Adsorption of Potassium on MoS2(100) Surface: First-Principles

ARTICLE pubs.acs.org/JPCC

Adsorption of Potassium on MoS2(100) Surface: A First-Principles Investigation Amity Andersen,*,† Shawn M. Kathmann,‡ Michael A. Lilga,§ Karl O. Albrecht,§ Richard T. Hallen,§ and Donghai Mei*,‡ †

Environmental Molecular Sciences Laboratory, ‡Fundamental and Computational Sciences Directorate, and §Energy and Environment Directorate, Pacific Northwest National Laboratory, Richland, Washington 99352, United States

bS Supporting Information ABSTRACT: Potassium- (K-) promoted MoS2 catalyst is a very promising catalyst used for synthesis of mixed higher alcohols from syngas. Herein, periodic density functional theory calculations were performed to investigate the interaction of potassium with the Mo and S edges of the MoS2(100) surface. Both neutral K- and charged “Kþ”-promoted MoS2(100) systems at different sulfur coverages were studied. Our calculations indicate that the adsorbed K atom readily donates its single 4s valence electron to the MoS2 structure, and the neutral K and charged Kþ systems show similar adsorption behavior. Isolated K atom/ion tends to maximize its interactions with the available S atoms on the edge surface, preferring the 4-fold S hollow site on fully sulfided Mo and S edges and the interstitial sites where K atom/ion binds with 24 S atoms on the edge surface. The presence of K atoms/ions affects the electronic and magnetic properties of the edge surface. As the K coverage increases, the average adsorption energy of the K atoms, surface work function, and amount of 4s electron transfer from the K atoms to the MoS2(100) surface all decrease, suggesting an increased metallization of the K adlayer. The tendency to form a chainlike K adlayer along the interstitial gap area of two edges is found. The KK distances in the K chains of the adlayer are 3.23.7 Å, which is notably less than that of bulk K metal. Density of states analysis for the K-saturated MoS2(100) surface suggests enhanced involvement of broad K 3d states beginning just above the Fermi level. The K promotional effects on the selectivity of mixed alcohol synthesis from CO hydrogenation can be rationalized as an increase in the surface basicity due to the increasing surface electron charge donated by K doping. The adsorbed K atoms/ions also provide active sites that facilitate CO hydrogenation, block Mo and S edge sites for CO dissociation leading to hydrocarbon formation, and limit H2 dissociative adsorption at the edge surface via the s-type electron repulsion from the transferred K 4s electron.

1. INTRODUCTION MoS2 is commonly used in the industrial hydrodesulfurization (HDS) of sulfur-contaminant compounds (e.g., thiophene) in crude oil.1 With the growing concerns of climate change and foreign energy dependence, alkali-promoted MoS2 has also been found as one of the promising catalysts for the production of higher (C2þ) mixed alcohols from renewable, biomass-derived syngas.2,3 MoS2 itself produces only hydrocarbons (primarily methane). With the addition of basic alkali metal promoters, however, the MoS2-based catalysts become more selective for the formation of linear-chained alcohols such as ethanol.3 Early independent studies by the Dow Chemical Company4,5 and the Union Carbide Corporation6 suggested that both supported and unsupported MoS2 catalysts are active for alcohol synthesis from syngas with an alcohol selectivity range of 7590%. Compared with other alcohol synthesis catalysts, the alkalipromoted MoS2 catalysts are sulfur-resistant, slow to coking deactivation, and have less CO2 sensitivity.2,7 Little is known about the catalytic role that alkali promoters play in alcohol selectivity and where these promoters are in the r 2011 American Chemical Society

MoS2-based catalysts. Alkali-promoted Mo-based catalysts (MoS2based or Mo/MoOx/MoC-based varieties treated with an H2S atmosphere) for alcohol synthesis have been characterized with different microscopic, spectroscopic, and diffraction characterization techniques, each providing limited insight into the complex reaction mechanism.6,818 The characterization results suggested that the addition of basic alkali promoters decreases the surface acidity of the catalyst, thus suppressing various side reactions such as isomerization, dehydrogenation, and coking, as well as reducing the active sites for CO dissociation that are responsible for hydrocarbon formation.11,14,15,18 The observed general trend for the alkali metal promoters in terms of increased higher alcohol production for MoS2-based catalysts is Li < Na < K < Cs < Rb, which is in line with the increasing basicity.3 Theoretical studies can provide atomistic and electronic structure details that are difficult to obtain experimentally. Until Received: October 20, 2010 Revised: March 7, 2011 Published: April 15, 2011 9025

dx.doi.org/10.1021/jp110069r | J. Phys. Chem. C 2011, 115, 9025–9040

The Journal of Physical Chemistry C now, most of the theoretical investigations of alkali-promoted MoS2 have been limited to Li-promoted MoS2 surface and Liintercalated interactions, which focused on lithium battery technology applications, not catalytic alcohol synthesis.19,20 Unlike heavy alkali metals (K, Rb, and Cs), light alkali metals (Li and Na) are known to profoundly disrupt the 2H phase structure of MoS2;21,22 thus they are not suitable for use as promoters in MoS2 for higher alcohol synthesis.23 Potassium is typically added to supported or unsupported MoS2 catalyst in salt form (e.g., K2CO3, KOH, K2S, K2SO4, KCl), either through aqueous solution11 or mechanically ground into the MoS2.9 For positive pKa salts such as KOH, K2S, and K2CO3, the anion is easily separated from the cationic K over MoS2.11 Energy-dispersive X-ray spectroscopy (EDS) measurements have shown that Kþ spreads uniformly over the catalyst after an induction period under syngas/H2S reactor conditions. For large negative pKa salts such as KCl and K2SO4, the anion is difficult to separate from the Kþ cation. In the latter case, the potassium salt particles largely stay intact and nonuniformly spread over the catalyst. Lee et al.11 suggested that uniform Kþ distribution over the catalyst is correlated with high selectivity toward alcohol synthesis. They argued that the subsurface K atoms/ions would migrate into the MoS2 bulk, resulting in a selectivity shift from alcohols to C2þ hydrocarbons as a consequence of electronic changes to the MoS2.11 A similar observation of K- and Cs-promoted Co-MoS2/clay catalysts was also reported by Iranmahboob et al.9,10,24 using a combination of X-ray photospectroscopy (XPS), scanning electron microscopy (SEM), and EDS measurements. In addition to the above experimental studies of unsupported and supported K-promoted MoS2, alkali-promoted catalyst materials such as oxide-supported Mo-, MoOx-,and Mo/C-based catalysts, which form MoSx phases including 2H MoS2 phase under sulfur-containing syngas/H2S, also provide insight into reactivity.1 Due to the transitional nature of these Mo-based catalysts (i.e., formation of intermediate domains of MoS2), a variety of chemical species with oxidation states and magnetic property signatures may be present and in higher concentrations, compared to catalyst materials based on MoS2 alone. For example, electron spin resonance (ESR) and XPS studies of MoO3/K2CO3/SiO2 and K2MoO3/SiO2 under syngas and H2S atmosphere for the production of methanethiol (a potential competing product to alcohols) show diverse chemical surface species. Yang et al.25,26 observed “oxo-Mo(V)”, “thio-Mo(V)”, and S species signatures in their ESR spectra and S2, [SS]2, S6þ, Mo4þ, Mo5þ, and Mo6þ species signatures in their XPS spectra. XPS results show that K2CO3 promoter enhances the sulfiding and reduction of high oxidation state Mo6þ species to MoS2 and “oxo-Mo(V)” species. The formation of disulfide [SS]2 and S2 species is enhanced as well. Chen et al.8 suggested that the promotion of the water-gas-shift reaction by use of the K-promoted Mo-based catalysts can be attributed in to a high Mo5þ/Mo4þ ratio. On the basis of scanning tunneling microscope (STM) experiments and theoretical density functional theory (DFT) calculations of thiophene adsorption on the single-layered, goldsupported MoS2 catalyst, the MoS2(100) edge surface sites are the most catalytically active sites for HDS.27 K can decorate the edge surfaces and intercalate the interstitial space of MoS2 under the operating conditions.23,28 Although the intercalation may have an electronic effect on the catalytic properties of MoS2, Fourier transform infrared spectroscopy (FTIR) studies

ARTICLE

indicated that surface K species provide additional active sites for CO adsorption and hydrogenation.11 Moreover, the most selective catalyst is formed after an induction period where the CO32 FTIR signature disappears.11,14,15 In the present work, we focus on the adsorption of neutral K and Kþ on the MoS2(100) edge surface and in the interstitial gap between two edge surfaces. To the best of our knowledge, the effects of the adsorbed K atoms/ions on the structural, electronic, and magnetic features of the MoS2(100) surface are addressed by DFT calculations for the first time. The K-promoting role in improving alcohol selectivity is briefly discussed.

