Abstract
Recent reports of C60-functionalized metal tips [Kelly, K. F., Sarkar, D., Hale, G. D., Oldenburg, S. J. & Halas, N. J. (1996) Science 273, 1371–1373] and carbon nanotube tips [Dai, H., Hafner, J. H., Rinzler, A. G., Colbert, D. T. & Smalley, R. E. (1996) Nature (London) 384, 147–151] demonstrate the potential of controlling the chemical identity and geometric structure of tip atoms in scanning tunneling microscopy (STM). This work reports the performance of a heteropolyacid (HPA)-functionalized Pt/Ir tip, which was formulated by contacting a mechanically formed tip with a solution of H3PW12O40 molecules. Attachment of an H3PW12O40 molecule on the metal tip was confirmed by observing the characteristic negative differential resistance (NDR) behavior of H3PW12O40 in tunneling spectroscopy. Atomic resolution images of bare graphite as well as of H6P2W18O62 HPA monolayers on graphite were successfully obtained with a Pt/Ir-HPA tip. In the H3PW12O40 molecule on a metal tip, it is likely that a terminal oxygen of W⩵O (an oxygen species projecting outward from the pseudospherical H3PW12O40 molecule) serves as an atomically sharp and stable tip. Additionally, superimposed superperiodic structures commensurate with the underlying graphite lattice were regularly observed with the modified tips. This result suggests that tip functionalization with these metal oxide molecules may enhance resolution in a fashion analogous to functionalization with C60.
Keywords: STM tip‖functionalization‖heteropolyacid‖terminal oxo group‖atomically sharp probe
Atomically sharp and stable tips are of great importance in scanning tunneling microscopy (STM). Many ways to produce sharp tips of various materials for STM have been developed, including electrolytic (electrochemical) polishing/etching, chemical polishing/etching, ion milling, cathode sputtering, whisker growth, electron-beam deposition, flame polishing, mechanical sharpening, cutting, machining, fragmenting, and so on (1). However, none of these production methods is universally applicable to all materials of interest. It is emphasized that the critical components of high-resolution tunneling tips are actually minitips with radii less than 100 Å, the structure of which is much more difficult to control (2). Minitips occur frequently as a result of the variability of tip preparation techniques, but the origin of these structures is not clear and there are no truly reproducible means of generating them. The shape and arrangement of the minitips are important factors that can affect the stability of the image (3). If several atoms or small clusters of atoms at the end of the tip act independently, the image will actually be a superposition of images. By positioning graphite-covered W tips consisting of four or fewer independent atoms in different arrangements and different orientations with respect to the graphite, for example, it has been shown that a variety of different periodic images of graphite can be produced (4). The chemical identity of the tip atom is also important in STM imaging. A simulation study on the imaging of graphite with various single-atom tips, such as Na and Ca, has shown that the graphite image is very sensitive to the identity of the tip atom, and that the corrugation amplitude is much smaller than expected (5).
Heteropolyacids (HPAs) are early-transition metal oxygen anion clusters that exhibit a wide range of sizes, compositions, and molecular architectures (6). One of the great advantages of HPAs is their well-defined molecular structure and tunable redox properties (7). A recent STM study of HPAs has shown that these inorganic molecules form two-dimensional ordered arrays on graphite surfaces and exhibit distinctive current-voltage (I–V) behavior, referred to as negative differential resistance (NDR) in their tunneling spectra (8). NDR peak positions observed for pure HPA arrays not only serve as fingerprints of the HPAs, but also correlate well with the reduction potentials of HPAs (9).
