Ultrafast Vibrational Relaxation Dynamics of a Rhenium Bipyridyl CO2

Nov 14, 2012 - Copyright © 2012 American Chemical Society ...... Yam , V. W. W.; Lau , V. C. Y.; Cheung , K. K. Organometallics 1995, 14, 2749– 275...
1 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/JPCC

Ultrafast Vibrational Relaxation Dynamics of a Rhenium Bipyridyl CO2−Reduction Catalyst at a Au Electrode Surface Probed by TimeResolved Vibrational Sum Frequency Generation Spectroscopy Chantelle L. Anfuso, Allen M. Ricks, William Rodríguez-Córdoba, and Tianquan Lian* Department of Chemistry, Emory University, Atlanta, Georgia 30322, United States S Supporting Information *

ABSTRACT: The vibrational relaxation dynamics of a rhenium bipyridyl CO2-reduction catalyst chemisorbed to a model Au electrode surface have been investigated using broadband time-resolved vibrational sum frequency generation (VSFG) spectroscopy. IR pump-VSFG probe spectra indicate a biexponential decay of the excited vibrational population, consisting of an ultrafast initial relaxation followed by complete recovery of the ground vibrational state within tens of picoseconds. The ultrafast decay is assigned to rapid population equilibration between the strongly coupled CO stretching modes of the rhenium bipyridyl complex (τυ−υ = 0.26 ps), whereas the slower decay is assigned to population relaxation from these coupled modes to lower frequency modes within the molecule (T1 = 14.8 ps). This study demonstrates the ability of time-resolved VSFG to selectively monitor the vibrational energy relaxation processes at the electrode surface in relevant electrocatalytic systems.

■

INTRODUCTION In recent years, several rhenium bipyridyl complexes have been developed and explored for the electrocatalytic reduction of CO2 to CO and formate, exhibiting promising efficiencies and selectivities.1−8 Although these catalysts are free to diffuse in homogeneous systems, the catalytic reaction is driven by the electron transfer process at the electrode.9,10 Additionally, numerous studies have reported enhanced catalytic turnover in heterogeneous systems, where these complexes are immobilized on the electrode surface.10−14 The coordination of these catalysts to electrode surfaces may affect the efficiency of electron transfer and electrocatalytic reduction,15−19 which are also intricately related to the rates and mechanisms of vibrational energy flow for adsorbates on electrodes.20 Vibrational energy relaxation dynamics can differ significantly between molecules in solution and at surfaces due to the interaction with the electrode in the latter case; this can subsequently stabilize or destabilize the catalytic intermediates toward reaction. Therefore, a detailed study of the vibrational relaxation of these complexes adsorbed on model electrode surfaces may shed light on the complicated reaction dynamics in these electrocatalytic systems. The electrode surface can be selectively accessed through the use of surface-specific nonlinear optical techniques such as time-resolved vibrational sum frequency generation spectroscopy (TR-VSFG). This technique has been used previously to monitor vibrational dynamics of molecules at liquid21−23 and solid surfaces.20,24−27 Although liquid surfaces are fundamentally interesting, a great deal of fascinating chemistry occurs at the air/solid and liquid/solid interfaces, such as many types of catalysis and electrochemical reactions. However, TR-VSFG studies on solid interfaces have primarily been limited to narrowband IR sources and thus picosecond time scales, © 2012 American Chemical Society

limiting their applicability to systems exhibiting picosecond molecular dynamics. Studies which have employed femtosecond broadband TR-VSFG techniques have been principally restricted to model diatomic molecules. We are interested in extending these studies to include broadband TR-VSFG of molecular catalysts on electrode surfaces. We herein report our investigation on the vibrational dynamics of the CO2-reduction catalyst Re(dcbpy)(CO)3Cl [dcbpy =2,2′-bipyridine-4,4′-(COOH)2] (ReC0A) modified to chemisorb on a gold surface through an alkane thiol spacer layer (ReC0-Au, shown in Figure 1a) using IR pump-VSFG probe techniques. We show that this technique is capable of probing subpicosecond vibrational relaxation dynamics. Comparison with results for ReC0A in DMF obtained using standard transient absorption measurements indicates that coordination of the catalyst to the electrode surface does not significantly alter the mechanism of vibrational relaxation at the catalytic rhenium center in this system. Interestingly, our analysis demonstrates that the large nonresonant Au background is actually beneficial for this pump−probe VSFG study, while it typically complicates static VSFG spectra. This may prove useful for studying other electrocatalysts on metal electrodes, which often exhibit large nonresonant SFG signals.