2. THEORETICAL APPROACH 2.1. DFT Calculations. Spin-polarized periodic DFT calculations were performed with the CP2K code.29 These calculations employ a mixed Gaussian and planewave basis set. Core electrons were represented with norm-conserving GoedeckerTeter Hutter pseudopotentials,3032 and the valence electron wave function was expanded in a double-ζ basis set with polarization functions33 along with an auxiliary plane-wave basis set used to represent the density. Semicore pseudopotentials were used for Mo (4s2 4p6 5s1 4d5) and K (3s2 3p6 4s1), and a valence-only pseudopotential was used for S (3s2 3p4). The generalized gradient approximation exchangecorrelation functional of Perdew, Burke, and Enzerhof (PBE)34 was used for all calculations. The energy cutoff for the plane-wave basis set was set at 300 Ry. Because of the relatively large size of the supercell models studied here, the Brillouin zone was sampled at the Γ-point. Bader analysis using the code developed by Henkelman and co-workers3537 and the projected density of states (PDOS) generated from a projection of the total density of states onto spherical harmonics were used in the electronic structure analysis. Work function calculations were also performed by subtraction of the vacuum electrostatic potential (halfway between the slab surface and its image slab surface) and the Fermi energy, that is, Φ = Vvacuum εFermi. 2.2. Construction of MoS2(100) Edge Surfaces with Different Surface Sulfur Coverages. The bulk MoS2 structure was constructed by use of crystallographic data of the hexagonal 2H polymorph structure of MoS2 (space group P63/mmc) in the literature.22,21 The lattice parameters are a = b = 3.160 Å, c = 12.294 Å, R = β = 90°, and γ = 120°. The bulk structure was then cleaved along the (100) plane, generating a Mo-terminated (1010) edge and an S-terminated (1010) edge. This edge surface was expanded to four units along the x direction and to four layers in the z direction. A vacuum layer of 20 Å was added above the cleaved surface slab to give a volume of 12.640 12.294 29.122 Å3, in a similar fashion to previous DFT studies that used a localized basis-set code.3840 Except for the atoms in the bottom layer, all the atoms in the model edge surface slab were allowed to relax. The resulting “as-cleaved” structure, referred to as Mo0S100 in this work, is shown in Figure 1a. By moving half of the surface S atoms from the S-terminated edge to the Mo-terminated edge on the Mo0S100 surface, that is, a 50%50% concentration of S at each of the edges, the Mo50S50 surface was created (Figure 1b). Previous DFT calculations indicated that the Mo50S50 surface is the most energetically stable under HDS working conditions.41 The third model surface considered in this work is a fully sulfur-covered edge surface (referred to as Mo100S100 surface) by adding a full coverage of sulfur at the bare Mo edge of the “as-cleaved” Mo0S100 system 9026

dx.doi.org/10.1021/jp110069r |J. Phys. Chem. C 2011, 115, 9025–9040

The Journal of Physical Chemistry C

ARTICLE

catalytic properties at the edge surface due to changes in electronic structure. In the latter case, it is unclear whether K can be considered neutral or cationic. It is reasonable to argue that electron donation from K to MoS2 substrate occurs indirectly through the anion of certain potassium salts. FTIR experiments suggest that CO32 is absent after an induction period when the alcohol selectivity is highest.14,15 Thus, we assume a model system representing the catalyst surface with only K (neutral and charged systems) on the four edge model surfaces. For the charged “Kþ” system, a uniform þ1 charge was applied to the system (i.e., “hole-doping”). As we will discuss later on, the edge surface of MoS2(100) possesses magnetic properties, which can be affected by the presence of alkali metal atoms/ions.4244 For the charged K-doped surface systems, only the adsorption of a single K atom/ion was considered. With increasing K atom coverages with two or more K atoms placed on the 4 1 model surfaces, only neutral systems were considered. Various sites for possible K/Kþ adsorption on four edge MoS2(100) model surfaces were shown in Figure 1. K/Kþ in the interstitial sites (1 and 2) on the surface and subsurface sites (3 and 4) were also explored in this work (Figure 1e). Each adsorption configuration was optimized with the Broyden FletcherGoldfarbShanno (BFGS) algorithm to reach the local energy minima. The convergence criteria for the rootmean-square (rms) and maximum changes in geometry between the previous and current structures were set to 1.5 103 and 3.0 103 bohr, respectively, and the rms and maximum force component convergence criteria were set to 3 104 and 4.5 104 Hartree/bohr, respectively. The adsorption energies of K adsorbates on the four model surfaces studied here were computed as follows: Ead ðKÞ ¼ Etot ðK þ surf Þ ½Etot ðKÞ þ Etot ðsurf Þ

Figure 1. Optimized MoS2(100) surfaces with different S coverages: (a) “as-cleaved” Mo0S100, (b) Mo50S50, (c) Mo100S100, and (d) Mo37.5vS50. (e) Basal view of the MoS2(100) surface with the interstitial sites. Side and top views are shown on the left and right, respectively. Possible adsorption sites are numerically labeled. Light blue-green spheres represent molybdenum atoms; yellow spheres represent S atoms.

(Figure 1c). The Mo100S100 surface mimics the initial state of a freshly synthesized MoS2 catalyst. Finally, we also considered a model surface with a sulfur “vacancy” at the Mo edge (Figure 1d). By eliminating one surface S atom at the Mo edge of the Mo50S50 surface, the fourth model with 37.5% sulfur coverage at the Mo edge and 50% sulfur coverage at the S edge was constructed (referred to as Mo37.5vS50 surface). 2.3. Placement of K/Kþ on Edge Surface Models. As we mentioned in the Introduction, K-promoted MoS2-based catalysts have been experimentally prepared with either neutral metal K (vapor phase or ammonia-dissolved) or potassium salts. In the former case, neutral K atoms are introduced to study the intercalation without any discussion of the promoting effect, although this basic research does have indirect implications for its

ð1Þ

where Etot(K þ surf) is the total energy for the adsorbed K adsorbates on the surface, Etot(K) is the energy for the isolated K atom in vacuum, and Etot(surf) is the energy of the clean edge surface. Based on this definition, a negative value of Ead(K) indicates stable adsorption. For the charged systems where a uniform charge for the charged Kþ system was applied, the [Etot(K) þ Etot(surf)] is replaced by the total energy of the surface slab with the K atom placed at 10 Å above the surface. For multiple K atoms adsorbed on the edge surfaces, the averaged adsorption energy per K atom was calculated as follows: Ead ðKÞ ¼ fEtot ðNK K þ surf Þ ½NK Etot ðKÞ þ Etot ðsurf Þg=NK ð2Þ where NK is the number of K atoms adsorbed on the surface. The adsorption energies of a few configurations were checked against a model surface where the bulk lattice was fully optimized with a plane-wave basis-set pseudopotential code (details to be included in another publication) and were found to be within the error of DFT (GGA-PBE). We note that the DFT does not model the van der Waals interaction between the MoS2 sheets very accurately, and indeed, full lattice expansion with GGA-PBE is 4% wider than experiment in the z direction.

3. RESULTS AND DISCUSSION 3.1. K/Kþ Adsorption on MoS2(100) Edge Surfaces. 3.1.1. “As-cleaved” Mo0S100 Surface. Different sites on the

Mo0S100 surface shown in Figure 1a for both neutral and 9027



ARTICLE

Table 1. Adsorption Energies, Bond Distances, and Bader Charges of K/Kþ on the “As-cleaved” Mo0S100 Edge Surface Kþ (charged system)

K (neutral system) Ead (eV)

charge (|e|)

Mo atop (site 1)

1.89

0.774

MoMo bridge (site 2)

1.90

0.765

4-S hollowa (site 5)

3.15

0.869

adsorption site

dKS (Å)

dKMo (Å)

Ead (eV)

charge (|e|)

dKS (Å)

dKMo (Å)

3.453

0.65

0.838

3.631

3.689, 3.692

0.69

0.816

3.846, 3.862

1.84

0.877

3.287, 3.318,

2.06

0.868

3.220, 3.239,

Mo Edge

S Edge 3.237, 3.238, 3.258, 3.260 4-S hollow with

3.24

0.853

disulfideb (site 5)

3.328, 3.333

3.165, 3.179, 3.359, 3.375

3.413, 3.433 Interstitial

above interstitial (site 8) above interstitial with

2.82 2.49

0.846 0.824

2.986, 2.987, 3.520 3.120, 3.550

3.531

1.21

0.870

3.223

3.753

disulfide (site 8)

All initial site placements of K/Kþ at the Mo edge that are neither Mo atop nor MoMo bridge sites are unstable. K/Kþ moves to the 4-fold hollow sites from an interstitial site on the S edge upon geometry optimization. b Dimerization on the S edge by creating a disulfide bond. a

