DOI: 10.1002/cphc.201500097
Communications
Linear Alkane C¢C Bond Chemistry Mediated by Metal Surfaces Zeying Cai,[a] Meizhuang Liu,[a] Limin She,[a] Xiaoli Li,[a] Jason Lee,[a] Dao-Xin Yao,[a] Haiming Zhang,[b] Lifeng Chi,[b, c] Harald Fuchs,[c] and Dingyong Zhong*[a] Linear alkanes undergo different C¢C bond chemistry (coupling or dissociation) thermally activated on anisotropic metal surfaces depending on the choice of the substrate material. Owing to the one-dimensional geometrical constraint, selective dehydrogenation and C¢C coupling (polymerization) of linear alkanes take place on Au(110) surfaces with missing-row reconstruction. However, the case is dramatically different on Pt(110) surfaces, which exhibit similar reconstruction as Au(110). Instead of dehydrogenative polymerization, alkanes tend to dehydrogenative pyrolysis, resulting in hydrocarbon fragments. Density functional theory calculations reveal that dehydrogenation of alkanes on Au(110) surfaces is an endothermic process, but further C¢C coupling between alkyl intermediates is exothermic. On the contrary, due to the much stronger C¢Pt bonds, dehydrogenation on Pt(110) surfaces is energetically favorable, resulting in multiple hydrogen loss followed by C¢C bond dissociation.
In order to utilize alkanes as a feedstock in chemical production, selective C¢H and C¢C bond activation is required.[1] In the past decades, examples of alkane C¢H and C¢C bond activation on transition metal surfaces have been intensively reported.[2] However, in such heterogeneous catalytic approaches, the molecules may undergo various competing processes simultaneously, resulting in poor selectivity. Unveiling surface-mediated chemical processes of alkanes on the atomic scale is of fundamental importance for achieving high selectivity and realizing efficient utilization.[3] Recently, linear alkane polymerization through the terminal/penultimate C¢H bond [a] Z. Cai,+ M. Liu,+ Dr. L. She, X. Li, J. Lee, Prof. Dr. D.-X. Yao, Prof. Dr. D. Zhong School of Physics and Engineering and State Key Laboratory of Optoelectronic Materials and Technologies Sun Yat-sen University Xingang Xi Road 135, 510275 Guangzhou (China) E-mail: [email protected] [b] Dr. H. Zhang, Prof. Dr. L. Chi Institute of Functional Nano and Soft Materials (FUNSOM) and Collaborative Innovation Center of Suzhou Science and Technology Soochow University 199 Ren-Ai Road, Suzhou, Jiangsu 215123 (China) [c] Prof. Dr. L. Chi, Prof. Dr. H. Fuchs Physikalisches Institut Westflische Wilhelms-Universitt Wilhelm-Klemm-Straße 10, 48149 Mìnster (Germany) [+] These authors contributed equally to the work Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cphc.201500097.
ChemPhysChem 2015, 16, 1356 – 1360
activation and direct C¢C coupling has been achieved on anisotropic Au(110) surfaces.[4] The missing-row reconstructions at the Au(110) surfaces result in parallel (1 Õ 2) grooves with atomically scaled width and depth. The reactant molecules are located in the groove and their orientation and diffusion are efficiently constrained in a one-dimensional (1D) manner. Although gold has long been considered to have poor catalytic reactivity,[5] in the last decades, nanostructured gold has been found to exhibit catalytic effect for certain processes.[6] Compared with gold, platinum has been considered as one of the most versatile heterogeneous catalysts.[7] A systematic study of Pt–Au alloys as catalysts for the conversion of nhexane has revealed that with increasing the gold concentration at the surface, the isomerization rate increases substantially, while the aromatization and hydrogenolysis rates decrease exponentially.[8] This remarkable change indicates that platinum is more reactive for C¢H and C¢C bond activation than gold. Given the eminent catalytic performance of platinum, one may ask whether the Pt(110) surface, which possesses similar missing-row reconstruction with atomic grooves as Au(110), is better than Au(110) for on-surface linear alkane polymerization. Here, we report a comparison study on the behaviors of linear alkanes adsorbed on Au(110) and Pt(110) surfaces. With the combination of scanning tunneling microscopy (STM) and density functional theory (DFT) calculations, the kinetic and dynamic analysis of the on-surface chemistry of linear alkanes on Pt(110) and Au(110) surfaces has been investigated on the atomic scale. Instead of selective dehydrogenation and C¢C coupling on Au(110), multiple C¢H bond cleavage with the formation of C¢Pt bonds took place on Pt(110). Due to the much stronger C¢Pt bonds, the diffusion of dehydrogenated intermediates along the 1D-constrained (1 Õ 2) grooves was dramatically hindered and subsequent C¢C cleavage is more energetically favorable. As a case study this work provides a microscopic insight into surface-mediated C¢C bond chemistry of linear alkanes. Linear alkane molecules (C32H66) were deposited on Au(110) surfaces at room temperature (~ 300 K). Figure 1 a shows the typical STM image of samples with a submonolayer coverage (about 0.2 monolayer) measured at 78 K. The molecules were adsorbed in the (1 Õ 2) reconstruction grooves and diffused rapidly along the [1–10] direction even at such low temperature. With a sample bias of ¢3.5 V, no molecule was visible in the STM image. When the sample bias was changed to ¢1.0 V with the STM tip closer to the surface, some bright features appeared near the surface defects, indicating an increased resi-
1356
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
Figure 1. STM images of n-dotriacontane (C32H66) on Au(110) and Pt(110), before and after annealing. a) Submonolayer coverage of C32H66 deposited on Au(110)-(1 Õ 2) at 300 K (U = ¢3.5 V; I = 100 pA). C32H66 molecules are invisible by STM due to the diffusion. Inset, monolayer coverage, the dashed black lines show the centerlines of the uppermost gold rows. The inset includes a diagram showing the cross section of the molecule–surface interface. b) After annealing at 490 K, polymeric chains formed in the Au(110)(1 Õ 3) reconstruction grooves (U = ¢0.02 V; I = 2 nA). c) Submonolayer coverage of C32H66 deposited on Pt(110)-(1 Õ 2) at 300 K (U = ¢0.05 V; I = 1 nA). C¢ C bond cleavage took place, resulting in shorter fragments. d) As-grown at 80 K (U = ¢2.1 V; I = 100 pA). Some molecules switched between the two equivalent adsorption sites.
