Shape and Size of Cobalt Nanoislands Formed Spontaneously on

May 13, 2016 - The size of the nanoislands is determined by the balance between energy gain from creating C/CO-covered edges and energy penalty to ...
1 downloads 0 Views 5MB Size
Download PDF
Letter pubs.acs.org/JPCL

Shape and Size of Cobalt Nanoislands Formed Spontaneously on Cobalt Terraces during Fischer−Tropsch Synthesis Arghya Banerjee,† Violeta Navarro,‡ Joost W. M. Frenken,‡ Alexander P. van Bavel,§ Herman P. C. E. Kuipers,§ and Mark Saeys*,∥ †

Department of Chemical and Biomolecular Engineering, 4 Engineering Drive 4, National University of Singapore, Singapore 117576 Kamerlingh Onnes Laboratory, Leiden University, Niels Bohrweg 2, 2333CA Leiden, The Netherlands § Shell Technology Centre Amsterdam, P.O. Box 38000, 1030 BN Amsterdam, The Netherlands ∥ Laboratory for Chemical Technology, Ghent University, Technologiepark 914, 9052 Gent, Belgium ‡

S Supporting Information *

ABSTRACT: Cobalt-based catalysts undergo a massive and spontaneous reconstruction to form uniform triangular nanoislands under Fischer−Tropsch (FT) conditions. This reconstruction is driven by the unusual and synergistic adsorption of square-planar carbon and CO at the 4-fold edge sites of the nanoislands, driving the formation of triangular islands. The size of the nanoislands is determined by the balance between energy gain from creating C/CO-covered edges and energy penalty to create C/COcovered corners. For carbon chemical potentials corresponding to FT conditions, triangular Co islands with 45 Co atoms (about 2 nm) are the most stable surface structure. Decreasing the carbon chemical potential and hence the stability of squareplanar carbon favors the formation of larger islands, until reconstruction becomes unfavorable and CO-covered terraces are thermodynamically the most stable. The predicted structure of the islands is consistent with in situ scanning tunneling microscopy images obtained for the first time under realistic FT reaction conditions on a Co(0001) surface.

T

seconds or faster. The structure of the catalyst surface under reaction conditions is hence governed by the thermodynamic stability of the reconstructed surface rather than the dynamics of the reconstruction. To understand the factors that drive the formation of Co nanoislands under FT conditions and to determine the thermodynamically stable size and shape of these islands, we studied the thermodynamic stability of Co islands of various sizes and shapes using density functional theory (DFT). The predicted shapes and structures are compared with in situ STM images obtained at 3 bar, pH2:pCO = 2, 483 K on a Co(0001) surface with ReactorSTM, a unique STM specially designed for in situ catalytic studies.19 In this system, the STM tip is integrated inside a small flow reactor, with the hot model catalyst surface serving as one of the reactor walls. In order to meet industrial conditions, the reactor pressure can be increased up to several bars and the sample heated up to 623 K. The setup allows switching between high-pressure conditions and ultrahigh vacuum, which enables the combination of high-pressure experiments with more conventional techniques to prepare and characterize a clean and atomically smooth surface as a well-defined starting configuration for the experiments. The in situ STM studies bring us a step closer

he surface structure of heterogeneous catalysts under reaction conditions often differs dramatically from ideal clean surfaces. Such reconstructions, driven by the strong chemisorption of reactants and reaction intermediates, leads to the formation of the catalytically active sites only under reaction conditions.1 Strong adsorption of CO, oxygen and carbon has been observed to reconstruct transition metal catalyst surfaces.2−11 Early ex situ scanning tunneling microscopy (STM) studies by Wilson and de Groot revealed a remarkable reconstruction when a Co single crystal is exposed to syngas (a mixture of CO and H2) under Fischer−Tropsch (FT) conditions of 4 bar and 520 K. After about 1 h under reaction conditions, the sample was transferred to a UHV STM cell and the formation of rather uniform Co nanoislands with a diameter of about 2 nm and a height of one atomic layer was observed.8 The formation of these nanoislands results in the creation of “defect” or step sites, which are often linked to FT activity and selectivity.12−16 However, it is important to study catalysts under working conditions to obtain a realistic insight in the catalytic sites in action and during reaction.1 Nanoisland formation depends on the reaction conditions, and no nanoislands were observed by in situ STM at 10 mbar and 493 K.17,18 The reconstruction of the Co surface during FT synthesis happens on a time scale of several minutes to hours, while kinetically controlled reactions such as diffusion of Co-carbonyl surface species and surface reactions occur on a time scale of © XXXX American Chemical Society