In this work we attempted to chemically modify a mechanically formed Pt/Ir tip by attaching HPA molecules to functionalize the Pt/Ir tip as an STM probe. Among various HPA structural classes, Keggin-type (10) H3PW12O40 was chosen for its structural simplicity. Fig. 1A shows the molecular structure of the soccer ball-like [PW12O40]3− heteropolyanion, constructed from the x-ray diffraction (XRD) data (11). In this representation, oxygen atoms are represented as spheres. The structure of Keggin-type heteropolyanion, [PW12O40]3−, consists of a heteroatom, P, at the center of the anion cluster, tetrahedrally coordinated to four oxygen atoms. This tetrahedron is surrounded by four groups of three edge-sharing WO6 octahedra, and the groups are interconnected by corner-sharing. A sphere at the outermost vertex of each WO6 octahedron represents a terminal oxo species. The W⩵O terminal group has a bond length of 1.7Å. A single H3PW12O40 molecule has twelve W⩵O terminal bonds at its outermost surface, depicted schematically in Fig. 1B. The molecular size of H3PW12O40 is 11 ≈ 12 Å as determined by XRD (11) and STM (12). The performance of Pt/Ir-H3PW12O40 tips was demonstrated by probing bare graphite as well as Wells–Dawson-type (13) H6P2W18O62 HPA monolayers on graphite.
Figure 1.
(A) Molecular structure of the pseudospherical Keggin-type [PW12O40]3− heteropolyanion. Oxygen atoms are represented as spheres. A sphere projecting outward in each WO6 octahedron represents terminal oxygen species. (B) A simplified representation of the W⩵O projection.
Materials and Methods
Preparation of Pt/Ir-H3PW12O40 Tip.
H3PW12O40 was purchased from Aldrich Chem (Metuchen, NJ), and H6P2W18O62 was synthesized according to published methods (14). About 0.01 M aqueous solutions of each sample were prepared. A mechanically formed Pt/Ir (90/10) tip was contacted with a drop of the H3PW12O40 solution, and then the tip was allowed to dry in air for 1 h at room temperature in order for H3PW12O40 molecules to be attached on the tip. The tip was maintained in a downward orientation during the entire processes.
STM Imaging and Tunneling Spectroscopy.
STM images of highly oriented pyrolytic graphite were obtained with Pt/Ir-H3PW12O40 tips in air by using a Topometrix TMX 2010 Discoverer scanning tunneling microscope. Scanning was done in the constant current mode at a positive sample bias of 100 mV and a tunneling current of 1 ≈ 1.5 nA. All STM images presented in this report are unfiltered, and the reported periodicities (lattice constants) represent average values determined by performing two-dimensional fast Fourier transform analyses on at least three images for each sample. Tunneling spectra were also obtained with Pt/Ir-H3PW12O40 tips in air on a graphite surface. Both Topometrix TMX 2010 and LK Technologies LK-1000 scanning tunneling microscopes were used to confirm consistency and reproducibility of tunneling spectra. To measure a tunneling spectrum, the sample bias was ramped from −2 to +2 V with respect to the tip and the tunneling current was monitored. Attachment of an H3PW12O40 molecule on the Pt/Ir tip was confirmed by observation of its characteristic NDR behavior in tunneling spectroscopy measurements before STM imaging. Several tunneling spectra were measured on the graphite surface with the Pt/Ir-H3PW12O40 tip to ensure the stability of the tip and the reproducibility of the tunneling spectra. The Pt/Ir-H3PW12O40 tip was also used as an STM probe in imaging H6P2W18O62 molecules on graphite to test its ability to probe inorganic monolayers. For this purpose, a drop of 0.01 M aqueous solution of H6P2W18O62 sample was deposited on a graphite surface and allowed to dry for 1 h at room temperature before imaging.
Results and Discussion
Confirmation of H3PW12O40 Attachment.
