Scanning Tunneling Microscopy Investigation of Substrate-Dependent

Mar 14, 2011 - National Center for Nanoscience and Technology, Beijing 100190, ... In-Plane Intermolecular Interaction Assisted Assembly and Modified ...
0 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/JPCC

Scanning Tunneling Microscopy Investigation of Substrate-Dependent Adsorption and Assembly of Metallofullerene Gd@C82 on Cu(111) and Cu(100) Shixiong Zhao,†,§ Jun Zhang,‡,§ Jinquan Dong,† Bingkai Yuan,‡ Xiaohui Qiu,*,‡ Shangyuan Yang,† Jian Hao,† Hong Zhang,† Hui Yuan,† Gengmei Xing,† Yuliang Zhao,† and Baoyun Sun*,† †

CAS Key Lab for Biomedical Effects of Nanomaterials and Nanosafety, Institute of High Energy Physics, Chinese Academy of Sciences, and National Center for Nanoscience and Technology of China, Beijing 100049, China ‡ National Center for Nanoscience and Technology, Beijing 100190, China ABSTRACT: The structural and electronic properties of Gd@C82 adsorbed on Cu(111) and Cu(100) surfaces were investigated by scanning tunneling microscopy (STM). STM images showed that the lattice structure of the underlying metal substrates affected the arrangement of the metallofullerene molecules. Preferred adsorption orientations of the molecules were found from STM images with submolecular resolution. Differences in the scanning tunneling spectroscopy of molecules adsorbed on Cu(111) and Cu(100) were observed and attributed to the work function difference between the substrates. In addition, the electronic properties of the fullerene monomer, dimer, and monolayer were characterized and compared, revealing the roles played by the molecule-substrate interaction and the intermolecular interaction.

’ INTRODUCTION Endohedral metallofullerenes have novel properties with potential applications in biomedicine and electronics.1-4 Many studies have been done to elucidate the variation of structural and electronic properties of metallofullerenes when interacting with other compounds or being adsorbed on substrates.5-8 It has been found that the electronic structures at molecular and cluster levels of metallofullerenes could be significantly modified by the substrates. Such knowledge is of importance for building solid-state devices involving single molecules or structured molecular assemblies. The intermolecular interaction and the interplay between fullerenes and substrates have been widely studied by STM and STS, including C60, Dy@C82, La@C82, La2@C80, Gd@C82, and M3N@C80, on various metal and semiconductor surfaces.9-15 The properties of the adsorbates were changed in varying degrees based on the nature of interactions with different substrates. For instance, the HOMO-LUMO gap of C60 molecules on Au(111) showed an increase of 0.6 eV compared with C60 on Ag(100) because Ag(100) transferred 0.2 electrons to the fullerenes while the electron transfer from Au(111) could be neglected.16 Dy@C82 and Dy@C60 molecules do not form ordered structures on the Si surface, suggesting strong interactions between Si atoms and fullerene molecules.17 The local density of states (LDOS) of single Gd@C82 on the edge of Ag(100) showed spatial dependence features, as revealed by scanning tunneling spectroscopy (STS).15 The C2v-La@C82 on the Si substrate had been proved to have a HOMO-LUMO gap of 0.5 eV, which is smaller than that of Ih-La2@C80 (1.2 eV).11 Here, we presented an STM/STS study of Gd@C82 molecules deposited onto Cu(111) and Cu(100). A difference in the r 2011 American Chemical Society

adsorption behavior and domain orientations of the molecules was observed. There is a noticeable difference in the electronic properties for the molecules adsorbed on the two substrates. To understand the interaction between the Gd@C82 molecule, the electronic properties of the fullerene monomer, dimer, and monolayer were systematically characterized.

’ EXPERIMENTAL SECTION The carbon soot-containing Gd-metallofullerenes was synthesized by the direct current arc discharge method using a composite anode made of graphite and Gd2O3 in a He atmosphere at a pressure of 200 Torr. The soot was collected, extracted, and then separated by two-stage high-performance liquid chromatography (HPLC, LC908-C60, Japan Analytical Industry Co.) with toluene as the mobile phase employing a Buckyprep column (20 mm  250 mm, Nacalai Co., Japan) and a 5PBB column (20 mm  250 mm, Nacalai Co., Japan). One of the Gd@C82 isomers (Gd@C82-I) used in the present study was separated by recycling HPLC isolation for about six times. The purity of Gd@C82-I was confirmed by MALDITOF-MS (AutoFlex, Bruker Co., Germany) and UV-vis-NIR to be 99.9%. The 50 mg powder of the metallofullerene was obtained by evaporating the solvent, recrystallizing, and then degassing at 473 K at 10-6 Torr for 24 h before being used. All STM and STS measurements were performed at liquid N2 temperature (78 K) using an Omicron STM with Pt-Ir tips. The experiments were operated in an ultra-high-vacuum STM Received: December 21, 2010 Published: March 14, 2011 6265