Received: March 22, 2016 Accepted: May 13, 2016

1996

DOI: 10.1021/acs.jpclett.6b00555 J. Phys. Chem. Lett. 2016, 7, 1996−2001

Letter

The Journal of Physical Chemistry Letters

Figure 1. Creation of Co islands on Co terraces. (a) Procedure to compute the island creation energy. Co atoms dissociate from a step edge (I) to create a Co6 island (III). Note that the number of step, terrace and bulk Co atoms remains constant from (I) to (II). The island creation energy is therefore the energy to remove 6 Co atoms from the bulk and place a Co6 island on a terrace. (b) Island creation energies for two types of Co15 islands and for a hexagonal Co19 island. The island creation energy depends on the shape and the size of the islands. Energies for other sizes and shapes can be found in the Supporting Information (SI.1). (c) Types of edge sites for Co islands, and creation energies for each type of site. The site creation energies can be used to calculate island creation energies for islands of any size. Corner atoms are black, terrace atoms light gray, B5 atoms dark gray, and F4 atoms white.

Since carbon-containing species bind strongly to 4-fold edge sites,9,20−23 it is natural to ask whether the adsorption of C or CH species can stabilize those sites sufficiently to make edge creation thermodynamically favorable (Figure 2). The strong carbon binding energy at 4-fold B5 sites, −676 kJ/mol, leads to a stability of −18 kJ/mol relative to a CO, H2, H2O gas-phase reservoir under FT conditions (Supporting Information, SI.3). This value is not sufficient to overcome the edge site creation penalty of 45 kJ/mol. Carbon is unstable at 3-fold terrace and F4 sites under FT conditions. The unique and unusual stability of square-planar carbon was analyzed in detail by Nandula et al.,24 and attributed to the σ-aromaticity of the bonding in the Co4C subunit. The carbon coverage at the B5 sites is limited to 50% and adsorption of carbon beyond 50% is highly unfavorable. Carbon also binds strongly at 4-fold sites next to a B5 corner, but not at the 3-fold sites of the B5 corner. Since no carbon adsorbs strongly at the 3-fold corner site, the corner creation energy is not compensated by carbon adsorption. The carbon stability depends on the carbon chemical potential in the system (μC), and is a strong function of the temperature and the gas phase composition, as we discuss below and in the Supporting Information (SI.3 and SI.4). Another strongly binding surface species during FT synthesis is the reactant CO13,14,25 (Figure 2). Experimental and theoretical studies have reported CO coverages of around 50% during FT synthesis.26,27 On hexagonally closest packed Co terraces, two dominant configurations are found: a √3 × √3 structure (1/3 ML) and a 2√3 × 2√3 structure (7/12 ML) with adsorption free energies of −65 and −35 kJ/mol, respectively, under FT conditions. CO adsorbs strongly at B5 and F4 step sites and for a step coverage of 100%, the bridge site becomes the preferred adsorption site.28 CO also adsorbs strongly and with high coverage at corner sites. Calculations however, indicate that dicarbonyl species and step coverages beyond 100% are not stable at edge and corner sites at FT conditions (Figure S2 in the Supporting Information).

