Nanoscale Organization of GaSe Quantum Dots on a Gold Surface

Oct 8, 2009 - Chikan and Kelley. 2002 2 (2), pp 141–145. Abstract: The synthesis, characterization, and purification of GaSe nanoparticles are descr...
0 downloads 0 Views 2MB Size
Download PDF
19102

J. Phys. Chem. C 2009, 113, 19102–19106

Nanoscale Organization of GaSe Quantum Dots on a Gold Surface Jingru Shao, Hoda Mirafzal, Jared R. Petker, Janice Lianne S. Cosio, David F. Kelley,* and Tao Ye* School of Natural Sciences, UniVersity of California, 5200 North Lake Road, Merced, California 95343 ReceiVed: August 4, 2009; ReVised Manuscript ReceiVed: September 14, 2009

Precise spatial organization and electronic coupling between quantum dots are pivotal for many potential applications. Typical spherical quantum dots in assemblies are separated by organic ligands and hence weakly coupled. GaSe nanoparticles are disk-like particles that are four atoms thick with tunable lateral dimensions. Previous spectroscopic investigations indicate the formation of nanoscale aggregates in which the quantum dots are strongly coupled. In this report, we show that the anisotropic properties of these particles may be exploited to assemble surface-stabilized superstructures with well-defined distances between the quantum dots. By changing the ligands adsorbed on the nanoparticle edges, three distinct aggregate morphologies can be produced. The surface chemistry of GaSe orients the nanoparticles on a gold surface and induces stacking in the surface normal direction. The discrete heights of such stacked aggregates suggest that the layers are held together by van der Waals interactions with a regular spacing. Such structures, with their well-defined electronic coupling, have potential implications in fundamental studies of photoinduced charge transfer and transport, as well as device fabrications. Introduction The optical and electronic properties of semiconductor nanoparticles (quantum dots) such as size dependent band gaps,1 very broad absorption bands, and possible multiexciton generation2-6 imply applications of quantum dots in technologies such as thin film field effect transistors, solar cells, and photodetectors.7-9 Utilization of such unique properties requires the nanoparticles to be spatially organized and coupled to each other in desired manners. Many reports on quantum dot superlattices and superstructures have appeared.7,8 In most cases, the nanoparticles in the assemblies are weakly coupled because of the relatively large interparticle separations. These large separations are caused by the presence of ligand shells that are required to stabilize and passivate the particle surfaces.7 Sintering9 and linking by bifunctional molecules10 are two strategies employed to increase the coupling between nanoparticles. However, the degree of control is limited as the morphologies of the assemblies are significantly affected in the process. In this report, we demonstrate that the anisotropic geometry of GaSe nanoparticles can be exploited to form one-dimensional aggregates with potentially improved electronic coupling. GaSe nanoparticles consist of a single two-dimensional GaSe sheet containing top and bottom selenium atomic layers sandwiching two gallium layers.11 Similar to graphite, a network of covalent interactions connects the atoms in the basal planes. In a bulk crystal of GaSe, van der Waals interactions dominate the bonding between the sheets with an interlayer spacing of 0.80 nm.11 Chikan et al. have prepared GaSe quantum dots with size tunable optical and electronic properties using a high temperature solution method.12 Because the top and bottom Se layers are relatively inert and only the edges are capped with organic ligands, GaSe nanoparticles have the potential to form stacked structures with well-defined electronic coupling, as these * Corresponding author. E-mail: [email protected] (T.Y.) and [email protected] (D.F.K.).

