b. Department of Physics, University of Science and Technology of China, Hefei 230026, China
In single-molecule transistors, the overlap between the wave function of the molecule and those of the electrodes plays an important role in determining the electron transport properties of the devices [1,2]. Generally, the source and drain electrodes in single-molecule devices usually use metal materials, such as Au, Ag and Cu. When a molecule is in direct contact with a metal surface, the strong coupling between them could broaden the molecular levels  and even lead to a strong distortion of the native molecular orbitals. Repp et al. demonstrated the direct images of the unperturbed molecular orbitals of the individual pentacene molecules by the electronic decoupling provided by the ultrathin insulating NaCl films . However, this technique by adding ultrathin insulating layer to separate the molecule from the metal surface is difficult to implement in single-molecule devices. Furthermore, compared to the case that the molecule is in direct contact with the metal surface, the electronic decoupling insulating films usually increase the electron residence time in the tunneling junction . This can induce dynamic electrostatic effects and alter the energy-level alignment between the molecular orbital energies and the Fermi level of substrate . In terms of producing good metal-molecule contacts suitable for molecular electronics applications, studies concerning the interaction between molecular orbitals and various metal surfaces are of great interest [5,7-11].
The phenanthrene-edge-type polycyclic aromatic hydrocarbons have attracted much attention owing to their excellent characteristics in organic field-effect transistors [12-20] and alkali (or alkali-earth)-metal-doped organic hydrocarbon superconductors [21-26]. Among them, the picene, which consists of five benzene rings in a zigzag arrangement (see Fig. 1(c)), is the first aromatic superconductor with a high superconducting transition temperature (Tc) through doping  and also possesses high hole mobilities in thin film transistors . To explore the mechanism of its superconductivity, the electronic structures of both pristine\newpage\noindent and doped solid picene have been widely investigated both from experiment [27-29] and theory [30-34]. All these studies focused on macroscopic solid picene, however, the local electronic structure of a single molecule mediated by interaction with various substrates has not yet been clarified so far. Recently, some scanning tunneling microscopy (STM) studies have revealed the topographic features and electronic properties of picene molecules adsorbed on Ag(111) and Au(111) [35,36]. This is an important step on the road toward understanding electron transport of single-molecule devices in molecular electronics.
In this work, we present an investigation of the electronic properties of individual picene molecules adsorbed on the Cu(111) surface by combining scanning tunneling microscopy/spectroscopy (STM/STS) and density functional theory (DFT) calculations. The dI/dV spectra and maps show bias voltage-dependent and spatial-resolved sub-molecular features of frontier states for the picene/Cu(111) system. With the help of theoretical calculations, we have attributed all of these features to the different molecular orbitals of picene with some intermixing.II. EXPERIMENTAL AND COMPUTATIONAL DETAILS
The experiments were carried out with a low-temperature STM (Omicron GmbH) in ultrahigh vacuum (UHV) condition. We use a polished copper single crystal (99.999%, MaTecK GmbH) as the substrate. Clean and atomically flat Cu(111) surface was obtained by several cycles of Ar+ sputtering (1000 eV) and annealing the crystal in the preparation chamber with a base pressure of 2×10-10 mbar. Picene molecules were deposited onto the Cu(111) surface by thermal evaporation from an organic source (OME, MBE-Komponenten GmbH), with the substrates at room temperature. No further annealing was carried out after the evaporation. The sample was immediately transferred to the STM chamber with a base pressure of 10-11 mbar and cooled down to 5 K for the further measurements. All the measurements were performed at 5 K to ensure the high quality and resolution of images and spectra of the molecules. We used an electrochemically etched tungsten tip which was cleaned by Ar+ sputtering. The tip was further cleaned by intentionally dipping the tip apex into the clean Cu surface prior to all measurements. Topographic STM images were recorded in constant-current mode and the dI/dV spectra were measured by a lock-in amplifier (SR830, Stanford Research Systems) with a small modulation signal (4 mV, 731 Hz) added to the bias voltage. All spectra were checked by measuring the Shockley surface state of Cu(111) with the same tip to ensure no obvious feature caused by tip effects. The bias voltage here refers to the sample voltage with respect to the tip.