toward understanding the dynamic catalyst structure under reaction conditions. Conceptually, islands are created by removing atoms from a step edge to create an island on a terrace, as illustrated in Figure 1a. The energy penalty for this step, the clean island creation energy, was computed for islands of different sizes and shapes. The detailed procedure is discussed in the Supporting Information (SI.1). In short, the calculation can be divided into two steps: the removal of a row of step edge atoms to a Co bulk reference state, and the subsequent creation of a Co island from the bulk reference state. As we demonstrate in the Supporting Information (SI.1), the reaction energy for the first step is zero and the chemical potential of the Co atoms is determined by the Co bulk. Three types of islands were considered, depending on the nature of the edge sites (Figure 1b). B5 islands have edges consisting of 4-fold B5 sites,12 F4 islands have edges of 3-fold F4 sites, and hexagonal islands have both F4 and B5 edge sites. Breaking Co−Co bonds to remove an atom from the step edge and create an island is unfavorable, and the energy penalty to create a Co15 B5 island, 608 kJ/mol, or 40.5 kJ/mol Coisland is indeed substantial. The island creation penalty is lower for F4 islands than for B5 islands, and F4 islands are more stable than B5 islands of the same size, even though the number of Co−Co bonds is identical. Since the island creation energy increases linearly with the number of edge sites, the island creation energy can be computed from the number and type of edge and corner sites (Figures 1c and SI.1). Consistent with the lower island creation energy for F4 islands, the edge site creation energy for a 3-fold F4 site is lower than for a 4-fold B5 site. Using the step creation energies in Figure 1c, the energy cost to create a 3 nm B5 Co66 island is 3 × 22 + 30 × 45 = 1416 or 21.5 kJ/mol Coislands. Note that the number of B5 sites (10 per edge) includes the B5 sites next to the B5 corner, and is larger than the number of dark gray B5 atoms. This definition leads to somewhat low values for the corner creation energies (Figure 1c). 1997

DOI: 10.1021/acs.jpclett.6b00555 J. Phys. Chem. Lett. 2016, 7, 1996−2001

Letter

The Journal of Physical Chemistry Letters

triangular B5 islands under FT conditions. To illustrate the procedure, we compute the stability of a triangular Co45 island with C/CO-covered 4-fold B5 edge sites and a terrace CO coverage of 1/3 ML. To create an island, CO molecules need to desorb from the Co terrace sites underneath the island. The footprint of the Co45 island is shown by the white triangle in Figure 3a and contains 66 Co atoms. The creation of a clean Co45 island adds 1146 kJ/mol to the CO desorption penalty of 1430 kJ/mol. The island is stabilized by the synergistic adsorption of 50% carbon and 100% CO at the B5 edge and corner sites, −2418 kJ/mol. Adsorption of 7 CO molecules on the central 21 terrace atoms of the Co45 island contributes 7 × −65 kJ/mol = −455 kJ/mol. Taken together, 297 or 6.6 kJ/mol Coisland is gained when a Co45 island is created on a CO-covered terrace under FT conditions. The stability of the C/COcovered B5 edge sites is expected to drive the formation of even smaller islands. However, two factors limit the stability of smaller islands, as illustrated for Co28. For this size, the energy cost to create the corners becomes a substantial fraction of the energy balance, and this energy cost is not compensated by the adsorption of C/CO. Moreover, the number of CO molecules on the central terrace of Co28 is only 3, and the small size of this island reduces the CO stability from −65 to −47 kJ/mol (SI.2). Co atoms are therefore less stable in Co28 islands than in Co45 islands under FT conditions. Larger islands are less stable because they gain less from the B5 edge stabilization by C/CO, as illustrated for Co66 in Figure 3a and in SI. 5 for larger islands. A plot of the island stability as a function of size is shown in Figure S3 in the Supporting Information. In the limit of very large islands, CO adsorption in the island center compensates CO desorption from the terrace, and the island stability is determined by the B5 edges. Under FT conditions, the creation of a B5 edge site is favorable by 16 kJ/mol, a value comparable with the difference in surface energy between Co(111) and Co(100), 20 kJ/mol. Assuming that the carbon chemical potential is determined by the CO + H2 + * ↔ C* + H2O reaction, we find that the most stable islands under FT conditions are triangular and contain 45 Co atoms (Figure 3a). Note that the stabilities of the islands reported in Figure 3a are normalized per Co atom in the island. The free energy gained when a Co66 island breaks up to form Co45 islands, C/COcovered Co66 + 66/5 CO(g) + 117/45 H2(g) → 66/45 C/COcovered Co45 + 117/45 H2O(g), is quite substantial at −26 kJ/ mol. The formation of a CO-covered surface carbide is thermodynamically favorable under FT conditions,28 but significantly less favorable than the creation of the cobalt nanoislands described here because CO adsorption on the p4g carbide is very weak.28 The predicted size of ∼2 nm agrees well with the size of the islands observed by Wilson and De Groot using ex situ STM on a Co(0001) catalyst after exposure to reaction conditions.8 It is, however, likely that carbon is not fully in equilibrium with CO, and the carbon stability and coverage (or the chemical potential) could be lower than expected. For example, a high CO dissociation barrier at the step edges would lead to C coverages lower than those predicted from the quasi-equilibrium assumption. The effect of this factor will be discussed below. Our in situ STM observations of a flat Co(0001) surface also show the formation of triangular islands with a monolayer height under reaction conditions at 3 bar, pH2:pCO = 2, 483 K. The observed islands, however, show a broader size distribution, which depends on the time of exposure to reaction conditions. Smaller 2 nm islands are very mobile at the reaction conditions, but appear