dence time of molecules near these sites, probably due to the strengthened tip–molecule interaction (Figure S1).[9] At a monolayer coverage, all molecules, which exhibited a straight stick shape with a length of ~ 4 nm, were adsorbed with the long axis along the (1 Õ 2) reconstructed grooves, that is, the [1–10] direction. There were five molecules packed in every three reconstruction units in the [001] direction, resulting in an averaged intermolecular distance of ~ 0.5 nm. (inset of Figure 1 a). This indicates that the parallel atomic grooves are an efficient template to constrain the orientation of the molecules. After annealing the sample at 490 K for 10 min, polymeric chains were observed in wider grooves (~ 1.2 nm) corresponding to missing-row (1 Õ 3) reconstruction,[4] as shown in Figure 1 b. Similarly, C32H66 molecules were deposited on Pt(110)-(1 Õ 2) surfaces at 300 K. However, this case is dramatically different from that on Au(110) (Figure 1 c). The molecules were adsorbed on the (1 Õ 2) reconstructed uppermost platinum rows and shorter fragments were formed, implying the occurrence of C¢C bond cleavage. To investigate the temperature effect, we deposited C32H66 molecules on Pt(110) at 80 K followed by annealing at different temperatures up to 300 K. On the asgrown sample, uniform and straight features were observed with a length of ~ 4 nm, corresponding to intact C32H66 molecules (Figure 1 d and Figure 2 a). They adsorbed in the grooves with the long axis along the [1–10] direction, implying much ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
Figure 2. STM images of n-dotriacontane (C32H66) on Pt(110) at different temperatures. a) As-grown at 80 K (U = ¢3 V; I = 30 pA). Intact C32H66 molecules adsorbed in the grooves with the long axis along the [1–10] direction. b) Annealing to 200 K (U = ¢0.03 V; I = 1.3 nA). Some molecules changed the adsorption configuration with one terminal adsorbed from the reconstruction grooves to the neighboring uppermost platinum rows, while the remaining still adsorbed in the grooves. c) Annealing to 225 K (U = ¢0.01 V; I = 2 nA). Most molecules underwent terminal dehydrogenation. d) After annealing at 250 K (U = ¢0.01 V; I = 1 nA). Molecules entirely transferred their adsorbed position from the grooves to the ridges. Darker features are marked by white arrows corresponding to the positions with H atoms dissociated. e) Further annealing to 300 K (U = ¢0.05 V; I = 1 nA). C¢C bond cleavage took place and shorter fragments were formed. In all figures, the dashed black lines show the centerlines of the uppermost platinum rows.
less diffusion ability on Pt(110) compared to Au(110). Nevertheless, diffusion along the grooves were occasionally observed (see Figure S2), indicating that no strong chemical bond is formed between the molecules and the surface. There are two equivalent adsorption sites located on the gradient {111} microfacets of the grooves. Due to rapid switching between the two adsorption sites, most molecules appear blurred in the STM images (Figure S3). After annealing at a temperature of 200 K for 90 min, some molecules changed their adsorption configuration with one terminal site adsorbed from the reconstruction grooves to the neighboring uppermost platinum row, while the other end remained adsorbed in the groove (Figure 2 b). When the sample was annealed to 225 K for 90 min, most molecules underwent such terminal transformation and some of them adsorbed with both terminals climbing up on the uppermost rows (Figure 2 c). After further annealing to 250 K for 90 min, the molecules were completely immobilized and anchored at the surface and changed their adsorbed position from the grooves to the ridges (Figure 2 d). They exhibited a tortuous shape and non-uniform contrast, with some parts brighter and some dimmer. The above observation indicates that one or more C¢ H bonds of each molecule were dissociated followed by the formation of C¢Pt bonds, resulting in stronger molecule–surface interaction. When increasing the annealing temperature to 300 K for 90 min, C¢C bond cleavage took place with the formation of shorter fragments (Figure 2 e). This is different from a previous study on the ethane/Pt(110)-(1 Õ 2) system,[2j] in which no C¢C bond cleavage was observed at 440 K. Our experiments demonstrate that the behavior of linear C32H66 molecules on Pt(110) is totally different from the case on Au(110). On Au(110), alkane molecules were aligned and rapidly diffused along the 1D grooves with a low energy barrier. Such a 1D constraint would increase the probability for neighboring molecules to meet end to end thus achieve C¢C cou-
1357
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications pling only at the terminal/penultimate sites. On the other hand, although the 1D constraint would still affect the molecular orientation and diffusion on Pt(110), multi-dehydrogenation followed by C¢C bond cleavage is the preferential pathway. Furthermore, the C¢H activation temperature on Pt(110) is much lower than on Au(110), in which the electronic effect of the metal surfaces plays a key role.[2c] Figure 3. Calculated adsorption configurations and diffusion barriers of n-hexane and n-hexyl intermediate on Density functional theory Au(110)-(1 Õ 2) and Pt(110)-(1 Õ 2). a),b) n-hexane and n-hexyl on Au(110)-(1 Õ 2), respectively. c) n-hexane on (DFT) calculations were carried Pt(110)-(1 Õ 2). Here, n-hexane tends to adsorb to the {111} microfacets. d) n-hexyl on Pt(110)-(1 Õ 2). A C¢Pt bond out to gain an insight into the is formed between the terminal CH2 and Pt atom from the upmost rows. Inset, top view. Adsorption energies are reaction mechanism. For simplic- listed in (a)–(d). e) Calculated energy diagrams of diffusion barriers along [1–10] (1D grooves) for n-hexane and hexyl. ity, we chose n-hexane (C6H14) as the model molecule in our simuintact n-hexane, the barriers for the n-hexyl significantly inlations. Metal surfaces were modeled by a slab consisting of crease, 1.02 eV on Pt(110)-(1 Õ 2) and 0.66 eV on Au(110)-(1 Õ 2). six atomic layers (for details, see the Supporting Information). The high diffusion barrier on Pt(110) significantly decreases the We first optimized the adsorption configurations for a single ncollision probability between the dehydrogenated species and hexane molecule and n-hexyl intermediate (one H atom dissotherefore limits the probability of C¢C bond coupling taking ciated from the terminal methyl group of n-hexane and adplace. sorbed at a bridge site) adsorbed on the Au(110)-(1 Õ 2) and To obtain a thermodynamic view on the surface-mediated Pt(110)-(1 Õ 2) surfaces, respectively. On Au(110)-(1 Õ 2) surfaces, C¢C bond chemistry, we compared the energy changes for the both n-hexane and n-hexyl adsorb nearly in the center of the fundamental surface processes of n-hexane (Figure 4). We groove. The dehydrogenated carbon atom is bound to chose two C6H14 molecules adsorbed in the grooves as the inia second-layer gold atom with a bonding distance of 2.13 æ (Figure 3 a and Figure 3 b). On Pt(110)-(1 Õ 2), n-hexane adsorbs tial state. Dehydrogenation from the terminal methyl group in the groove and aslant to the second layer of platinum with the formation of n-hexyl and two H atoms adsorbed at atoms (Figure 3 c), which is similar to ethane as reported elsebridge sites increases the total energy on both Au(110)-(1 Õ 2) where.