Figure 2. Adsorbed K atom on the Mo0S100 surface: (a) 4-fold S site on the S edge and (b) with an adjacent disulfide from reconstruction. The same color code as in Figure 1 is applied. K atom is represented by the purple sphere. All distances shown in the figure are in angstroms.

cationic charged systems were studied. The calculated adsorption energies are listed in Table 1. For the Mo0S100 system, the most energetically stable site for K/Kþ adsorption is the 4-fold hollow site over the fully sulfided S edge surface (Figure 2a). Adsorption at the subsurface interstitial sites between two edges is not energetically stable. We found that the K atom/ion moves from the initial tetragonal subsurface sites during optimization to an adjacent 4-fold hollow S site on the S edge upon adsorption. Meanwhile, an SS dimer is formed during the optimization (shown in Figure 2b). Compared to the 4-fold S hollow site with no SS dimerization, the adsorption of K with neighboring SS dimer configuration is slightly more stable. This is consistent with a previous DFT study, which showed that alternating SS

dimerization patterns along the S edge were energetically more stable than the undimerized S edge.42 The KS distances for the neutral K adsorption (3.243.26 Å) are slightly shorter than those for the charged Kþ adsorption (3.293.33 Å) at the 4-fold hollow S site on the S-edge surface. The difference in averaged KS lengths between K and Kþ adsorption configurations are summarized in Table 1. With the disulfide species adjacent to the 4-fold hollow S site, the KS bond lengths are broader for the neutral K system (3.183.38 Å) than for the charged Kþ system (3.223.43 Å). Two KS bond lengths on the side of the hollow site closest to the disulfide dimer are longer than those farthest from the disulfide dimer. This is due to steric interaction between the positively charged K and disulfide species. Our calculated 9028



ARTICLE

Figure 3. Adsorbed K atoms on the Mo0S100 surface: (a) Mo atop site on the Mo edge and (b) MoMo bridge site on the Mo edge. The same color code as in Figure 1 is applied. K atom is represented by the purple sphere. All distances shown in the figure are in angstroms.

KS distances are close to the KS distance of 3.20 Å in the antifluorite K2S structure.45 Another stable site for K/Kþ adsorption on the Mo0S100 surface was found at the interstitial sites between two edges. Both the neutral K and Kþ move out of the subsurface hexagonal interstitial site during optimization. The K atom/ion induces the dimerization of two opposing S edge sulfur atoms. Unlike the optimization from the octahedral subsurface site discussed earlier, the K atom/ion remains at the interstitial site and interacts with two sulfur atoms below the terminal Mo atom on the Mo edge. Similarly, we note that the KS distances for neutral K adsorption are slightly shorter than those of the charged Kþ system. Since the trend in KS distances in the neutral K systems are shorter, the constrained distance between two MoS2 edges is too short for a stable interstitial configuration for Kþ adsorption. An energetically stable site is found over the interstitial with no dimer formed on the S-edge for only the neutral K system. If the interstitial was allowed to expand, both the neutral K and charged Kþ systems would prefer to coordinate with subsurface S atoms below the two edges. The least energetically stable sites on the Mo0S100 surface were found at the atop Mo and bridge MoMo sites on the bare Mo edge (Figure 3 panels a and b, respectively). The stability of these two adsorption configurations is similar (Table 1). The MoK distances are longer than the KS distances where the K atom/ion is adsorbed at a 4-fold hollow S site on the S-edge or at the interstitial site. A number of ultra-high-vacuum experiments addressing the adsorption of heavy alkali metals on the MoS2(0001) basal surface can be found in the literature.4651 Though these experiments address only alkali metal adsorption at the basal surface, some comparisons with the S edge surface and the sulfurcovered basal surface can be made. Kamaratos et al.48 measured a ∼3 eV desorption energy at low K coverage (θK < 0.5 ML) in their combined low-temperature (100 K) electron diffraction spectroscopy (LEED), atomic emission spectroscopy (AES), electron energy loss spectroscopy (EELS), and thermal desorption spectroscopy (TDS) experiments. Our lowest K adsorption energies for the neutral system are similar to the basal surface estimation. Kamaratos et al.48 noted that this binding energy is close to that for ionic bonding of K, implying that the

low-coverage adsorbed K is likely to be in an ionic state. The ∼3 eV binding energy at low coverage is similar to the ionic cohesion energy of K to Cl in the KCl lattice stated by Ciraci and Batra52 in their theoretical work on the semiconducting Si(001) system. In all cases, the Bader charge of K reflects a cationic K that has donated charge to the MoS2 edge surface system. We have calculated the bulk body-centered cubic (bcc) metallic K cohesive energy to be 0.93 eV, in good agreement with literature. The binding energy with respect to this value would decrease the binding energy to ∼2 eV with low coverage and ∼1 eV at high coverage. We also note that the work function is greater for the charged K-doped systems (5.36.5 eV) compared to the neutral K-doped and -undoped systems. For the neutral K-doped system, there is a lowering of the work function from the donation of the K 4s electron to the substrate and a favorable ionic dipole interaction would form. In the charged case, the surface charge rearranges to form a screening charge and would pay an energy penalty for this rearrangement. Finally, Alexiev et al.53 published the only electronic structure study of an alkali-metal (Li) doped MoS2(100) surface (Mo0S100 only). In contrast to the K-doped surface with multiple favorable K-adsorption configurations, the only energetically favorable adsorption site for Li was found over the interstitial space, which would be favorable for facile diffusion of the small lithium atom/ion into the interstitial spaces. However, Alexiev et al.’s calculations cannot be readily compared with experimental observations since Li (and Na) promotion is known to cause profound structural change to 2H-MoS2, which is not the case for heavy alkali metals K, Rb, and Cs.23,49 We note here that our calculations consider only the adsorption situation at 0 K. Climbing image nudged-elastic band54 transition state searches we conducted suggest that surface diffusion can readily occur at temperatures relevant to catalytic conditions. The maximum barrier to surface diffusion is ∼3 kcal/mol. 3.1.2. Mo50S50 Surface. The energetically stable sites for K/Kþ at the Mo50S50 surface are those over the interstitial where the K/Kþ coordinates with three S atoms as shown in Figure 4. At these stable sites, K triply coordinates either with two surface S atoms on the Mo edge and one surface S atom on the S edge (Figure 4a) or with one S atom on the Mo edge, one subsurface 9029



ARTICLE

Figure 4. Adsorbed K atom on the interstitial sites of the Mo50S50 surface: (a) binding with two surface S atoms on the Mo edge and one S atom on the S edge, (b) binding with two S atoms on the Mo edge (one surface S atom and one subsurface S atom) and one S atom on the S edge, and (c) binding with two S atoms on the Mo edge (one surface S atom and one subsurface S atom) and one atom on the S edge with lattice distortion.

S on the Mo edge, and one S atom on the S edge (Figure 4b). Energetically, these two configurations are nearly equal. As a result, adsorbed K atoms/ions are expected to linearly decorate the

space over the interstitials at the (100) edge surface. Since the K atom is confined to the sites over the interstitial, the S and Mo sites at the MoS2(100) edges are relatively open to approach by 9030



ARTICLE

Table 2. Adsorption Energies, Bond Distances, and Bader Charges of K/Kþ on the Mo50S50 and Mo37.5vS50 Edge Surfaces Kþ (charged system)


adsorption site

charge (|e|)

dKS (Å)

Ead (eV)

charge (|e|)

dKS (Å)

Mo50S50 Interstitiala coordinated with 1 S of Mo edge and 2 S of S edge (site 5)

3.46

0.894

3.167, 3.254, 3.286

1.64

0.890

3.190, 3.372, 3.529

coordinated with 2 S of Mo edge and 1 S of S edge 1c (site 5)

3.37

0.863

3.169, 3.197, 3.370

1.72

0.893

3.196, 3.427, 3.498

coordinated with 2 S of Mo edge and 1 S of S edge 2d (site 5)

3.30

0.868

3.194, 3.275, 3.313

1.72

0.893

3.373

interstitiale (site 1)

3.06

0.795

3.222, 3.428, 3.502

1.42

0.392

3.256, 3.296, 3.377

over “vacancy” (site 2)

2.55

0.854

3.199, 3.202

1.04

0.898

3.304, 3.304

b

Mo37.5vS50

All stable sites for K/Kþ on the Mo50S50 surface are above the interstitial. b Distortion of bridging S on the S edge facilitates coordination of two S atoms with K. c K coordination with two topmost bridging S of Mo edge as well as one bridge S on top of S edge. d K coordination with one bridge S on top of Mo edge and one prismatic S directly below on the Mo edge, as well as one bridge S on top of S edge. e K migrates away from the “vacancy” toward sulfur-rich region. a

