP calculates the adsorption energy of pyridine on Au(111) to be 0.28 eV in a 3x3 unit cell in a vertical adsorption site with the molecule’s N atom bonded to an Au atom [58]. We find a maximum adsorption energy of 0.34 eV in our 3x4 unit cell using PBE, and find the same equilibrium adsorption configuration. Furthermore, they calculate the vertical configuration to be 0.22 eV more stable than any flat configuration [58]. We find that in 3x4 unit cell using PBE the most energetically stable vertical configuration to be 0.27 eV more stable than any flat configuration. They also claim there is little difference in adsorption energy, at most 2 meV, among flat configurations [58]. We also calculate using PBE little variation among adsorption energy for the flat configurations, at most 3 meV. On the other hand, Bilić et al. studies pyridine on Au(111) using the PW91 functional. They use a 4 layer slab with a pyridine on each side of the slab. They allow the top and bottom layer to

relax and in between slabs they position 10 Å of vacuum [59]. They compute the most energetically favorable adsorption site to be identical to ours, a vertical configuration in which the molecule’s N atom bonds to an Au atom. They run calculations in both a 3x3 and 2x2 unit cell. Mollenhauer et al. also studies pyridine on Au(111) in a 3x3 unit cell using PBE, PW91, PBEsol, and PBE0. They add the dispersion interaction using the DFT-D3 approach to the PBE and PBE0 functionals. To model their system they utilize a 3 layer slab with 19 Å of vacuum in between slabs, and they hold the Au layers fixed at their experimental bulk lattice parameters. They find, in agreement with our results, the highest binding energy configuration to be a vertical configuration with the molecule’s N atom bonded to an Au atom. Their binding energies, without dispersion corrections, range from 0.21 eV for PBE to 0.46 eV for PBEsol. Adding dispersion corrections increases the binding energy of pyridine on Au(111) from 0.21 eV to 0.77 eV for PBE and from 0.22 eV to 0.85 eV for PBE0 [60]. We observe a similar phenomenon when we include dispersion interactions going from PBE to any of our other vdW inclusive functionals. Next Buimaga-Iarinca et al. using the Spanish Initiative for Electronic Simulations with Thousands of Atoms (SIESTA) code with localized basis sets based on short-ranged pseudoatomic orbitals and several van der Waals inclusive functionals calculates that pyridine adsorbs with a binding energy ranging from 0.28 eV to 0.64 eV in a 5x5 unit cell. They find the most stable flat configuration to be when the center of the pyridine atom sits over a hollow site with the nitrogen atom over a top site. The most stable vertical configuration occurs when the nitrogen atom sits over a hollow site [61]. They compute the flat configurations to possess slightly larger or identical adsorption energy to that of the vertical configurations. These results for the most part disagree with ours. The adsorption energies are in range of our results but we find the vertical configurations with the N atom situated over an Au atom to be the most stable adsorption sites. We never calculate the vertical configurations with the N atom over a hollow site to possess a large adsorption energy. These differences in predicted geometric properties are probably due to the fact that we use different methods. As the authors of study point out, the overlap of basis set for gold and the basis set used for organic molecules can lead to an artificial decrease in total energy [61]. Lastly, Ferrighi et al. using the meta-GGA M06-L and PBE studies the adsorption of pyridine on Au(111). They find the adsorption energies for v-a and v-a30 to be 0.34 eV and 0.33 eV using PBE in exact agreement with our results [62]. They also notice the large gap in adsorption energy between flat and vertical sites using PBE. Using M06-L they calculate the highest binding energy to be v-a with an adsorption energy of 0.49 eV. However in contrast to PBE, the flat configurations possess adsorption energies only at most 4 meV off of the vertical configurations [62]. A decrease in the energy gap between flat and vertical configurations can also be seen in our results when we move up in accuracy from PBE to any of the vdW inclusive functionals.