[10] For n-hexyl, the terminal CH2 group prefers to bond and (1 Õ 3) surfaces. Further dehydrogenation from the penultimate CH2 groups on the Au(110)-(1 Õ 2) surface results in furwith an uppermost Pt atom while the remainder of the molecule stays in the groove (Figure 3 d). The result agrees well with our experiments and is consistent with a previous study on ethyl (C2H5).[11] The adsorption energies of n-hexane on Au(110)-(1 Õ 2) and Pt(110)-(1 Õ 2) are ¢0.94 and ¢1.29 eV, respectively. However, the values change significantly after H dissociation. The adsorption of the n-hexyl intermediate plus the released H atom results in an energy increase of 1.15 eV on Au(110)-(1 Õ 2), implying an endothermic dehydrogenation process. On the other hand, dehydrogenation is exothermic on Pt(110)-(1 Õ 2) with a heat of 0.19 eV, resulting from a slightly shorter C¢Pt bond (2.08 æ). Here, what happens on the Au(110)-(1 Õ 3) surface, which also has been calculated, has the same trend as Au(110)-(1 Õ 2) (see Figure S4). The stronger molecule–surface interaction on Pt(110) was further confirmed by calculating the diffusion barriers for nhexane and n-hexyl along the reconstruction grooves (Figure 3 e). The very low barriers of intact n-hexane (0.03 eV for gold and 0.09 eV for platinum) agree with our experimental observations that alkane monomers exhibit strong diffusion ability in the 1D channels. Meanwhile, the diffusion barrier on Figure 4. Calculated energy diagrams for C¢H activation and C¢C bond couPt(110)-(1 Õ 2) is nearly three times larger than that on Au(110)pling of n-hexane on Au(110) and Pt(110) surfaces. The values are the (1 Õ 2), which can explain why the diffusion on the Au surface energy difference with respect to the initial state (two intact C6H14 molecules is much easier than on Pt surface at 78 K. Compared with adsorbed in the reconstruction groove). ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
1358
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications ther energy increase (0.94 eV for two molecules). However, dimerization of two n-hexyl intermediates through C¢C coupling with the formation of n-dodecane dramatically decreases the total energy [¢1.61 eV for (1 Õ 2) and ¢0.95 eV for (1 Õ 3)]. Thus, we believe the exothermic feature of C¢C coupling is the driving force to promote alkane polymerization on Au(110). On Pt(110)-(1 Õ 2) surfaces, an energy decrease of ¢0.29 eV is obtained when two n-hexane molecules convert into two n-hexyl intermediates. By further dehydrogenation, the energy of the system continues to decrease (¢0.47 eV). Similarly, partial dehydrogenation of ethane on Pt(110)-(1 Õ 2) is an exothermic process as reported elsewhere,[11] in contrast to the Pt(111) case, where dissociation is endothermic.[2e] Further comparison between two n-hexyl intermediates and the dimerized product n-dodecane indicates a small energy difference, 0.08 eV lower for the latter. However, the decomposition of ndodecane into CH3 and C11H23 fragments will slightly decrease the energy by ¢0.08 eV, probably owing to the stronger C¢Pt bonding of the CH3 species.[12] Furthermore, considering the facility of C¢H bond dissociation on Pt(110), we chose CH2(CH)5 to mimic the multi-dehydrogenated species. Terminal C¢C coupling of CH2(CH)5 species on Au(110)-(1 Õ 2) results in an energy decrease of ¢2.53 eV. In contrast, an energy increase of 0.10 eV is obtained on Pt(110)-(1 Õ 2), indicating that multi-dehydrogenation under ultrahigh vacuum conditions may further facilitate C¢C bond cleavage. Indeed, under H2 environment, C¢C coupling has been observed for methane adsorbed on platinum catalyst to produce higher weight hydrocarbons.[13] We have investigated the on-surface C¢C bond chemistry of linear alkanes and found that C¢C bond coupling and dissociation are mediated at different metal surfaces. In particular, dehydrogenative C¢C coupling occurs on anisotropic Au(110) surfaces, while C¢C bond dissociation occurs on Pt(110) surfaces. Dehydrogenation on Pt(110) surfaces is energetically favorable due to the strong C¢Pt interactions, resulting in multiple hydrogen loss, which further facilitates C¢C cleavage. Our work implies that the competition between C¢C and C¢metal bonding is a crucial factor determining the pathway of alkane onsurface chemistry.
to the tip and the images were taken in the constant-current mode. Calculations were carried out in the framework of density functional theory (DFT) by using the Vienna Ab Initio Simulation Package (VASP).[14] Electron–electron exchange and correlation interactions were described within the generalized gradient approximation (GGA) by employing the Perdew, Burke, and Ernzerhof (PBE) functional.[15] The valence–core interactions were described using the Projector Augmented Wave (PAW) method.[16] Semiempirical dispersion corrections were added using the scheme by Grimme.[17] The plane-wave energy cutoff used for all calculations was 400 eV. The convergence criterion for the forces of all structural relaxations was 0.01 eV æ¢1. All the metal surfaces were modeled by a slab consisting of six atomic layers and separated by a vacuum region of 15 æ. The geometry of the adsorption complex was optimized by relaxing all atoms of the adsorbate and the metal atoms of four uppermost layers of the surface. A (5 Õ 1) extension periodicity of the reconstructed Au(110)-(1 Õ 2), Au(110)-(1 Õ 3), and Pt(110)-(1 Õ 2) surfaces and a 4 Õ 2 Õ 1 K-mesh were used for our calculations. The adsorption energy Ea is defined as Ea = EX/M¢(EM + EX), where EX/M, EM, and EX are the total energies of the relaxed adsorption system, the metal slab and the gas-phase species, respectively.