Figure 2. Average thermodynamic stability (kJ/mol) of carbon, CH and CO at the different sites in Figure 1c, relative to a gas phase reservoir at typical FT conditions (20 atm, 500 K, pH2:pCO = 2; 60% conversion). Note that the stabilities depend strongly on the reaction conditions (Figure 3). Additional structures and coverages can be found in the Supporting Information (SI.2). #Average CO stability on a 50% CHx-covered B5 site.

Interestingly, the presence of square-planar carbon at the B5 site strengthens adsorption of CO at this site by 10 kJ/mol. This is remarkable since the Co atoms now bind both C and CO. The enhanced stability of CO is caused by the oxidation of the edge Co atoms by square-planar carbon.29 The reduced electron density at the Co atom strengthens CO adsorption by reducing Pauli repulsion with the 5σ HOMO of CO.30 A similar effect is found at the B5 corners. In the most stable configuration, one of the two B5 corner sites is occupied by square-planar carbon and three CO molecules adsorb at the corner (Figure 2). Additional, less stable, configurations can be found in the Supporting Information (SI.2). A similar attractive interaction is not found for CO/CH coadsorption. In the presence of CO, the hydrogenation of square-planar carbon at the B5 site is therefore unfavorable by 12−14 kJ/mol. The island creation energies in Figure 1 and the C/CO stabilities in Figure 2 can be combined to evaluate the thermodynamic stability of C/CO-covered Co islands of various sizes and shapes as a function of the reaction conditions (Figure 3). The strong and synergistic adsorption of square-planar carbon and CO at the B5 sites stabilizes the formation of 1998

DOI: 10.1021/acs.jpclett.6b00555 J. Phys. Chem. Lett. 2016, 7, 1996−2001

Letter

The Journal of Physical Chemistry Letters

Figure 3. Creation of C/CO-covered Co islands (a) Thermodynamic analysis (kJ/mol) for the creation of Co28, Co45, and Co66 islands on COcovered terraces, using the values in Figure 1 and 2 and under FT conditions. Triangular Co45 islands are the most stable under FT conditions. (b) Effect of the carbon chemical potential (μC) and the CO chemical potential (μCO) on the size of the most stable islands. The (0,0) point corresponds to standard FT conditions (20 bar, 500 K, pH2/pCO = 2, 60% conversion). μC is calculated for the reaction CO + H2+ *↔ C* + H2O, and depends on the total pressure, the temperature, the pH2/pCO ratio, and the conversion. Co45 B5 islands are the most stable when μC is high and square-planar carbon is highly stable. As the stability of square-planar carbon decreases, the most stable islands become larger, and eventually island formation becomes unfavorable. The dotted line indicates the conditions where infinitely large B5 islands become more stable than CO-covered terraces. At low μC, square-planar carbon is no longer stable, and the island stability depends only on the CO chemical potential. Several experimental conditions are indicated, for a CO conversion of 60%.

from early stages (Figure 4a,b and Figure S4 in the Supporting Information for an overview image). This is consistent with the creation and removal of square-planar carbon species at the edge sites, suggesting quasi-equilibrated square-planar carbon