As just shown, on Cu(111) and Ag(111) a large amount of experimental evidence points to pyridine adsorbing flat on these surfaces at low coverage. If pyridine behaves in a similar fashion on Au(111) we should expect the molecule to adsorb flat on Au(111). Every functional we test though predicts pyridine to adsorb with its molecular plane perpendicular to Cu, Ag, and Au(111). As every functional fails, we attribute this error to the exchange term. The exchange part of these functional may be too repulsive when pyridine sits flat on these surfaces causing an underbinding of the flat configurations. Adding dispersion corrections helps remedy this issue by binding the flat configurations more strongly, but does not completely solve problem. Clearly, we need a better functional to tackle this specific class of problems. 4. Electronic Structure Results To supplement our geometric analysis, we calculate some of the electronic properties, which we list in Table 3, of the molecule/substrate system for the highest adsorption energy configuration for each functional/metal combination. The first property we list in Table 3 is the change transfer from the substrate to the molecule, which we calculate using Bader charge analysis [74-77]. As seen from Table 3, little charge transfer occurs from the substrate to the molecule during the adsorption process except perhaps as calculated using SCAN+rVV10. Charge transfer never exceeds |0.1-e|, except using SCAN+rVV10, and most functional/metal combinations in fact predict close to zero charge transfer. While all the functionals, except SCAN+rVV10, predict no charge transfer or a little charge transfer from the molecule to the substrate, SCAN+rVV10 predicts pyridine to take 0.2 -e from the substrate. This difference in charge transfer is an important distinction between the performance of SCAN+rVV10 and the performance of rest of the functionals.

Table 3: Charge transfer from the substrate to the molecule, change in the width (ΔW d) and center (ΔEd) of the d-band of the atoms that compose the first layer of the surface, and change in the surface’s work function (ΔΦ) upon the adsorption of pyridine on the Cu, Ag, and Au(111) surfaces in the 3x4 unit cell. Calculated clean work function values, Φ clean, are given for reference. Metal

Funtional/Adsorption Site

Ag(111)

PBE/v-a30 optB86b-vdW/v-a30 optB88-vdW/v-a30 optPBE-vdW/v-a revPBE-vdW/v-a rPW86-vdW2/v-a30 SCAN+rVV10/v-a30 PBE/v-a optB86b-vdW/v-a30 optB88-vdW/v-a30 optPBE-vdW/v-a revPBE-vdW/v-a rPW86-vdW2/v-a SCAN+rVV10/v-a30 PBE/v-a optB86b-vdW/v-a30 optB88-vdW/v-a30 optPBE-vdW/v-a30 revPBE-vdW/v-a30 rPW86-vdW2/v-a SCAN+rVV10/v-a30

Au(111)

Cu(111)

Charge transfer (-e) -0.1 -0.1 0 0 0 0 0.2 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 0.2 0 -0.1 0 -0.1 0 0 0.2

ΔWd(eV)

ΔEd(eV)

0.02 0.05 0.06 0.04 0.01 0.03 0.01 0.04 0.06 0.07 0.05 0.02 0.04 0.01 0.02 -0.03 0.02 0.02 0.02 0.02 0.00

-0.04 -0.07 -0.07 -0.05 -0.03 -0.03 -0.03 -0.04 -0.05 -0.06 -0.04 -0.02 -0.03 -0.06 -0.03 0.04 0.11 -0.03 -0.03 -0.03 -0.03

ΔΦ(eV)

Φclean(eV)

-1.02 -1.11 -1.11 -1.02 -0.91 -0.94 -1.14 -1.20 -1.27 -1.26 -1.17 -1.03 -1.07 -1.29 -1.37 -1.44 -1.27 -1.36 -1.28 -1.28 -1.48

4.41 4.66 4.76 4.63 4.54 4.85 4.55 5.15 5.37 5.50 5.37 5.32 5.62 5.43 4.75 4.97 5.03 4.92 4.82 5.04 4.99