Acknowledgements The work was financially supported by NSFC (Project 11374374, 11275279 and 91227201) and supported by National Supercomputer Center in Guangzhou. Keywords: alkanes · C¢C coupling · C¢C dissociation · density functional calculations · scanning probe microscopy
Experimental Section The experiments were performed on an Omicron low-temperature STM at 78 K with a base vacuum below 1 Õ 10¢10 mbar. Monocrystalline Pt(110) and Au(110) surfaces (from MaTecK GmbH) with diameter of 9–10 mm were cleaned by repeated Ar + sputtering and annealing at 550 8C for 10–15 min under ultrahigh vacuum (base vacuum < 3 Õ 10¢10 mbar). Subsequently, n-dotriacontane (C32H66, purchased from Alfa Aesar, 98 %) molecules were thermally evaporated at 100 8C from a quartz crucible onto the sample surfaces. The deposition time is 1 to 2 min depending on the desired coverage to be obtained. During deposition, the Pt(110) surface was kept at 300 K or 80 K, followed by annealing at different temperatures. The Au(110) surface was kept at 300 K, followed by annealing at 490 K. The as-prepared samples were transferred from the preparation chamber to the STM chamber without exposure to air and were analyzed by STM at 78 K using electrochemically etched W tips. All given voltages refer to the bias on samples with respect ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
1359
[1] a) J. A. Labinger, J. E. Bercaw, Nature 2002, 417, 507 – 514; b) J. Roithov, D. Schrçder, Chem. Rev. 2010, 110, 1170 – 1211; c) A. E. Shilov, G. B. Shul’ pin, Activation and Catalytic Reactions of Saturated Hydrocarbons in the Presence of Metal Complexes, Kluwer, Dordrecht, 2000. [2] a) D. F. Johnson, W. H. Weinberg, J. Chem. Phys. 1995, 103, 5833 – 5847; b) D. Kelly, W. H. Weinberg, J. Chem. Phys. 1996, 105, 11313 – 11318; c) D. F. Johnson, W. H. Weinberg, Science 1993, 261, 76 – 78; d) W. H. Weinberg, Y.-K. Sun, Surf. Sci. Lett. 1992, 277, L39 – L46; e) G. Papoian, J. K. Nørskov, R. Hoffmann, J. Am. Chem. Soc. 2000, 122, 4129 – 4144; f) A. M. Contreras, M. Montano, S. J. Kweskin, M. M. Koebel, K. Bratlie, K. Becraft, G. A. Somorjai, Top. Catal. 2006, 40, 19 – 34; g) D. Kelly, W. H. Weinberg, J. Chem. Phys. 1996, 105, 271 – 278; h) D. F. Johnson, W. H. Weinberg, J. Chem. Soc.Faraday Trans. 1995, 91, 3695 – 3702; i) M. R. A. Blomberg, P. E. M. Siegbahn, U. Nagashima, J. Wennerberg, J. Am. Chem. Soc. 1991, 113, 424 – 433; j) J. J. W. Harris, V. Fiorin, C. T. Campbell, D. A. King, J. Phys. Chem. B 2005, 109, 4069 – 4075; k) G. A. Somorjai, A. L. Marsh, Phil. Trans. R. Soc. A 2005, 363, 879 – 900; l) A. L. Marsh, K. A. Becraft, G. A. Somorjai, J. Phys. Chem. B 2005, 109, 13619 – 13622; m) P. D. Szuromi, J. R. Engstrom, W. H. Weinberg, J. Phys. Chem. 1985, 89, 2497 – 2502; n) A. V. Walker, D. A. King, Phys. Rev. Lett. 1999, 82, 5156 – 5159; o) D. T. P. Watson, S. Titmuss, D. A. King, Surf. Sci. 2002, 505, 49 – 57; p) W. H. Weinberg, Langmuir 1993, 9, 655 – 662; q) P. Kratzer, B. Hammer, J. K. Nørskov, J. Chem. Phys. 1996, 105, 5595 – 5604; r) W. H. Weinberg, J. Vac. Sci. Technol. A 1992, 10, 2271 – 2276; s) M. C. McMaster, R. J. Madix, J. Chem. Phys. 1993, 98, 9963 – 9976; t) G. W. Cushing, J. K. Navin, S. B. Donald, L. Valadez, V. Johnek, I. Harrison, J. Phys. Chem. C 2010, 114, 17222 – 17232. [3] J. F. Weaver, A. F. Carlsson, R. J. Madix, Surf. Sci. Rep. 2003, 50, 107 – 199. [4] D. Y. Zhong, J.-H. Franke, S. K. Podiyanachari, T. Blçmker, H. M. Zhang, G. Kehr, G. Erker, H. Fuchs, L. F. Chi, Science 2011, 334, 213 – 216. [5] G. C. Bond, D. T. Thompson, Catal. Rev. 1999, 41, 319 – 388. [6] a) A. S. K. Hashmi, G. J. Hutchings, Angew. Chem. Int. Ed. 2006, 45, 7896 – 7936; Angew. Chem. 2006, 118, 8064 – 8105; b) L. Grill, M. Dyer, L. Laffer-
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
[7] [8]
[9] [10]
[11]
entz, M. Persson, M. V. Peters, S. Hecht, Nat. Nanotechnol. 2007, 2, 687 – 691; c) O. Daz Arado, H. Mçnig, H. Wagner, J.-H. Franke, G. Langewisch, P. A. Held, A. Studer, H. Fuchs, ACS Nano 2013, 7, 8509 – 8515. a) J. W. A. Sachtler, G. A. Somorjai, J. Catal. 1983, 81, 77 – 94; b) G. A. Somorjai, J. Carrazza, Ind. Eng. Chem. Fundam. 1986, 25, 63 – 69. a) S. Vajda, M. J. Pellin, J. P. Greeley, C. L. Marshall, L. A. Curtiss, G. A. Ballentine, J. W. Elam, S. Catillon-Mucherie, P. C. Redfern, F. Mehmood, P. Zapol, Nat. Mater. 2009, 8, 213 – 216; b) M. Lersch, M. Tilset, Chem. Rev. 2005, 105, 2471 – 2526. N. Jiang, Y. Y. Zhang, Q. Liu, Z. H. Cheng, Z. T. Deng, S. X. Du, H.-J. Gao, M. J. Beck, S. T. Pantelides, Nano Lett. 2010, 10, 1184 – 1188. a) J. A. Stinnett, M. C. McMaster, S. L. M. Schroeder, R. J. Madix, Surf. Sci. 1996, 365, 683 – 700; b) J. A. Stinnett, R. J. Madix, J. Chem. Phys. 1996, 105, 1609 – 1620. A. T. Anghel, S. J. Jenkins, D. J. Wales, D. A. King, J. Phys. Chem. B 2006, 110, 4147 – 4156.
ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
[12] M. A. Petersen, S. J. Jenkins, D. A. King, J. Phys. Chem. B 2004, 108, 5909 – 5919. [13] M. Belgued, P. Pareja, A. Amariglio, H. Amariglio, Nature 1991, 352, 789 – 790. [14] a) G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558 – 561; b) G. Kresse, J. Furthmìller, Comput. Mater. Sci. 1996, 6, 15 – 50; c) G. Kresse, J. Furthmìller, Phys. Rev. B 1996, 54, 11169 – 11186. [15] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865 – 3868. [16] a) P. E. Blçchl, Phys. Rev. B 1994, 50, 17953 – 17979; b) G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758 – 1775. [17] S. Grimme, J. Comput. Chem. 2006, 27, 1787 – 1799.