Figure 5. Adsorbed K atom on the Mo37.5vS50 surface: (a) binding at the vacancy site with two S atoms on the Mo edge or (b) binding at the interstitial site with two S atoms on the Mo edge and one S atom on the S edge.

adsorbates although the long-ranged interaction of the large K atoms/ions could affect the adsorbates. This is in contrast to the case where the K atoms/ions are adsorbed at the 4-fold S-coordinated hollow sites on the fully sulfided S edge surface, blocking

potential S and Mo adsorption sites. The calculated adsorption energies and structural parameters are summarized in Table 2. If potassium migrates into the subsurface (as would occur in the intercalation process), hexagonal features at the subsurface 9031



ARTICLE

Figure 6. Adsorbed K atom on the Mo100S100 surface: (a) 4-fold S site on the Mo edge; (b) binding at the interstitial site with two S atoms on the Mo edge and two S atoms on the S edge; (c) binding at the interstitial site with two S atoms on the Mo edge and two S atoms on the S edge (one of which is included in an SS dimer). 9032



ARTICLE

Figure 7. K atom deeply embedded in the interstitial of the Mo100S100 surface.

can be distorted by the embedded K. This is shown in Figure 4c. This configuration is slightly more energetically stable (3.46 eV) compared to the configuration with K being merely above the interstitial and inducing minimal strain to the MoS2 lattice. The local subsurface structure distortion to a hexagonal structure is expected to affect the adsorption of syngas species such as CO and H2 by subtly changing the electronic structure of the edge. 3.1.3. Mo37.5vS50 Surface. The Mo37.5vS50 surface model represents an S “vacancy” on the 50% sulfided Mo edge, resulting in a bare MoMo bridge site. Two stable K adsorption sites were identified. In the first configuration shown in Figure 5a, K directly adsorbs at the vacancy site by binding with the two surface S atoms on either side of the vacancy. The second configuration involves K atom/ion over the interstitial site (Figure 5b). In the second configuration, the K atom/ion was initially placed over the interstitial close to the vacancy. Upon optimization, the K moved from the more positively charged vacancy site to the more negatively charged sulfur-rich domain over the interstitial site. Our calculations show that the K atom/ ion adsorbed over the sulfur-rich interstitial surface domain is energetically more favorable. The same trend in KS distances of the neutral K system being slightly shorter than that of the charged Kþ system holds (3.20 Å for the neutral K system versus 3.30 Å for the charged Kþ system). However, this trend does not precisely hold for the case where the K is over the interstitial away from the vacancy (3.223.50 Å for the neutral case versus 3.243.38 Å for the charged Kþ system). Since K prefers sulfur-rich domains away from vacancy sites, the vacancy sites with exposed bare Mo atoms would favor hydrocarbon formation reactions over CO hydrogenation. The calculated adsorption energies and structural parameters are summarized in Table 2. 3.1.4. Mo100S100 Surface. On the Mo100S100 surface, K adsorbs at the 4-fold hollow S site on the fully sulfided Mo and S edges. The disulfide dimers of the fully sulfided Mo edge provide pseudo-4-fold hollow sites for K adsorption (Figure 6a). This configuration is not as stable as the 4-fold hollow site on the S edge. Figure 6b shows another stable configuration where K adsorbs over the interstitial with 4-fold coordinating S atoms. Subsurface sites for K adsorption were attempted. Similar to the case of the subsurface sites in the Mo0S100 system, the subsurface sites for K atom/ion are not stable. In the case of the charged Kþ system, K does not stay in the interstitial over the optimization run and moves to a four-coordinated site with one

surface (in a SS dimer) and two subsurface S atoms at the Mo edge and one surface S (in a SS dimer) at the S edge. The K atom/ion is not fully over the interstitial site, and there is some apparent strain from the KMoS2 wall repulsion pushing the MoS2 sheets outward. As with the Mo0S100 case, an SS dimer at the S edge is formed from the K atom pushing the two S edge S atoms together on the way out of the subsurface hexagonal site. On the other hand, K moving to the top over the interstitial during optimization was observed for the neutral system (Figure 6c). As with the cationic charged system case, an SS dimer is formed during the optimization where K moves out of the unstable subsurface site. Finally, we want to briefly discuss the possibility of K pseudointercalation. K, like other alkali metals (Li, Na, Rb, and Cs), is known to intercalate into the MoS2 bulk between the van der Waals-held MoS2 sheets. For the heavy alkali metals (K, Rb, Cs), the 2H structure of MoS2 remains well-defined with a 35% increase in the c-axis length (from 12.294 to 16.580 Å).23 Figure 7 shows the pseudointercalation of a single K atom well below the surface/subsurface in the fully sulfided model system. With the crystallographic spacing of the MoS2 sheets, the insertion is only slightly favorable (0.06 eV) for the neutral K system and unfavorable (1.54 eV) for the charged Kþ system. Structurally, there is a slight bulging outward of the MoS2 where the K/Kþ is inserted. The KS distances are much shorter than the preferred 3.13.4 Å distances found at the edge surface, indicating that the interstitial spacing has to be expanded in order to accommodate intercalated K. 3.2. Electronic Structure and Magnetic Properties Analysis. For the bulk MoS2, the S atoms coordinate the Mo atoms in a trigonal prismatic configuration. This coordination results in a ligand-field splitting of the Mo 4d orbitals into three groups: degenerate dxz and dyz, nondegenerate dz2, and degenerate dxy and dx2y2.55,56 The bottom of the conduction band is primarily composed of a combination of Mo 4d dz2, dxy, and dx2y2states.21,49 The S 3p orbitals are split into two groups with px and py in one group and pz in the other group.21 Previous DFT studies have shown that a low-lying band of states around 16 to 13 eV, which is primarily attributed to the S 3s.57 The valence band, typically extending from 8 eV to the Fermi level, is primarily composed of a mixture of Mo 4d states and S 3p states (the filled Mo dz2 mixes strongly with the S 3pz orbitals).57,58 The conduction band from the top of the band gap to ∼4 eV is also primarily 9033



ARTICLE

Figure 8. Total (DOS) and local (atomic) density of states (LDOS) for the most stable configuration of adsorbed K atom on (a) Mo0S100, (b) Mo50S50, (c) Mo100S100, and (d) Mo37.5v-S50 surfaces. (A) Unpromoted DOS, (B) K-promoted DOS, (C) LDOS of Mo, (D) LDOS of S, and (E) LDOS of K. All energies are shifted to the Fermi energy of the unpromoted system at 0 eV.

Figure 9. Projected density of states (PDOS) for the neutral K doped Mo0S100 system with the formation of disulfide: (a) d band for the Mo atom under the 4-fold S hollow site on the S edge, (b) d band for the two Mo atoms adjacent to the newly formed disulfide (inset shows splitting of R and β spin DOS), (c) sp band for four S atoms (hollow site) binding with K atom, and (d) sp band for the disulfide S atoms. (A) Unpromoted; (B) K-promoted. All energies are shifted to the Fermi energy of the unpromoted system at 0 eV.

composed of a mixture of Mo 4d and S 3p states. The S 3p states are dominant in the valence band up to the Fermi level and the Mo 4d states dominate the conduction band above the band gap. The optical band gap energy has been reported to be between 1 and 2 eV22,21,59 while DFT calculations typically underestimate this band gap (0.80.9 eV).57,59 For the MoS2(100) edge surface, the differences in the coordination of Mo and S atoms at the edge (unsaturation and/or distortion) compared with the bulk reduce the ligand-field

splitting.42,57 The metallic behavior of the edge surface from these “brim states” (as indicated by the closing of the band gap) has been confirmed by STM experiments60 and DFT calculations of Au-supported MoS2 nanoparticles.57 For the unpromoted MoS2(100) edge surface, our calculated total DOS shown in Figure 8 is in agreement with the previous DFT results by Raybaud et al.57 The lack of a band gap shown by our DOS analysis is in agreement with those earlier observations60 and calculations.57 9034



ARTICLE

Table 3. Adsorption Energies, Bond Distances, and Bader Charges of K/Kþ on the Mo100S100 Edge Surface Kþ (charged system)


charge (|e|)

4-fold disulfidedisulfide “hollow” (site 1)

2.61

0.833

4-S hollowa (site 4)

3.16

0.862

4-fold coordinationb (site 3)

3.00

0.856

adsorption site

dKS (Å)

Ead (eV)

charge (|e|)

dKS (Å)

3.074, 3.084, 3.094, 3.097

1.16

0.894

3.194, 3.204, 3.220, 3.221

1.72

0.879

3.310, 3.318, 3.387, 3.410

0.97

0.836

1.64

0.882

Mo Edge

S Edge 3.242, 3.247, 3.317, 3.337 Interstitial 3.107, 3.139, 3.180, 3.250

4-fold coordination with S-edge disulfide 1c (site 3) 4-fold coordination with S-edge disulfide 2d (site 3)

3.13

0.842

3.070, 3.121, 3.245, 3.298

2.939, 2.978, 3.147, 3.203 3.128, 3.254, 3.339, 3.431

a

Identical to S edge 4-fold hollow sites on the S edge in the Mo0S100 system. b K coordinated with S atoms from two different disulfide at Mo edge and two S on the same side of S edge. c K coordinated with one S atom of disulfide at Mo edge and two S on the subsurface layer below, as well as one S from newly formed disulfide at S edge. d K coordinated with two S atoms of two different disulfide at Mo edge, one S atom of the newly formed disulfide at the S edge and one S atom from nondimer sulfur at S edge.