For change in the work function, we observe a decrease from the clean substrate work function ranging from 0.91 eV to 1.14 eV for Ag(111), 1.03 eV to 1.29 eV for Au(111), and finally 1.27 eV to 1.48 eV for Cu(111) depending on the functional used. Comparing with the work of Zhong et al for Cu(111), we note good agreement. They observe a clean Cu(111) work function of 4.87 eV which decreases continuously to 2.55 eV at 1 ML [47]. We note a 1.27 eV to 1.48 eV decrease in the surface’s work function at 1/12 ML of pyridine, in the 3x4 unit cell, which is well within the range of the continuous 2.32 eV decrease in the Cu(111) work function witnessed by Zhong et al. going from a clean surface to 1 ML of pyridine [47]. Overall the magnitude of the work function decrease follows the trend Ag < Au < Cu, which is the exact same trend for

the adsorption energy. We also note that SCAN+rVV10 and optB88-vdW consistently predict the largest change in the surface’s work function. Moving on, looking at the change in the width, ΔWd, and center, ΔEd, of the d-band of the first layer of the substrate, which can be a good indicator of the strength of the interaction between the adsorbate and the surface, we see little change in the d-band going from the clean surface to the surface with the adsorbate. Over all of the functional and metal combinations, the maximum |ΔWd| is 0.07 eV and the maximum | ΔEd| is 0.11 eV. This may be expected though as pyridine interacts rather weakly with the (111) coinage metal surfaces and, additionally, only the N atom interacts with the surface in the highest adsorption energy configuration leading to only one atom in the surface interacting strongly with the pyridine molecule, the metal atom below the N atom. Finally, we note that in almost all cases the d-band widens and shifts left away from the Fermi energy except for pyridine on Cu(111) using the optB86b-vdW and optB88-vdW functionals. Using optB86-vdW we calculate ΔWd to be -0.03 eV and ΔEd to be 0.04 for Cu(111) and using optB88-vdW we calculate ΔWd to be 0.02 eV and ΔEd to be 0.11 eV for Cu(111). Also SCAN+rVV10 predicts the width of the d-band of the Cu(100) surface to remain unchanged. The reasons for these abnormalities are unknown. 5. Effects of Coverage As numerous studies claim to observe that pyridine undergoes a coverage driven phase transition on Ag(111) [44,45,49] and Cu(111) [47] from a relatively flat phase to a vertical phase, we run calculations of pyridine on Cu, Ag, and Au(111) in a 4x4, 5x4, and 4x6 unit cell. We calculate the adsorption properties of v-a30, usually the vertical adsorption configuration with the highest adsorption energy in the 3x4 unit cell, and bB1, usually the flat adsorption configuration with the highest adsorption energy in the 3x4 unit cell. We additionally only run calculations using SCAN+rVV10, due its promising position as a meta-GGA functional, and optB88-vdW to have one functional to compare with. Fig. 5 plots the adsorption energy difference, ΔE, between v-a30 and bB1 versus the number of atoms in the substrate or coverage 1 . As one can see from Fig. 5 decreasing the coverage does not appear to change the energy gap, which is around 0.05 eV, between v-a30 and bB1 for pyridine on Ag(111). For pyridine on Au(111) decreasing the coverage seems to slightly increase the energy gap between v-a30 and bB1, and for pyridine on Cu(111) decreasing the coverage seems to slightly decrease the energy gap between v-a30 and bB1. Out of all three surfaces, only on Cu(111) does decreasing the coverage of pyridine lead to a flat configuration being more preferred, albeit very slightly more preferred. Nonetheless, it appears that even on Cu(111) a flat configurations may never be outright preferred over a vertical configuration. Even more consequential though, if we assume that pyridine should adsorb flat at these coverages then optB88-vdW outperforms SCAN+rVV10 on Au(111) and Cu(111) with optB88-vdW consistently predicting bB1 to be 0.03 eV to 0.05eV closer in adsorption energy to v-a30 than SCAN+rVV10.

Figure 5. Difference in adsorption energy, ΔE, between v-a30 and bB1 for pyridine adsorbed on Cu, Ag, and Au(111) as a function of the number of atoms in the substrate. Lines are provided to guide one’s eyes.

If we look more broadly at the effects of coverage we note several additional interesting results. Fig. 6 a) plots the adsorption energy versus the number of atoms in the substrate. See Table S2 in the supporting information for the listed values. From Fig. 6 a) we note that generally decreasing the coverage leads to a larger adsorption energy. The one possible exception to this rule is pyridine on Au(111) using SCAN+rVV10 where the adsorption energy first decreases, in the 4x4 unit cell, before increasing. Overall our results suggest that the molecule-molecule interactions are repulsive. Decreasing the coverage leads to weaker molecule-molecule interactions and higher adsorption energies. Despite adsorption energy increasing as one decreases the coverage of pyridine on these surfaces, N-Metal distances remain relatively constant, changing at most 0.03 Å from one unit cell to another. See Fig. 6 b) for a plot of N-Metal distances as a function of the number of atoms in the substrate. See Table S2 in the supporting information for their listed values.