The first-principles calculations were performed based on DFT in the generalized gradient approximation proposed by Perdew, Burke, and Ernzerhof  implemented with DMOL3 [38-40]. We used a semiempirical way following the scheme proposed by Tkatchenko and Scheffler  for van der Waals correction, and employed density functional semicore pseudopotential  with the basis set consisting of double numerical atomic orbits augmented by polarization functions. The calculation model is a periodic slab consisting of the adsorbed picene molecule and five layers of Cu atoms with each layer containing 50 atoms which represents the copper substrate. And a vacuum space of ~15 Å perpendicular to Cu surface was used. Brillouin zone was sampled only at γ-point. The criterion for the total energy and force was 1×10-5 Hartree and 1×10-3 Hartree/Å respectively, and all the atoms except those in the bottom three Cu layers were fully relaxed. The STM images were simulated by Tersoff-Hamann formula [43, 44] at the height of 3.0 Å from the molecular plane.III. RESULTS AND DISCUSSION
Figure 1(a) displays a representative STM image of picene molecules on Cu(111) at low coverage. The picene molecules are mono-dispersedly adsorbed on copper substrate. A close-up image of the single picene molecule marked by a white box in Fig. 1(a) is shown in Fig. 1(b). The picene molecule is asymmetric bean-shaped protrusion with the benzene ring planes parallel to the Cu(111) surface, which agrees well with the molecular structure and orientation in Fig. 1(c).
To know clearly the electronic structure of a single picene molecule adsorbed on Cu(111), we measured the dI/dV spectra obtained by ramping the bias voltage while keeping the tip at a constant height over the picene molecule. As shown in Fig. 2(a), the dI/dV spectrum taken on an individual picene molecule exhibits two conductance shoulders around -1.2 and 1.6 V. These characters can be assigned to the highest occupied state (labeled as O1) and the lowest unoccupied state (labeled as U1) of the single picene adsorbed on Cu(111). This spectrum exhibits a gap of ~2.8 eV around the Fermi level, which is obviously smaller than the previous results for picene on Ag(111) and Au(111) [35-36]. This difference for the three picene-substrate systems arises from a difference in molecular charging energies for electron tunneling . The relatively stronger screening behavior for picene on Cu(111) induced by the stronger interaction between picene and Cu substrate leads to the lower charging energy and smaller conductance gap for picene/Cu(111). In the gap region, a step-like feature with an onset around -0.4 eV can be seen, whose shape and energy position are quite similar to the Shockley surface state observed on bare Cu(111) substrate , we thus attribute this feature observed on a singly adsorbed picene molecule to the surface states of Cu(111) substrate.
For investigating the real-space properties of the frontier molecular orbitals of a single picene molecule adsorbed on a metal surface, we additionally performed energy-resolved dI/dV mapping experiments. This is accomplished by spatially mapping the dI/dV signal at a constant sample bias over the molecular surface, which permits the observation of the electron probability distribution of the molecular orbitals. The bias voltages were chosen to match the O1 and U1 energies as determined from the dI/dV spectrum in Fig. 2(a). In the upper panel of Fig. 2(b), we show the constant-current STM images of a single picene molecule recorded at -1.2 and 1.6 V. The STM image at -1.2 V shows a narrow feature similar to the molecular structure shown in Fig. 1(c), and by contrast, the molecule appears much broader for voltages at 1.6 V. This bias dependence of the STM images should originate from the different frontier orbitals of picene, and can be interpreted as the orbital-mediated tunneling out of occupied molecular orbital or into unoccupied molecular orbital . The corresponding dI/dV maps recorded at -1.2 and 1.6 V are shown in the lower panel of Fig. 2(b). The dI/dV map of O1 shows the strongest intensity over the short edge of the molecule, whereas the dI/dV map of U1 shows the dI/dV signal spreading over the whole molecule and with the largest conductivity found over the outermost periphery of the molecular long edge.