particles can form direct van der Waals contact with minimal hindrance from the ligands. Previous spectroscopic investigations indicate the formation of nanoscale aggregates in which the quantum dots are strongly coupled, resulting in a number of interesting phenomena such as exciton migration,13 charge transfer,14 and enhanced luminescence.15 However, the precise structure of such aggregates and how they may be integrated into potential devices remain unclear. To address these questions, we have deposited GaSe nanoparticles on a gold surface, allowing the use of surface chemistry to direct the assembly of superstructures and interrogation of the structures with atomic force microscopy (AFM). We find that the GaSe nanoparticles have the tendency to be adsorbed in a “flat-on” orientation on gold surface, because of the strong interaction between bottom Se layer and the gold substrate. Also, we find that the molecules (either trioctylphosphine oxide or dodecyl aldehyde) ligated to the edge of the nanoparticles play an important role in controlling the aggregation state of these nanoparticles. Three different morphologies are observed, which can be potentially applied in studies of charge separation and transport in nanoparticles on surfaces. Experimental Section GaSe nanoparticles are synthesized using methods similar to those reported previously.12 We now know that in the previous syntheses, there were small amounts of octyl phosphonic acid (OPA) in the trioctyl phosphine (TOP), which is the solvent for the reaction mixture. OPA is a common impurity in TOP and trioctyl phosphine oxide (TOPO).16 Under reaction conditions, OPA undergoes dehydration to form phosphonic anhydrides,17 much as what occurs in the synthesis of CdSe nanoparticles.18,19 If present in significant amounts, the phosphonic anhydrides will strongly ligate the particle edges. In the present case, particles were synthesized from TOP/TOPO mixtures lacking significant amounts of OPA and, hence, phosphonic anhydrides. This results in particles that are ligated with weakly bound TOPO ligands rather than the more strongly

10.1021/jp907516j CCC: $40.75  2009 American Chemical Society Published on Web 10/08/2009

Nanoscale Organization of GaSe Quantum Dots bound polydentate phosphonic anhydride ligands that were present in earlier syntheses. As prepared, the nanoparticles are well-dispersed in trioctylphosphine (TOP) with a small amount of trioctylphosphine oxide (TOPO), which stabilizes the nanoparticles via ligation to the edges.12 A previous study used TEM and electronic spectroscopy to establish the relationship between the size and the absorption onset of GaSe nanoparticles.20 With this relationship, we determine the average diameter of the GaSe nanoparticles used in the present study to be about 5-6 nm. Static fluorescence spectra were obtained at an excitation wavelength of 410 nm, using a Jobin-Yvon Fluorolog-3 spectrometer, with a xenon lamp and double monochromator excitation source and a CCD detector. The spectra are corrected for instrument response, using correction curves generated from the spectrum of an Optronix spectrally calibrated lamp. Absorption spectra were collected on a Varian 50 spectrometer. The adsorption of the nanoparticles to gold surfaces was carried out in a glovebox to minimize exposure to oxygen and moisture. To achieve an optimum surface coverage, the stock solution was diluted to approximately 30 µM in toluene (Sigma Aldrich HPLC grade). A single crystal gold bead was prepared by melting a gold wire and then mounted on a platinum foil.21 Prior to each adsorption experiment, the bead was cleaned in the piranha solution (1:3 v/v 30% H2O2 and H2SO4) and rinsed with ultrapure water produced by a Barnstead Nanopure water purification system. (CAUTION: Piranha solution can react Violently with organic materials, and should be handled with extreme caution. Piranha solution should not be stored in tightly sealed containers.) After annealing with hydrogen flame, atomically flat (111) terraces that are several hundred nanometers wide are routinely observed. The flame annealed gold bead was transferred to the glovebox and immersed into the nanoparticle solution for 20 s to 2 min. The bead was briefly rinsed with toluene to remove unbound nanoparticles and blown dry with nitrogen. Tapping mode AFM images were obtained with an Agilent 5500 SPM system with silicon noncontact tips from MikroMasch. The tip radius of curvature is about 10-15 nm. The cantilever typically has a spring constant ∼5 N/m and a resonance frequency ∼160 kHz. Images are visualized and analyzed with the Gwyddion SPM image analysis software22 as well as MatLab. After a median filter is applied to reduce noise, protrusions are masked by thresholding, and the maximum heights of the protrusions are extracted to calculate the height histograms. Results and Discussion Figure 1a shows AFM images of freshly prepared TOPO capped GaSe particles adsorbed on Au(111). As synthesized, TOPO molecules are ligated to the edge of these nanoparticles, preventing aggregation.12 The particles are resolved as protrusions approximately 20-30 nm in lateral size and 1.0 ( 0.1 nm high on atomically flat terraces of Au(111). These particles are approximately 5 nm in lateral size, and the presence of the TOPO ligands is expected to increase the diameter by 2 nm. The larger apparent lateral sizes observed by AFM are attributed to the convolution of the finite tip size and surface topography. The radius of curvature of the AFM tip is about 10-15 nm, according to the manufacturer’s specification. With this tip, convolution with the surface topography is expected to increase the apparent lateral size of a 1 nm high particle by 15 nm. The observed heights (Figure 1b) are much smaller than the lateral sizes of the nanoparticles, close to 0.80 nm, the interlayer spacing of bulk GaSe.11 This shows that the nano-

