NANO LETTERS
Manipulating Surface Diffusion Ability of Single Molecules by Scanning Tunneling Microscopy
2009 Vol. 9, No. 1 132-136
D. Y. Zhong,†,‡ J. Franke,† T. Blo¨mker,§ G. Erker,§ L. F. Chi,*,† and H. Fuchs†,‡ Physikalisches Institut, UniVersita¨t Mu¨nster, Wilhelm-Klemm-Strasse 10, 48149 Mu¨nster, Germany, Center for Nanotechnology (CeNTech), UniVersita¨t Mu¨nster, Heisenbergstrasse 11, 48149 Mu¨nster, Germany, Institut fu¨r Nanotechnologie, Forschungszentrum Karlsruhe, 76021 Karlsruhe, Germany, and Organisch-Chemisches Institut, UniVersita¨t Mu¨nster, Corresstrasse 40, 48149 Mu¨nster, Germany Received September 3, 2008; Revised Manuscript Received December 3, 2008
ABSTRACT The bonding of single diferrocene [Fc(CH2)14Fc, Fc ) ferrocenyl] molecules on a metal surface can be enhanced by partial decomposition of Fc groups induced by the tunneling current in scanning tunneling microscopy. Although the isolated intact molecule is mobile on the terrace of Cu(110) at 78 K, the modified molecule is immobilized on the terrace. Calculations based on density functional theory indicate that the hollow site of the Cu(110) surface is the energetically favorable adsorption site for both ferrocene and the Fe-cyclopentadienyl complex, but the latter one possesses a much higher binding energy with the substrate.
The manipulation of single atoms or molecules with atomic precision, including the control of the spatial location and movement, the electronic, magnetic, optical, and chemical properties, is a fundamental requirement in nanotechnology. By means of scanning tunneling microscopy (STM), different basic events at single molecular/atomic level such as lateral1,2 and rolling movement,3 desorption,4 conformation change,5 dissociation,6 and chemical reaction7 have been achieved. Thermally activated diffusion of molecules on surfaces,8 which is determined by the interactions of the molecules with the substrate surface, is one of the important aspects that should be considered in fabrication of single-molecule based devices. Adsorbed atoms or molecules that involve strong interactions, such as covalent bonding, with the substrate, are immobilized on surfaces at ambient conditions or even at elevated temperatures. Weakly adsorbed molecules, on the other hand, perform thermally activated diffusion on surfaces.8 Immobilization of weakly adsorbed molecules may be associated by intermolecular interactions accompanied with the formation of molecular aggregates. Single molecules can also be trapped by certain active sites on surfaces, for example, defects and step edges.9,10 By a voltage pulse between STM tip and substrate, molecules can be pinned at a liquid-solid interface.11 Although it is well-known that * Corresponding author,
[email protected]. † Physikalisches Institut and Center for Nanotechnology (CeNTech), Universita¨t Mu¨nster. ‡ Institut fu¨r Nanotechnologie, Forschungszentrum Karlsruhe. § Organisch-Chemisches Institut, Universita¨t Mu¨nster. 10.1021/nl802677c CCC: $40.75 Published on Web 12/23/2008
2009 American Chemical Society
the bonding of single molecules on a surface can be modified by STM manipulation, little work has been done to probe and manipulate the diffusion ability of single molecules in nanoscale so far. Here, we report that the diffusion ability of single molecules on a metal surface can be manipulated by STM, resulting in immobilization of individual molecules. The immobilization is associated by the tunneling current induced partial decomposition, which strengthens the bonding and elevates the diffusion barriers of molecules on the surface, as confirmed by density functional theory (DFT) calculations. We select a ferrocene derivate as our model molecule. Ferrocene has long been employed on a large scale as a fuel additive for the in situ generation of Fe catalysts for better fuel burning.12 Recently, ferrocene and its derivatives have been intriguing in nanoelectronics due to their novel electronic and electrochemical properties, for example, stochastic conductance variation,13 reversible switching,14 negative differential resistance,15 etc. Control of these molecules on a single-molecule level will aid in the investigation of relevant phenomena such as catalytic and oxidation-reduction reactions on the nanoscale. Experiments were conducted under ultrahigh vacuum. Single-crystalline Cu(110), cleaned by sputtering-annealing cycles, was used as the substrate. Diferrocene molecules (diFc-14), Fc(CH2)14Fc (Fc ) ferrocenyl),16 a ferrocene derivative consisting of two Fc groups bridged by an oligoethylene chain (Figure 1a), were deposited on the substrate at room temperature. The substrate was then
Figure 1. Diferrocenes (diFc-14) adsorbed on Cu(110) surface. (a) Molecular structure of diFc-14 [Fc(CH2)14Fc]. (b) diFc-14 deposited on Cu(110) at room temperature with a nominal coverage of 1-5 × 104 µm-2. The molecules prefer adsorbing at the step edges. Monomer (M) and dimmers (D) are marked. Inset of (b) shows the pseudo-three-dimensional image of the monomer. The two Fc groups exhibit ringlike feature with their 5-fold axis nearly perpendicular to the substrate. STM was conducted at V ) 1 V and I ) 10 pA at 78 K.