Figure 6. a) Adsorption energy versus the number of atoms in the substrate and b) N-Metal distances versus the number of atoms in the substrate for pyridine adsorbed on Cu, Ag, and Au (111). Lines are provided to guide one’s eyes. Overall we demonstrate that the pyridine-pyridine interactions are repulsive. More importantly, we illustrate that for all the coverages studied here pyridine prefers to adsorb in a vertical configurations on the (111) coinage metal surfaces. In addition, pyridine shows no signs that it will eventually prefer a flat adsorption at even lower coverages with the energy difference between flat and vertical configurations changing only small amounts as we decrease coverage. This contradicts the available experimental studies which suggest that pyridine should adsorb with its molecular plane parallel, not perpendicular, to the (111) coinage metal surfaces at low coverage. This is reminiscent of the so-called CO adsorption puzzle [78], where a number of different DFT functionals predict the wrong adsorption site for CO on a number of different transition metal surfaces. Pyridine, like CO, looks to be a difficult problem for current DFT methods. Evidently we need better methods to handle the adsorption of pyridine on the (111) coinage metals surfaces. 6. Conclusions Throughout our detailed DFT study of the adsorption of pyridine on (111) coinage metal surfaces we fail to observe any strong coverage driven phase transition. In all the explored unit cells, we unanimously observe the most favorable adsorption site to be a vertical configuration with the pyridine’s N atom located over a metal atom. Introducing the vdW interactions does not change the most favorable adsorption site, but rather enhances pyridine’s adsorption energy. Finally, decreasing the pyridine coverage increases the adsorption energy demonstrating the pyridine-pyridine interactions are repulsive. Despite failing to predict the coverage driven phase transition, out results reveal interesting properties of the electronic structure of the molecule/substrate system. We observe a healthy change in the substrate’s work function upon the adsorption of pyridine. We also observe a

small change in the d-band of the first layer substrate atoms upon the adsorption of pyridine. Moreover we calculate a generally low charge transfer from the molecule to the substrate except for SCAN+rVV10. The SCAN+rVV10 functional, in contrast, predicts the molecule to take 0.2 –e from the surface. This large, by comparison, charge transfer in the opposite direction of the other functionals appears to be an important defining characteristic of SCAN+rVV10. Besides this large charge transfer, SCAN+rVV10 and the other vdW inclusive functionals perform similarly. Given SCAN+rVV10’s large computational cost, due to it being a meta-GGA functional, and the similarly of its results with the other vdW inclusive functionals we recommend using the less expensive optB86b-vdW or optB88-vdW functional. Although due to all the functionals predicting the incorrect adsorption site we must again call for the development of more accurate methods for this class of problems. Taken all together, we can peg the adsorption of pyridine on these coinage metals surfaces to range from weak to strong physisorption as evident by low adsorption energies and small changes in the geometric properties of both the substrate and molecule upon adsorption. The key to going from weak to strong physisorption appears to be the vdW interaction as evident by the vdW interaction enhancing the adsorption energy by up to 540 meV. The inclusion of the vdW interaction, in addition to greatly enhancing the adsorption energy, also leads to final adsorption sites where the molecule can be tilted off the surface normal without greatly changing adsorption energies. Furthermore, the vdW interaction in general lowers the gap between flat and vertical configurations. It may be that the vdW interaction plays some role in the coverage driven phase transitions mentioned in the literature. CONFLICT OF INTEREST There is no conflict of interest related to the present publication.

ACKNOWLEDGEMENTS We acknowledge support from the U.S. Department of Energy Basic Energy Science under Contract No DE-FG02-11ER16243. 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.

