Rotation and dissociation dynamics of a single O2 molecule on the Pt(111) surface determined from a first principles study
RRotation and dissociation dynamics of a single O molecule on the Pt(111) surfacedetermined from a first principles study Qiang Fu , , Jinlong Yang , Yi Luo , ∗ Hefei National Laboratory for Physical Science at the Microscale,University of Science and Technology of China, Hefei, Anhui 230026, P.R. China and Department of Theoretical Chemistry, School of Biotechnology,Royal Institute of Technology, SE-10691 Stockholm, Sweden (Dated: October 27, 2010)The STM induced rotation and dissociation dynamics of a single oxygen molecule on the Pt(111)surface have been finally determined by first principles calculations together with a newly developedstatistical model for inelastic electron tunneling. Several long-standing puzzles associated with thesedynamic processes in this classic system have been fully resolved. It is found that the unexpected lowenergy barrier of the O rotation is originated from an ingenious pathway, while the prior occupationof the metastable hcp-hollow site after the O dissociation can be attributed to a dynamic processof surface accommodation. The experimentally observed non-integer power-law dependence of therotation rate as a function of the current can be perfectly explained by taking into account therandomness of multi-electron inelastic tunneling processes. PACS numbers:
Rich dynamic processes of a single oxygen molecule onthe Pt(111) surface were achieved a decade ago by elegantscanning tunneling microscopy (STM) experiments [1, 2].These experiments demonstrated for the first time thatthe inelastic electron tunneling could be used to manip-ulate and control chemical reactions of a single moleculeon the metal surface [1, 2]. Such a superb ability of STMhas been widely employed in recent years to study sin-gle molecular chemistry of many different systems[3–5].However, the underlying mechanisms of many dynamicprocesses are still largely unexplored even for the firstexperiments of this kind. It was observed that under themanipulation of the STM tip, the single oxygen moleculecan be dissociated into two oxygen atoms [1] or rotateamong three equivalent orientations [2] on the Pt(111)surface. But, the ultralow energy barrier of the O ro-tation is counter-intuitive in view of its large adsorptionenergy on the Pt(111) surface. It is also difficult to under-stand the prior occupation of the metastable hcp-hollowsite after the O dissociation since the fcc-hollow site ismore stable by about 0.4eV in energy. Another outstand-ing problem is the non-integer power-law dependence ofthe rotation rate as a function of the tunneling current.It is known that the exponent of this power-law relationrepresents the number of inelastic electrons needed fora rotation event. Non-integer exponent means, accord-ing to the conventional theory [6], that the event of afractional electron tunneling could take place, but it issimply impossible in reality. A good solution to all theseproblems could be very useful for understanding the STMinduced single molecular chemistry in general since theenergy barrier and the reaction rate are the most funda-mental parameters for all chemical processes.In this letter, we demonstrate that with a systematicfirst principles study, it is possible to explain the exper-imental findings through identifying the most favorable rotation and dissociation pathways for a single O on thePt(111) surface. We find that the single oxygen moleculecan rotate on the Pt(111) surface in an ingenious waythat only requires very little energy to overcome the bar-rier. We also discover an interesting dynamic processof the surface accommodation that rationalizes the ob-served prior occupation of the metastable hcp-hollow siteafter the O dissociation. With the application of a re-cently developed statistical model, the non-integer expo-nent of the power-law relationship between the rotationrate and the current can be fully reproduced and ex-plained, which comes from the statistical average of allpossible n-electron events. It is noted that the statisticalproperty of the inelastic multi-electrons tunneling pro-cess holds the key for understanding the peculiar switch-ing behavior of a single molecular switch in a recent STMexperiment [7].Spin-polarized calculations were performed with theVienna Ab-initio Simulation Package (VASP) [8, 9]. Pro-jector augmented wave (PAW) potentials were employed[10] within a plane wave basis set expanded up to a cut-off energy of 400eV. Exchange correlation was describedby the generalized gradient approximation (GGA-PW91)[11]. GGA-PBE [12] was also used for comparison. ThePt(111) surface was modeled by a periodic supercell offour layers separated by a 12˚A thick vacuum region. Arectangular 2 √ × × × FIG. 1: Optimized structures (A and D), simulated STM im-ages (B and E), and the STM topographic images from theexperiments of Ref. [2] and [1] (C and F) for the adsorptionof oxygen molecule and atoms. Copyright 1998 American As-sociation for the Advancement of Science and 1997 AmericanPhysical Society. retical STM images were simulated with Tersoff-Hamannformula [14], i.e. integrating spatially resolved density ofstate (DOS) in energy from a bias potential to the Fermilevel.We first determine the adsorption structures of an oxy-gen atom and a single molecule on the Pt(111) surface.This is an issue that has been extensively investigatedboth experimentally [1, 2, 15, 16] and theoretically [17–20]. The oxygen molecule was found in experiments toadsorb on the face centered cubic (fcc) threefold hollowsites with a lightly canted top-hollow-bridge configura-tion [2, 17] as shown in Fig. 1A. The adsorption energy E ad was estimated to be 0.79eV, which agrees well withthe theoretical result of 0.68eV [17]. The correspondingSTM image has also been simulated, as shown in Fig.1B, which resembles very well with the ”pear” shape ob-served in the experiment [2] (Fig. 1C). The asymmetricappearance of the molecule in the STM image is resultedfrom the tilted adsorption structure. For the adsorptionof an oxygen atom, we only consider the fcc- and hcp-hollow sites, which are the ones identified in the STMexperiments [1]. Our calculated adsorption energies forthe fcc- and hcp- hollow sites are 4.66 and 4.24eV, respec-tively, which are in good agreement with the experiments[1, 16] and previous calculations [19, 20]. We also con-sider the co-adsorption of two oxygen atoms, i.e. onelocates at the fcc-hollow and another on the hcp-hollowsite, to mimic the final product of the dissociation (Fig.1F). In this case, the adsorption energy is about 1.05eVmore than that of the molecular adsorption (this valueincreases to 1.42eV when the two atoms are far awayfrom each other). The large energy difference causes non-thermal motion of the oxygen as ”hot atoms” [21], andtherefore explains the fact that the two oxygen atoms canseparate as far as three lattice constants after the dissoci-ation [1]. The simulated STM image as displayed in Fig. FIG. 2: Energy profile and the corresponding structures forthe rotation process of a single O molecule on the Pt(111)surface. The red line is fitted according to the energy and theforce (first derivative of energy) of each images (blue dots). Itshould be noted that ’A’ and ’F’ should have the same energy,and the two barriers should have the same height because ofthe symmetry. The slight asymmetry of the potential energysurface is non-physically introduced by the slab model used inthe simulations. The orange star denotes the exact locationof the paramagnetic state.
1E fits well with the one obtained from the experiment[1](Fig. 1F).As we have mentioned, it was found experimentallythat a single oxygen molecule can rotate reversibly amongthree equivalent orientations on the Pt(111) surface un-der the excitation of inelastic electrons [2]. In the ex-periment only very small energy is needed to inducethe rotation [2]. It is quite surprising to see the ultra-low rotation barrier because the interaction between theoxygen molecule and the Pt(111) surface is known tobe strong. Besides, the experimentally estimated value(0.15eV This work is supported bythe National Key Basic Research Program of China(2010CB923300, 2006CB922004), by the Swedish Re-search Council (VR), and by the National NaturalScience Foundation of China (Grant Nos. 20773112,50721091, 20925311). We acknowledge the supportfrom the Swedish National Infrastructure for Computing(SNIC), the USTC-HP HPC project, the SCCAS and theShanghai Supercomputer Center. ∗ Corresponding author. E-mail: [email protected][1] B. C. Stipe et al. , Phys. Rev. Lett. , 4410 (1997)[2] B. C. Stipe et al. , Science , 1907 (1998)[3] W. Ho, J. Chem. Phys. , 11033 (2002)[4] A. J. Mayne et al. , Chem. Rev. , 4355 (2006)[5] Q. Fu et al. , Phys. Chem. Chem. Phys. , 12012 (2010)[6] S. W. Gao et al. , Solid State Commun. , 271 (1992); S.W. Gao et al. , J. Elec. Spec. Relat. Phenom. , 665(1993); S. W. Gao et al. , Phys. Rev. B , 4825 (1997);R. E. Walkup et al. , Phys. Rev. B , 1858 (1993); R.E. Walkup et al. , J. Elec. Spec. Relat. Phenom. ,523 (1993); G. P. Salam et al. , Phys. Rev. B , 10655(1994)[7] S. Pan et al. , Proc. Natl. Acad. Sci. , 15259 (2009)[8] G. Kresse et al. , Comput. Mater. Sci. , 15 (1996)[9] G. Kresse et al. , Phys. Rev. B , 11169 (1996)[10] P. E. Blochl, Phys. Rev. B , 17953 (1994)[11] J. P. Perdew et al. , Phys. Rev. B , 13244 (1992)[12] J. P. Perdew et al. , Phys. Rev. Lett. , 3865 (1996)[13] G. Henkelman et al. , J. Chem. Phys. , 9901 (2000)[14] J. Tersoff et al. , Phys. Rev. B , 805 (1985)[15] C. Puglia et al. , Surf. Sci. , 119 (1995)[16] D. H. Parker et al. , Surf. Sci. , 489 (1989)[17] A. Eichler and J. Hafner, Phys. Rev. Lett. , 4481(1997)[18] L. Qi et al. , Phys. Rev. Lett. , 146101 (2008)[19] P. J. Feibelman et al. , Phys. Rev. Lett. , 2257 (1996)[20] A. Bogicevic et al. , Phys. Rev. B , R4289 (1998)[21] J. Wintterlin et al. , Phys. Rev. Lett. , 123 (1996)[22] V. A. Ranea et al. , Phys. Rev. Lett. , 136104 (2004)[23] T. Kumagai et al. , Phys. Rev. Lett. , 166101 (2008)[24] J. Wintterlin et al. , Phys. Rev. Lett. , 123 (1996)[25] C. Sprodowski et al. , J. Phys.: Condens. Matter ,264005 (2010)[26] J. A. Stroscio et al. , Science , 948 (2006)[27] J. I. Pascual et al. , Nature , 525 (2003)[28] P. A. Sloan et al. , Nature434