Figure 2. Manipulation on a single diFc-14 molecule at 78 K. The arrows show the manipulation directions. (a) Before manipulation. The molecule is adsorbed at a step edge. (b, c) The position and orientation of the molecule are changed by two manipulation processes. (d) The molecule diffused to a neighboring step edge. The tunneling parameters for imaging are V ) 1 V and I ) 10 pA, and for manipulation V ) 5 mV and I ) 2 nA (2.5 MΩ).
transferred onto the STM stage at 78 or 5 K for imaging and manipulating. At a sparse coverage (1-5 × 104 µm-2), the molecules are preferentially adsorbed at the step edges of the Cu(110) surface at 78 K (Figure 1b). Monomers and aggregates consisting of two to tens of molecules were observed at the step edges. The measured center-center distance of the two Fc groups in a molecule is 2.1 nm, implying that the oligoethylene chains are almost straight without bending. Molecules in larger aggregates are oriented with their chains roughly along the Cu[1 -1 0] direction, whereas the chains of dimers and monomers are along the step edge. The two Fc groups of a monomer appear as protrusions with ringlike features in STM images (inset of Figure 1b), indicating that both are adsorbed with the cyclopentadienyl (Cp) rings toward the substrate.17 The interactions of ferrocene with metal surfaces such as Ag(100) and Cu(100) are very weak.18-20 At a temperature above 250 K, ferrocene molecules are desorbed from the Ag(100) surface.18 As for diferrocene molecules, the interaction is enhanced owing to doubled Fc groups and oligoethylene chains. As a result, stable monolayer films with ordered structures are formed at room temperature on metal surfaces including Ag(110), Cu(110), Cu(111), and Au(111).21,22 However, an isolated diFc-14 molecule is still mobile on the terraces of Cu(110) at room temperature, as suggested by the preferential adsorption at step edges (Figure 1b). In order to investigate the diffusion ability of a single diFc-14 molecule on terraces at 78 K, lateral manipulation by an STM tip was conducted. To realize the lateral manipulation, the STM tip was located upon one of the two Fc groups and approached closer to the molecule by increasing the tunneling current typically from 10 pA to 1-2.5 nA and decreasing the gap voltage to 5 mV (tunneling resistance of 2-5 MΩ) with the feedback loop on. Then, the tip was moved to a destination site. The target molecule was pushed by the tip. A manipulation sequence is shown in Figure 2. Before manipulating, the molecule was adsorbed with both Fc groups at the step edge (Figure 2a). After two sequential manipulation processes, one of the Fc groups moved to the lower terrace but still stayed at or by the step
edge (Figure 2panels b and c). The following manipulation (Figure 2, panels c and d) had the molecule released from the step edge and diffuse to a neighboring step edge. The experiments indicate that, although the lateral location and/ or the configuration of the molecule was changed by manipulation, it stayed at a step edge. We have conducted similar experiments with the sample cooled down to 5 K. In that case, however, the molecule stayed at the terrace after being released from a step edge by manipulation. Therefore, we conclude that a single diFc molecule is mobile at 78 K on Cu(110) terraces while it is frozen at 5 K. To change the bonding status of diFc-14 molecules with the surface, one of the two Fc groups from a molecule was modified by tunneling electrons from the STM tip. As shown in Figure 3b, the STM tip was positioned over one of the Fc groups (arrowed) of an isolated molecule adsorbed at a step edge. Subsequently, 2.2 V of sample bias was applied with a tunneling current of about 1.5 nA. The voltage was then gradually increased from 2.2 to 2.4 V in a time interval of 0.5 s with the feedback loop suspended (Figure 3b). The current increased accordingly when a sudden decrease happened (Figure 