To further elucidate the dI/dV spectrum and maps in Fig. 2, we performed theoretical DFT calculations for the free and adsorbed single picene molecule. The top and side view of the theoretically optimized geometric structure for a single picene molecule adsorbed on Cu(111) are shown in Fig. 3 (a) and (b), respectively. The picene molecule was basically kept planar with a averaged distance of about 3.2 Å from the topmost Cu(111) surface layer (Fig. 3(b)). Figure 3(c) shows the theoretically calculated partial density of states (PDOS) for the picene molecule in the free case (black curve) and adsorption case (red curve). The conductance gaps for these two cases are almost the same in spite of the peak positions for the picene/Cu shows an overall shift about 0.2 eV towards the high energy side. This energy shift can be related to a charge transfer of about 0.4 electron to Cu substrate from picene molecule (see Fig. 3(d)) according to our Millikan charge analysis based on the calculation results. For the picene/Cu case, DFT calculations yield the highest occupied state and lowest unoccupied state of the adsorbed molecule at about -1.2 and 1.7 eV, which are basically in agreement with the STS result in Fig. 2(a). In the left panel of Fig. 3(e), we display the calculated frontier molecular orbitals of a free picene molecule, i.e. the highest occupied molecular orbital (HOMO), the lowest unoccupied molecule orbital (LUMO), HOMO-1, HOMO-2, LUMO+1, LUMO+2. On the other hand, the simulated dI/dV maps at -1.2 and 1.7 eV relative to the Fermi level for the picene/Cu adsorption system are shown in the right panel of Fig. 3(e). Here, the occupied state at -1.2 eV of picene/Cu system exhibits stronger intensity of local density of states (LDOS) over the short edge of the molecule, whereas the unoccupied state at 1.7 eV shows the LDOS spreading over the whole molecule and makes the molecule much broader. Our further analyses find that the former is mainly related to the HOMO orbital of individual picene molecule, in addition, the HOMO-1 and HOMO-2 orbitals also make small contribution to this state. The latter seems to be the result of the intermixing between LUMO and LUMO+1 states of individual picene molecule. Generally, the agreement between the simulated dI/dV maps for picene/Cu system in Fig. 3(e) and the experimental dI/dV maps in Fig. 2(b) is good, thus this allows us to ascribe the origin of states at -1.2 and 1.6 V to the HOMO and LUMO/LUMO+1 of the individual picene molecule, respectively. When the picene molecule is in direct contact with Cu(111) surface, the inherent molecular orbitals of the free molecule are perturbed due to the molecule-substrate interaction; this interaction is mainly charge transfer and brings energy shift of the gap region as shown before; but it also possesses small orbital hybridization between the molecular orbitals and substrate states, which possibly results that the formed hybrid states within small energy range have components from multiple molecular orbitals, i.e. the intermixing between the molecular orbitals as displayed above.