J. Phys. Chem. C, Vol. 113, No. 44, 2009 19103

Figure 1. (a) AFM topography image of trioctylphosphine oxide (TOPO) capped GaSe nanoparticles adsorbed on Au(111) surface. The nanoparticle solution (TOPO capped GaSe nanoparticles dispersed in toluene) was freshly prepared. The particles are resolved as protrusions approximately 20-30 nm in lateral size and 1.0 ( 0.1 nm high in the image. (b) Histogram of topographic heights of the nanoparticles, indicating a single population with the height centering around 1 nm. (c) Structure of trioctylphosphine oxide (TOPO). (d) Proposed schematic of TOPO capped GaSe nanoparticles adsorbed on gold surface. The nanoparticles are adsorbed in a flat-on orientation, while TOPO ligands prevent aggregation of the nanoparticles.

particles are single-layer structures and are adsorbed in a flat-on orientation, that is, with the basal planes parallel to the substrate surface. It is noteworthy that the GaSe nanoparticles deposited from the solution assume an orientation that is similar to that of another layered compound MoS2 nanoclusters, produced by surface reaction on gold.23,24 Although the MoS2 nanoclusters are grown by surface reactions on gold23,24 instead of being synthesized in a solution then assembled on a surface, both GaSe and MoS2 semiconductor nanostructures are terminated with chalcogenide atomic layers that interact with the gold substrate. This suggests that orientation control is not unique to GaSe, but most likely general to quantum dots of a variety of chalcogenide layered semiconductors, such as WS2 and InSe.11 The adsorption orientation of the nanoparticles can be explained by the tendency to maximize surface interactions. Selenium is a soft Lewis base that can chemisorb on gold surfaces, a soft Lewis acid.25 If the particles were adsorbed in an edge-on orientation, then only a small number of selenium atoms would be in contact with the gold surface, and the adsorption energy would be small. However, with the two-dimensional particles adsorbed in a flat-on orientation, an entire selenium atomic layer is allowed to interact with the gold surface, significantly lowering the energy. We find that these particles are completely immobile on the surface after repeated scans by AFM over hours. In addition, ultrasonication of the gold surface with the adsorbed particles in toluene for 5 min results in no detectable change in the surface density of the particles, also indicating that the particles were strongly adsorbed on the surface. In contrast, nanoparticles deposited on graphite, for which no strong particle/surface bonding is expected, could not be resolved. The evidence above is an indication of the chemisorption characteristics on gold. The exact nature of the interactions between the gold substrate and a layered compound semiconductor remains unclear. The valence saturated Se outerlayers are largely regarded as inert and mainly interact with molecules and other surfaces through van der Waals contact.26,27 Although coordination bonds due to interaction between the selenium lone pair electrons and the metallic substrate are likely, XPS studies of the gold-GaSe and MoS2 interfaces so far revealed no chemical

19104

J. Phys. Chem. C, Vol. 113, No. 44, 2009

Figure 2. (a) AFM topography image of trioctylphorsphine oxide (TOPO) capped GaSe nanoparticles adsorbed on Au(111) surface. The solution was retrieved from the supernatant of a TOPO capped nanoparticles solution prepared 3 months prior to use. (b) Crosssectional profiles of GaSe nanoparticles marked in (a). The measured features are about 100 nm in diameter and 1.0 nm thick. (c) Proposed schematic of aggregated GaSe nanoparticles adsorbed on gold surface. Desorption of TOPO ligands induces irreversible agglomeration and/ or Ostwald ripening, as indicated by the dimension of features in (b).