3d). As shown in Figure 3c, after the above process, the apparent height of the Fc group decreased from about 2.4 Å to 1.6 Å. Figure 3e presents the profile of the modified molecule along the long axis. Certain modifications on adsorbates, such as conformation,23,24 orientation, dissociation,25,26 or charging,27,28 alter the apparent height of the adsorbates. As reported elsewhere, Fc groups of diFc-14 may perform reversible orientational switching from the tilted configuration (5-fold axis nearly parallel to the surface normal) to the edge-on configuration (5-fold axis perpendicular to the surface normal), resulting in a decrease of the apparent height.22 However, we believe that the current-induced modification described above is not due to the orientational switching, considering the facts that the Fc group is permanently modified by the tunneling current and no reversible switching has been observed. More importantly, the modification is accompanied by the occurrence of a smaller fragment, which has been occasionally observed on the surface nearby the modified Fc group. By
Nano Lett., Vol. 9, No. 1, 2009
133
Figure 3. Partial decomposition of a single diFc-14 molecule induced by tunneling current. (a) The diFc-14 molecule before decomposition. (b) The tip was moved upon one of the Fc groups (arrowed) and the voltage was increased gradually from 2.2 to 2.4 V with a large tunneling current (typically >1 nA). (c) Partially decomposed molecule. The apparent height of the decomposed Fc group decreased. (d) The tunneling current during the decomposition process. A sudden decrease of current took place at the voltage of 2.24 V. (d) Height histogram of a partially decomposed molecule shown in Figure 4g. The tunneling parameters are V ) 1 V and I ) 5 pA at 78 K for (a) to (c). (f) dI/dV curves obtained from Cu(110) substrate (black square), Fc group (red circle), and FeCp complex (green triangle), respectively. The black line is a Fano fitting to the resonance peak from the CpFe complex.
comparing the height difference between the intact Fc group and the modified one with the theoretical simulation in ref 17, we believe that the modification is due to the removal of a Cp ring and the formation of a FeCp complex. For removal of one Cp ring from a ferrocene molecule (partial decomposition) in gaseous state, an energy barrier about 4 eV is encountered.29 On metal surfaces, this process may be facilitated by the coupling with the substrate and can be induced by light and electron irradiation and thermal activation.29,30 Ferrocene molecules deposited on Au(111) are partially decomposed by thermal activation even below room temperature.17 Scanning tunneling spectroscopy (STS) was performed on the intact Fc and partially decomposed FeCp complex at 5 K. A lock-in amplifier was used with a sinusoidal modulation signal (10 mV, 1108 Hz). Figure 3f shows the dI/dV curves near the Fermi energy obtained on the bare Cu(110) surface, Fc group, and FeCp complex of a partially decomposed molecule. A resonance peak with a Fano-type shape located near zero bias clearly appeared on the curve from the FeCp complex. We believe that the peak is due to the Kondo effect originating from the coupling between the conduction elections of the substrate and the magnetic moment of the FeCp complex. Considering the half-width of the resonance peak, Γ/2 ) 23.7 ( 1.4 mV obtained by Fano fitting, one 134
Figure 4. Manipulation on partially decomposed molecule at 78 K. The arrows show the manipulation directions. (a) Before manipulation. (b) The molecule was manipulated from the step edge to the terrace. (c-h) The location and orientation of the molecule were changed by manipulation. The dashed lines, which are in the [11j0] direction of Cu(110), show the trace of the partially decomposed Fc group. The tunneling parameters for imaging are V ) 1 V and I ) 10 pA, and for manipulation V ) 5 mV and I ) 2 nA (2.5 MΩ).