In order to understand the local electronic structure of individual picene molecule adsorbed on Cu(111), especially information of the unoccupied states which may be important to superconductivity study of the alkali-doped picene molecule , dI/dV spectra obtained at different sites of picene molecule were measured, as shown in Fig. 4(a). We find that the energy position of unoccupied electronic state displays a remarkable dependence on the surface positions of single picene molecule. The intramolecular spatial variation of these spectra for picene should arise from the LDOS distribution of the molecular orbitals. The constant-current dI/dV maps of the same picene molecule in the inset of Fig. 4(a) acquired at different sample biases in the range between 0.8 and 2.0 V are shown in Fig. 4(b). These maps exhibit bias-dependent sub-molecular features. The dI/dV map recorded at 0.8 V shows the dI/dV signal only on the interior region of picene molecule, whereas the dI/dV map at 1.6 V shows the molecular state extend over the whole molecule with large intensity on the molecular long edge. When the bias voltage varies from 1.6 V to 2.0 V, the main contributions to dI/dV signal change from the center of the molecular long edge to the short edge. In order to understand the nature of this spatially varied electronic structure observed in the experimental dI/dV maps, we have calculated PDOS of some atoms (as indicated in the red and blue boxes of the inset molecular structure) of long (red curve) and short (blue curve) edge of an adsorbed picene molecule, as shown in Fig. 4(c). It can be found that there exists a reversion of the contrast between the PDOS of two edges. This is consistent with our experimentally measured dI/dV spectra in Fig. 4(a). As presented in Fig. 4(d), we also simulate the energy-resolved dI/dV maps for picene on Cu(111), aiming at understanding remarkable spatial inhomogeneity of the experimental dI/dV maps in Fig. 4(b). The simulated map at energy around 0.9 eV relative to the Fermi level shows PDOS intensity over the interior skeleton of picene molecule, which reproduces the experimental image at 0.8 V. This state is the hybrid state between Cu substrate states and the LUMO/LUMO+1 of picene, in which the Cu substrate states make the main contribution so that the molecular pattern seems smaller. The simulated map at 1.7 eV is related to the intermixing LUMO and LUMO+1 of picene, which is in consistence with the experimental map at 1.6 V, as discussed before. The simulated map at energy around 2.2 eV, which is mainly related to the LUMO+2 of picene (see Fig. 3(e)), reproduces the experimental image at 2.0 V, especially the enhancement of the outmost protrusion. The simulated map at energy around 2.0 eV displaying the contributions to PDOS both from the long and short edge of the molecule agrees well with the experimental map at 1.7 V, and can be seen as the intermixing of the LUMO+1 and LUMO+2 of picene. The agreement between the experimental images shown in Fig. 4(b) and the theoretical images in Fig. 4(d) is good, which suggests that the STS measurements detect the convolution of energetically adjacent molecular orbitals for picene molecules directly adsorbed on Cu(111).IV. CONCLUSION
In summary, we have studied the electronic structures of single picene molecules adsorbed on Cu(111) surface by means of a combination of STM/STS measurements and theoretical calculations based on DFT. Our combined experimental and theoretical study reveals the following points: (i) The experimental results are well-reproduced by the DFT calculations taking the molecule-substrate interaction into account. The electron transfer from picene to Cu substrate can be responsible for the energy shift of peak positions between the free picene and adsorbed single picene molecule. (ii) The observed deviation between the conductance gaps of picene on Au(111), Ag(111) and Cu(111) indicates that there is a stronger interaction between the Cu substrate and the picene molecule in contrast to that of the picene/Au and picene/Ag system. (iii) The stronger molecule-substrate interaction can also induce orbital hybridization between the Cu substrate states and the molecular orbitals of picene, hence makes the STS measurements detect the convolution of energetically adjacent molecular orbitals. This work shows the way in which a combination of STM/STS measurements with DFT calculations has allowed one to extract valuable information about the molecular orbitals of single molecule adsorbed on a bare metal surface.V. ACKNOWLEDGMENTS