shift of core electron binding energies of selenium or sulfur atoms in the outer layer.26 However, since the bottom selenium layer of the GaSe nanoparticle and the gold surface have different lattice constants and symmetries, not all of the selenium atoms on the bottom layer will form bonds with the gold underneath. As a result, it is difficult to reveal the chemical interactions between the selenium and the gold with XPS because of the small fraction of bonding selenium atoms. This may explain why there was no chemical shift detected in the studies before.26 Prolonged storage (over 2 months) of the TOPO capped nanoparticles results in some degree of aggregation, as indicated by the formation of a visible precipitate in the stock solution. After immersing the substrate into this supernatant, AFM images display sheets that were about 100 nm in diameter and 1.0 nm thick (Figure 2a). The measured thickness of the sheet is quite uniform and close to the 0.80 nm layer thickness of the bulk GaSe semiconductor (Figure 2b). This provides additional evidence of the tetra-layer structure of the GaSe nanoparticles. The formation of larger particles is attributed to the partial desorption of the weakly adsorbed and hence labile TOPO ligands on the edges of original GaSe nanoparticles, which may cause irreversible agglomeration and/or Ostwald ripening.1 As the edges are exposed, the nanoparticles aggregate at their edges (Figure 2c). The possibility that the particles aggregate on the surface to form sheet like structures in a nucleation and growth mode can also be considered. This has been observed on the growth of atomic28 and molecular thin films.29 However, this is unlikely in the present case, because it would require significant lateral surface diffusion of these adsorbates to be incorporated into the peripherals of the thin film islands. We expect the lateral mobility of these particles to be quite low because of the strong Au-Se interactions as discussed above. Addition of dodecylaldehyde (dodecanal) to a solution of TOPO-capped GaSe nanoparticles has large effects on the particle absorption and fluorescence spectroscopy (Figure 3). Significant red shifts are observed in both spectra, which are attributed to the absorption and fluorescence of strongly interacting GaSe nanoparticle aggregates. A large increase in the fluorescence intensity is also observed. These spectral

Shao et al.

Figure 3. Absorption and fluorescence spectra of TOPO-ligated (solid dots), and dodecanal-ligated (open dots) GaSe nanoparticles. Red shift in both spectra indicates the presence of GaSe nanoparticle aggregates caused by addition of dodecanal.

Figure 4. (a) AFM topography image of dodecanal ligands capped GaSe nanoparticles adsorbed on Au(111) surface. (b) Cross-sectional profiles of GaSe nanoparticles marked in (a), showing different heights that are integer times of about 1.0 nm. (c) Structure of dodecylaldehyde (dodecanal). (d) Proposed schematic of dodecanal capped GaSe nanoparticles adsorbed on gold surface. The nanoparticles are also adsorbed in a flat-on orientation.

changes are completely analogous to those seen in organic dyes upon the formation of J-aggregates.30,31 In both cases, dipolar coupling of the transition oscillators shifts the lowest allowed transition to lower energy. Along with the spectral shift, higher intensity in the fluorescence spectrum is also observed in Figure 3. This is because of “superradiance”, an increase in the oscillator strength of the transition, which results in an increase in the fluorescence quantum yield.15 The spectroscopy of these aggregates has been discussed in detail in a recent publication.32 The important point here is that these spectral changes indicate that the presence of dodecanal results in the formation of very strongly interacting GaSe nanoparticle aggregates in solution: dodecanal replaces the far more bulky TOPO ligands, enabling the nanoparticles to stack into tightly bound aggregates. The spectroscopic studies indicate alignment of the transition diploes in these aggregates.32 However, spectroscopic investigations provide limited information about the morphology of these aggregates. Hence, we used AFM to image nanoparticle aggregates adsorbed on the surface. AFM results in Figure 4a show that, after adding dodecanal to a TOPO capped GaSe nanoparticle solution, the observed morphology of the adsorbed nanoparticles undergoes significant changes. In addition to 1 nm high particles, 2-6 nm high features are also observed. A histogram of the particle heights shows that the observed heights

Nanoscale Organization of GaSe Quantum Dots

J. Phys. Chem. C, Vol. 113, No. 44, 2009 19105

C0 ) C1/(1-KC1)2

Figure 5. Histogram of occurrences versus apparent height of dodecanal capped nanoparticles (bars), Gaussian fit to the height distribution (solid curves), and linear fit to the local maxima (inset). All of the results indicate that the lack of steric hindrance caused by dodecanal ligands induces aggregation as depicted in Figure 4d.