■

EXPERIMENTAL METHODS Sample Preparation. Gold films were prepared on clean sapphire substrates via sputter coating. Sapphire windows (1″ diameter) were first sonicated in piranha solution (3:1 H2SO4/

Received: September 7, 2012 Revised: November 13, 2012 Published: November 14, 2012 26377

dx.doi.org/10.1021/jp3089098 | J. Phys. Chem. C 2012, 116, 26377−26384

The Journal of Physical Chemistry C

Article

thiol adsorbed on gold and the carboxylic groups of the ReC0A complex. The procedure was performed in inert argon atmosphere. A freshly prepared SAM was immersed in a ReC0A dry tetrahydrofuran (THF) solution (0.1 mg/mL) containing N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC; 50 mg/mL) and 4-dimethylaminopyridine (DMAP; 0.1 mg/mL). The reaction was stirred at 50 °C for 24 h. The substrate was subsequently rinsed several times with THF, taken out of the argon box, thoroughly washed with DMF, and dried in a flow of pure argon gas. The schematic structure of the final ReC0-Au system is shown in Figure 1a. VSFG Experimental Details. A detailed description of our static VSFG experimental setup has been described elsewhere.31 The time-resolved measurements are based on the same VSFG system but with several modifications. All measurements were carried out on a 1 kHz Spitfire Ti:sapphire regenerative amplifier system (Spectra Physics) producing 150 fs pulses at 800 nm with a pulse energy of 4 mJ. Half of the fundamental was filtered with a narrowband interference filter (CVI F01800-UNBLK-1.00) to narrow the spectral bandwidth to ∼12 cm−1 (25). However, since the nonresonant Au response is large compared to the resonant ReC0 response and the excited state population is small, we find that only three terms, which contain ANR and are linear in N1(t) and NC1 (t), dominate. The IR pump-VSFG probe response becomes (2) 2 ΔS(t , ω) ∝ |χp(2) (t )|2 − |χun |

⎡ −A 1 Γ1 (2N (t ) + N C(t )) 01 01 1 1 = ANR ⎢ 2 1 ⎢⎣ (ω01 − ωIR )2 + Γ101 +

+

1 1 A12 Γ12N1(t ) 1 − ωIR )2 + Γ112 (ω12

2

⎤ ⎥ 2 C (ω01 − ωIR )2 + Γ C01 ⎥⎦ C C C A 01 Γ 01N1 (t )

(9)

In the harmonic oscillator approximation, both the IR transition moment and the Raman transition moment are proportional to (j + 1)1/2.26,46−48 Therefore, we find that A112/ A101 = 2 and A101 = AC01. According to the 2DIR spectrum of a similar rhenium bipyridyl complex,38 transitions 2,3, and 4 have similar frequencies and cannot be resolved in our measurement; therefore we assume that ω112 = ωC01 and Γ112 = ΓC01. With these assumptions, eq 9 can be further simplified (2) 2 ΔS(t , ω)∝|χp(2) (t )|2 − |χun |

⎛ A 1 (N (t ) − N (t )) 1 + ⎜ 011 0 1 ⎝ ω01 − ωIR + i Γ 01

⎡ ⎤ 1 1 C C −A 01 Γ 01 Γ 01 A 01 ⎥ = ANR ⎢ + 2 2 1 C ⎢⎣ (ω01 − ωIR )2 + Γ101 − ωIR )2 + Γ C01 ⎥⎦ (ω01

1 A12 (N1(t ) − N2(t )) 1 ω12 − ωIR + i Γ112 C A 01 (N1C(t ) − N11(t )) ⎞ iπ /2 ⎟e C − ωIR + i Γ C01 ⎠ ω01

(7)

(2N1(t ) + N1C(t ))

(10)

It should be noted that only those terms that contain the NR response of Au remain in eqs 9 and 10. Thus one interesting consequence of the large NR Au background in this system is that it serves to amplify the smaller resonant signal of the adsorbate but is completely eliminated in the completely

(6)

where N2(t) is the population in the |200⟩ state, N11(t) is the sum population of the |110⟩ and |101⟩ states, and AC01, ωC01, and 26381