dx.doi.org/10.1021/jp1121454 | J. Phys. Chem. C 2011, 115, 6265–6268

The Journal of Physical Chemistry C

ARTICLE

Figure 2. STM images of Gd@C82 islands on Cu(111) (a) and Cu(100) (b) surfaces. The different lattice structures of the domains were labeled by A-D and a on the two substrates, respectively.

Figure 1. STM images of Gd@C82 molecules adsorbed on Cu(111) with increasing molecular coverage (a-c). The inset in (c) is the zoomin image of a Gd@C82 island taken at the bias voltage of 2.0 V and current of 1 nA. (d) Cross-sectional profile along the line labeled as A-B.

chamber at 10-10 Torr. The single-crystal substrates Cu(111) and Cu(100) were cleaned by repetitive Neþ sputtering for three times (about 20 min/time) and then annealed at 850 K for 2030 min and confirmed by STM observations. The Gd@C82 powder was put into an Al2O3 crucible and sublimated onto the copper substrate, which was kept at room temperature. The experiment was performed in the current-constant mode. The dI/dV spectra were measured through lock-in detection of the ac tunneling current driven by a 1700 Hz, 90 mV signal added to the junction bias under open loop conditions. In the STS measurement, the tip was held at a fixed position above the sample.

’ RESULTS AND DISCUSSION The typical STM images of Gd@C82 molecules adsorbed on Cu(111) are shown in Figure 1. The spherical features in the pictures correspond to individual Gd@C82 molecules. Figure 1a shows the STM image in the initial stage of Gd@C82 deposition, revealing that small amounts of molecules were preferentially adsorbed on the step edges of the Cu(111) surface and most of them were located at the bottoms of the edges. Some metallofullerenes were clustered together to form dimers, whereas others existed as monomers on the edges, similar to the previously reported result.18 With the increasing of the coverage, the adsorbed molecules took almost all the edge positions and formed cluster chains along the step edges, as shown in Figure 1b. The Gd@C82 islands were thereafter formed on the terraces of the Cu(111) surface after the step edges were completely covered by the molecules (Figure 1c). Enlarging one molecule island (the inset image), we found that these molecules aggregated into a quasi-hexagonal closed-packed domain, similar to the fcc (111) structure. The larger bright feature on the island was a molecule adsorbed at the second layer. The lattice direction of the island was along the close-packed direction of the underlying Cu surface. The cross-sectional profile labeled as A-B through the molecules (Figure 1d) demonstrated that the apparent height of the metallofullerene Gd@C82 adsorbed on the substrate was 0.75 nm and the nearest neighbor distance of the molecules was 1.10 nm at the bias voltage of 2.0 V and current of 1 nA, similar to the value of [email protected]

In the experiment, because Gd@C82 molecules were deposited onto the metal surface, which was kept at room temperature, the molecules had enough kinetic energy and tended to move from the high-energy site (the terrace) to the low-energy site (the step edge or the defect sites). Thus, the molecules preferred to take the step edge position and then to form domains on the terrace around the defect site following adsorption. As reported in previous studies, Gd@C82 molecules were electrophilic and there was charge transfer from the substrate to the fullerene molecules.19 Therefore, due to the Smoluchowski effect,20,21 which creates a charge distribution that leaves the top of the step edge slightly electropositive and the bottom electronegative, the bottom of the step edge was preferred by the Gd@C82 molecules. In addition, the bottom of the step edge has a higher coordination number than the upper sites and should also be consequently favored. Deformed molecules could frequently be seen in almost all of the observed islands, which were speculated to reside on these reactive defect sites (Figure 2). Gd@C82 assemblies on Cu(111) displayed several orientations, as shown in Figure 2a, which are different from the La@C82 configurations on the Ag/Si(111) surface where the principal domain is oriented to the substrate axis by an angle of 30.22 Various domain orientations on the Cu(111) surface are labeled as A-D. Domain A had the configuration with an angle of 30 between the molecule lattice vectors and those of the Cu(111) surface. Within domains B-D, the molecules were at an angle of less than 30 (25, 20, 10) to the substrate lattice vectors. Therefore, the growth of Gd@C82 islands at room temperature was not epitaxial on the Cu(111) surface, and the arrangement of these islands was determined by the molecule-molecule interaction. With the increase of the molecule coverage on the surface, the Gd@C82 molecules tended to form the second layer of the islands rather than to form a complete monolayer on the surface. This observation indicates that the growth of the Gd@C82 molecules on Cu(111) was in a Volmer-Weber growth mode, similar to that of La@C82 grown on the Ag/Si surface.22 The adsorbate-adsorbate interaction in this case was stronger than that between the adsorbate and the substrate surface, thus leading to the formation of multilayered molecule islands upon further deposition, which was different from the case for C60 on the Ag(110) surface.23 The growth of Gd@C82 islands might start from the defect on the Cu(111) surface, and the arrangement of molecules was mainly dominated by the intermolecular interaction. The growth process of Gd@C82 on Cu(100) is similar to that on Cu(111), where the preferential adsorption sites of the molecule were the step edges and the defects. The molecules on Cu(100) formed quasi-hexagonal closed-packed arrays with a neighbor distance and molecular height similar to those on Cu(111). However, the molecular arrays on Cu(100) were found to have only one domain orientation (see Figure 2b) with one of 6266

