NANO LETTERS
Reduced Metallic Properties of Ligand-Stabilized Small Metal Clusters
2003 Vol. 3, No. 3 305-307
Huijing Zhang,† Gu1 nter Schmid,‡ and Uwe Hartmann*,† Institute of Experimental Physics, UniVersity of Saarbru¨ cken, P.O. Box 151150, D-66041 Saarbru¨ cken, Germany, and Institute of Inorganic Chemistry, UniVersity of Essen, UniVersita¨ tsstrasse 5-7, D-45117 Essen, Germany Received November 15, 2002; Revised Manuscript Received January 10, 2003
ABSTRACT Low-temperature ultrahigh-vacuum scanning tunneling microscopy was employed to analyze the electronic behavior of Au55 clusters stabilized by [P(C6H5)3]12Cl6 ligands. At 7 K, the actual arrangement of the C6H5 rings of the ligand molecules could be imaged. Spectroscopic data reveal discrete energy levels with an average spacing of 170 meV that can be attributed to the Au55 core. Additionally, characteristic charge-quantization phenomena were observed. Energy and charge quantization both support the view that the clusters consist of a metallic core extending slightly beyond the first close-packed shell of Au atoms.
Small metal clusters of a few up to a few hundred atoms are of fundamental interest because their physical and chemical properties generally differ from those of individual molecules as well as from those of bulk solids.1 Whereas bigger ligandcoated metal clusters, such as Pt309, clearly exhibit metallic behavior,2 it can not yet be clarified to what extent this holds for the smaller cluster Au55. This is surprising because the cluster compound Au55[P(C6H5)3]12Cl6 certainly belongs to the most intensively investigated species. Clusters permit the exploration of how atomic-scale properties gradually develop into bulk properties. Because of their unique physical properties, they have also been considered for technical applications.3,4 The Au55 cluster belongs to the family of magic-atom-number metal clusters that consist of a central atom and a number of close-packed atomic shells, where the nth one is composed of 10n2 + 2 atoms. The clusters are wet-chemically synthesized and stabilized by an organic ligand shell.5 The latter makes the compound largely chemically inert in comparison to naked clusters, which cannot be investigated under ambient conditions. The two-shell compound Au55[P(C6H5)3]12Cl6 exhibits particularly interesting optical6 and electronic properties.3,7,8 An obvious question is to what extent these properties are determined by the ligand molecules and atoms. Although a variety of different experimental methods and theoretical approaches have been applied to characterize the cluster’s structure and electronic properties,9-16 it remained unclear whether the properties of the ligand-coated cluster deviate from those of the naked one and what the properties of the * Corresponding author. E-mail:
[email protected]. Phone: ++49 (681) 302-3798. Fax: ++49 (681) 302-3790. † University of Saarbru ¨ cken. ‡ University of Essen. 10.1021/nl0258980 CCC: $25.00 Published on Web 02/08/2003
© 2003 American Chemical Society
latter are. Earlier Mo¨ssbauer investigations indicated largely nonmetallic behavior in contrast to that of Pt309.10 A vanishing density of electronic states at the Fermi level has also been concluded from the absence of a linear contribution to the specific heat.11 However, X-ray-induced photoelectron spectroscopy,12 X-ray absorption spectroscopy,13 and ac and dc conductivity measurements14 all point toward metallic behavior. Furthermore, molecular spectral features known from P(C6H5)3-stabilized Au13 are not found for Au55.15 In recent work again based on photoelectron spectroscopy, it was even shown that Au55 behaves nonmetallically with an intact ligand shell, whereas a transition to metallic behavior occurs after partially decomposing the ligand shell by the removal of the Cl atoms.16 Scanning tunneling microscopy and spectroscopy (STM/ STS) are appropriate methods with which to explore electronic phenomena on a local scale.17 There have been many attempts to image the Au55[P(C6H5)3]12Cl6 cluster at high spatial resolution.7,8,18,19 Although it was possible to determine the diameter of the cluster fairly precisely,7 atomic or molecular details could not be imaged in a convincing way. At elevated temperatures, this is due to the thermal motion of the highly mobile ligand molecules.20 Additionally, the clusters exhibit a high mobility on top of substrates, where they are physisorbed. A dedicated UHV-STM21 was employed for the present investigations. Au films (50 nm thick) were deposited on highly oriented pyrolytic graphite (HOPG) by thermal evaporation under UHV conditions. Subsequent annealing results in large (111) terraces of atomic flatness.21 The clusters were then ex situ deposited on the Au(111) films from solution by spin coating. Afterward, the whole sample
Figure 2. Tunneling spectra acquired at the two distinct locations marked in Figure 1A. The dashed curve (a) was taken right above a C6H5 ring, and the solid one (b), next to the ring. The bias refers to the substrate potential. The arrows indicate conductivity peaks that precisely coincide for both spectra. The inset shows the current versus voltage curve on top of a phenyl ring for almost the entire tunneling regime.