Received: February 4, 2015 Published online on March 6, 2015
1360
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
Linear Alkane C¢C Bond Chemistry Mediated by Metal Surfaces Zeying Cai,[a] Meizhuang Liu,[a] Limin She,[a] Xiaoli Li,[a] Jason Lee,[a] Dao-Xin Yao,[a] Haiming Zhang,[b] Lifeng Chi,[b, c] Harald Fuchs,[c] and Dingyong Zhong*[a] Linear alkanes undergo different C¢C bond chemistry (coupling or dissociation) thermally activated on anisotropic metal surfaces depending on the choice of the substrate material. Owing to the one-dimensional geometrical constraint, selective dehydrogenation and C¢C coupling (polymerization) of linear alkanes take place on Au(110) surfaces with missing-row reconstruction. However, the case is dramatically different on Pt(110) surfaces, which exhibit similar reconstruction as Au(110). Instead of dehydrogenative polymerization, alkanes tend to dehydrogenative pyrolysis, resulting in hydrocarbon fragments. Density functional theory calculations reveal that dehydrogenation of alkanes on Au(110) surfaces is an endothermic process, but further C¢C coupling between alkyl intermediates is exothermic. On the contrary, due to the much stronger C¢Pt bonds, dehydrogenation on Pt(110) surfaces is energetically favorable, resulting in multiple hydrogen loss followed by C¢C bond dissociation.
In order to utilize alkanes as a feedstock in chemical production, selective C¢H and C¢C bond activation is required.[1] In the past decades, examples of alkane C¢H and C¢C bond activation on transition metal surfaces have been intensively reported.[2] However, in such heterogeneous catalytic approaches, the molecules may undergo various competing processes simultaneously, resulting in poor selectivity. Unveiling surface-mediated chemical processes of alkanes on the atomic scale is of fundamental importance for achieving high selectivity and realizing efficient utilization.[3] Recently, linear alkane polymerization through the terminal/penultimate C¢H bond [a] Z. Cai,+ M. Liu,+ Dr. L. She, X. Li, J. Lee, Prof. Dr. D.-X. Yao, Prof. Dr. D. Zhong School of Physics and Engineering and State Key Laboratory of Optoelectronic Materials and Technologies Sun Yat-sen University Xingang Xi Road 135, 510275 Guangzhou (China) E-mail: [email protected] [b] Dr. H. Zhang, Prof. Dr. L. Chi Institute of Functional Nano and Soft Materials (FUNSOM) and Collaborative Innovation Center of Suzhou Science and Technology Soochow University 199 Ren-Ai Road, Suzhou, Jiangsu 215123 (China) [c] Prof. Dr. L. Chi, Prof. Dr. H. Fuchs Physikalisches Institut Westflische Wilhelms-Universitt Wilhelm-Klemm-Straße 10, 48149 Mìnster (Germany) [+] These authors contributed equally to the work Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cphc.201500097.
ChemPhysChem 2015, 16, 1356 – 1360
activation and direct C¢C coupling has been achieved on anisotropic Au(110) surfaces.[4] The missing-row reconstructions at the Au(110) surfaces result in parallel (1 Õ 2) grooves with atomically scaled width and depth. The reactant molecules are located in the groove and their orientation and diffusion are efficiently constrained in a one-dimensional (1D) manner. Although gold has long been considered to have poor catalytic reactivity,[5] in the last decades, nanostructured gold has been found to exhibit catalytic effect for certain processes.[6] Compared with gold, platinum has been considered as one of the most versatile heterogeneous catalysts.[7] A systematic study of Pt–Au alloys as catalysts for the conversion of nhexane has revealed that with increasing the gold concentration at the surface, the isomerization rate increases substantially, while the aromatization and hydrogenolysis rates decrease exponentially.[8] This remarkable change indicates that platinum is more reactive for C¢H and C¢C bond activation than gold. Given the eminent catalytic performance of platinum, one may ask whether the Pt(110) surface, which possesses similar missing-row reconstruction with atomic grooves as Au(110), is better than Au(110) for on-surface linear alkane polymerization. Here, we report a comparison study on the behaviors of linear alkanes adsorbed on Au(110) and Pt(110) surfaces. With the combination of scanning tunneling microscopy (STM) and density functional theory (DFT) calculations, the kinetic and dynamic analysis of the on-surface chemistry of linear alkanes on Pt(110) and Au(110) surfaces has been investigated on the atomic scale. Instead of selective dehydrogenation and C¢C coupling on Au(110), multiple C¢H bond cleavage with the formation of C¢Pt bonds took place on Pt(110). Due to the much stronger C¢Pt bonds, the diffusion of dehydrogenated intermediates along the 1D-constrained (1 Õ 2) grooves was dramatically hindered and subsequent C¢C cleavage is more energetically favorable. As a case study this work provides a microscopic insight into surface-mediated C¢C bond chemistry of linear alkanes. Linear alkane molecules (C32H66) were deposited on Au(110) surfaces at room temperature (~ 300 K). Figure 1 a shows the typical STM image of samples with a submonolayer coverage (about 0.2 monolayer) measured at 78 K. The molecules were adsorbed in the (1 Õ 2) reconstruction grooves and diffused rapidly along the [1–10] direction even at such low temperature. With a sample bias of ¢3.5 V, no molecule was visible in the STM image. When the sample bias was changed to ¢1.0 V with the STM tip closer to the surface, some bright features appeared near the surface defects, indicating an increased resi-
1356
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
Figure 1. STM images of n-dotriacontane (C32H66) on Au(110) and Pt(110), before and after annealing. a) Submonolayer coverage of C32H66 deposited on Au(110)-(1 Õ 2) at 300 K (U = ¢3.5 V; I = 100 pA). C32H66 molecules are invisible by STM due to the diffusion. Inset, monolayer coverage, the dashed black lines show the centerlines of the uppermost gold rows. The inset includes a diagram showing the cross section of the molecule–surface interface. b) After annealing at 490 K, polymeric chains formed in the Au(110)(1 Õ 3) reconstruction grooves (U = ¢0.02 V; I = 2 nA). c) Submonolayer coverage of C32H66 deposited on Pt(110)-(1 Õ 2) at 300 K (U = ¢0.05 V; I = 1 nA). C¢ C bond cleavage took place, resulting in shorter fragments. d) As-grown at 80 K (U = ¢2.1 V; I = 100 pA). Some molecules switched between the two equivalent adsorption sites.