When promoted with K, the local density of states of the K (Figure 8) shows a weak, highly delocalized set of states around the Fermi level with a clearly discernible K 3p state resonance peak between 15 and 14 eV, overlapping the MoS2 valence states that are dominated by the S 3s states. For the Mo0S100, the lowest configuration with K over a 4-fold S hollow adjacent to a SS dimer (shown in Figure 2b) has a few resonance peaks overlapping with the conduction band Mo d and S sp states that can be assigned to a mixture of 4s (∼3.25 eV), 4p (∼5 eV), and 3d (57 eV) states. The 3d states at 57 eV connect the upper part of the conduction band to a higher manifold of states (dominated by the S 3d states), indicating hybridization of the K 3d states with those of MoS2 (DOS shown in Figure 8a). All the total DOS plots for the four different model edge surfaces (Figure 8) are qualitatively similar, with an overall 0.10.3 eV shift in the total DOS for the K-promoted MoS2 systems compared with that of the unpromoted MoS2 systems. This shift is qualitatively consistent with that reported by Alexiev et al.53 for the Li-promoted MoS2(100) (Mo0Mo100) surface from periodic HartreeFock calculations. The presence of K promoter, however, does not profoundly change the overall profile of the DOS. This is in qualitative agreement with the “rigid band” observations of heavy alkali metal ns electron charge injection into the bottom of the conduction band of MoS2 in angle-resolved X-ray photoelectron spectroscopy (ARXPS) experiments.49 Electron donation of the highly delocalized K 4s charge to the MoS2 surface may result in a repulsion of the p and d electrons leading to a shift in the p and d DOS, as seen in K-doped transition metal surfaces like K/Rh(111).61 DOS plots for the neutral and charged Kþ systems have nearly the same profiles.61 Recent experimental observations and DFT calculations indicate that the MoS2 edge surface has notable magnetic properties.4244 Though these magnetic properties are not expected to significantly influence the catalytic properties of the surface, the magnetization of the edge is important in the interpretation of characterization data from ESR spectroscopy of alkali-doped MoS2 catalysts. Vojvodic et al.42 found that changes in S coverage and orientation (e.g., undimerized, dimerized, trimers) at the unprompted MoS2 edge can induce changes in the magnetic properties of the edge surface. The hybridization of the Mo 4d and the undimerized S 3p states results in a spin

splitting of the S states at the S edge that is associated with a small local magnetic moment. Similar behavior was observed in our calculations. Figure 9a shows the projected density of states (PDOS) for the Mo atom directly below the adsorbed K atom at the 4-fold hollow S site on the S edge for the p and d angular momentum channels, respectively. Although there are examples of state spin splitting in most of the DOS analyses considered here, the inset in Figure 9c demonstrates one of the more profound examples. Bader charge analysis shows that the K charge ranges from þ0.39 to þ0.89 |e| for both neutral and charged systems (Tables 14). Most Mo atoms on the edge show small charge changes of ∼0.01 |e|. For K adsorbed at the Mo0S100 surface with no SS dimer formation (Figure 2a), the S atoms at the 4-fold hollow S site show a small increase in negative charge of 0.10 to 0.12 |e| for the neutral K system and 0.07 to 0.11 |e| for the charged cationic system. The negative charge of the Mo atom below the adsorbed K increases 0.09 |e| for the K and 0.08 |e| for the charged cationic systems, respectively. Upon sulfided dimer formation from the subsurface K optimization, the charge differences are nearly identical to the Kpromoted, undimerized configuration with respect to the 4-fold hollow S atoms coordinated with K. Similar charge changes are found for the other surface models studied here. The creation of a sulfur dimer results in some charge redistribution among the Mo atoms that are bonded to the SS dimer. Each Mo atom gains 0.07 and 0.11 |e| for the neutral K system and 0.05 to 0.10 |e| for the charged cationic system while the S atoms of the dimer lose þ0.19 to þ0.30 |e| for the neutral system and þ0.21 to þ0.31 |e| for the charged cationic system. There is also a small charge gain for the Mo atoms that is not directly associated with the 4-fold S hollow on the S edge. Interestingly, for the Mo0S100 K-promoted, undimerized charged system with K over an S edge 4-fold S hollow (depicted in the top figure in Figure 2a), an adjacent pair of flanking S edge sulfur atoms is close to each other in what appears to be a “pre-dimer” configuration (2.73 Å opposed to 3.00 Å for the unpromoted system) with a decrease in negative charge of þ0.05 and þ0.07 |e| for the two sulfur atoms. Bader charge analysis of the spin density difference shows a large local magnetic moment of 0.37μB on each of the predimer S atoms while the other S atoms on the S edge show very small magnetic moments ranging 9035



ARTICLE

Table 4. Average Adsorption Energies, Bader Charges, and Work Functions at Different K Coverages on Mo0S100, Mo50S50, and Mo100S100 Edge Surfaces K coverage (ML)

Ead (eV/K)

charge (|e|/K)

Φ (eV)

Mo0S100 Surface 0.0

5.0

0.125a

3.24

0.853

4.4

0.25

2.68

0.807

3.8

0.5 1

2.51 2.04

0.774 0.598

3.4 3.1

Mo50S50 Surface 0.0

5.6

0.125a

3.46

0.863

5.0

0.25

3.21

0.854

4.4

0.375

3.00

0.848

4.1

0.5

2.88

0.834

3.9

0.625

2.72

0.816

3.8

0.75 1

2.60 2.35

0.802 0.777

3.8 3.3

Mo100S100 Surface 0.0

5.0

0.125a

3.16

0.842

4.6

0.25

2.62

0.780

3.9

0.5

2.61

0.742

3.2

1

2.00

0.398

3.3

2b

2.29

0.557

3.3

a

Lowest energy configurations were used. b Sulfur coverage of 100% sulfided Mo edge reduces to 50% with KxSy species formation over the edge surface.

from 0.003 to 0.04μB, which is close to the average magnetic moment of 0.03μB for S atoms on the S edge of the unpromoted Mo0S100 system (see Table S1 in the Supporting Information). These results are in qualitative agreement with the calculations of Vojvodic et al.42 They postulated that the drive for lower-energy SS dimer configurations may be due to Peierlstype instabilities that are potentially pronounced at the lowdimension edge surface of MoS2.42 Indeed, the presence of K in the charged cationic Mo0S100 system appears to modulate the magnetic configuration of the edge surface toward the favorable SS dimerization on the S edge. Also consistent with previous results,42 the bare Mo edge tends to have a ferromagnetic arrangement with a local magnetic moment of 0.12μB per Mo atom for the unpromoted Mo0S100 system. Addition of K on the S edge affects the local magnetic moments of the Mo atoms on the Mo edge. For the undimerized neutral case, the local Mo atom magnetic moments decrease slightly in magnitude with the four Mo moments ranging from 0.08μB to 0.09μB, and for the charged case, they increase with an alternating pattern ranging from 0.11μB to 0.27μB. Similar behavior is seen for the fully sulfided Mo100S100 system (Figure 1c). However, we also note that the fully sulfided Mo edge induces some notable differences in the magnetic properties of the MoS2(100) system in contrast to the bare Mo edge system (Mo0S100). In the charged K-promoted Mo100S100 system, the alternating predimer pairs at the S edge have 0.12μB and 0.13μB magnetic moments for one undimerized SS pair and 0.46μB and 0.46μB for the other pair,