Figure 1. (A) STM image of a single Au55[P(C6H5)3]12Cl6 cluster on a Au(111) surface at 7 K, obtained at a bias of 2 V and a current of 100 pA using a Pt-Ir tip. Image size: 3.3 nm × 2.9 nm. (B) Space-filling model of the cluster compound: cuboctahedral core with 55 Au atoms (yellow), 12 P(C6H5)3 molecules (with P pink, C gray, and H blue) bound to the 12 edges of the cuboctahedron, and 6 Cl atoms (violet) located in the center of the 6 square faces of the Au core.
is transferred into the UHV chamber for STM/STS investigations. The STM images confirm that partially disordered monolayers of clusters are typical. Within many local areas, the clusters are arranged in a rather perfect hexagonal closest packing. Figure 1A shows an individual cluster within such an arrangement imaged at 7 K. Thermal motion is sufficiently reduced, and the molecular structure of the ligand is partially visible. The latter corresponds fairly well to the space-filling model (Figure 1B), which was constructed such that the cuboctahedral geometry of the Au55 core and the locations of the 12 P(C6H5)3 molecules and the 6 Cl atoms agree with previous assumptions.5 A comparison between the STM image and the model in Figure 1 shows that the P(C6H5)3 molecules are resolved and that their mutual spacing and location coincide with the expected values. We performed locally resolved STS at well-defined sites, such as those marked in Figure 1A, on individual clusters. The deduced differential conductivity dI/dV as a function of the tunneling bias is related to the density of occupied and unoccupied electronic states close to the Fermi level.17 The spectra shown in Figure 2 result from the two distinct locations indicated in Figure 1A: the solid curve was taken 306
next to the C6H5 rings (i.e., in tunneling between the tip and the naked Au shell). The dashed curve was acquired precisely on top of a C6H5 ring. Around zero bias, both spectra exhibit a Coulomb blockade2 with a total width of about 1.2 V where the conductivity is largely suppressed. Since, however, we have a finite temperature of 7 K, the conductivity does not completely vanish. There is also a certain probability of cotunneling within the Coulomb blockade.22 One identifies small conductivity oscillations within the blockade regime, where most of the maxima precisely coincide for the two spectra. It can thus be concluded that the molecular electronic states of the ligand do not dominate the conductivity in this regime. Electronic transport is evidently much more complex than it is for a completely nonmetallic cluster.16 There is a finite density of electronic states close to the Fermi level rather than a perfect gap. Upon varying the bias voltage between the tip and the substrate, we shift the potential differences across the two involved tunneling junctions established between the tip and the cluster and the cluster and the substrate. Thus, we selectively trace the discrete energy levels of the cluster compound. The occupied as well as unoccupied levels are visible in Figure 2 in terms of conductivity oscillations. Since the electric potential of the cluster is floating somewhere between that of the substrate and that of the tip, the interpeak spacing in Figure 2 does not precisely correspond to the energy-level spacing of the cluster. If one takes into account, however, that the electrical capacitance between the sample and the cluster equals 3 times the value between the tip and the cluster, a ratio that was determined from accompanying current-distance STM measurements, then the average level spacing of the Au55[P(C6H5)3]12Cl6 cluster in the vicinity of the Fermi level amounts to about 170 meV. If we compare this value to that expected from a simple free-electron model,23 then we deduce an “electronically apparent” cluster diameter of 0.9 nm. The reduction in width of the cluster’s valence band with respect to that of bulk Au24 has been accounted for in this estimation. If we assume that the clusters have on average a 2-fold degeneracy Nano Lett., Vol. 3, No. 3, 2003
Figure 3. Conductivity spectrum taken on top of a phenyl ring. P(C6H5)3 ligand molecules were deposited as a monolayer on top of a Au(111) substrate.