dence time of molecules near these sites, probably due to the strengthened tip–molecule interaction (Figure S1).[9] At a monolayer coverage, all molecules, which exhibited a straight stick shape with a length of ~ 4 nm, were adsorbed with the long axis along the (1 Õ 2) reconstructed grooves, that is, the [1–10] direction. There were five molecules packed in every three reconstruction units in the [001] direction, resulting in an averaged intermolecular distance of ~ 0.5 nm. (inset of Figure 1 a). This indicates that the parallel atomic grooves are an efficient template to constrain the orientation of the molecules. After annealing the sample at 490 K for 10 min, polymeric chains were observed in wider grooves (~ 1.2 nm) corresponding to missing-row (1 Õ 3) reconstruction,[4] as shown in Figure 1 b. Similarly, C32H66 molecules were deposited on Pt(110)-(1 Õ 2) surfaces at 300 K. However, this case is dramatically different from that on Au(110) (Figure 1 c). The molecules were adsorbed on the (1 Õ 2) reconstructed uppermost platinum rows and shorter fragments were formed, implying the occurrence of C¢C bond cleavage. To investigate the temperature effect, we deposited C32H66 molecules on Pt(110) at 80 K followed by annealing at different temperatures up to 300 K. On the asgrown sample, uniform and straight features were observed with a length of ~ 4 nm, corresponding to intact C32H66 molecules (Figure 1 d and Figure 2 a). They adsorbed in the grooves with the long axis along the [1–10] direction, implying much ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
Figure 2. STM images of n-dotriacontane (C32H66) on Pt(110) at different temperatures. a) As-grown at 80 K (U = ¢3 V; I = 30 pA). Intact C32H66 molecules adsorbed in the grooves with the long axis along the [1–10] direction. b) Annealing to 200 K (U = ¢0.03 V; I = 1.3 nA). Some molecules changed the adsorption configuration with one terminal adsorbed from the reconstruction grooves to the neighboring uppermost platinum rows, while the remaining still adsorbed in the grooves. c) Annealing to 225 K (U = ¢0.01 V; I = 2 nA). Most molecules underwent terminal dehydrogenation. d) After annealing at 250 K (U = ¢0.01 V; I = 1 nA). Molecules entirely transferred their adsorbed position from the grooves to the ridges. Darker features are marked by white arrows corresponding to the positions with H atoms dissociated. e) Further annealing to 300 K (U = ¢0.05 V; I = 1 nA). C¢C bond cleavage took place and shorter fragments were formed. In all figures, the dashed black lines show the centerlines of the uppermost platinum rows.
less diffusion ability on Pt(110) compared to Au(110). Nevertheless, diffusion along the grooves were occasionally observed (see Figure S2), indicating that no strong chemical bond is formed between the molecules and the surface. There are two equivalent adsorption sites located on the gradient {111} microfacets of the grooves. Due to rapid switching between the two adsorption sites, most molecules appear blurred in the STM images (Figure S3). After annealing at a temperature of 200 K for 90 min, some molecules changed their adsorption configuration with one terminal site adsorbed from the reconstruction grooves to the neighboring uppermost platinum row, while the other end remained adsorbed in the groove (Figure 2 b). When the sample was annealed to 225 K for 90 min, most molecules underwent such terminal transformation and some of them adsorbed with both terminals climbing up on the uppermost rows (Figure 2 c). After further annealing to 250 K for 90 min, the molecules were completely immobilized and anchored at the surface and changed their adsorbed position from the grooves to the ridges (Figure 2 d). They exhibited a tortuous shape and non-uniform contrast, with some parts brighter and some dimmer. The above observation indicates that one or more C¢ H bonds of each molecule were dissociated followed by the formation of C¢Pt bonds, resulting in stronger molecule–surface interaction. When increasing the annealing temperature to 300 K for 90 min, C¢C bond cleavage took place with the formation of shorter fragments (Figure 2 e). This is different from a previous study on the ethane/Pt(110)-(1 Õ 2) system,[2j] in which no C¢C bond cleavage was observed at 440 K. Our experiments demonstrate that the behavior of linear C32H66 molecules on Pt(110) is totally different from the case on Au(110). On Au(110), alkane molecules were aligned and rapidly diffused along the 1D grooves with a low energy barrier. Such a 1D constraint would increase the probability for neighboring molecules to meet end to end thus achieve C¢C cou-
1357
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications pling only at the terminal/penultimate sites. On the other hand, although the 1D constraint would still affect the molecular orientation and diffusion on Pt(110), multi-dehydrogenation followed by C¢C bond cleavage is the preferential pathway. Furthermore, the C¢H activation temperature on Pt(110) is much lower than on Au(110), in which the electronic effect of the metal surfaces plays a key role.[2c] Figure 3. Calculated adsorption configurations and diffusion barriers of n-hexane and n-hexyl intermediate on Density functional theory Au(110)-(1 Õ 2) and Pt(110)-(1 Õ 2). a),b) n-hexane and n-hexyl on Au(110)-(1 Õ 2), respectively. c) n-hexane on (DFT) calculations were carried Pt(110)-(1 Õ 2). Here, n-hexane tends to adsorb to the {111} microfacets. d) n-hexyl on Pt(110)-(1 Õ 2). A C¢Pt bond out to gain an insight into the is formed between the terminal CH2 and Pt atom from the upmost rows. Inset, top view. Adsorption energies are reaction mechanism. For simplic- listed in (a)–(d). e) Calculated energy diagrams of diffusion barriers along [1–10] (1D grooves) for n-hexane and hexyl. ity, we chose n-hexane (C6H14) as the model molecule in our simuintact n-hexane, the barriers for the n-hexyl significantly inlations. Metal surfaces were modeled by a slab consisting of crease, 1.02 eV on Pt(110)-(1 Õ 2) and 0.66 eV on Au(110)-(1 Õ 2). six atomic layers (for details, see the Supporting Information). The high diffusion barrier on Pt(110) significantly decreases the We first optimized the adsorption configurations for a single ncollision probability between the dehydrogenated species and hexane molecule and n-hexyl intermediate (one H atom dissotherefore limits the probability of C¢C bond coupling taking ciated from the terminal methyl group of n-hexane and adplace. sorbed at a bridge site) adsorbed on the Au(110)-(1 Õ 2) and To obtain a thermodynamic view on the surface-mediated Pt(110)-(1 Õ 2) surfaces, respectively. On Au(110)-(1 Õ 2) surfaces, C¢C bond chemistry, we compared the energy changes for the both n-hexane and n-hexyl adsorb nearly in the center of the fundamental surface processes of n-hexane (Figure 4). We groove. The dehydrogenated carbon atom is bound to chose two C6H14 molecules adsorbed in the grooves as the inia second-layer gold atom with a bonding distance of 2.13 æ (Figure 3 a and Figure 3 b). On Pt(110)-(1 Õ 2), n-hexane adsorbs tial state. Dehydrogenation from the terminal methyl group in the groove and aslant to the second layer of platinum with the formation of n-hexyl and two H atoms adsorbed at atoms (Figure 3 c), which is similar to ethane as reported elsebridge sites increases the total energy on both Au(110)-(1 Õ 2) where.