giving a ferrimagnetic configuration (see Table S2 in the Supporting Information). The associated MoMo atom pairs on the S edge that bond to these predimer SS pairs have magnetic moments of 0.22μB and 0.23μB for the first SS pair and negligible magnetic moments of 0.01μB for the second SS pair. However, the SS dimers at the Mo edge have a ferromagnetic configuration with each S atom having a local magnetic moment of 0.110.12μB. Upon the addition of one neutral K atom (Figure 2b), the local magnetic moments for the two “pre-dimers” decrease to 0.11μB per S in the first pair and 0.430.48μB in the second pair (closest to the adsorbed K atom). The neutral K-doped Mo100S100 system also has a significant change in the magnetic moments for adjacent MoMo pairs in a ferrimagnetic configuration (0.04μB for each Mo in the first pair and 0.17μB and 0.19μB for the second pair). The magnetic moments of the SS dimers on the Mo edge become zero (see Table S2 in the Supporting Information). A net charge gain from 0.03 to 0.13 |e| is obtained for most of the S atoms on the Mo edge sulfur except for one S atom in the predimer adjacent to the adsorbed K atom, which has a small þ0.03 charge (the flanking S pairs appear to be polarized with the S closest to the positively charged K being more negative than the S on the opposing side). For the charged cationic system, the only significant change in local magnetic moments is seen with the predimer SS pair closest to the potassium where the local magnetic moment is zero for both S atoms and one of the associated Mo atoms (the other Mo changed sign to 0.01 μB). Finally, the unpromoted Mo50S50 surface (Figure 1b) has local magnetic moments of 0.470.50μB for Mo atoms on the S edge and 0.170.26μB for Mo atoms on the Mo edge (see Table S3 in the Supporting Information). Upon K/Kþ addition, both neutral and charged cationic systems show virtually no magnetic moment for any of the edge atoms (Figure 4c). When an edge S vacancy is opened up on the Mo edge over the Mo37.5vS100 surface (Figure 1d), the Mo edge has a small 0.02μB magnetic moment for all edge atoms except for S atoms around the S vacancy (0.06μB to 0.07μB). The magnetic moments of S atoms on the S edge range, in an alternating pattern, from 0.16μB to 0.21μB (see Table S4 in the Supporting Information). The magnetic moments of Mo atoms that comprise the S edge range, also in an alternating pattern, from 0.05μB to 0.14μB. Upon addition of K over the interstitial, neutral and charged systems exhibit nearly the same magnetic behavior. The magnetic moments of S atoms around the S vacancy on the Mo edge slightly decrease (0.05μB to 0.06μB) for both neutral and charged systems. Mo atoms on the S edge show an increase in magnetic moments (0.460.52μB for neutral and 0.470.51μB for charged systems), while S atoms on the S edge have a very small magnetic moment of 0.01μB. Thus, the overall edge surface has a ferrimagnetic configuration. 3.3. Coverage Effects. With increasing K coverage on the surface, it is expected that charge is built up in the conduction band, leading to less charge transfer from K to the surface with formation of a metallic K adlayer. A number of K atoms were added to the Mo0S100, Mo50S50, and Mo100S100 surfaces on the basis of the number of available adsorption sites. One monolayer of K coverage is defined as one K atom per surface MoS2 unit. Figure 10 depicts the optimized geometry of K adlayer at its highest coverage on each surface. After optimization, the KK distances on the 1.0 ML K-covered Mo50S50 and 2.0 ML K-covered Mo100S100 model surfaces are 9036



ARTICLE

Figure 10. Optimized structures of MoS2(100) surfaces at high K coverage: (a) Mo0S100 at 1.0 ML K coverage, (b) Mo50S50 at 1.0 ML K coverage, and (c) Mo100S100 at 2.0 ML K coverage. 9037


The Journal of Physical Chemistry C considerably shorter (3.66 and 3.69 Å, respectively) than the KK distance of 4.61 Å in bcc bulk K.62 For the 2.0 ML K-covered Mo100S100 model surface, K atoms were added to every available 4-fold S hollow site on each edge as well as every pseudo-four-coordinated S site above the interstitial site. Four surface S atoms on the Mo edge were dislodged, forming a sulfur-poor KS phase on top of the K adlayer. This indicates that the sulfur dimers are unstable on the Mo edge when the K coverage is greater than 1.0 ML. Upon closer inspection of Mo0S100, Mo50S50, and Mo100S100 structures at high K coverage, we found that the K atoms tend to organize as chains along the top of the van der Waals gap and over the fully sulfided S edge. For the Mo100S100 surface at 2.0 ML K coverage, the KK distances of the chain structures are very short with an average distance of 3.16 Å (Figure 10c). This is also seen with the Mo50S50 model at the highest coverage, but the KK chains are zigzagged with a longer average distance of 3.66 Å at 1 ML (Figure 10b). The unusually short KK distances can be attributed to the increased 3d (transition-metal-like) bonding character, reminiscent of the anomalously short KK distances for pure K metal and K alloys under high pressure.63,64 The PDOS of K shows a dramatic increase in the number and intensity of accessible K 3d states in the conduction band, which begins at ∼1 eV above the Fermi level. This “chemical pressure” effect of compressed KK distances relative to bcc bulk KK distances may be a contributing factor in the favorable intercalation of K and other heavy alkali metals. Bader charge analysis listed in Table 4 suggests a decrease in charge transfer from the K to the surface as the K coverage increases. This is accompanied by a decreasing adsorption energy per K atom. Kamaratos et al.48 found that K most likely adsorbs at the graphite-like hexagonal sites on the basal surface at K coverages below 0.5 ML. Above 0.5 ML K coverage, an observed 3 eV K plasmon and a low-energy desorption of 0.9 eV were observed, indicating the formation of metallic K adlayers. A dramatic decrease in the work function for adsorbed K at the basal plane at low coverage also indicates charge transfer of the K 4s electron to MoS2 edge surface.46 The calculated work functions for the unpromoted Mo0S100 and Mo100S100 edge surfaces (see Table 4) are within the error bars of the experimentally measured work function for the MoS2 basal surface (4.8 ( 0.5 eV).65 The calculated work function for the unpromoted Mo50S50 edge surface is slightly above the upper boundary (5.6 eV). Herein, we must mention that there are no measurements of work functions related to the MoS2 edge surface, and there are some important differences between K adsorption on the basal surface and on the edge surface. The basal surface is similar to the semiconducting bulk system, while the edge surface is more metallic. At low K coverage, the addition of K results in a gap state just above the MoS2 valence band from the transfer of the K 4s electron to MoS2.49 This is accompanied by a sharp decrease in the work function (by ∼1.1 eV) at low coverage (θK < 0.5 ML).46 As more K is added at low temperature, a metallic K layer forms, and the reduction in work function is much more gradual,46 and the work function is still ∼1 eV above that of pure K metal (2.3 eV66). In the cases of Mo100S0, Mo50S50, and Mo100S100 edge surfaces, a decrease in the work function with increasing K coverage is observed in this work. The decrease is less dramatic at low K coverage compared to that for the basal surface. This may be due to the edge surface not having the transition from insulating to conducting since it is inherently

ARTICLE

metallic. The Mo0S100 and Mo100S100 systems have more dramatic decreases in work function than the Mo50S50 system. The coverage effect on K interaction with the MoS2 edge surface is found to be similar to other semiconductor and metal systems. For example, MoS2 shares some structural similarities with the hexagonal graphite basal plane. There is similar behavior with respect to K adsorption over the hexagonal sites of the graphite sheet, where DFT calculations have shown that there is a reduced 4s charge transfer from the K atom to the graphite conduction band with increasing K coverage.67,68 The graphite system at low K coverage shows completely ionized Kþ, and their delocalized screening charge from the donated electron in the graphite surface forms large dipoles and a large work function reduction of the graphite surface.67 At low coverage, there is a long-range repulsive interaction between the K cations due to electrostatic repulsion and to the filling of the graphite π* states by charge donation from the K cations.67 This similar behavior at low K coverage observed for graphite is also seen at metal surfaces such as Rh(111) (but this involves metal d orbitals). At high K coverage, metal and semiconductor (e.g., MoS2) surfaces show similar behavior with respect to the formation of metallic K adlayers and clustering.48,61 3.4. Discussion of K Promotion and Alcohol Selectivity. In this study, we focused only on the interactions of K at the edge surfaces and near-surface interstitial sites where the most prevalent catalytic activity occurs for MoS2-based catalysts. K may play multiple roles in alcohol-selective MoS2-based catalysts. K can block the Mo and S surface sites that favor CO dissociation leading to hydrocarbon formation; K can also serve as an adsorption site for CO and other reaction intermediates to promote CO hydrogenation over dissociation. Moreover, K changes the surface basicity by transferring its delocalized 4s charge to the MoS2 system. This excess charge on the MoS2 surface may promote chain growth reactions for higher alcohols and inhibit H2 dissociation. From experimental work with related Mo-based catalysts, there is spectroscopic evidence that K may also stabilize surface sulfur species such as disulfides.25,26 Open Mo sites are likely to favor CO dissociation, and retention of S is expected to be an important factor in the alcohol selectivity of K-promoted MoS2-based catalysts. We have shown that K can physically block species that can remove surface sulfur. K can also cause electronic changes to the surface that render S-removing surface reactions unfavorable. Our calculations indicate that K can influence the creation of S dimer species and show that K can influence the S surface coverage in the case of Mo100S100 system with 2 ML K coverage. The hygroscopic character of K atom when ionized may also aid in the adsorption of H2S gas required to replenish lost surface S species during the catalytic process. Alkali metal promoters like K metal and K2O are well-known in the group VIII metal catalysts literature.69,70 As with the MoS2 edge surface studied here, K atom donates the 4s electron to group VIII metals, and this electron donation is then postulated to promote back-bonding into π* orbitals of adsorbate systems such as CO, N2, and NO.70 As a result, this facilitates the dissociation of diatomic species such as adsorbed CO and enhances the stability of reaction intermediates. For the semiconductor MoS2 material with metallic edge state behavior, injection of the 4s electron of K into the MoS2 conduction band can be considered to be a quasi-metallic catalyst material with catalytic behavior similar to that of K-promoted transition metal 9038