This work was supported by the National Basic Research Program of China (No.2011CB921400), the Strategic Priority Research Program (B) of the Chi-nese Academy of Sciences (No.XDB01020100), the Key Research Program of the Chinese Academy of Sci-ences (No.KJCX2-EWJ02), the Youth Innovation Pro-motion Association of the Chinese Academy of Sci-ences (No.2011322), and the National Natural Science Foundation of China (No.21473174, No.21273210, and No.51132007).
|||M. Perrin, Burzurí E., and S. J. van der Zant H., Chem. Soc. Rev. 44 , 902 (2015). DOI:10.1039/C4CS00231H|
|||Moth-Poulsen and T. Bjørnholm K., Nat. Nanotechnol. 4 , 551 (2009). DOI:10.1038/nnano.2009.176|
|||E. Goiri, P. Borghetti, El-Sayed A., Enrique Ortega J., and G. de Oteyza D., Adv. Mater. 28 , 1340 (2016). DOI:10.1002/adma.v28.7|
|||J. Repp, G. Meyer, M. Stojković S., A. Gourdon, and C. Joachim, Phys. Rev. Lett. 94 , 026803 (2005). DOI:10.1103/PhysRevLett.94.026803|
|||W. H. Soe, C. Manzano, De Sarkar A., N. Chandrasekhar, and C. Joachim, Phys. Rev. Lett. 102 , 176102 (2009). DOI:10.1103/PhysRevLett.102.176102|
|||J. B. Neaton, M. S. Hybertsen, and S. G. Louie, Phys. Rev. Lett. 97 , 216405 (2006). DOI:10.1103/PhysRevLett.97.216405|
|||X. Lu, M. Grobis, K. H. Khoo, S. G. Louie, and M. F. Crommie, Phys. Rev. Lett. 90 , 096802 (2003). DOI:10.1103/PhysRevLett.90.096802|
|||K. Wang, J. Zhao, S. Yang, L. Chen, Q. Li, B. Wang, S. Yang, J. G. Hou, and Q. Zhu, Phys. Rev. Lett. 91 , 185504 (2003). DOI:10.1103/PhysRevLett.91.185504|
|||X. Lu, M. Grobis, K. H. Khoo, S. G. Louie, and M. F. Crommie, Phys. Rev B70 , 115418 (2004).|
|||M. Feng, J. Zhao, and H. Petek, Science 320 , 359 (2008). DOI:10.1126/science.1155866|
|||Müllegger S., Schöfberger W., M. Rashidi, L. M. Reith, and R. Koch, J. Am. Chem. Soc. 131 , 17740 (2009). DOI:10.1021/ja908157j|
|||H. Okamoto, N. Kawasaki, Y. Kaji, Y. Kubozono, A. Fujiwara, and M. Yamaji, J. Am. Chem. Soc. 130 , 10470 (2008). DOI:10.1021/ja803291a|
|||N. Kawasaki, Y. Kubozono, H. Okamoto, A. Fujiwara, and M. Yamaji, Appl. Phys. Lett. 94 , 043310 (2009). DOI:10.1063/1.3076124|
|||Y. Kaji, N. Kawasaki, X. Lee, H. Okamoto, Y. Sugawara, S. Oikawa, A. Ito, H. Okazaki, T. Yokoya, A. Fujiwara, and Y. Kubozono, Appl. Phys. Lett. 95 , 183302 (2009). DOI:10.1063/1.3257373|
|||Y. Kaji, R. Mitsuhashi, X. Lee, H. Okamoto, T. Kambe, N. Ikeda, A. Fujiwara, M. Yamaji, K. Omote, and Y. Kubozono, Org. Electron. 10 , 432 (2009). DOI:10.1016/j.orgel.2009.01.006|
|||N. Kawasaki, W. L. Kalb, T. Mathis, Y. Kaji, R. Mitsuhashi, H. Okamoto, Y. Sugawara, A. Fujiwara, Y. Kubozono, and B. Batlogg, Appl. Phys. Lett. 96 , 113305 (2010). DOI:10.1063/1.3360223|
|||Y. Sugawara, Y. Kaji, K. Ogawa, R. Eguchi, S. Oikawa, H. Gohda, A. Fujiwara, and Y. Kubozono, Appl. Phys. Lett. 98 , 013303 (2011). DOI:10.1063/1.3540648|
|||N. Komura, H. Goto, X. He, H. Mitamura, R. Eguchi, Y. Kaji, H. Okamoto, Y. Sugawara, S. Gohda, K. Sato, and Y. Kubozono, Appl. Phys. Lett. 101 , 083301 (2012). DOI:10.1063/1.4747201|
|||N. Kawai, R. Eguchi, H. Goto, K. Akaike, Y. Kaji, T. Kambe, A. Fujiwara, and Y. Kubozono, J. Phys. Chem C116 , 7983 (2012).|