formation at the steps. Over time, larger islands also develop, such as the one in Figure 4c, possibly due to coalescence of smaller ones. The smallest islands resolved in the STM topographic images are 2.0 ± 0.7 nm, Figure 4a,b and S4, and agree with the theoretical size. These small islands are, however, highly mobile and only visible for 2 or 3 frames during STM imaging, i.e., for approximately 2 min. The larger islands are less mobile, and they do not disappear nor seem to be altered during imaging and are hence more easily observed at the high temperature of the reaction conditions. Wilson and De Groot also observed extensive island structures in their ex situ observations of the Co(0001) crystal after 2 h of exposure to reaction conditions. At their imaging conditions, the mobility of the nanoislands is greatly reduced, making them more easily observable. Furthermore, it should be noted that vacancy islands of similar sizes are observed on the surface, and they appear to be less mobile than islands of similar sizes. A completely analogous thermodynamic analysis can be developed for such vacancy islands. The stability of carbon and CO depends on the reaction conditions and can be described by their chemical potential (Supporting Information, SI.3). The island stability calculations in Figure 3a can be repeated for a range of carbon and CO chemical potentials (Supporting Information, SI.4) and the resulting stability diagram is shown in Figure 3b. Increasing the carbon chemical potential (and hence the stability of squareplanar carbon) or the CO chemical potential, favors the formation of smaller islands, as expected. Reducing the carbon and CO stability favors larger islands, and eventually islands become unstable. At sufficiently low carbon chemical potentials, the formation of square-planar carbon at the B5 edges becomes

Figure 4. (a) and (c) STM topographic images of triangular islands formed on a Co(0001) surface under reaction conditions, 3 bar, pH2:pCO = 2, 483 K. (b) Line scan along the red dashed line in panel a. I = 400 pA; V = 0.4 V. 1999

DOI: 10.1021/acs.jpclett.6b00555 J. Phys. Chem. Lett. 2016, 7, 1996−2001

Letter

The Journal of Physical Chemistry Letters unfavorable (ΔμC ∼ −0.2 eV). For sufficiently high CO chemical potentials, the stability and the high coverage of CO at the edge sites becomes sufficient to stabilize Co islands without the presence of square planar carbon, as was also observed on Pt surfaces.3 Without carbon, CO-covered 3-fold F4 edge sites are more stable than CO-covered 4-fold B5 sites (Figure 1 and 2), and the most stable islands are terminated by 3-fold F4 edge sites. Note that the 45 atom B5 and F4 islands are stable for a large range of the C and CO chemical potentials. As the chemical potentials decrease, the stabilization by carbon adsorption reduces and larger islands become more stable than smaller islands, as illustrated in SI.4, until island creation is no longer favorable and CO-covered terraces are the most stable phase. The latter transition is indicated by the “infinite islands” line. More detailed stability calculations can be found in the Supporting Information (SI.3 and SI.4). As indicated, the carbon and CO chemical potentials depend on the reaction conditions, and several points are indicated for a pH2:pCO ratio of 2 and a conversion of 60%. Under FT conditions and assuming that the carbon chemical potential is determined by the CO + H2 + * ↔ C* + H2O reaction, 2 nm Co45 islands are thermodynamically the most stable. Reducing the pressure to 300 mbar and to 10 mbar reduces the CO and C chemical potentials. Under these conditions, island formation is calculated to be unfavorable, in line with in situ STM images obtained at those conditions.17 Reducing the temperature increases the chemical potentials and would favor smaller islands. However, at lower temperatures the reaction kinetics are likely too slow to equilibrate the carbon forming reaction that defines the chemical potential. In conclusion, the thermodynamic stability, shape, and size of Co nanoislands were studied using DFT and compared with in situ STM images at reaction conditions. Our thermodynamic analysis shows that the synergistic adsorption and high stability of square-planar carbon and CO at 4-fold B5 edge sites stabilizes triangular, 2 nm Co45 islands under FT conditions. The cost of creating corners and the reduced CO adsorption energy limit the stability of even smaller islands. The stability of carbon and CO depends on the reaction conditions and the carbon and CO chemical potentials determine the shape and size of the most stable islands. In situ STM images of a Co(0001) surface under realistic reaction conditions provide experimental evidence for the formation of nanoislands. Under reaction conditions, 2 nm islands are highly mobile and difficult to observe. Therefore, somewhat larger triangular islands dominate the in situ STM images. The STM images for the first time illustrate the restructuring of the Co surface under realistic FT reaction conditions.