References [1] H. E. Katz, J. Huang, Thin-Film Organic Electronic Devices, Annu. Rev. Mater. Res. 39 (2009) 71–92. [2] J. R. Heath, M. A. Ratner, Molecular Electronics, Phys. Today 56 (2003) 43–49. [3] G. A. Somorjai, Introduction to Surface Chemistry and Catalysis, Wiley, New York, 1994.

[4]J. Klimeš, D. R. Bowler, A. Michaelides, Van der Waals Density Functionals Applied to Solids, Phys. Rev. B 83 (2011) 195131 . [5] J. Carrasco, J. Klimeš, A. Michaelides, The Role of Van Der Waals Forces in Water Adsorption on Metals, J. Chem. Phys. 138 (2013) 024708. [6] J. Carrasco, B. Santra, J. Klimeš, A. Michaelides, To Wet or Not to Wet? Dispersion Forces Tip the Balance for Water Ice on Metals, Phys. Rev. Lett. 106 (2011) 026101. [7] J. Klimeš, A. Michaelides, Perspective: Advances and Challenges in Treating Van Der Waals Dispersion Forces in Density Functional Theory, J. Chem. Phys. 137 (2012) 120901. [8] W. Liu, J. Carrasco, B. Santra, A. Michaelides, M. Scheffler, A. Tkatchenko, Benzene Adsorbed on Metals: Concerted Effect of Covalency and Van Der Waals Bonding, Phys. Rev. B 86 (2012) 245405. [9] A. Tkatchenko, L. Romaner, O. T. Hofmann, E. Zojer, C. Ambrosch-Draxl, M. Scheffler, Van der Waals Interactions Between Organic Adsorbates and at Organic/Inorganic Interfaces, MRS Bull. 35 (2010) 435–442. [10] H. Yildirim, T. Greber, A. Kara, Trends in Adsorption Characteristics of Benzene on Transition Metal Surfaces: Role of Surface Chemistry and van der Waals Interactions, J. Phys. Chem. C 117 (2013) 20572–20583. [11] J. Matos, H. Yildirim, A. Kara, Insight into the Effect of Long Range Interactions for the Adsorption of Benzene on Transition Metal (110) Surfaces, J. Phys. Chem. C 119 (2015) 1886– 1897. [12] H. Yildirim, A. Kara, 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 117 (2013) 2893–2902. [13] J. Matos, T. Rojas, H. Yildirim, A. Kara, On the Role of Long Range Interactions for the Adsorption of Sexithiophene on Ag(110) Surface, J. Chem. Phys. 140 (2014) 144703. [14] H. Yildirim, J. Matos, A. Kara, Role of Long-Range Interactions for the Structure and Energetics of Olympicene Radical Adsorbed on Au(111) and Pt(111) Surfaces, J. Phys. Chem. C 119 (2015) 25408–25419. [15] S. Gottardi, K. Müller, J. C. Moreno-López, H. Yildirim, U. Meinhardt, M. Kivala, A. Kara, M. Stöhr, Cyano-Functionalized Triarylamines on Au(111): Competing Intermolecular versus Molecule/Substrate Interactions, Adv. Mater. Interfaces 1 (2013) 1300025.