Fig. 2A shows the typical I–V spectrum obtained on bare graphite using a Pt/Ir-H3PW12O40 tip. The voltage resolution of our STM instruments is 0.05 V. The spectrum did not represent a typical I–V response of graphite, but instead showed NDR behavior, which was also observed consistently for H3PW12O40 deposited on graphite and probed by a bare metal tip (12). This result indicates that H3PW12O40 molecules were attached to the Pt/Ir tip and that electrons tunneled through these molecules. The NDR peak position measured on bare graphite with the Pt/Ir-H3PW12O40 tip is also comparable to that of a H3PW12O40 monolayer on graphite probed by a normal Pt/Ir tip. Fig. 2 B and C show the distribution of NDR peak voltages of H3PW12O40 measured for each case (NDR peak voltage was defined as the voltage at which the local maximum current was observed in the tunneling spectrum). In the normal case (H3PW12O40 monolayer probed with a bare Pt/Ir tip), the most frequent NDR peak voltage measured was found to be −1.20 V, and the statistical average of NDR peak voltages was −1.14 ± 0.09 V. In the case of Pt/Ir-H3PW12O40 system, the modified tip exhibited the most frequent NDR peak at −1.15 V with a statistical average of −1.15 ± 0.07 V. The values observed in both cases are nearly indistinguishable, indicating the presence of H3PW12O40 on the modified tip in the latter case.
Figure 2.
(A) A typical I–V spectrum obtained on bare graphite by using a Pt/Ir-H3PW12O40 tip. (B) Distribution of NDR peak voltages of H3PW12O40 on graphite obtained with a normal Pt/Ir tip (total number of tunneling spectra = 24). (C) Distribution of NDR peak voltages measured on bare graphite with a Pt/Ir-H3PW12O40 tip (total number of tunneling spectra = 62).
Atomically Sharp Pt/Ir-H3PW12O40 Probe.
Controlling the number and orientation of H3PW12O40 molecules attached on a Pt/Ir tip is still challenging. Despite this difficulty, atomic resolution images of graphite were frequently obtained with a Pt/Ir-H3PW12O40 tip. The atomic resolution suggests that electron tunneling occurs through a sharp feature of the H3PW12O40 molecule on the Pt/Ir tip, presumably a W⩵O terminal bond. Considering the molecular size of H3PW12O40 and its geometric structure shown in Fig. 1, a terminal oxo group projecting outward from the pseudospherical polyanion can possibly serve as an atomically sharp probe.
It has been demonstrated that a single electronegative adatom, such as sulfur or oxygen, on a W tip contributes more to the filled density of states on the tip than to the empty states (15). Atomic resolution images are obtained when electrons flow primarily from the single adatom on top of the metal tip (the occupied density of states) to the sample (the empty density of states). However, no atomic resolution image is obtained for the opposite bias, when electrons tunnel from a sample (the occupied density of states) to a large number of tip-metal surface atoms (the empty density of states), because the electronegative adatom contributes only very weakly to the unoccupied density of states (15). In imaging single adsorbed species on metal surfaces, it was also demonstrated that spatial resolution was increased in topographical scans by attaching a C⩵O molecule on the tip (oxygen atom is projecting outward on attachment) (16). Our STM imaging condition and H3PW12O40-functionalized tip meet the above requirements for obtaining atomic resolution images; electrons flow from tip to sample in the imaging mode, and the Pt/Ir-H3PW12O40 tip has an electronegative oxygen end atom chemically bound to tungsten. Moreover, the rigid and well-defined molecular structure of H3PW12O40 provides no need to control the position of W⩵O, which makes Pt/Ir-H3PW12O40 a highly stable and atomically sharp probe for STM imaging.
The performance of Pt/Ir-H3PW12O40 as an atomically sharp and stable tip was demonstrated by probing bare graphite as well as Wells–Dawson-type (13) H6P2W18O62 HPA monolayers on graphite. Fig. 3A shows the atomic resolution image of graphite obtained with a Pt/Ir-H3PW12O40 tip. This image clearly shows the hexagonal symmetry of graphite with a periodicity of 2.46 Å. Fig. 3B shows a molecular-resolution STM image of a H6P2W18O62 monolayer on graphite probed by a Pt/Ir-H3PW12O40 tip. This image clearly shows a two-dimensional well-ordered array with rugby-ball-like (ellipsoidal) features. The unit cell of H6P2W18O62 arrays constructed on the basis of lattice constants determined from two-dimensional fast Fourier transform shows that the arrays have the primitive unit cell (rhombus) and the conventional unit cell (centered oblique rectangle), shown in Fig. 3C. The primitive cell has sides of 14.6 Å with an included angle of 35.2°, and the conventional unit cell has a minor axis of 10.4 Å. The molecular structure of the [P2W18O62]6− heteropolyanion constructed from the x-ray diffraction data (14) is shown in Fig. 3D. This molecule consists of two defect-Keggin-type [PW9O34]9− fragments, and has a rugby-ball-like (ellipsoidal) shape with dimensions of 11 Å × 14.5 Å. This level of agreement between molecular dimensions and the periodicity of heteropolyanion monolayer is comparable to that obtained with metal tips (12). Thus, a well-formulated Pt/Ir-H3PW12O40 tip can provide high-resolution images; the tunneling tip is of atomic scale rather than the molecular scale (1 nm) of the HPA.