CO(g) + (x /2 + 1)H 2(g) + * ↔ CHx* + H 2O(g)

(1)

ΔGCHx (T , p) = ΔG 0(500 K) + RT ln(pH2O /pCO pH2 x /2 + 1 ) (2)

The Gibbs free energy for gas phase and adsorbed species was calculated by combining the electronic and zero point vdW-DF energies with standard enthalpy and entropy corrections obtained from frequency calculation for the full structure. The Co surface and islands were modeled using a 2layer, p(6 × 6) Co(111) slab with an optimized Co lattice constant of 3.56 Å, and an interslab spacing of 10 Å. The Brillouin zone was sampled with a (3 × 3 × 1) Monkhorst− Pack grid. The effect of slab thickness was considered in detail, (SI.1) but was found to have only a minor effect on the optimal shape and size of the nanoislands. Experimental Details. Experiments were performed on flat Co single crystals, Co(0001), manufactured at Surface Preparation Lab. Samples were prepared by cycles of sputtering with Ar ions and annealing under UHV at 620 K. The annealing temperature needs to be below the first order martensitic phase transition of Co37 (at 695 K) to prevent the formation of bulk defects such as dislocations.38 For one-third of the cycles, a 1 min long pulse of oxygen was applied during annealing to remove carbon contamination on the surface. Prior to experiments, the quality and composition of the Co(0001) surface were checked by LEED and Auger spectroscopy. Before exposure to the reactive gases, the sample was stabilized at the reaction temperature, 483 K, under a flow of a 2:1 mixture of H2 and Ar at 3 bar and imaged with the STM to verify the surface quality. After 2 h, the reactor environment was changed to the reaction conditions (3 bar of a mixture of H2 and CO, pH2:pCO = 2, 483 K) to perform the experiments. The image analysis was carried out with WSXM software.39

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpclett.6b00555. (1) Procedure to calculate clean island and step/corner site creation energies; (2) thermodynamic stability of CHx and CO species at step and corner sites; (3) procedure to calculate the C (ΔμC) and CO (ΔμCO) chemical potential for a range of conditions; (4) thermodynamic stability of various islands for a range of (μC, μCO) conditions. (5) Thermodynamic stability of different size islands on CO-covered terraces under FT conditions. (6) Overview STM image, showing the formation of multiple Co nanoislands under FT conditions (PDF)

■

METHODS Computational Details. Periodic spin-polarized DFT calculations were performed using the vdW-DF functional,31,32 a plane-wave basis set with a cutoff kinetic energy of 450 eV, and the projector-augmented wave method as implemented in the Vienna Ab-initio Simulation Package (VASP).33,34 Previous studies indicate that this approach provides an accurate description of CO adsorption on Co(0001),25 on Pt(111),35 of carbon adsorption on Co,28 of CHx, OH, and CH3O adsorption on Pt(111).36 The thermodynamic stability of CO, C, and CH adsorbates relative to a CO, H2 and H2O gas phase reservoir was evaluated from the Gibbs free reaction energy for (1) under FT conditions (500 K, 20 bar, 60% conversion).

■

AUTHOR INFORMATION

Corresponding Author

*E-mail address: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by Shell Global Solutions and a Faculty Strategic Funding initiative from the National University of Singapore. V.N. would like to thank the Netherlands Organisation for Scientific Research (NWO) and 2000

DOI: 10.1021/acs.jpclett.6b00555 J. Phys. Chem. Lett. 2016, 7, 1996−2001

Letter

The Journal of Physical Chemistry Letters the Technology Foundation (STW) for the financial support as part of the research program Veni with project number 11920.