[16] K. Müller, J. C. Moreno-López, S. Gottardi, U. Meinhardt, H. Yildirim, A. Kara, M. Kivala, M. Stöhr, Cyano-Functionalized Triarylamines on Coinage Metal Surfaces: Interplay of Intermolecular and Molecule-Substrate Interactions, Chem. Eur. J. 22 (2015) 581–589. [17] W. Malone, H. Yildirim, J. Matos, A. Kara, Van der Waals Inclusive Density Functional Theory Study of the Nature of Bonding for Thiophene Adsorption on Ni(100) and Cu(100) Surfaces, J. Phys. Chem. C 121 (2017) 6090–6103. [18] R. M. Dreizler, E. K. U. Gross, Density Functional Theory. An Approach to the Quantum Many-Body Problem, Springer-Verlag, Berlin, 1990. [19] A. D. Becke, Density-Functional Exchange-Energy Approximation with Correct Asymptotic Behavior, Phys. Rev. A 38 (1988) 3098–3100. [20] D. C. Langreth, M. J. Mehl, Beyond the Local-Density Approximation in Calculations of Ground-State Electronic Properties, Phys. Rev. B 28 (1983) 1809–1834. [21] J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J. Singh, C. Fiolhais, Atoms, Molecules, Solids, and Surfaces: Applications of the Generalized Gradient Approximation for Exchange and Correlation, Phys. Rev. B 46 (1992) 6671–6687. [22] S. Grimme, Semiempirical GGA-Type Density Functional Constructed with a Long-Range Dispersion Correction, J. Comput. Chem. 27 (2006) 1787–1799. [23] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, A Consistent and Accurate Ab Initio Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu, J. Chem. Phys. 132 (2010) 154104. [24]A. Tkatchenko, M. Scheffler, Accurate Molecular Van Der Waals Interactions from GroundState Electron Density and Free-Atom Reference Data, Phys. Rev. Lett. 102 (2009) 073005. [25] A. D. Becke, E. R. Johnson, A density-functional model of the dispersion interaction, J. Chem. Phys. 123 (2005) 154101. [26] J. Klimeš, D. R. Bowler, A. Michaelides, Chemical Accuracy for the van der Waals Density Functional, J. Phys. Condens. Matter 22 (2009) 022201. [27] K. Lee, É. D. Murray, L. Kong, B. I. Lundqvist, D. C. Langreth, Higher-Accuracy van der Waals Density Functional, Phys. Rev. B 82 (2010) 081101. [28] T. Thonhauser, V. R. Cooper, S. Li, A. Puzder, P. Hyldgaard, D. C. Langreth, Van der Waals Density Functional: Self-Consistent Potential and the Nature of the van der Waals Bond, Phys. Rev. B 76 (2007) 125112.

[29]M. Dion, H. Rydberg, E. Schröder, D. C. Langreth, B. I. Lundqvist, Van der Waals Density Functional for General Geometries, Phys. Rev. Lett. 92 (2004) 246401. [30] J. Wellendorf, K. T. Lundgaard, A. Mogelhoj, V. Petzold, D. Landis, J. K. Norskov, T. Bligaard, K. W. Jacobsen, Density functionals for surface science: Exchange-correlation model development with Bayesian error estimation, Phys. Rev. B 85 (2012) 235149. [31] E. D. Murray, K. Lee, D. C. Langreth, Investigation of Exchange Energy Density Functional Accuracy for Interacting Molecules, J. Chem. Theory Comput. 5 (2009) 2754–2762. [32] J. Sun, A. Ruzsinszky, J. P. Perdew, Strongly Constrained and Appropriately Normed Semilocal Density Functional, Phys. Rev. Lett. 115 (2015) 036402. [33] J. Tao, J. P. Perdew, V. N. Staroverov, G. E. Scuseria, Climbing the Density Functional Ladder: Nonempirical Meta–Generalized Gradient Approximation Designed for Molecules and Solids, Phys. Rev. Lett. 9 (2003) 146401. [34] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, L. A. Constantin, J. Sun, Workhorse Semilocal Density Functional for Condensed Matter Physics and Quantum Chemistry, Phys. Rev. Lett. 103 (2009) 026403. [35] J. Sun, B. Xiao, A. Ruzsinszky, Communication: Effect of the Orbital-Overlap Dependence in the Meta Generalized Gradient Approximation, J. Chem. Phys. 137 (2012) 051101 (2012). [36] Y. Zhao, D. G. Truhlar, A New Local Density Functional for Main-Group Thermochemistry, Transition Metal Bonding, Thermochemical Kinetics, and Noncovalent Interactions, J. Chem. Phys. 125 (2006) 194101. [37] J. M. del Campo, J. L. Gázquez, S. B. Trickey, A. Vela, A New Meta-GGA Exchange Functional Based on an Improved Constraint-Based GGA, Chem. Phys. Lett. 543 (2012) 179. [38] R. Sabatini, T. Gorni, S. D. Gironcoli, Nonlocal van der Waals density functional made simple and efficient, Phys. Rev. B 87 (2013) 041108®. [39] H. Peng, Z.-H.Yang, J. P. Perdew, J. Sun, Versatile Van Der Waals Density Functional Based on a Meta-Generalized Gradient Approximation, Phys. Rev. X 6 (2016) 041005. [40] D. Langreth, J. Perdew, The exchange-Correlation energy of a metallic surface, Solid State Commun. 17 (1975) 1425–1429. [41] O. Gunnarsson, B. I. Lundqvist, Exchange and correlation in atoms, molecules, and solids by the spin-Density-Functional formalism, Phys. Rev. B 13 (1976) 4274–4298. [42] D. C. Langreth, J. P. Perdew, Exchange-Correlation energy of a metallic surface: WaveVector analysis, Phys. Rev. B 15 (1977) 2884–2901.