fall into sub-populations centered at discrete height values. The discrete heights are assigned to integral numbers of particles stacked on the surface, consistent with the proposed adsorption orientation of the GaSe particles. Consistent with this assignment, linear regression shows that the most probable height is increased by 0.97 ( 0.02 nm for each additional layer. (Figure 5) This result indicates that the particles stack face to face (Figure 4d) up to a maximum of six layers in the solution. The stacking of the particles is enabled by the displacement of the bulky TOPO ligands with dodecanal, a less bulky ligand with a single alkyl chain. Dodecanal is a harder Lewis base with stronger interactions with the exposed gallium atoms at the edge and, therefore, are able to displace the TOPO ligands. The TOPO ligands introduce steric hindrance that prevents close contact required for stacking, as depicted in Figure 1. The observed apparent heights suggest that the dodecanal ligand does not significantly impede stacking, which can be rationalized by the smaller steric interactions of the dodecanal ligand. In sharp contrast to the aged sample with significant TOPO ligand desorption (Figure 2), large flat sheets were not observed for this sample. This indicates that the tightly bound dodecanal ligands effectively suppress lateral aggregation. The formation of van der Waals bound cofacial aggregates is not the same as collapse to bulk GaSe. The difference between the aggregate structure and the bulk is seen in the height increment: the observed height increment, 0.97 nm, is larger than the interlayer spacing of bulk GaSe crystal, 0.80 nm. Spectroscopic investigations have shown that the TOPO capped nanoparticles aggregate at high concentrations.30 This aggregation disappears as the solution is diluted by a factor of 20 in toluene.30 In contrast, the aggregation induced by dodecanal appears significantly more stable, and the aggregation remains in a solution diluted by ×10 000 (ca. 30 µM) for a period of at least hours to days as indicated by significant number of the features higher than 1 nm. The relative abundances of different aggregates may be inferred from the peak areas in Figure 5. This distribution is not monotonic, having maxima at 1 (monomers) and 4 to 5 particles. We suggest that this distribution may be due to a nonequilibrium distribution of aggregate sizes in solution. If we assume that, for equilibrium between one-dimensional aggregates and single particles NP, NPn-1 + NP T NPn, the equilibrium constant K is independent of the aggregation number n, and the concentration of aggregates with n particles Cn is determined by eq 1. A summation of Cn yields the total concentration of nano-

Cn ) C1(KC1)n-1 particles C0, eq 2.

(1)

(2)

The concentration of the monomer C1 should be the highest, and the Cn should progressively decrease with increasing n, because KC1 must be less than 1.0 for a finite C0. The peak areas display a nonmonotonic dependence on the aggregation number, n, suggesting that the distribution is not reflective of an equilibrium population. The aggregate dissociation process takes a long time (over many hours) to reach equilibrium as indicated by the evolution of absorption spectra of diluted solutions.30,31 We further suggest that the nonequilibrium population reflects the kinetics of this chain-like dissociation process. Accordingly, the distribution of stack heights may vary with the time the sample is made following dilution. The histogram here reveals the distribution halfway in the whole process: the dissociation of large aggregates can produce local maxima at intermediate numbers (e.g., 4 or 5), and this local maxima will shift to lower numbers over time. The high monomers count is reflective of the higher tendency of the chainlike solution aggregates to dissolve by dissociating monomers off the ends. Systematic studies on the morphologies of particles adsorbed after different durations of dilution may shed additional light on the dissociation kinetics. Conclusion The results presented here show that the adsorption orientation of the GaSe nanoparticle aggregates is controlled by the interaction between the selenium layer and the gold surface. The aggregation state is significantly influenced by the steric hindrance (or the lack thereof) of the ligands stabilizing the nanoparticles. Stacking can be induced by the introduction of smaller ligands. Further control of the surface chemistry should allow for the growth of larger stacked aggregates in the surface normal direction as well as heterojunctions of two-dimensional quantum dots.33 Optical spectroscopic studies have already established the strong coupling between GaSe nanoparticles in aggregates.13,14,30 Such nanostructures, with well-defined orientation and coupling on surfaces, should allow more efficient charge separation and directional charge transport important for a number of technological applications of quantum dots, such as photovoltaic devices. Acknowledgment. We acknowledge the support of UC Merced, Petroleum Research Foundation, Grant 48335-G5 (J.S., J.P., J.C., and T.Y.) and the U.S. Department of Energy, Grant DE-FG02-04ER15502 (H.M. and D.F.K.). References and Notes (1) Bawendi, M. G.; Steigerwald, M. L.; Brus, L. E. Annu. ReV. Phys. Chem. 1990, 41, 477. (2) Schaller, R. D.; Klimov, V. I. Phys. ReV. Lett. 2004, 92, 4. (3) Murphy, J. E.; Beard, M. C.; Norman, A. G.; Ahrenkiel, S. P.; Johnson, J. C.; Yu, P. R.; Micic, O. I.; Ellingson, R. J.; Nozik, A. J. J. Am. Chem. Soc. 2006, 128, 3241. (4) Nair, G.; Bawendi, M. G. Phys. ReV. B 2007, 76, 4. (5) Nair, G.; Geyer, S. M.; Chang, L. Y.; Bawendi, M. G. Phys. ReV. B 2008, 78, 10. (6) Beard, M. C.; Midgett, A. G.; Law, M.; Semonin, O. E.; Ellingson, R. J.; Nozik, A. J. Nano Lett. 2009, 9, 836. (7) Collier, C. P.; Vossmeyer, T.; Heath, J. R. Annu. ReV. Phys. Chem. 1998, 49, 371. (8) Fu, A. H.; Micheel, C. M.; Cha, J.; Chang, H.; Yang, H.; Alivisatos, A. P. J. Am. Chem. Soc. 2004, 126, 10832. (9) Gur, I.; Fromer, N. A.; Geier, M. L.; Alivisatos, A. P. Science 2005, 310, 462. (10) Luther, J. M.; Law, M.; Song, Q.; Perkins, C. L.; Beard, M. C.; Nozik, A. J. ACS Nano 2008, 2, 271.