Figure 3.
(A) Atomic resolution STM image of graphite obtained with a Pt/Ir-H3PW12O40 tip. (B) STM image of H6P2W18O62 monolayer on graphite obtained with a Pt/Ir-H3PW12O40 tip. (C) Schematic representation of unit cell of the H6P2W18O62 array. (D) Molecular structure of the ellipsoidal Wells–Dawson-type [P2W18O62]6− heteropolyanion. The polyanion consists of two defect-Keggin-type fragments, [PW9O34]9−. Each fragment consists of a central PO4 tetrahedron sharing corners with nine WO6 octahedra—the octahedra are somewhat distorted from an ideal octahedron. Three WO6 octahedra form a compact group by sharing edges, whereas the remaining six octahedra in each of the [PW9O34]9− fragments form a zigzag ring by alternately sharing edges and corners. The two fragments are linked by six nearly linear W—O—W bonds.
Superimposed Images of Superperiodic Structures and Underlying Graphite.
Fig. 4 shows a set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip in air with varying scan sizes. Fig. 4A clearly shows well-ordered hexagonal superperiodic features with a periodicity of 70.4 Å. The periodicity and orientation of the superperiodic structures were consistent, regardless of the variation of scanning parameters such as set point current, scan angle (rotation), and scan rate. More importantly, the superperiodic and real-size graphite structures were probed simultaneously by decreasing the scan size (Fig. 4B). Even in a very small scan area (Fig. 4C), we could observe bright domains representing the superperiodic structure. The simultaneous observation of superimposed images of superperiodic structures and real-size graphite obtained with a Pt/Ir-H3PW12O40 tip in air is quite interesting.
Figure 4.
(A-C) A set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip with varying scan size. (D) A schematic representation of unit cells of the superperiodic structure and underlying graphite arrays for φ = 27°. φ represents an azimuthal angle between lattice vectors, a1 and b1. Superperiodic lattice vectors can be expressed in terms of the graphite lattice vectors by b1 = 18a1 + 15a2 and b2 = −15a1 + 33a2 for φ = 27°, and b1 = 15a1 + 18a2 and b2 = −18a1 + 33a2 for φ = 33°.
One more important feature of the superimposed images shown in Fig. 4 is that both the superperiodic structure and real-size graphite have hexagonal symmetry, but the orientations of the two are rotated with respect to each other. Fig. 4D shows the schematic representation of unit cells of the superperiodic structure and underlying graphite. The superperiodic structure has hexagonal symmetry (β = 60°) with lattice constants of b1 = b2 = 70.4 Å. The underlying graphite also has hexagonal symmetry (α = 60°) with lattice constants of a1 = a2 = 2.46 Å. Because both lattices have hexagonal symmetry, two reciprocal array representations can be built within the angle of 0° to 60°. The measured azimuthal angle (φ), between two lattice vectors, a1 and b1, can be either 27° or 33°. When applying the experimentally determined lattice parameters to the epicalc software developed at the University of Minnesota (17), we obtained a dimensionless potential (V/Vo) value of 0.5, indicating that the superperiodic structure is commensurate with the underlying hexagonal graphite.