[10] For n-hexyl, the terminal CH2 group prefers to bond and (1 Õ 3) surfaces. Further dehydrogenation from the penultimate CH2 groups on the Au(110)-(1 Õ 2) surface results in furwith an uppermost Pt atom while the remainder of the molecule stays in the groove (Figure 3 d). The result agrees well with our experiments and is consistent with a previous study on ethyl (C2H5).[11] The adsorption energies of n-hexane on Au(110)-(1 Õ 2) and Pt(110)-(1 Õ 2) are ¢0.94 and ¢1.29 eV, respectively. However, the values change significantly after H dissociation. The adsorption of the n-hexyl intermediate plus the released H atom results in an energy increase of 1.15 eV on Au(110)-(1 Õ 2), implying an endothermic dehydrogenation process. On the other hand, dehydrogenation is exothermic on Pt(110)-(1 Õ 2) with a heat of 0.19 eV, resulting from a slightly shorter C¢Pt bond (2.08 æ). Here, what happens on the Au(110)-(1 Õ 3) surface, which also has been calculated, has the same trend as Au(110)-(1 Õ 2) (see Figure S4). The stronger molecule–surface interaction on Pt(110) was further confirmed by calculating the diffusion barriers for nhexane and n-hexyl along the reconstruction grooves (Figure 3 e). The very low barriers of intact n-hexane (0.03 eV for gold and 0.09 eV for platinum) agree with our experimental observations that alkane monomers exhibit strong diffusion ability in the 1D channels. Meanwhile, the diffusion barrier on Figure 4. Calculated energy diagrams for C¢H activation and C¢C bond couPt(110)-(1 Õ 2) is nearly three times larger than that on Au(110)pling of n-hexane on Au(110) and Pt(110) surfaces. The values are the (1 Õ 2), which can explain why the diffusion on the Au surface energy difference with respect to the initial state (two intact C6H14 molecules is much easier than on Pt surface at 78 K. Compared with adsorbed in the reconstruction groove). ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
1358
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications ther energy increase (0.94 eV for two molecules). However, dimerization of two n-hexyl intermediates through C¢C coupling with the formation of n-dodecane dramatically decreases the total energy [¢1.61 eV for (1 Õ 2) and ¢0.95 eV for (1 Õ 3)]. Thus, we believe the exothermic feature of C¢C coupling is the driving force to promote alkane polymerization on Au(110). On Pt(110)-(1 Õ 2) surfaces, an energy decrease of ¢0.29 eV is obtained when two n-hexane molecules convert into two n-hexyl intermediates. By further dehydrogenation, the energy of the system continues to decrease (¢0.47 eV). Similarly, partial dehydrogenation of ethane on Pt(110)-(1 Õ 2) is an exothermic process as reported elsewhere,[11] in contrast to the Pt(111) case, where dissociation is endothermic.[2e] Further comparison between two n-hexyl intermediates and the dimerized product n-dodecane indicates a small energy difference, 0.08 eV lower for the latter. However, the decomposition of ndodecane into CH3 and C11H23 fragments will slightly decrease the energy by ¢0.08 eV, probably owing to the stronger C¢Pt bonding of the CH3 species.[12] Furthermore, considering the facility of C¢H bond dissociation on Pt(110), we chose CH2(CH)5 to mimic the multi-dehydrogenated species. Terminal C¢C coupling of CH2(CH)5 species on Au(110)-(1 Õ 2) results in an energy decrease of ¢2.53 eV. In contrast, an energy increase of 0.10 eV is obtained on Pt(110)-(1 Õ 2), indicating that multi-dehydrogenation under ultrahigh vacuum conditions may further facilitate C¢C bond cleavage. Indeed, under H2 environment, C¢C coupling has been observed for methane adsorbed on platinum catalyst to produce higher weight hydrocarbons.[13] We have investigated the on-surface C¢C bond chemistry of linear alkanes and found that C¢C bond coupling and dissociation are mediated at different metal surfaces. In particular, dehydrogenative C¢C coupling occurs on anisotropic Au(110) surfaces, while C¢C bond dissociation occurs on Pt(110) surfaces. Dehydrogenation on Pt(110) surfaces is energetically favorable due to the strong C¢Pt interactions, resulting in multiple hydrogen loss, which further facilitates C¢C cleavage. Our work implies that the competition between C¢C and C¢metal bonding is a crucial factor determining the pathway of alkane onsurface chemistry.
to the tip and the images were taken in the constant-current mode. Calculations were carried out in the framework of density functional theory (DFT) by using the Vienna Ab Initio Simulation Package (VASP).[14] Electron–electron exchange and correlation interactions were described within the generalized gradient approximation (GGA) by employing the Perdew, Burke, and Ernzerhof (PBE) functional.[15] The valence–core interactions were described using the Projector Augmented Wave (PAW) method.[16] Semiempirical dispersion corrections were added using the scheme by Grimme.[17] The plane-wave energy cutoff used for all calculations was 400 eV. The convergence criterion for the forces of all structural relaxations was 0.01 eV æ¢1. All the metal surfaces were modeled by a slab consisting of six atomic layers and separated by a vacuum region of 15 æ. The geometry of the adsorption complex was optimized by relaxing all atoms of the adsorbate and the metal atoms of four uppermost layers of the surface. A (5 Õ 1) extension periodicity of the reconstructed Au(110)-(1 Õ 2), Au(110)-(1 Õ 3), and Pt(110)-(1 Õ 2) surfaces and a 4 Õ 2 Õ 1 K-mesh were used for our calculations. The adsorption energy Ea is defined as Ea = EX/M¢(EM + EX), where EX/M, EM, and EX are the total energies of the relaxed adsorption system, the metal slab and the gas-phase species, respectively.