can derive the Kondo temperature TK ∼ 194 K, which is similar to that of the cobalt phthalocyanine/Au(111) system but much higher than that of the Co/Au(111) system.31,32 Except for the feature near zero bias, no other obvious difference of the STS was found in the range from -2 to 2 V. We then conducted STM manipulation on the modified molecules. It has been found that the diffusion ability of the molecules was dramatically changed after current induced partial decomposition. As shown in Figure 4, a modified molecule adsorbed at a step edge (Figure 4a) was released from the step edge by STM lateral manipulation and stayed at the terrace of the Cu(110) surface (Figure 4b). Although the orientation and location of the molecule are changed by the subsequent manipulation process (Figure 4, panels c-f), it was found that the modified Fc group preferentially moves along the [11j0] (atomic rows) direction of the Cu(110) surface (dashed lines). This constrained diffusion behavior we observed is similar to a previous study on the surface diffusion of 4-trans-2-(pyrid-4-yl-vinyl)benzoic acid on Pd(110) by variable-temperature STM.33 Furthermore, we found it is more difficult to move the FeCp complex than the intact Fc group by the STM tip, indicating the stronger bonding of FeCp to the surface. Both lateral movement (Figures 4, panels a and b and panels f and g) and rotation centered at the modified head (Figures 4, panels b-e and panels g and h) have been achieved in our experiments. The above results imply that the molecule-substrate interaction Nano Lett., Vol. 9, No. 1, 2009
Figure 5. Energetically favorable configurations and spin-polarized partial density of states (PDOS) of iron atoms in ferrocene and FeCp complex adsorbed on Cu(110) simulated by DFT. The binding energies at hollow, top, short-bridge, and long-bridge sites are presented. Insets, side view. (a) and (c) Ferrocene. (b) and (d) FeCp complex.
is strengthened by the current-induced modification. It should be noted that with the creation of a FeCp complex possessing unique magnetic properties, an intact Fc group keeping pristine electrochemical properties was retained on the partially decomposed molecule. DFT calculations were performed to understand the bonding enhancement associated by partial decomposition using the VASP code.34-36 We considered the adsorption of an intact ferrocene and a FeCp complex at high-symmetry sites including top, hollow, long-bridge, and short-bridge sites of the Cu(110) surface, with the Cp ring(s) parallel to the substrate surface. The surface was modeled by a three-layer slab of Cu(110) which was found to be sufficient for total energy calculations by comparing results to five-layer slabs. The exchange correlation energy functional of Perdew, Burke, and Ernzerhof37 was employed, and the valence-core interactions were modeled using the projector augmented wave method (PAW)38 as implemented in VASP.39 To reduce molecule-molecule interactions, a 4 × 6 surface unit cell of the Cu(110) substrate was chosen. Only the Γ-point was used to sample the k-mesh, and a plane wave basis set cutoff of 400 eV was found to be sufficient. All structural relaxations were converged to forces smaller than 10 meV/Å for ferrocene and 20 meV/Å for FeCp systems. Spinpolarized calculations converged to a spinless state for all ferroceneconfigurations.Thus,duetotheweakferrocene-substrate interaction in these configurations the expected singlet state of the ferrocene is retained.40 After geometrical optimization, we found that the hollow site is energetically favorable for both ferrocene and FeCp complexes (Figure 5). The binding energy of ferrocene on the hollow site is 0.57 eV, which is similar to the value on the long-bridge site and slightly higher than on-top and shortbridge sites. As for the FeCp complex, the most favorable configuration is on the hollow site with the iron atom toward the surface, which possesses a binding energy of 3.66 eV. Nano Lett., Vol. 9, No. 1, 2009