Fig. 5 also shows another set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip, showing superimposed images of a well-ordered hexagonal superperiodic structure (β = 60°, b1 = b2 = 14.97 Å) and underlying real-size graphite (α = 60°, a1 = a2 = 2.46 Å), with array rotation with respect to each lattice (φ = 25.3° or 34.7°). These lattice constants also produced the dimensionless potential value of 0.5 by the epicalc simulation, indicating that the superperiodic structure is again commensurate with the underlying graphite.
Figure 5.
Another set of unusual STM images of graphite obtained with a Pt/Ir-H3PW12O40 tip with varying scan size, showing a superimposed hexagonal superperiodic structure (β = 60°, b1 = b2 = 14.97 Å) and underlying real-size graphite (α = 60°, a1 = a2 = 2.46 Å). φ is either 25.3° or 34.7°. Superperiodic lattice vectors can be expressed in terms of the graphite lattice vectors by b1 = 4a1 + 3a2 and b2 = −3a1 + 7a2 for φ = 25.3°, and b1 = 3a1 + 4a2 and b2 = −4a1 + 7a2 for φ = 34.7°.
As demonstrated in Figs. 4 and 5, hexagonal superperiodic structures commensurate with underlying graphite were observed in imaging graphite with Pt/Ir-H3PW12O40 tips, although the periodicity and rotation angle of the superperiodic structures with reference to underlying graphite varied. These images are similar to some of the reported Moiré-pattern images of graphite (with superperiodic structures ranging from 17 to 148 Å) obtained with bare metal tips (18–24). One of the most accepted explanations for the observation of superperiodic structures along with the underlying graphite lattice is the Moiré-rotation hypothesis, which assumes a simple rotation (misorientation) of surface graphite layer with respect to the underlying layer(s) (18–20). Moiré patterns are interference patterns arising from rotation between two layers of any repeating lattice, which can cause a superperiodic structure having the same symmetry as the original lattice. The images presented in Figs. 4 and 5 are particularly reminiscent of those reported by Bernhardt et al. (23) and Beyer et al. (24). In the both cases, the appearance of superperiodic structures was produced by manipulation of graphite sheets with the tip to generate rotation relative to the underlying graphite. As noted by Bernhardt et al. (23), the orientation angle φ of the superperiodic structure relative to the atomic lattice of the graphite layer is given by φ = 30° ± (θ/2), where D = d/2sin(θ/2), with d and D representing the lattice constants of the graphite and the superperiodic structure, respectively. The values of φ and D for the images in Fig. 5 are consistent with these equations. The larger superperiodicity observed in Fig. 4 should be associated with a value of φ nearer 30°, as observed, although the expected values of φ based on the superperiodicity in Fig. 4 (70.4 Å) are 29° and 31°.
It has been reported that three-dimensional electron tunneling (electron scattering) in lattice-mismatched systems can also produce the superperiodic and underlying lattice images at the same time (25). Surface layers often reconstruct or relax to minimize the surface energy, so it is not unreasonable to think the outermost layer might have a slightly different lattice constant than the layers beneath it, giving rise to a lattice-mismatched system. However, this mechanism may not account for the rotation of the superperiodic structure with respect to the graphite lattice.
As can be inferred from the molecular structure of H3PW12O40 shown in Fig. 1, another possibility for the origin of superimposed images is that the terminal W⩵O groups on a single H3PW12O40 molecule can serve as multiple tips depending on the molecular orientation, providing multiple pathways for electron tunneling. Simulation studies were carried out to test this hypothesis. The simulation allowed for up to four tips in any spatial arrangement desired, as well as arbitrary weights and heights for each tip. A hexagonal surface was created with a function reported in the literature for the simulation (21). Tips were examined that had the same symmetry as the surface but were scaled with integer and noninteger factors. Simulations were performed for both constant current topography and constant height topography STM modes. However, no superperiodic structures resulted from any of the simulations, indicating that independent multiple tips are not sufficient to explain the observed superperiodicity. Indeed, we conclude that tunneling with multiple tips can produce variations in the STM images at the length scale of the surface lattice constant or smaller—no long-range periodicity is produced in the STM images in such cases.