dx.doi.org/10.1021/jp3089098 | J. Phys. Chem. C 2012, 116, 26377−26384

The Journal of Physical Chemistry C

Article

spectra can be described very well by the proposed model. We can now see that the GSB results from the first term in eq 10, while the EA results from the second term. The best fits yields ω101 = 2023 cm−1 and Γ101 = 13.5 cm−1, consistent with the parameters obtained from the static spectra obtained with different IR central wavelengths shown in Figure 2a,c. As shown in Figure 4a, the GSB and EA kinetics of ReC0-Au are well described by an instantaneous rise (τ0 ≈ 40 − 50 fs) and biexponential decay. The decay kinetics contain a smaller ultrafast component (τv−v = 0.26 ps), followed by full recovery of the ground state within tens of picoseconds (T1 = 14.8 ps). Three mechanisms can be considered for vibrational relaxation in this system: intramolecular relaxation, intermolecular relaxation, and direct energy transfer to the substrate. To evaluate the relative contributions of these pathways, we compared the vibrational dynamics of ReC0-Au with those of ReC0A in DMF. The vibrational dynamics of ReC0A in DMF were monitored using a standard IR-pump IR-probe transient absorption technique. The transient IR absorption spectra (shown in the Supporting Information) are similar to the transient VSFG difference spectra for ReC0-Au (Figure 3a), exhibiting a GSB at 2025 cm−1 and EA at 2005 cm−1. Fitting of the GSB and EA (Figure 4b and Table 3) shows that the kinetics are also well-characterized by a biexponential decay. Comparison of the GSB kinetics of ReC0-Au and ReC0A in DMF (Figure 4b) indicate nearly identical kinetics, implying that their relaxation pathways are the same. This rules out the pathway of direct energy transfer to the substrate for ReC0-Au, since this pathway is not available for ReC0A in DMF. This is likely due to the long distance between ReC0A and the Au surface (∼10 Å). Ongoing studies are examining the dependence of T1 relaxation rate on the alkane thiol spacer length. Furthermore, due to the very different local environments, we expect that the intermolecular relaxation rate would be different in the two systems; the similar kinetics therefore suggest that both the slow and fast decays of the |100⟩ population are due to intramolecular vibrational energy redistribution (IVR). The IVR kinetics are well described by an ultrafast intramolecular vibrational energy redistribution caused by strong coupling between the CO stretching modes of ReC0A (τυ−υ = ∼0.26 ps) followed by vibrational population relaxation from the coupled modes, which is facilitated by multiquantum vibrational energy transfer to lower frequency modes within the molecule (T1 = 14.8 ps). It should be noted that the IR-pump IR-probe and the IRpump-VSFG-probe experiments do not necessarily sample the same vibrational populations. These two similar but distinct methods and systems (ReC0A in DMF and ReC0-Au, respectively) have very different selection rules. In the IRpump/IR-probe transient absorption measurement for ReC0A in solution, all excited molecules are probed. However, as seen in eq 3, VSFG intensity depends on both the infrared and Raman transition dipole moments and probes molecules with nonrandom orientation distributions. In addition, the ReC0-Au system is subject to the metal-surface-selection rule. This rule states that for adsorbate/metal systems, molecules with transition dipole moments parallel to the metal surface are infrared (and thus VSFG) inactive and will not be probed in the SFG measurement.50 It is also interesting to compare our result with a recent elegant study of vibrational relaxation dynamics of a related Re carbonyl complex, fac-Re(phenanthroline)(CO)3Cl, immobilized onto a SiO2 surface by heterodyne-detected transient grating spectroscopy.51 A significant shortening of the

eliminated shown in Figure 3a. The large NR signal is therefore beneficial in this pump−probe VSFG study, although it often complicates the analysis in static VSFG experiments.49 Furthermore, because of the π/2 phase difference between the Au and adsorbate signal, only the changes in the imaginary part of the molecular response are observed in the difference spectra, which resemble transient IR absorption spectra of the same complex in solution (see below). It should be noted that this is only true when the phase difference between the nonresonant Au and resonant molecular response is π/2 or 3π/ 2. A phase difference of 0 or π would selectively amplify the real part of the molecular response. Other phase values would lead to mixing of the real and imaginary parts and more complicated peak shapes in the static and pump−probe VSFG spectra; however, knowledge of the phase between the resonant and nonresonant contributions should allow for proper spectral fitting and kinetics extraction. Therefore we anticipate that a large NR background may still be beneficial in systems with other relative phase values, provided that the phase relationships and magnitudes of the resonant and nonresonant contributions can be obtained from fitting the static spectra. Another interesting consequence of the large NR Au background in this system is that the change in VSFG intensity becomes linearly dependent on the population change of the involved vibration states, simplifying the extraction of kinetics. Within the model shown in Figure 3b, the time-dependent populations N1(t) and NC1 (t) can be described by biexponential kinetics41 N1(t ) = ΔN e−t / τv −v + N1,eq e−t / T1

(11)

C −t / T1 N1C(t ) = −ΔN e−t / τv −v + N1,eq e

(12)

Here, ΔN is the population that is transferred between the three CO stretching modes and N1,eq and NC1,eq are the quasisteady populations in the resonant mode and the coupled modes, respectively, after the initial population equilibration and prior to their T1 relaxation. The overall time evolution of the GSB or EA is thus given by ΔS(t , ω) ∝ 2N1(t ) + N1C(t ) C ∝ ΔN e−t / τv −v + (2N1,eq + N1,eq )e−t / T1