The values for short-bridge and long-bridge sites are 2.20 and 3.07 eV, respectively. We have also considered the configuration with the iron atom pointing upward and found this to be less stable with a binding energy of only 0.26 eV. From the calculated binding energies at different sites, one can estimate the surface diffusion barriers, which determine the diffusion ability along different directions. For an intact ferrocene, no or negligible diffusion barriers exist in both [001] and [11j0] directions, indicating free diffusion on the surface. But for the FeCp complex, the diffusion barriers in [001] and [11j0] directions are 1.46 and 0.59 eV, respectively. Thus, we conclude that the tunneling current induced immobilization of the diFc-14 molecule is due to the elevated diffusion barriers of the FeCp complex in comparison to ferrocene. Furthermore, the lower diffusion barrier in [11j0] direction than in the [001] direction agrees with our finding that the partially decomposed molecule is easier to be manipulated in the [11j0] direction. A closer look at the electronic structure of the ferrocene and FeCp systems also shows that the molecule-substrate coupling of the FeCp complex is much stronger than in the case of the ferrocene molecule. The spin-resolved density of states projected onto the molecule shows sharp peaks resembling closely the molecular peaks for the ferrocene system (Figure 5c). This system is diamagnetic since both spin components are equal. For the FeCp complex the situation differs (Figure 5d): The interaction with the substrate is much stronger, yielding generally broadened peaks and a net magnetic moment of 1.86 µB. This magnetic moment, which results in the Kondo resonance as observed experimentally, is almost entirely concentrated on the Fe atom. In summary, we demonstrate here the possibility to manipulate the diffusion ability of a single molecule on surfaces by an STM tip. In particular, tunneling-currentinduced partial decomposition of a single diferrocene enhances the bonding with the underlying Cu(110) surface. Theoretical simulations indicate the immobilization of partially decomposed molecules is due to the increase of surface diffusion barriers. In addition, Kondo effect originating from the coupling between the conduction electrons of the substrate and the magnetic moment of the FeCp complex was observed in STS measurement. Acknowledgment. The work was financially supported by the Deutsche Forschungsgemeinschaft through SFB 424 and TRR 61. J.F. is indebted to Vasile Caciuc for helpful discussions. References (1) Eigler, D. M.; Schweizer, E. K. Nature 1990, 344, 524. (2) Bartels, L.; Meyer, G.; Rieder, K.-H. Phys. ReV. Lett. 1997, 79, 697. (3) Grill, L.; Rieder, K.-H.; Moresco, F.; Rapenne, G.; Stojkovic, S.; Bouju, X.; Joachim, C. Nat. Nanotechnol. 2007, 2, 95. (4) Stokbro, K.; Thirstrup, C.; Sakurai, M.; Quaade, U.; Hu, B. Y.-K.; Perez-Murano, F.; Grey, F. Phys. ReV. Lett. 1998, 80, 2618. (5) Iancu, V.; Deshpande, A.; Hla, S.-W. Nano Lett. 2006, 6, 820. (6) Stipe, B. C.; Rezaei, M. A.; Ho, W.; Gao, S.; Persson, M.; Lundqvist, B. I. Phys. ReV. Lett. 1997, 78, 4410. (7) Hla, S.-W.; Bartels, L.; Meyer, G.; Rieder, K.-H. Phys. ReV. Lett. 2000, 85, 2777. (8) Naumovets, A. G.; Vedula, Y. S. Surf. Sci. Rep. 1985, 4, 365. 135