We conclude, therefore, that the images in Figs. 4 and 5 are Moiré-rotation patterns caused by misalignment of the top few graphite layers of the substrate. We observe superperiodic structures approximately 20% of the time when imaging graphite with Pt/Ir-H3PW12O40 tips. We have never observed such structures when imaging graphite with bare Pt/Ir tips in air with our instruments. The question then arises as to why the HPA-functionalized tips are more sensitive to these structures than are the corresponding bare metal tips. Some insight may be offered by the unusual sensitivity of C60-functionalized tips to electronic perturbations of the surface reported by Halas and coworkers (26–28), who demonstrated that C60-functionalized tips were much more sensitive to electron scattering from point defects on graphite than were bare metal tips. Although C60-terminated tips were able to image anisotropic threefold scattering from point defects at room temperature, cryogenic temperatures are required to produce similar images with a bare metal tip (29). This difference has been explained in terms of the differences in the density of states of clean versus C60-functionalized tips. With the bare metal tip, thermal broadening of the density of states about the Fermi level of the tip reduces its ability to resolve the electronic perturbations at point defects; low temperatures are required to sharpen the occupancy of the tip density of states at the Fermi level (29). In contrast, the tip functionalized with molecular C60 is characterized by electronic states of the molecule that are narrow in energy compared with the bulk bands of the metallic tip (26), and is able to resolve electron scattering from surface structures at ambient conditions. It is clear from the present work that this enhanced sensitivity to perturbations of surface electronic structure is not unique to C60 functionalization of STM tips. HPAs, pseudospherical metal oxide molecules of comparable dimensions to C60, may afford similar enhanced sensitivity to surface electronic structure in STM. Thus, this work extends the range of molecular tip-functionalization agents from conductive C60 to semiconducting metal oxides (HPAs). The observation of negative differential resistance behavior for HPAs [and the ability to vary the potential at which it occurs by choice of HPA structure and composition (8, 9)] suggests further opportunities for tuning the electronic properties of STM tips to probe a range of surface and subsurface electronic properties of imaged samples.
Conclusions
H3PW12O40 molecules were attached to a mechanically formed Pt/Ir tip to functionalize the metal tip as an STM probe. Attachment of H3PW12O40 on the Pt/Ir tip was confirmed by observing NDR behavior of the H3PW12O40 in tunneling spectra. Atomic resolution images of graphite as well as of H6P2W18O62 monolayers on graphite could be obtained with a Pt/Ir-H3PW12O40 tip. STM images of graphite probed by a Pt/Ir-H3PW12O40 tip showed the superimposed images of superperiodic structures and the real-size graphite lattice. The superperiodic structures have hexagonal symmetry commensurate with underlying graphite. These images represent Moiré patterns produced by misorientation of the top-most layers of the graphite surface. The HPA-functionalized tips appear to be particularly sensitive to these electronic perturbations of the surface owing to the discrete electronic levels provided by the semiconducting oxide (HPA) molecules.
Acknowledgments
The Topometrix TMX 2010 was acquired by means of an equipment grant from the United States Department of Energy. I.K.S. acknowledges fellowship support from the Seoam Scholarship Foundation.
Abbreviations
- STM
scanning tunneling microscopy
- HPA
heteropolyacid
- NDR
negative differential resistance
Footnotes
This paper results from the Arthur M. Sackler Colloquium of the National Academy of Sciences, “Nanoscience: Underlying Physical Concepts and Phenomena,” held May 18–20, 2001, at the National Academy of Sciences in Washington, DC.
This paper was submitted directly (Track II) to the PNAS office.