Acknowledgements The work was financially supported by NSFC (Project 11374374, 11275279 and 91227201) and supported by National Supercomputer Center in Guangzhou. Keywords: alkanes · C¢C coupling · C¢C dissociation · density functional calculations · scanning probe microscopy
Experimental Section The experiments were performed on an Omicron low-temperature STM at 78 K with a base vacuum below 1 Õ 10¢10 mbar. Monocrystalline Pt(110) and Au(110) surfaces (from MaTecK GmbH) with diameter of 9–10 mm were cleaned by repeated Ar + sputtering and annealing at 550 8C for 10–15 min under ultrahigh vacuum (base vacuum < 3 Õ 10¢10 mbar). Subsequently, n-dotriacontane (C32H66, purchased from Alfa Aesar, 98 %) molecules were thermally evaporated at 100 8C from a quartz crucible onto the sample surfaces. The deposition time is 1 to 2 min depending on the desired coverage to be obtained. During deposition, the Pt(110) surface was kept at 300 K or 80 K, followed by annealing at different temperatures. The Au(110) surface was kept at 300 K, followed by annealing at 490 K. The as-prepared samples were transferred from the preparation chamber to the STM chamber without exposure to air and were analyzed by STM at 78 K using electrochemically etched W tips. All given voltages refer to the bias on samples with respect ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
1359
[1] a) J. A. Labinger, J. E. Bercaw, Nature 2002, 417, 507 – 514; b) J. Roithov, D. Schrçder, Chem. Rev. 2010, 110, 1170 – 1211; c) A. E. Shilov, G. B. Shul’ pin, Activation and Catalytic Reactions of Saturated Hydrocarbons in the Presence of Metal Complexes, Kluwer, Dordrecht, 2000. [2] a) D. F. Johnson, W. H. Weinberg, J. Chem. Phys. 1995, 103, 5833 – 5847; b) D. Kelly, W. H. Weinberg, J. Chem. Phys. 1996, 105, 11313 – 11318; c) D. F. Johnson, W. H. Weinberg, Science 1993, 261, 76 – 78; d) W. H. Weinberg, Y.-K. Sun, Surf. Sci. Lett. 1992, 277, L39 – L46; e) G. Papoian, J. K. Nørskov, R. Hoffmann, J. Am. Chem. Soc. 2000, 122, 4129 – 4144; f) A. M. Contreras, M. Montano, S. J. Kweskin, M. M. Koebel, K. Bratlie, K. Becraft, G. A. Somorjai, Top. Catal. 2006, 40, 19 – 34; g) D. Kelly, W. H. Weinberg, J. Chem. Phys. 1996, 105, 271 – 278; h) D. F. Johnson, W. H. Weinberg, J. Chem. Soc.Faraday Trans. 1995, 91, 3695 – 3702; i) M. R. A. Blomberg, P. E. M. Siegbahn, U. Nagashima, J. Wennerberg, J. Am. Chem. Soc. 1991, 113, 424 – 433; j) J. J. W. Harris, V. Fiorin, C. T. Campbell, D. A. King, J. Phys. Chem. B 2005, 109, 4069 – 4075; k) G. A. Somorjai, A. L. Marsh, Phil. Trans. R. Soc. A 2005, 363, 879 – 900; l) A. L. Marsh, K. A. Becraft, G. A. Somorjai, J. Phys. Chem. B 2005, 109, 13619 – 13622; m) P. D. Szuromi, J. R. Engstrom, W. H. Weinberg, J. Phys. Chem. 1985, 89, 2497 – 2502; n) A. V. Walker, D. A. King, Phys. Rev. Lett. 1999, 82, 5156 – 5159; o) D. T. P. Watson, S. Titmuss, D. A. King, Surf. Sci. 2002, 505, 49 – 57; p) W. H. Weinberg, Langmuir 1993, 9, 655 – 662; q) P. Kratzer, B. Hammer, J. K. Nørskov, J. Chem. Phys. 1996, 105, 5595 – 5604; r) W. H. Weinberg, J. Vac. Sci. Technol. A 1992, 10, 2271 – 2276; s) M. C. McMaster, R. J. Madix, J. Chem. Phys. 1993, 98, 9963 – 9976; t) G. W. Cushing, J. K. Navin, S. B. Donald, L. Valadez, V. Johnek, I. Harrison, J. Phys. Chem. C 2010, 114, 17222 – 17232. [3] J. F. Weaver, A. F. Carlsson, R. J. Madix, Surf. Sci. Rep. 2003, 50, 107 – 199. [4] D. Y. Zhong, J.-H. Franke, S. K. Podiyanachari, T. Blçmker, H. M. Zhang, G. Kehr, G. Erker, H. Fuchs, L. F. Chi, Science 2011, 334, 213 – 216. [5] G. C. Bond, D. T. Thompson, Catal. Rev. 1999, 41, 319 – 388. [6] a) A. S. K. Hashmi, G. J. Hutchings, Angew. Chem. Int. Ed. 2006, 45, 7896 – 7936; Angew. Chem. 2006, 118, 8064 – 8105; b) L. Grill, M. Dyer, L. Laffer-
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Communications
[7] [8]
[9] [10]
[11]
entz, M. Persson, M. V. Peters, S. Hecht, Nat. Nanotechnol. 2007, 2, 687 – 691; c) O. Daz Arado, H. Mçnig, H. Wagner, J.-H. Franke, G. Langewisch, P. A. Held, A. Studer, H. Fuchs, ACS Nano 2013, 7, 8509 – 8515. a) J. W. A. Sachtler, G. A. Somorjai, J. Catal. 1983, 81, 77 – 94; b) G. A. Somorjai, J. Carrazza, Ind. Eng. Chem. Fundam. 1986, 25, 63 – 69. a) S. Vajda, M. J. Pellin, J. P. Greeley, C. L. Marshall, L. A. Curtiss, G. A. Ballentine, J. W. Elam, S. Catillon-Mucherie, P. C. Redfern, F. Mehmood, P. Zapol, Nat. Mater. 2009, 8, 213 – 216; b) M. Lersch, M. Tilset, Chem. Rev. 2005, 105, 2471 – 2526. N. Jiang, Y. Y. Zhang, Q. Liu, Z. H. Cheng, Z. T. Deng, S. X. Du, H.-J. Gao, M. J. Beck, S. T. Pantelides, Nano Lett. 2010, 10, 1184 – 1188. a) J. A. Stinnett, M. C. McMaster, S. L. M. Schroeder, R. J. Madix, Surf. Sci. 1996, 365, 683 – 700; b) J. A. Stinnett, R. J. Madix, J. Chem. Phys. 1996, 105, 1609 – 1620. A. T. Anghel, S. J. Jenkins, D. J. Wales, D. A. King, J. Phys. Chem. B 2006, 110, 4147 – 4156.
ChemPhysChem 2015, 16, 1356 – 1360
www.chemphyschem.org
[12] M. A. Petersen, S. J. Jenkins, D. A. King, J. Phys. Chem. B 2004, 108, 5909 – 5919. [13] M. Belgued, P. Pareja, A. Amariglio, H. Amariglio, Nature 1991, 352, 789 – 790. [14] a) G. Kresse, J. Hafner, Phys. Rev. B 1993, 47, 558 – 561; b) G. Kresse, J. Furthmìller, Comput. Mater. Sci. 1996, 6, 15 – 50; c) G. Kresse, J. Furthmìller, Phys. Rev. B 1996, 54, 11169 – 11186. [15] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865 – 3868. [16] a) P. E. Blçchl, Phys. Rev. B 1994, 50, 17953 – 17979; b) G. Kresse, D. Joubert, Phys. Rev. B 1999, 59, 1758 – 1775. [17] S. Grimme, J. Comput. Chem. 2006, 27, 1787 – 1799.
Received: February 4, 2015 Published online on March 6, 2015
1360
Ó 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