[43] C.-Q. Ma, E. Mena-Osteritz, T. Debaerdemaeker, M. M. Wienk, R. A. J. Janssen, P. Bäuerle, Functionalized 3D Oligothiophene Dendrons and Dendrimers— Novel Macromolecules for Organic Electronics. Angew. Chem. Int. Ed. 46 (2007) 1679–1683. [44] M. Bader, J. Haase, K.-H. Frank, A. Puschmann, A. Otto, Orientational Phase Transition in the System Pyridine/Ag(111): A Near-Edge X-Ray-Absorption Fine-Structure Study, Phys. Rev. Lett. 56 (1986) 1921–1924. [45] J. Demuth, K. Christmann, P. Sanda, The vibrations and structure of pyridine chemisorbed on Ag(111): the occurrence of a compressional phase transformation, Chem. Phys. Lett. 76 (1980) 201–206. [46] P. Avouris, J. E. Demuth, Electronic excitations of benzene, pyridine, and pyrazine adsorbed on Ag(111), J. Chem. Phys. 75 (1981) 4783–4794. [47] Q. Zhong, C. Gahl, M. Wolf, Two-Photon photoemission spectroscopy of pyridine adsorbed on Cu(111), Surf. Sci. 496 (2002) 21–32. [48] W.-B. Cai, L.-J. Wan, H. Noda, Y. Hibino, K. Ataka, M. Osawa, Orientational Phase Transition in a Pyridine Adlayer on Gold(111) in Aqueous Solution Studied by in situ Infrared Spectroscopy and Scanning Tunneling Microscopy, Langmuir 14 (1998) 6992–6998. [49] A. Otto, K. Frank, B. Reihl, Inverse photoemission of pyridine on silver (111), Surf. Sci. 163 (1985) 140–152. [50] P. N. Sanda, J. M. Warlaumont, J. E. Demuth, J. C. Tsang, K. Christmann, J. A. Bradley, Surface-Enchanced Raman Scattering from Pyridine on Ag(111), Phys. Rev. Lett. 45 (1980) 1519–1523. [51] J. E. Demuth, P. N. Sanda, Observation of Charge-Transfer States for Pyridine Chemisorbed on Ag(111), Phys. Rev. Lett. 47 (1981) 57–60. [52] R. Dudde, E. Koch, N. Ueno, R. Engelhardt, Comparative study of angle resolved photoemission spectra from pyridine adsorbed on Ag(111) and on Ag polycrystalline substrates, Surf. Sci. 178 (1986) 646–656. [53] A. Campion, Surface Raman spectroscopy without enhancement: Pyridine on Ag(111), J. Electron Spectrosc. Relat. Phenom, 29 (1983) 397–400. [54] J. E. Rowe, C. V. Shank, D. A. Zwemer, C. A. Murray, Ultrahigh-Vacuum Studies of Enhanced Raman Scattering from Pyridine on Ag Surfaces, Phys. Rev. Lett. 44 (1980) 1770– 1773. [55] P. Avouris, J. E. Demuth, Vibrational overtone spectroscopy of benzene and pyridine on Ag(111), J. Chem. Phys. 75 (1981) 5953–5954.