References
- 1.Melmed A J. J Vac Sci Technol B. 1991;9:601–608. [Google Scholar]
- 2.Rohrer G. In: Scanning Tunneling Microscopy and Spectroscopy. Bonnell D, editor. New York: VCH; 1993. pp. 155–187. [Google Scholar]
- 3.Park S, Nogami J, Quate C F. Phys Rev B. 1987;36:2863–2866. doi: 10.1103/physrevb.36.2863. [DOI] [PubMed] [Google Scholar]
- 4.Colton R J, Baker S M, Driscoll R J, Youngquist M G, Baldeschwieler J D, Kaiser W J. J Vac Sci Technol A. 1988;6:349–353. [Google Scholar]
- 5.Tersoff J, Lang N D. Phys Rev Lett. 1990;65:1132–1135. doi: 10.1103/PhysRevLett.65.1132. [DOI] [PubMed] [Google Scholar]
- 6.Pope M T. Heteropoly and Isopoly Oxometalates. New York: Springer; 1983. pp. 58–100. [Google Scholar]
- 7.Okuhara T, Mizuno N, Misono M. Adv Catal. 1996;41:113–252. [Google Scholar]
- 8.Kaba M S, Song I K, Barteau M A. J Phys Chem. 1996;100:19577–19581. [Google Scholar]
- 9.Song I K, Kaba M S, Barteau M A, Lee W Y. Catal Today. 1998;44:285–291. [Google Scholar]
- 10.Keggin J F. Nature (London) 1933;131:908–909. [Google Scholar]
- 11.Brown G M, Noe-Spirlet M R, Busing W R, Levy H A. Acta Crystallogr B. 1977;33:1038–1046. [Google Scholar]
- 12.Kaba M S, Song I K, Duncan D C, Hill C L, Barteau M A. Inorg Chem. 1998;37:398–406. doi: 10.1021/ic9705655. [DOI] [PubMed] [Google Scholar]
- 13.Dawson B. Acta Crystallogr. 1953;6:113–126. [Google Scholar]
- 14.Strandberg R. Acta Chem Scand A. 1975;29:350–358. [Google Scholar]
- 15.Tromp R M, van Loenen E J, Demuth J E, Lang N D. Phys Rev B. 1988;37:9042–9046. doi: 10.1103/physrevb.37.9042. [DOI] [PubMed] [Google Scholar]
- 16.Lee H J, Ho W. Science. 1999;286:1719–1722. doi: 10.1126/science.286.5445.1719. [DOI] [PubMed] [Google Scholar]
- 17.Hillier A C, Ward M D. Phys Rev B. 1996;54:14037–14051. doi: 10.1103/physrevb.54.14037. [DOI] [PubMed] [Google Scholar]
- 18.Kuwabara M, Clarke D R, Smith D A. Appl Phys Lett. 1990;56:2396–2398. [Google Scholar]
- 19.Liu C, Chang H, Bard A L. Langmuir. 1991;7:1138–1142. [Google Scholar]
- 20.Xhie J, Sattler K, Ge M, Venkateswaran N. Phys Rev B. 1993;47:15835–15841. doi: 10.1103/physrevb.47.15835. [DOI] [PubMed] [Google Scholar]
- 21.Cee V J, Patrick D L, Beebe T P., Jr Surf Sci. 1995;329:141–148. [Google Scholar]
- 22.Osing J, Shvets I V. Surf Sci. 1998;417:145–150. [Google Scholar]
- 23.Bernhardt T M, Kaiser B, Rademann K. Surf Sci. 1998;408:86–94. [Google Scholar]
- 24.Beyer H, Müller M, Schimmel Th. Appl Phys A. 1999;68:163–166. [Google Scholar]
- 25.Kobayashi K. Phys Rev B. 1996;53:11091–11099. doi: 10.1103/physrevb.53.11091. [DOI] [PubMed] [Google Scholar]
- 26.Kelly K F, Sarkar D, Hale G D, Oldenburg S J, Halas N J. Science. 1996;273:1371–1373. [Google Scholar]
- 27.Kelly K F, Halas N J. Surf Sci. 1998;416:L1085–L1089. [Google Scholar]
- 28.Kelly K F, Mickelson E T, Hauge R H, Margrave J L, Halas N J. Proc Natl Acad Sci USA. 2000;97:10318–10321. doi: 10.1073/pnas.190325397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Kushmerick J G, Kelly K F, Rust H-P, Halas N J, Weiss P S. J Phys Chem B. 1999;103:1619–1622. [Google Scholar]