|||X. He, R. Eguchi, H. Goto, E. Uesugi, S. Hamao, Y. Takabayashi, and Y. Kubozono, Org. Electron. 14 , 1673 (2013). DOI:10.1016/j.orgel.2013.03.035|
|||R. Mitsuhashi, Y. Suzuki, Y. Yamanari, H. Mitamura, T. Kambe, N. Ikeda, H. Okamoto, A. Fujiwara, M. Yamaji, N. Kawasaki, Y. Maniwa, and Y. Kubozono, Nature 464 , 76 (2010). DOI:10.1038/nature08859|
|||Y. Kubozono, H. Mitamura, X. Lee, X. He, Y. Yamanari, Y. Takahashi, Y. Suzuki, Y. Kaji, R. Eguchi, K. Akaike, T. Kambe, H. Okamoto, A. Fujiwara, T. Kato, T. Kosugi, and H. Aoki, Phys. Chem. Chem. Phys. 13 , 16476 (2011). DOI:10.1039/c1cp20961b|
|||X. F. Wang, R. H. Liu, Z. Gui, Y. L. Xie, Y. J. Yan, J. J. Ying, X. G. Luo, and X. H. Chen, Nat. Commun. 2 , 507 (2011). DOI:10.1038/ncomms1513|
|||X. F. Wang, Y. J. Yan, Z. Gui, R. H. Liu, J. J. Ying, X. G. Luo, and X. H. Chen, Phys. Rev B84 , 214523 (2011).|
|||X. F. Wang, X. G. Luo, J. J. Ying, Z. J. Xiang, S. L. Zhang, R. R. Zhang, Y. H. Zhang, Y. J. Yan, A. F. Wang, P. Cheng, G. J. Ye, and X. H. Chen, J. Phys:Condens. Matter 24 , 345701 (2012). DOI:10.1088/0953-8984/24/34/345701|
|||M. Xue, T. Cao, D. Wang, Y. Wu, H. Yang, X. Dong, J. He, F. Li, and G. F. Chen, Sci. Rep. 2 , 389 (2012).|
|||Q. Xin, S. Duhm, F. Bussolotti, K. Akaike, Y. Kubozono, H. Aoki, T. Kosugi, S. Kera, and N. Ueno, Phys. Rev. Lett. 108 , 226401 (2012). DOI:10.1103/PhysRevLett.108.226401|
|||H. Okazaki, T. Wakita, T. Muro, Y. Kaji, X. Lee, H. Mitamura, N. Kawasaki, Y. Kubozono, Y. Yamanari, T. Kambe, T. Kato, M. Hirai, Y. Muraoka, and T. Yokoya, Phys. Rev B82 , 195114 (2010).|
|||F. Roth, B. Mahns, Büchner B., and M. Knupfer, Phys. Rev B83 , 144501 (2011).|
|||A. Ruff, M. Sing, R. Claessen, H. Lee, Tomić M., H. O. Jeschke, and R. Valent, Phys. Rev. Lett. 110 , 216403 (2013). DOI:10.1103/PhysRevLett.110.216403|
|||T. Kato, T. Kambe, and Y. Kubozono, Phys. Rev. Lett. 107 , 077001 (2011). DOI:10.1103/PhysRevLett.107.077001|
|||M. Casula, M. Calandra, G. Profeta, and F. Mauri, Phys. Rev. Lett. 107 , 137006 (2011). DOI:10.1103/PhysRevLett.107.137006|
|||M. Kim, B. I. Min, G. Lee, H. J. Kwon, Y. M. Rhee, and J. H. Shim, Phys. Rev B83 , 21450 (2011).|
|||T. Kosugi, T. Miyake, S. Ishibashi, R. Arita, and H. Aoki, Phys. Rev B84 , 214506 (2011).|
|||Y. Yoshida, H. Yang, H. Huang, S. Guan, S. Yanagisawa, T. Yokosuka, M. Lin, W. Su, C. Chang, G. Hoffmann, and Y. Hasegawa, J. Chem. Phys. 141 , 114701 (2014). DOI:10.1063/1.4894439|
|||C. Zhou, H. Shan, B. Li, A. Zhao, and B. Wang, Appl. Phys. Lett. 108 , 171601 (2016). DOI:10.1063/1.4947283|
|||J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77 , 3865 (1996). DOI:10.1103/PhysRevLett.77.3865|
|||B. Delley, J. Chem. Phys. 113 , 7756 (2000). DOI:10.1063/1.1316015|
|||Kohn and L. J. Sham W., Phys. Rev. 140 , A1133 (1965). DOI:10.1103/PhysRev.140.A1133|
|||J. Sham and W. Kohn L., Phys. Rev. 145 , 561 (1966). DOI:10.1103/PhysRev.145.561|
|||Tkatchenko and M. Scheffler A., Phys. Rev. Lett. 102 , 073005 (2009). DOI:10.1103/PhysRevLett.102.073005|
|||D. R. Hamann, M. Schluter, and C. Chiang, Phys. Rev. Lett. 43 , 1494 (1979). DOI:10.1103/PhysRevLett.43.1494|
|||Tersoff and D. R. Hamann J., Phys. Rev. Lett. 50 , 1998 (1983). DOI:10.1103/PhysRevLett.50.1998|
|||Tersoff and D. R. Hamann J., Phys. Rev B31 , 805 (1985).|
|||J. Kliewer, R. Berndt, E. V. Chulkov, V. M. Silkin, P. M. Echenique, and S. Crampin, Science 288 , 1399 (2000). DOI:10.1126/science.288.5470.1399|
|||L. Scudiero, D. E. Barlow, U. Mazur, and K. W. Hipps, J. Am. Chem. Soc. 123 , 4073 (2001). DOI:10.1021/ja0100726|
b. 中国科学技术大学物理系, 合肥 230026