(9) Schunack, M.; Rosei, F.; Naitoh, Y.; Jiang, P.; Gourdon, A.; Laegsgaard, E.; Stensgaard, I.; Joachim, C.; Besenbacher, F. J. Chem. Phys. 2002, 117, 6259. (10) Chambliss, D. D.; Wilson, R. J.; Chiang, S. Phys. ReV. Lett. 1991, 66, 1721. (11) Foster, J. S.; Frommer, J. E.; Arnett, P. C. Nature 1988, 331, 324. (12) Cornils, B.; Herrmann, W. A. Applied homogeneous catalysis with organometallic compounds; Wiley-VCH: Weinheim, 2002. (13) Wassel, R. A.; Fuierer, R. R.; Kim, N.; Gorman, C. B. Nano Lett. 2003, 3, 1617. (14) Engtrakul, C.; Sita, L. R. Nano Lett. 2001, 1, 541. (15) Gorman, C. B.; Carroll, R. L.; Fuierer, R. R. Langmuir 2001, 17, 6923. (16) Wedeking, K.; Mu, Z.; Kehr, G.; Sierra, J. C.; Lichtenfeld, C. M.; Grimme, S.; Erker, G.; Fro¨hlich, R.; Chi, L.; Wang, W.; Zhong, D.; Fuchs, H. Chem. Eur. J. 2006, 12, 1618. (17) Braun, K.-F.; Iancu, V.; Pertaya, N.; Rieder, K.-H.; Hla, S.-W. Phys. ReV. Lett. 2006, 96, 246102. (18) Welipitiya, D.; Dowben, P. A.; Zhang, J.; Pai, W. W.; Wendelken, J. F. Surf. Sci. 1996, 367, 20. (19) Waldfried, C.; Welipitiya, D.; Hutchings, C. W.; de Silva, H. S. V.; Gallup, G. A.; Dowben, P. A.; Pai, W. W.; Zhang, J.; Wendelken, J. F.; Boag, N. M. J. Phys. Chem. B 1997, 101, 9782. (20) Woodbridge, C. M.; Pugmire, D. L.; Johnson, R. C.; Boag, N. M.; Langell, M. A. J. Phys. Chem. B 2000, 104, 3085. (21) Zhong, D. Y.; Wang, W. C.; Dou, R. F.; Wedeking, K.; Erker, G.; Chi, L. F.; Fuchs, H. Phys. ReV. B 2007, 76, 205428. (22) Zhong, D. Y.; Wedeking, K.; Blo¨mker, T.; Erker, G.; Fuchs, H.; Chi, L. F. Submitted. (23) Moresco, F.; Meyer, G.; Rieder, K.-H.; Tang, H.; Gourdon, A.; Joachim, C. Phys. ReV. Lett. 2001, 86, 672.
136
(24) Comstock, M. J.; Levy, N.; Kirakosian, A.; Cho, J.; Lauterwasser, F.; Harvey, J. H.; Strubbe, D. A.; Fre´chet, J. M. J.; Trauner, D.; Louie, S. G.; Crommie, M. F. Phys. ReV. Lett. 2007, 99, 038301. (25) Lauhon, L. J.; Ho, W. J. Phys. Chem. A 2000, 104, 2463. (26) Komeda, T.; Kim, Y.; Fujita, Y.; Sainoo, Y.; Kawai, M. J. Chem. Phys. 2004, 120, 5347. (27) Repp, J.; Meyer, G.; Olsson, F. E.; Persson, M. Science 2004, 305, 493. (28) Olsson, F. E.; Paavilainen, S.; Persson, M.; Repp, J.; Meyer, G. Phys. ReV. Lett. 2007, 98, 176803. (29) Stauf, G. T.; Driscoll, D. C.; Dowben, P. A. Thin Solid Films 1987, 153, 421. (30) Lin, J.-L.; Petrovykh, D. Y.; Kirakosian, A.; Rauscher, H.; Himpsel, F. J.; Dowben, P. A. Appl. Phys. Lett. 2001, 78, 829. (31) Zhao, A.; Li, Q.; Chen, L.; Xiang, H.; Wang, W.; Pan, S.; Wang, B.; Xiao, X.; Yang, J.; Hou, J. G.; Zhu, Q. Science 2005, 309, 1542. (32) Madhavan, V.; Chen, W.; Jamneala, T.; Crommie, M. F.; Wingreen, N. S. Science 1998, 280, 567. (33) Weckesser, J.; Barth, J. V.; Kern, K. J. Chem. Phys. 1999, 110, 5351. (34) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 47, 558. (35) Kresse, G.; Furthmu¨ller, J. Comput. Mater. Sci. 1996, 6, 15. (36) Kresse, G.; Furthmu¨ller, J. Phys. ReV. B 1996, 54, 11169. (37) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865. (38) Blo¨chl, P. E. Phys. ReV. B 1994, 50, 17953. (39) Kresse, G.; Joubert, D. Phys. ReV. B 1999, 59, 1758. (40) Swart, M. Inorg. Chim. Acta 2007, 360, 179.
NL802677C
Nano Lett., Vol. 9, No. 1, 2009