The Journal of Physical Chemistry C systems such as rhodium.61 However, unlike group VIII metal systems, the availability of Mo metal centers for strong adsorption of CO is hindered by the presence of S and K species. At moderate or high S coverage with adsorbed surface K, weakly adsorbed CO may favor direct hydrogenation of CO to HCO, leading to oxygenates, instead of CO dissociation, leading to hydrocarbons. K-promoted MoS2 catalyst is thought to be a bifunctional catalyst where H2 dissociation occurs on the MoS2 edge surfaces and CO is molecularly adsorbed by Kþ with direct CO hydrogenation dominant.11,18,71 With regard to the H2 dissociative adsorption at MoS2 edge sites, however, the delocalized electron of s orbital character injected by the K into the metal system may have the ability to “poison” H2 dissociative adsorption due to the repulsion between the delocalized s charge of the surface and the 1s character of the approaching H2. This phenomenon has been noted in the experimental literature of K-promoted Rh(111) and other group VIII metals and may apply to the metallic edge of MoS2(100).61 We also note, unlike transition metal promoters such as Ni and Co, which have short-range interactions with sulfur at the edge (∼2.4 Å)72 and are integrated tightly into the edge surface (changing the phase and morphology of MoS2 crystallites),73,74 the interaction of potassium with the MoS2 edge surface is longrange with very little perturbation to the edge surface structures. As noted earlier, the K interaction is also expected to be dynamic with facile surface diffusion of K atoms/ions. Though the K atoms/ions are transient, obstruction of Mo sites by adsorbed K atoms/ions can be partly responsible for decreased hydrocarbon formation via CO dissociation and increased alcohol selectivity.

4. CONCLUSION K-promoted MoS2 is a promising catalyst for the production of C2þ alcohols. In this work, the interaction of K with the MoS2 edge surface was investigated by density functional theory. Among the various sites on the Mo and S edge surfaces, as well as the interstitial and subsurface sites, the most optimal sites for K atom/ion to occupy are those that maximize the interaction of K with S atoms. K atoms/ions are least likely to bind with Mo atoms on the edge surface. The KS distance in the K-promoted MoS2 is comparable to the KS distance in the antifluorite K2S lattice. In contrast to transition-metal promoters, the interaction of K with S and Mo atoms is long-range. High mobility of K on the MoS2 catalyst is also found. The preferred configuration for K/ Kþ is to maximize its interactions with S atoms. On the bare Mo-terminal edge surface of the Mo0S100 model, stable local energy minima were found for K atom/ion directly atop Mo atoms or directly over bridging MoMo sites. These minima, however, are ∼1 eV higher than those for K atom/ion adsorbed at the 4-fold hollow S site on the S edge surface. Thermal fluctuations can easily drive K migration to more sulfur-rich regions of the surface. Sorption of K atoms/ions in subsurface hexagonal or tetrahedral sites is not stable with the experimental crystallographic spacing of the MoS2. Expansion of the interstitial is required to accommodate K at the interstitial sites. As K coverage increases, the adsorption energy per K atom and work function decreases. The decrease in work function when K is added to MoS2 directly correlates with increased basicity of the system. At 2.0 ML K coverage, K atoms tend to form chains of K atoms over the interstitial and fully sulfided S edge with KK distances of 3.23.7 Å, much less than that for bcc K metal. Moreover, the appearance of K 3d states in the DOS

ARTICLE

just above the Fermi level becomes more pronounced, indicating a possible transition-metal character to the K adlayer. At low K coverage, adsorbed K injects most of its 4s electron charge into the MoS2 surface for the neutral K-promoted MoS2 system. As K coverage increases, the transfer of charge from K to the MoS2 surface decreases. Bader charge analysis indicates that the transferred K 4s electron charge is delocalized over the metallic MoS2(100) edge surface. In the charged Kþ system, the MoS2(100) edge surface shows a slight electron transfer to K with a screening charge layer at the MoS2(100) edge. In agreement with previous DFT studies, our DOS analyses show a metallic edge system with various magnetization configurations for many of the edge K-doped sulfur configurations. On the basis of our calculations, the promotional effects that K has on MoS2’s selectivity toward mixed alcohols synthesized via CO hydrogenation may be due to increased surface basicity from the increased K-donated surface charge. This increased s-type electron surface charge can prevent dissociative H2 adsorption over the MoS2 edge via repulsion. Adsorbed K can also physically block the approach of H2 toward the MoS2 surface. Finally, K atoms at the surface can provide physical barriers to Mo and S edge CO adsorption and dissociation sites and provide weaker binding sites for CO adsorption that do not lead to dissociation of CO.

’ ASSOCIATED CONTENT

bS Supporting Information. Four tables listing Mo0S100 edge surface Bader spin magnetic moments and Mo100S100, Mo50S50, and Mo37.5vS50 edge surface Bader spin charges. This material is available free of charge via the Internet at http:// pubs.acs.org. ’ AUTHOR INFORMATION Corresponding Author

*E-mail [email protected] (A.A.) or donghai.mei@pnl. gov (D.M.).

’ ACKNOWLEDGMENT This work was funded by a CRADA project (PNNL/297) with Range Fuels. The research was performed by use of EMSL, a national scientific user facility sponsored by the Department of Energy’s Office of Biological and Environmental Research and located at Pacific Northwest National Laboratory. We thank Dr. Roger Rousseau for his insight into the intricacies of the MoS2 system, its interaction with alkali metals, and the application of electronic structure methods. We also thank Ron Stites and Karl Kharas at Range Fuels for their insightful discussions concerning alkali-promoted MoS2 as an alcohol synthesis catalyst. ’ REFERENCES (1) Afanasiev, P. C. R. Chim. 2008, 11, 159. (2) Spivey, J. J.; Egbebi, A. Chem. Soc. Rev. 2007, 36, 1514. (3) Subramani, V.; Gangwal, S. K. Energy Fuels 2008, 22, 814. (4) Quarderer, G. J.; Cochran, G. A.; Dow Chemical Company. Catalytic process for producing mixed alcohols from hydrogen and carbon monoxide. European Patent 0119609, September 26, 1984. (5) (a) Jackson, G. R.; Mahajan, D.; PowerEnerCat, Inc.. Method for production of mixed alcohols from synthesis gas. U.S. Patent 6753353, June 22, 2004. (b) Jackson, G. R.; Mahajan, D.; PowerEnerCat, Inc.. Method for production of mixed alcohols from synthesis gas. U.S. Patent 6248796, June 19, 2001. 9039