■

pressure, high-temperature catalytic reaction conditions. Rev. Sci. Instrum. 2014, 85, 083703−083710. (20) Tan, K.; Xu, J.; Chang, J.; Borgna, A.; Saeys, M. Carbon deposition on Co catalysts during Fischer−Tropsch synthesis: A computational and experimental study. J. Catal. 2010, 274, 121−129. (21) Gong, X. Q.; Raval, R.; Hu, P. CO dissociation and O removal on Co(0 0 0 1): A density functional theory study. Surf. Sci. 2004, 562, 247−256. (22) Gong, X. Q.; Raval, R.; Hu, P. CHx hydrogenation on Co(0001): A density functional theory study. J. Chem. Phys. 2005, 122, 024711. (23) Weststrate, C. J.; Ciobica, I. M.; Saib, A. M.; Moodley, D. J.; Niemantsverdriet, J. W. Fundamental issues on practical Fischer− Tropsch catalysts: How surface science can help. Catal. Today 2014, 228, 106−112. (24) Nandula, A.; Trinh, Q. T.; Saeys, M.; Alexandrova, A. N. Origin of extraordinary stability of square-planar carbon atoms in surface carbides of cobalt and nickel. Angew. Chem., Int. Ed. 2015, 54, 5312− 5316. (25) Gunasooriya, G. T. K. K.; van Bavel, A. P.; Kuipers, H. P. C. E.; Saeys, M. CO adsorption on cobalt: Prediction of stable surface phases. Surf. Sci. 2015, 642, L6−L10. (26) Den Breejen, J. P.; Radstake, P. B.; Bezemer, G. L.; Bitter, J. H.; Frøseth, V.; Holmen, A.; De Jong, K. P. On the origin of the cobalt particle size effects in Fischer−Tropsch catalysis. J. Am. Chem. Soc. 2009, 131, 7197−7203. (27) Zhuo, M.; Borgna, A.; Saeys, M. Effect of the CO coverage on the Fischer−Tropsch synthesis mechanism on cobalt catalysts. J. Catal. 2013, 297, 217−226. (28) Banerjee, A.; van Bavel, A. P.; Kuipers, H. P. C. E.; Saeys, M. Origin of the formation of nanoislands on cobalt catalysts during Fischer−Tropsch synthesis. ACS Catal. 2015, 5, 4756−4760. (29) Tsakoumis, N. E.; Voronov, A.; Ronning, M.; van Beek, W.; Borg, O.; Rytter, E.; Holmen, A. Fischer−Tropsch synthesis: An XAS/ XRPD combined in situ study from catalyst activation to deactivation. J. Catal. 2012, 291, 138−148. (30) Banerjee, A. Structure of cobalt catalysts during Fischer− Tropsch synthesis: Role of adsorbates and mechanistic consequences. Ph.D. Thesis, National University of Singapore, Singapore, 2016. (31) Dion, M.; Rydberg, H.; Schröder, E.; Langreth, D. C.; Lundqvist, B. I. Van der Waals density functional for general geometries. Phys. Rev. Lett. 2004, 92, 246401. (32) Klimeš, J.; Bowler, D. R.; Michaelides, A. Van der Waals density functionals applied to solids. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 195131. (33) Kresse, G.; Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 1996, 6, 15−50. (34) Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. (35) Lazic, P.; Alaei, M.; Atodiresei, N.; Caciuc, V.; Brako, R.; Blugel, S. Density functional theory with nonlocal correlation: A key to the solution of the CO adsorption puzzle. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81, 045401. (36) Lew, W.; Crowe, M. C.; Campbell, C. T.; Carrasco, J.; Michaelides, A. The energy of hydroxyl coadsorbed with water on Pt(111). J. Phys. Chem. C 2011, 115, 23008−23012. (37) Betteridge, W. Properties of metallic cobalt. Prog. Mater. Sci. 1980, 24, 51−192. (38) Erbudak, M.; Wetli, E.; Hochstrasser, M.; Pescia, D.; Vvedensky, D. D. Surface phase transitions during martensitic transformations of single-crystal Co. Phys. Rev. Lett. 1997, 79, 1893−1896. (39) Horcas, I.; Fernandez, R.; Gomez-Rodriguez, J. M.; Colchero, J.; Gomez-Herrero, J.; Baro, A. M. WSXM: A software for scanning probe microscopy and a tool for nanotechnology. Rev. Sci. Instrum. 2007, 78, 013705.