dx.doi.org/10.1021/jp1121454 |J. Phys. Chem. C 2011, 115, 6265–6268

The Journal of Physical Chemistry C

Figure 3. High-resolution STM images of Gd@C82 molecules adsorbed on Cu(111) obtained at a bias voltage of -2.0 (a), -0.75 (b), 0.5 (c), and 1.0 V (d).

the lattice vectors oriented along Æ100æ crystallographic directions of the Cu(100) substrate. The different domain orientations of Gd@C82 on Cu(111) and Cu(100) might indicate that the bonding of the molecules to the substrate was affected by the lattice structure of the underlying surface. Previous studies have suggested that the fullerenes bonded to the metal surfaces most likely via weak ionic bonds or van der Waals forces.24 It has also been reported that the work function of the substrate should play an important role in the moleculesubstrate interaction.16 The work function of Cu(111) is 4.94 eV, 0.35 eV higher than that of Cu(100) (4.59 eV).25 Therefore, charge transfer between Gd@C82 and Cu(100) might be more significant than that for the Cu(111) substrate, suggesting that the interaction between the Gd@C82 and the Cu(100) substrate was stronger than that between the molecule and Cu(111). Thus, preferred orientation of Gd@C82 on Cu(100) substrates is expected. High-resolution STM images of the Gd@C82 molecules observed at 78 K are shown in Figure 3. Though the intramolecular feature of the molecules varied with the bias voltage (-2.0 to 1.0 V), most of the Gd@C82 molecules showed similar characters at the same image conditions. The similar feature of individual molecules in the arrays indicated that the molecules adopted the same adsorption configurations, which is believed to be the stable energy orientation. However, the defect on the substrate may affect the orientation of the surrounding molecules. At the second layer, no intramolecular structure was resolved, indicating that fullerene molecules are inclined to rotate. Differential conductance spectra (dI/dV) on individual Gd@C82 molecules were acquired to investigate the electronic structures of molecules on Cu(111) and Cu(100), as shown in Figure 4. The most striking features in the spectra were the peaks centered at around 1.0 and 1.75 V, which were broadened due to intermolecular and molecule-substrate interactions. The observed STS peak positions on the Cu(100) substrate was slightly moved to more negative bias voltage, compared with those on Cu(111). Cu(100) has a lower work function and, therefore, favors charge transfer to the metallofullerene molecule. This could account for the observation that the two electronic states of

ARTICLE

Figure 4. dI/dV spectra of Gd@C82 molecules located in the islands on Cu(111) (a) and Cu(100) (b) taken at 78 K. Statistical analysis shows that more than 80% of the molecules in the islands had nearly identical peak positions on both Cu(111) and Cu(100) surfaces.

Figure 5. dI/dV spectra taken at various positions on a single Gd@C82 molecule (a) and a Gd@C82 dimer (b) adsorbed on the edge of Cu(111) at 78 K. Spectra shown as solid lines were taken at the corresponding spots indicated in the inset images and shifted vertically for clarity. The spectrum shown as a dashed line was obtained on the bare Cu(111) surface.