The Journal of Physical Chemistry C (6) Kinkade, N. E.; Union Carbide Corporation. Process for producing alcohols from hydrogen and carbon monoxide using an alkalimolybdenum sulfide catalyst. European Patent 0149255, May 6, 1987. Catalytic process for the production of alcohols from carbon monoxide, hydrogen, and olefins. European Patent 0149256, November 19, 1987. (7) Subramani, V.; Gangwal, S. K. Energy Fuels 2008, 22, 814. (8) Chen, A.; Wang, Q.; Li, Q.; Hao, Y.; Fang, W.; Yang, Y. J. Mol. Catal. A: Chem. 2008, 283, 69. (9) Iranmahboob, J.; Hill, D. O.; Toghiani, H. Appl. Surf. Sci. 2001, 185, 72. (10) Iranmahboob, J.; Toghiani, H.; Hill, D. O. Appl. Catal., A 2003, 247, 207. (11) Lee, J. S.; Kim, S.; Lee, K. H.; Nama, I. S.; Chung, J. S.; Kim, Y. G.; Woob, H. C. Appl. Catal., A 1994, 110, 11. (12) Lee, J. S.; Kim, S.; Kim, Y. G. Top. Catal. 1995, 2, 127. (13) Surisetty, V. R.; Tavasoli, A.; Dalai, A. K. Appl. Catal., A 2009, 365, 243. (14) Woo, H. C.; Nam, I. S.; Lee, J. S.; Chung, J. S.; Lee, K. H.; Kim, Y. G. J. Catal. 1992, 138, 525. (15) Woo, H. C.; Kim, Y. G.; Nam, I. S.; Chung, J. S.; Lee, J. S. Catal. Lett. 1993, 20, 221. (16) Xiang, M.; Li, D.; Li, W.; Zhong, B.; Sun, Y. Catal. Commun. 2007, 8, 513. (17) Xiao, H.; Lia, D.; Lia, W.; Sun, Y. Fuel Process. Technol. 2010, 91, 383. (18) Santiesteban, J. G.; Bogdan, C. E.; Herman, R. G.; Klier, K. Mechanism of C1C4 alcohol synthesis over alkali/MoS2 and alkali/ Co/MoS2 catalysts. Proceedings of the 9th International Congress on Catalysis, 1988. (19) Reshak, A. H.; Auluck, S. Phys. Rev. B 2003, 68, 125101. (20) Rocquefelte, X.; Boucher, F.; Gressier, P.; Ouvrard, G. Phys. Rev. B 2000, 62, 2397. (21) Mattheiss, L. F. Phys. Rev. B 1973, 8, 3719. (22) Wilson, J. A.; Yoffe, A. D. Adv. Phys. 1969, 18, 193. (23) Somoano, R. B.; Hadek, V.; Rembaum, A. J. Chem. Phys. 1973, 58, 697. (24) Iranmahboob, J.; Hill, D. O.; Toghiani, H. Appl. Catal., A 2002, 231, 99. (25) Yang, Y. Q.; Yuan, Y. Z.; Dai, S. J.; Wang, B.; Zhang, H. B. Catal. Lett. 1998, 54, 64. (26) Yang, Y. Q.; Yang, H.; Wang, Q.; Yu, L. J.; Wang, C.; Dai, S. J.; Yuan, Y. Z. Catal. Lett. 2001, 74, 221. (27) Lauritsen, J. V.; Kibsgaard, J.; Helveg, S.; Topsøe, H.; Clausen, B. S.; Lægsgaard, E.; Besenbacher, F. Nat. Nanotechnol. 2007, 2, 53. (28) Zak, A.; Feldman, Y.; Lyakhovitskaya, V.; Leitus, G.; PopovitzBiro, R.; Wachtel, E.; Cohen, H.; Reich, S.; Tenne, R. J. Am. Chem. Soc. 2002, 124, 4747. (29) CP2K is freely available from http://cp2k.berlios.de/. (30) Goedecker, S.; Teter, M.; Hutter, J. Phys. Rev. B 1996, 54, 1703. (31) Hartwigsen, C.; Goedecker, S.; Hutter, J. Phys. Rev. B 1998, 58, 3641. (32) Krack, M. Theor. Chem. Acc. 2005, 114, 145. (33) VandeVondele, J.; Hutter, J. J. Chem. Phys. 2007, 127, 114105. (34) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (35) Henkelman, G.; Arnaldsson, A.; Jonsson, H. Comput. Mater. Sci. 2006, 36, 254. (36) Sanville, E.; Kenny, S. D.; Smith, R.; Henkelman, G. J. Comput. Chem. 2007, 28, 899. (37) Tang, W.; Sanville, E.; Henkelman, G. J. Phys.: Condens. Matter 2009, 21, No. 084204. (38) Shi, X. R.; Jiao, H.; Hermann, K.; Wang, J. J. Mol. Catal. A: Chem. 2009, 312, 7. (39) Shi, X. R.; Wang, S. G.; Hu, J.; Wang, H.; Chen, Y. Y.; Qin, Z.; Wang, J. Appl. Catal., A 2009, 365, 62. (40) Sun, M.; Nelsona, A. E.; Adjaye, J. Catal. Today 2005, 105, 36. (41) Raybaud, P.; Hafner, J.; Kresse, G.; Kasztelan, S.; Toulhoat, H. J. Catal. 2000, 189, 129.

ARTICLE

(42) Vojvodic, A.; Hinnemann, B.; Nørskov, J. K. Phys. Rev. B 2009, 80, 125416. (43) Zhang, J.; Soon, J. M.; Loh, K. P.; Yin, J.; Ding, J.; Sullivian, M. B.; Wu, P. Nano Lett. 2007, 7, 2370. (44) Botello-Mendez, A. R.; Lopez-Uría, F.; Terrones, M.; Terrones, H. Nanotechnology 2009, 20, 325703. (45) Zintl, E.; Harder, A.; Dauth, B. Z. Elektrochem. 1934, 40, 588. (46) Papageorgopoulos, C. A.; Kamaratos, M.; Kennou, S.; Vlachos, D. Surf. Sci. 1991, 251252, 1057. (47) Papageorgopoulos, C.; Kamaratos, M.; Kennou, S.; Vlachos, D. Surf. Sci. 1992, 277, 273. (48) Kamaratos, M.; Vlachos, D.; Papageorgopoulos, C. A. J. Phys.: Condens. Matter 1993, 5, 535. (49) Park, K. T.; Kong, J. Top. Catal. 2002, 18, 175. (50) Park, K. T.; Kong, J.; Klier, K. J. Phys. Chem. B 2000, 104, 3145. (51) Park, K. T.; Richards-Babb, M.; Freund, M. S.; Weiss, J.; Klier, K. J. Phys. Chem. 1996, 100, 10739. (52) Ciraci, S.; Batra, I. P. Phys. Rev. B 1988, 37, 2955. (53) Alexiev, V.; Prins, R.; Weber, T. Phys. Chem. Chem. Phys. 2000, 2, 1815. (54) Henkelman, G.; Uberuaga, B. P.; Jonsson, H. J. Chem. Phys. 2000, 113, 9901. (55) Huisman, R.; De Jonge, R.; Haas, C.; Jellinek, F. J. Solid State Chem. 1971, 3, 56. (56) Caragiu, M.; Finberg, S. J. Phys.: Condens. Matter 2005, 17, R995. (57) Raybaud, P.; Hafner, J.; Kresse, G.; Toulhoat, H. Surf. Sci. 1998, 407, 237. (58) Mehran, F.; Title, R. S.; Shafer, M. W. Solid State Commun. 1976, 20, 369. (59) Reshak, A. H.; Auluck, S. Phys. Rev. B 2005, 71, No. 155114. (60) Helveg, S.; Lauritsen, J. V.; Lægsgaard, E.; Stensgaard, I.; Norskov, J. K.; Clausen, B. S.; Topsøe, H.; Besenbacher, F. Phys. Rev. Lett. 2000, 84, 951. (61) Liu, Z. P.; Hu, P. J. Am. Chem. Soc. 2001, 123, 12596. (62) Wyckoff, R. W. G. Crystal Structure, 2nd ed.; Interscience Publishers: New York, 1965. (63) Parker, L. J.; Atou, T.; Badding, J. V. Science 1996, 273, 95. (64) Atou, T.; Hasegawa, M.; Parker, L. J.; Badding, J. V. J. Am. Chem. Soc. 1996, 118, 12104. (65) Papageorgopoulos, C. A. Surf. Sci. 1978, 75, 17. (66) Tipler, P. A.; Llewellyn, R. A. Modern Physics, 3rd ed.; W. H. Freeman: New York, 1999. (67) Lamoen, D.; Persson, B. N. J. J. Chem. Phys. 1998, 108, 3332. € (68) Lou, L.; Osterlund, L.; Hellsing, B. J. Chem. Phys. 2000, 112, 4788. (69) Crowell, J. E.; Somorjai, G. A. Appl. Surf. Sci. 1984, 19, 73391. (70) Hagen, J. Industrial Catalysis: a Practical Approach; Wiley: New York, 1999. (71) Youchang, X.; Naasz, B. M.; Somorjai, G. A. Appl. Catal. 1986, 27, 233. (72) Sierraalta, A.; Herize, A.; A~ nez, R. J. Phys. Chem. A 2001, 105, 6519. (73) Lauritsen, J. V.; Kibsgaard, J.; Olesen, G. H.; Moses, P. G.; Hinnemann, B.; Helveg, S.; Nørskov, J. K.; Clausen, B. S.; Topsøe, H.; Lægsgaard, E.; Besenbacher, F. J. Catal. 2007, 249, 220. (74) Krebs, E.; Daudin, A.; Raybaud, P. Oil Gas Sci. Technol. 2009, 64, 1.

9040


Adsorption of Potassium on MoS2(100) Surface: First-Principles

Recommend Documents