REFERENCES

(1) Schulz, H. Major and minor reactions in Fischer−Tropsch synthesis on cobalt catalysts. Top. Catal. 2003, 26, 73−85. (2) Tao, F.; Dag, S.; Wang, L. W.; Liu, Z.; Butcher, D. R.; Salmeron, M.; Somorjai, G. A. Restructuring of hex-Pt(100) under CO gas environments: Formation of 2-D nanoclusters. Nano Lett. 2009, 9, 2167−2171. (3) Tao, F.; Dag, S.; Wang, L. W.; Liu, Z.; Butcher, D. R.; Bluhm, H.; Salmeron, M.; Somorjai, G. A. Break-up of stepped platinum catalyst surfaces by high CO coverage. Science 2010, 327, 850−852. (4) Hansen, P. L.; Wagner, J. B.; Helveg, S.; Rostrup-Nielsen, J. R.; Clausen, B. S.; Topsøe, H. Atom-resolved imaging of dynamic shape changes in supported copper nanocrystals. Science 2002, 295, 2053− 2055. (5) Yoshida, H.; Kuwauchi, Y.; Jinschek, J. R.; Sun, K.; Tanaka, S.; Kohyama, M.; Shimada, S.; Haruta, M.; Takeda, S. Visualizing gas molecules interacting with supported nanoparticulate catalysts at reaction conditions. Science 2012, 335, 317−319. (6) Eren, B.; Zherebetskyy, D.; Patera, L. L.; Wu, C. H.; Bluhm, H.; Africh, C.; Wang, L. W.; Somorjai, G. A.; Salmeron, M. Activation of Cu(111) surface by decomposition into nanoclusters driven by CO adsorption. Science 2016, 351, 475−478. (7) Vendelbo, S. B.; Elkjaer, C. F.; Falsig, H.; Puspitasari, I.; Dona, P.; Mele, L.; Morana, B.; Nelissen, B. J.; van Rijn, R.; Creemer, J. F.; Kooyman, P. J.; Helveg, S. Visualization of oscillatory behaviour of Pt nanoparticles catalysing CO oxidation. Nat. Mater. 2014, 13, 884−890. (8) Wilson, J.; De Groot, C. Atomic-scale restructuring in highpressure catalysis. J. Phys. Chem. 1995, 99, 7860−7866. (9) Ciobîcǎ, I. M.; van Santen, R. A.; van Berge, P. J.; van de Loosdrecht, J. Adsorbate induced reconstruction of cobalt surfaces. Surf. Sci. 2008, 602, 17−27. (10) Zhang, X.-Q.; van Santen, R. A.; Hensen, E. J. M. Carboninduced surface transformations of cobalt. ACS Catal. 2015, 5, 596− 601. (11) Corral Valero, M.; Raybaud, P. Stability of carbon on cobalt surfaces in Fischer−Tropsch reaction conditions: A DFT Study. J. Phys. Chem. C 2014, 118, 22479−22490. (12) Van Hardeveld, R.; Hartog, F. The statistics of surface atoms and surface sites on metal crystals. Surf. Sci. 1969, 15, 189−230. (13) Beitel, G. A.; Laskov, A.; Oosterbeek, H.; Kuipers, E. W. Polarization modulation infrared reflection absorption spectroscopy of CO adsorption on Co(0001) under a high-pressure regime. J. Phys. Chem. 1996, 100, 12494−12502. (14) Beitel, G. A.; De Groot, C. P. M.; Oosterbeek, H.; Wilson, J. H. A combined in-situ PM-RAIRS and kinetic study of single-crystal cobalt catalysts under synthesis gas at pressures up to 300 mbar. J. Phys. Chem. B 1997, 101, 4035−4043. (15) Shetty, S.; Van Santen, R. A. CO dissociation on Ru and Co surfaces: The initial step in the Fischer−Tropsch synthesis. Catal. Today 2011, 171, 168−173. (16) Kruse, N.; Schweicher, J.; Bundhoo, A.; Frennet, A.; Visart de Bocarme, T. Catalytic CO hydrogenation: Mechanism and kinetics from chemical transients at low and atmospheric pressures. Top. Catal. 2008, 48, 145−152. (17) Ehrensperger, M.; Wintterlin, J. In situ high-pressure hightemperature scanning tunneling microscopy of a Co(0001) Fischer− Tropsch model catalyst. J. Catal. 2014, 319, 274−282. (18) Ehrensperger, M.; Wintterlin, J. In situ scanning tunneling microscopy of the poisoning of a Co(0001) Fischer−Tropsch model catalyst by sulfur. J. Catal. 2015, 329, 49−56. (19) Herbschleb, C. T.; van der Tuijn, P. C.; Roobol, S. B.; Navarro, V.; Bakker, J. W.; Liu, Q.; Stoltz, D.; Canas-Ventura, M. E.; Verdoes, G.; van Spronsen, M. A.; Bergman, M.; Crama, L.; Taminiau, I.; Ofitserov, A.; van Baarle, G. J. C.; Frenken, J. W. M. The ReactorSTM: Atomically resolved scanning tunneling microscopy under high2001

DOI: 10.1021/acs.jpclett.6b00555 J. Phys. Chem. Lett. 2016, 7, 1996−2001