19106

J. Phys. Chem. C, Vol. 113, No. 44, 2009

(11) Levy, F. Crystallography and crystal chemistry of materials with layered structures; Reidel: Holland, 1976. (12) Chikan, V.; Kelley, D. F. Nano Lett. 2002, 2, 141. (13) Tu, H.; Mogyorosi, K.; Kelley, D. F. J. Chem. Phys. 2005, 122, 13. (14) Tu, H. H.; Kelley, D. F. Nano Lett. 2006, 6, 116. (15) Mogyorosi, K.; Kelley, D. F.J. Phys. Chem. C 2007, 111, 579. (16) Wang, F.; Tang, R.; Kao, J. L.-F.; Dingman, S. D.; Buhro, W. E. J. Am. Chem. Soc. 2009, 131, 4983. (17) Wang, Y.; Mirafzal, H.; Kelley, D. F., to be published. (18) Kopping, J. T.; Patten, T. E. J. Am. Chem. Soc. 2008, 130, 5689. (19) Liu, H. T.; Owen, J. S.; Alivisatos, A. P. J. Am. Chem. Soc. 2007, 129, 305. (20) Tu, H.; Chikan, V.; Kelley, D. F. J. Phys. Chem. B 2003, 107, 10389. (21) Clavilier, J.; Faure, R.; Guinet, G.; Durand, R. J. Electroanal. Chem. 1980, 107, 205. (22) Available from http://gwyddion.net/.

Shao et al. (23) Helveg, S.; Lauritsen, J. V.; Laegsgaard, E.; Stensgaard, I.; Norskov, J. K.; Clausen, B. S.; Topsoe, H.; Besenbacher, F. Phys. ReV. Lett. 2000, 84, 951. (24) Lauritsen, J. V.; Kibsgaard, J.; Helveg, S.; Topsoe, H.; Clausen, B. S.; Laegsgaard, E.; Besenbacher, F. Nat. Nanotechnol. 2007, 2, 53. (25) Monnell, J. D.; Stapleton, J. J.; Jackiw, J. J.; Dunbar, T.; Reinerth, W. A.; Dirk, S. M.; Tour, J. M.; Allara, D. L.; Weiss, P. S. J. Phys. Chem. B 2004, 108, 9834. (26) Lince, J. R.; Carre, D. J.; Fleischauer, P. D. Phys. ReV. B 1987, 36, 1647. (27) Tambo, T.; Tatsuyama, C. Surf. Sci. 1989, 222, 332. (28) Zhang, Z. Y.; Lagally, M. G. Science 1997, 276, 377. (29) Schwartz, D. K.; Steinberg, S.; Israelachvili, J.; Zasadzinski, J. A. N. Phys. ReV. Lett. 1992, 69, 3354. (30) Tu, H.; Yang, S.; Chikan, V.; Kelley, D. F. J. Phys. Chem. B 2004, 108, 4701. (31) Tu, H.; Mogyorosi, K.; Kelley, D. F. J. Chem. Phys. 2005, 122. (32) Mirafzal, H.; Kelley, D. F. J. Phys. Chem. C 2009, 113, 7139. (33) Chen, X. B.; Kelley, D. F. J. Phys. Chem. B 2006, 110, 25259.

JP907516J