of the levels, then the apparent diameter would be 1.1 nm. These rough estimates are significantly below the geometrically determined diameter of 1.4 nm for Au55 and are slightly above the geometrical diameter of its first atomic shell consisting of 12 atoms. A further confirmation of the metallic behavior of the cluster compound is obtained from the current-voltage tunneling curve shown in the inset of Figure 2. Tunneling through the cluster clearly involves the Coulomb blockade and staircase, both being characteristic for single-electron tunneling through small metallic particles. The difference between the geometrical and the “electronically apparent” diameter is most likely caused by the six Cl atoms being located on the six square faces of the cuboctahedral Au55 surface, as indicated in Figure 1B. Each of the strongly electronegative Cl atoms establishes a chemical bond that involves covalent and ionic contributions. The net effect is a depletion of valence electrons in the outer cluster shell. It has recently been shown that Au55[P(C6H5)3]12Cl6 behaves like a regular metal particle upon removing the Cl atoms.16 Hence, electron depletion and the electrostatic potential created by these ligand atoms effectively reduce the free-electron diameter of the cluster. A similar phenomenon was found for ligand-coated Pt309 clusters investigated by Mo¨ssbauer spectroscopy.2 Upon tunneling through the P(C6H5)3 ligand molecules, these contribute only through their dielectrical properties rather than in terms of pronounced unoccupied or occupied electronic states. To determine why this is the case, we deposited monolayers of ligand molecules onto the Au(111) substrates as the clusters were deposited before. At 7 K, the ligands were imaged with molecular resolution. Figure 3 shows a conductivity spectrum taken on top of a phenyl ring. Significant unoccupied states at energies above 0.25 eV are visible as well as occupied ones below -1.85 eV. It can, a priori, not totally be excluded that the ligand molecules adsorbed to the perfect Au(111) surface exhibit electronic states that deviate from those of the ligands attached to the cuboctahedral cluster cores. However, from symmetry considerations, the degeneracy of the energy levels should not be too different in both cases. This degeneracy is mainly determined by the internal symmetry of the P(C6H5)3 molecule rather than by the particular bond between P and Au55 or Au(111). It is thus unlikely that the respective energy Nano Lett., Vol. 3, No. 3, 2003
levels in Figure 2b are introduced by the particular bond between the ligand and the cluster. Current-distance measurements show that, under conditions under which spectrum b in Figure 2 was taken, roughly one-third of the applied bias between the tip and the substrate drops across the involved ligand molecule. It is thus evident that only spectral features beyond an energy of 0.75 eV could result from molecular electronic states. In turn, those states within the Coulomb blockade regime that largely coincide for both spectra in Figure 2 are clear fingerprints of the ligandstabilized cluster compound rather than of the ligand molecules. The long-standing controversy of whether the Au55 compound is metallic or not is from our point of view removed by the presented results. The cluster exhibits metallic properties, where the diameter of the metallic core is, however, reduced with respect to the geometrical diameter of Au55. Acknowledgment. We thank H. Gao and D. Mautes from the Experimental Physics Department of the University of Saarbru¨cken for processing the images in Figure 1 and M. Springborg and S. Pranab from the Department of Physical Chemistry of the University of Saarbru¨cken for helpful discussions on the electronic structure of the cluster. References (1) Cohen, M. L.; Knight, W. D. Phys. Today 1990, 42-50 (December). (2) Mulder, F. M.; Stegink, T. A.; Thiel, R. C.; de Jongh, L. J.; Schmid, G. Nature (London) 1994, 367, 716-718. (3) Kastner, M. A. Phys. Today 1993, 24-31 (January). (4) Scho¨n, G.; Simon, U. Colloid Polym. Sci. 1995, 273, 101-117, 202218. (5) Clusters and Colloids; Schmid, G., Ed.; VCH: Weinheim, Germany, 1994. (6) Fauth, K.; Kreibig, U.; Schmid, G. Z. Phys. D: At., Mol. Clusters 1991, 20, 297-300. (7) Houbertz, R. et al. Europhys. Lett. 1994, 28, 641-646. (8) Chi, L. F. et al. Appl. Phys. A 1998, 66, 187-190. (9) Feigenspan, Th.; Houbertz, R.; Hartmann, U. Nanostruct. Mater. 1997, 9, 367-370. (10) Mulder, F. M.; Thiel, R. C.; de Jongh, L. J.; Gubbens, P. C. M. Nanostruct. Mater. 1996, 7, 269-292. (11) Goll, G.; Lo¨hneysen, H. V.; Kreibig, U.; Schmid, G. Z. Phys. D: At., Mol. Clusters 1991, 20, 329-331. (12) van der Putten, D.; Zanoni, R. Phys. Lett. A 1995, 208, 345-350. (13) Marcus, M. A.; Andrews, M. P.; Zegenhagen, J.; Bommannavar, A. S.; Montano, P. Phys. ReV. B 1990, 42, 3312-3316. (14) Brom, H. B.; van Staveren, M. P. J.; de Jongh, L. J. Z. Phys. D: At., Mol. Clusters 1991, 20, 281-287. (15) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; Springer: Berlin, 1995. (16) Boyen, H.-G. et al. Phys. ReV. Lett. 2001, 87, 276401. (17) Chen, C. J. Introduction to Scanning Tunneling Microscopy; Oxford University Press: New York, 1993. (18) Wierenga, H. A. et al. AdV. Mater. 1990, 2, 482-484. (19) Becker, C.; Fries, Th.; Wandelt, K.; Kreibig, U.; Schmid, G. J. Vac. Sci. Technol., B 1991, 9, 810-813. (20) Schmid, G. Struct. Bonding (Berlin) 1985, 62, 51-83. (21) Zhang, H.; Memmert, U.; Houbertz, R.; Hartmann, U. ReV. Sci. Instrum. 2001, 72, 2613-2617. (22) Hanna, A. E.; Tuominen, M. T.; Tinkham, M. Phys. ReV. Lett. 1992, 68, 3228-3231. (23) Halperin, W. P. ReV. Mod. Phys. 1986, 58, 533-606. (24) Mansikka-aho, J.; Manninen, M.; Hammare´n, E. Z. Phys. D: At., Mol. Clusters 1991, 21, 271-279.
NL0258980 307