Gd@C82’s on Cu(100) were experimentally about 0.05 and 0.1 eV lower than those of molecules on Cu(111). The electronic structures of a single molecule, two adjacent molecules located at the step edge, or molecules in the monolayer islands were further elucidated by STS. Figure 5a shows typical dI/dV spatial spectra acquired at different positions on a single Gd@C82 located at the Cu(111) step edge. Three main peaks were identified at 0.80, 1.20, and 1.75 V. The intensity of these peaks displayed strongly spatial inhomogeneity from point to point, which is similar to the previous STS results for Gd@C82 on Ag(001)15 and C60 on Ag(001).26 The two peaks at around 0.8 and 1.20 V give rise to the peak at 1.0 V for molecules in the islands. The peaks might be related to interaction of the single 6267

dx.doi.org/10.1021/jp1121454 |J. Phys. Chem. C 2011, 115, 6265–6268

The Journal of Physical Chemistry C molecule with the underlying Cu substrate, as well as the step edge effect, resulting in a modulation of the molecule's electronic structure. The spatial inhomogeneity of the spectra of Gd@C82 reflects the spatial distribution of different molecular orbits. Pronounced differences were observed for the dI/dV spectra taken on molecular dimers shown in Figure 5b. Two obvious peaks could be observed at around 0.95-1.15 and 1.75 V. The peaks at around 0.8 and 1.20 V for the individual molecule were shifted to around 1.0 V. Although there is a strong spatial inhomogeneity on the individual Gd@C82 molecules and the dimers, the dI/dV spectra on molecules in the islands exhibited broadened band features. The uniform electronic structure observed in Gd@C82 islands might suggest the strong intermolecular interaction. Different from C60, there exist stronger dipole-dipole and charge-transfer interactions between the metallofullerenes. These strong interactions had previously been proved by the head-to-tail alignment in the metallofullerene crystal.27

’ CONCLUSIONS We have investigated the adsorption, orientation, and electronic structure of Gd@C82 molecules on Cu(111) and Cu(100) substrates via STM and spectroscopy. By comparing the growth process of the metallofullerene and the orientation of the islands on the two surfaces, we found a preferred orientation of the 3-fold Gd@C82 domain and a slightly negative STS peak position on the Cu(100) substrate, indicating stronger interaction between the molecules and the Cu(100) substrate, probably due to the lower work function of Cu(100) compared with that of the Cu(111) surface. However, the intermolecular interaction on both of the surfaces was much stronger than the mutual effect between the fullerenes and the substrate. High-resolution STM images suggested that Gd@C82 took a preferred molecular orientation on the substrates. Although we observed spatial inhomogeneity on a single molecule adsorbed at the step edge of Cu(111), a uniform electronic structure was found as the metallofullerenes formed monolayer islands on the terrace. These results emphasized the important role played by the interaction among the adsorbed metallofullerenes. ’ AUTHOR INFORMATION Corresponding Author

*Tel: 86-10-8823-3595. E-mail: [email protected] (B.S.), xhqiu@ nanoctr.cn. (X.Q.). Author Contributions §

These authors contributed equally.

’ ACKNOWLEDGMENT The authors acknowledge the funding support from the Chinese Academy of Sciences (CAS) “Hundred Talents Program” (H8291310S3), the National Natural Science Foundation of China (20971124 and 20973046), the MOST 973 Program (2009CB930204) and the Innovation Project of High Energy Physics Institute (K651551U11 and H75453BOU2).