[56] G. Zylka, A. Otto, Search for dynamical charge transfer excitations of CO and pyridine on copper by EELS, Surf. Sci. 475 (2001) 118–130. [57] P. Davies, N. Shukla, The adsorption of pyridine at clean, oxidised and hydroxylated Cu(111) surfaces, Surf. Sci. 322 (1995) 8–20. [58] K. Tonigold, A. Groß, 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. 132 (2010) 224701. [59] A. Bilić, J. R. Reimers, N. S. Hush, Adsorption of Pyridine on the Gold(111) Surface: Implications for ―Alligator Clips‖ for Molecular Wires, J. Phys. Chem. B 106 (2002) 6740–6747. [60] D. Mollenhauer, N. Gaston, E. Voloshina, B. Paulus, Interaction of Pyridine Derivatives with a Gold (111) Surface as a Model for Adsorption to Large Nanoparticles, J. Phys. Chem. C 117 (2013) 4470–4479. [61] L. Buimaga-Iarinca, C. Morari, Adsorption of small aromatic molecules on gold: a DFT localized basis set study including van der Waals effects, Theor. Chem. Acc. 133 (2014) 1502. [62] L. Ferrighi, G. K. H. Madsen, B. Hammer, Self-Consistent meta-Generalized gradient approximation study of adsorption of aromatic molecules on noble metal surfaces, J. Chem. Phys. 135 (2011) 084704. [63] G. Kresse, J. Furthm ller, Efficient Iterative Schemes for ab initio Total-Energy Calculations Using a Plane-Wave Basis Set, Phys. Rev. B, 54 (1996) 11169−11186. [64] G. Kresse, J. Furthm ller, Efficiency of ab-initio Total Energy Calculations for Metals and Semiconductors Using a Plane-Wave Basis Set, Comput. Mater. Sci. 6 (1996) 15−50. [65] G. Kresse, J. Hafner, Ab initio Molecular Dynamics for Liquid Metals, Phys. Rev. B, 47 (1993) 558−561. [66] . E. Bl chl, Projector Augmented-Wave Method, Phys. Rev. B, 50 (1994) 17953−17979. [67] G. Kresse, D. Joubert, From Ultrasoft Pseudopotentials to the Projector Augmented-Wave Method, Phys. Rev. B, 59 (1999) 1758−1775. [68] J. P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made Simple, Phys. Rev. Lett. 77 (1996) 3865−3868. [69] M. R. Hestenes, E. Stiefel, Methods of Conjugate Gradients for Solving Linear Systems, J. Res. Natl. Bur. Stand, 49 (1952) 409−436.

[70] M. . Teter, M. C. ayne, D. C. Allan, Solution of Schr dinger’s Equation for Large Systems, Phys. Rev. B, 40 (1989) 12255−12263. [71] W. Malone, J. von der Heyde, A. Kara, Competing adsorption mechanisms of pyridine on Cu, Ag, Au, and Pt(110) surfaces, J. Chem Phys. 149 (2018) 214703 [72] W. Malone, J. Matos, A. Kara, Adsorption of thiophene on transition metal surfaces with the inclusion of van der Waals effects, Surf. Sci. 669 (2018) 121. [73] S. Gautier, S. N. Steinmann, C. Michel, P. Fleurat-Lessard, P. Sautet, Molecular adsorption at Pt(111). How accurate are DFTfunctionals?, Phys. Chem. Chem. Phys. 17 (2015) 2892128930. [74] W. Tang, E. Sanville, G. Henkelman, A Grid-Based Bader Analysis Algorithm without Lattice Bias, J. Phys.: Condens. Matter, 21 (2009) 084204. [75] E. Sanville, S. D. Kenny, R. Smith, G. Henkelman, Improved Grid-Based Algorithm for Bader Charge Allocation, J. Comput. Chem. 28 (2007) 899−908. [76] G. Henkelman, A. Arnaldsson, H. A. Jonsson, Fast and Robust Algorithm for Bader Decomposition of Charge Density, Comput. Mater. Sci. 36 (2006) 354−360. [77] M. Yu, D. R. Trinkle, Accurate and Efficient Algorithm for Bader Charge Integration, J. Chem. Phys. 134 (2011) 064111. [78] . Lazić, M. Alaei, N. Atodiresei, V. Caciuc, R. Brako, S. Bl gel, Density functional theory with nonlocal correlation: A key to the solution of the CO adsorption puzzle, Phys. Rev. B 81 (2010) 045401.

GRAPHICAL ABSTRACT