ARTICLE

(2) Chen, C. Y.; Xing, G. M.; Wang, J. X.; Zhao, Y. L.; Li, B.; Tang, J.; Jia, G.; Wang, T. C.; Sun, J.; Xing, L.; Yuan, H.; Gao, Y. X.; Meng, H.; Chen, Z.; Zhao, F.; Chai, Z. F.; Fang, X. H. Nano Lett. 2005, 5, 2050. (3) Yasutake, Y.; Shi, Z. J.; Okazaki, T.; Shinohara, H.; Majima, Y. Nano Lett. 2005, 5, 1057. (4) Kato, H.; Kanazawa, Y.; Okumura, M.; Taninaka, A.; Yokawa, T.; Shinohara, H. J. Am. Chem. Soc. 2003, 125, 4391. (5) Tsuchiya, T.; Sato, K.; Kurihara, H.; Wakahara, T.; Nakahodo, T.; Maeda, Y.; Akasaka, T.; Ohkubo, K.; Fukuzumi, S.; Kato, T.; Mizorogi, N.; Kobayashi, K.; Nagase, S. J. Am. Chem. Soc. 2006, 128, 6699. (6) Iizumi, K.; Uchino, Y.; Ueno, K.; Koma, A.; Saiki, K.; Inada, Y.; Nagai, K.; Iwasa, Y.; Mitani, T. Phys. Rev. B 2000, 62, 8281. (7) Ton-That, C.; Dowd, A.; Shard, A. G.; Dhanak, V. R.; Taninaka, A.; Shinohara, H.; Welland, M. E. Phys. Rev. B 2007, 76, 165429. (8) Ton-That, C.; Shard, A. G.; Dhanak, V. R.; Shinohara, H.; Bendall, J. S.; Welland, M. E. Phys. Rev. B 2006, 73, 205406. (9) Treier, M.; Ruffieux, P.; Fasel, R.; Nolting, F.; Yang, S. F.; Dunsch, L.; Greber, T. Phys. Rev. B 2009, 80, 081403. (10) Shi, B. R.; Wang, X. S.; Huang, H. J.; Yang, S. H.; Heiland, W.; Cue, N. J. Phys. Chem. B 2001, 105, 11414. (11) Taninaka, A.; Shino, K.; Sugai, T.; Heike, S.; Terada, Y.; Hashizume, T.; Shinohara, H. Nano Lett. 2003, 3, 337. (12) Leigh, D. F.; Owen, J. H. G.; Lee, S. M.; Porfyrakis, K.; Ardavan, A.; Dennis, T. J. S.; Pettifor, D. G.; Briggs, G. A. D. Chem. Phys. Lett. 2005, 414, 307. (13) Leigh, D. F.; Norenberg, C.; Cattaneo, D.; Owen, J. H. G.; Porfyrakis, K.; Bassi, A. L.; Ardavan, A.; Briggs, G. A. D. Surf. Sci. 2007, 601, 2750. (14) Hosokawa, T.; Fujiki, S.; Kuwahara, E.; Kubozono, Y.; Kitagawa, H.; Fujiwara, A.; Takenobu, T.; Iwasa, Y. Chem. Phys. Lett. 2004, 395, 78. (15) Grobis, M.; Khoo, K. H.; Yamachika, R.; Lu, X. H.; Nagaoka, K.; Louie, S. G.; Crommie, M. F.; Kato, H.; Shinohara, H. Phys. Rev. Lett. 2005, 94, 136802. (16) Lu, X. H.; Grobis, M.; Khoo, K. H.; Louie, S. G.; Crommie, M. F. Phys. Rev. B 2004, 70, 115418. (17) Fujiki, S.; Kubozono, Y.; Hosokawa, T.; Kanbara, T.; Fujiwara, A.; Nonogaki, Y.; Urisu, T. Phys. Rev. B 2004, 69, 045415. (18) Hasegawa, Y.; Ling, Y.; Yamazaki, S.; Hashizume, T.; Shinohara, H.; Sakai, A.; Pickering, H. W.; Sakurai, T. Phys. Rev. B 1997, 56, 6470. (19) Sun, B. Y.; Li, M. X.; Luo, H. X.; Shi, Z. J.; Gu, Z. N. Electrochim. Acta 2002, 47, 3545. (20) Smoluchowski, R. Phys. Rev. 1941, 60, 661. (21) Chen, Y. M.; Deng, K.; Qiu, X. H.; Wang, C. ChemPhysChem 2010, 11, 379. (22) Butcher, M. J.; Nolan, J. N.; Hunt, M. R. C.; Beton, P. H.; Dunsch, L.; Kuran, P.; Georgi, P.; Dennis, T. J. S. Phys. Rev. B 2001, 64, 195401. (23) David, T.; Gimzewski, J. K.; Purdie, D.; Reihl, B.; Schlittler, R. R. Phys. Rev. B 1994, 50, 5810. (24) Maxwell, A. J.; Bruhwiler, P. A.; Arvanitis, D.; Hasselstrom, J.; Johansson, M. K. J.; Martensson, N. Phys. Rev. B 1998, 57, 7312. (25) Russier, V.; Badiali, J. P. Phys. Rev. B 1989, 39, 13193. (26) Lu, X. H.; Grobis, M.; Khoo, K. H.; Louie, S. G.; Crommie, M. F. Phys. Rev. Lett. 2003, 90, 096802. (27) Takata, M.; Umeda, B.; Nishibori, E.; Sakata, M.; Saito, Y.; Ohno, M.; Shinohara, H. Nature 1995, 377, 46.

’ REFERENCES (1) Meng, H.; Xing, G. M.; Sun, B. Y.; Zhao, F.; Lei, H.; Li, W.; Song, Y.; Chen, Z.; Yuan, H.; Wang, X. X.; Long, J.; Chen, C. Y.; Liang, X. J.; Zhang, N.; Chai, Z. F.; Zhao, Y. L. ACS Nano 2010, 4, 2773. 6268

dx.doi.org/10.1021/jp1121454 |J. Phys. Chem. C 2011, 115, 6265–6268