(13)

The IR pump-VSFG probe spectra and kinetics were simultaneously fit according to eqs 10 and 13; the results are shown in Figures 3a and 4a and the fitting parameters are listed in Tables 2 and 3. Overall, the fit shows that the transient Table 2. Fitting Parameters for the ReC0-Au IR PumpVSFGS Probe Spectra Shown in Figure 3 According to eq 10b ω101 (cm−1)

Γ101 (cm−1)

ωC01 (cm−1)

ΓC01 (cm−1)

2023

13.5

2009

22

Table 3. Fitting Parameters for the GSB and EA Kinetics for ReC0-Au and ReC0A in DMF system ReC0-Au ReC0A in DMF

τ0 (ps)

ΔN

τν‑ν (ps)

0.055 (±0.01) 0.042 (±0.02)

0.21

0.26 (±0.1) 0.33 (±0.4)

0.22

2N1,eq+ NC1,eq 0.79 0.78

T1 (ps) 14.8 (±6) 18.0 (±4)

26382

dx.doi.org/10.1021/jp3089098 | J. Phys. Chem. C 2012, 116, 26377−26384

The Journal of Physical Chemistry C

Article

(5) Hayashi, Y.; Kita, S.; Brunschwig, B. S.; Fujita, E. J. Am. Chem. Soc. 2003, 125, 11976−11987. (6) Fujita, E.; Muckerman, J. T. Inorg. Chem. 2004, 43, 7636−7647. (7) Takeda, H.; Koike, K.; Inoue, H.; Ishitani, O. J. Am. Chem. Soc. 2008, 130, 2023−2031. (8) Smieja, J. M.; Kubiak, C. P. Inorg. Chem. 2010, 49, 9283−9289. (9) Kumar, B.; Smieja, J. M.; Sasayama, A. F.; Kubiak, C. P. Chem. Commun. 2012, 48, 272−274. (10) Kumar, B.; Smieja, J. M.; Kubiak, C. P. J. Phys. Chem. C 2010, 114, 14220−14223. (11) Otoole, T. R.; Margerum, L. D.; Westmoreland, T. D.; Vining, W. J.; Murray, R. W.; Meyer, T. J. J. Chem. Soc., Chem. Commun. 1985, 1416−1417. (12) Christensen, P.; Hamnett, A.; Muir, A. V. G.; Timney, J. A.; Higgins, S. J. Chem. Soc. Faraday Trans. 1994, 90, 459−469. (13) Cecchet, F.; Alebbi, M.; Bignozzi, C. A.; Paolucci, F. Inorg. Chim. Acta 2006, 359, 3871−3874. (14) Cheung, K. C.; Guo, P.; So, M. H.; Lee, L. Y. S.; Ho, K. P.; Wong, W. L.; Lee, K. H.; Wong, W. T.; Zhou, Z. Y.; Wong, K. Y. J. Organomet. Chem. 2009, 694, 2842−2845. (15) O’Regan, B.; Gratzel, M. Nature 1991, 353, 737−740. (16) Kamat, P. V.; Meisel, D. Curr. Opin. Colloid Interface Sci. 2002, 7, 282−287. (17) Adams, D. M.; Brus, L.; Chidsey, C. E. D.; Creager, S.; Creutz, C.; Kagan, C. R.; Kamat, P. V.; Lieberman, M.; Lindsay, S.; Marcus, R. A.; et al. J. Phys. Chem. B 2003, 107, 6668−6697. (18) Gratzel, M. MRS Bull. 2005, 30, 23−27. (19) Prezhdo, O. V.; Duncan, W. R.; Prezhdo, V. V. Acc. Chem. Res. 2008, 41, 339−348. (20) Arnolds, H.; Bonn, M. Surf. Sci. Rep. 2010, 65, 45−66. (21) McGuire, J. A.; Shen, Y. R. Science 2006, 313, 1945−1948. (22) Smits, M.; Ghosh, A.; Sterrer, M.; Muller, M.; Bonn, M. Phys. Rev. Lett. 2007, 98, 098302. (23) Eftekhari-Bafrooei, A.; Borguet, E. J. Am. Chem. Soc. 2009, 131, 12034−+. (24) Harris, A. L.; Levinos, N. J. J. Chem. Phys. 1989, 90, 3878−3879. (25) Guyotsionnest, P.; Dumas, P.; Chabal, Y. J. J. Electron Spectrosc. 1990, 54, 27−38. (26) Chin, R. P.; Blase, X.; Shen, Y. R.; Louie, S. G. Europhys. Lett. 1995, 30, 399−404. (27) Xiong, W.; Laaser, J. E.; Mehlenbacher, R. D.; Zanni, M. T. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 20902−20907. (28) Pfennig, B. W.; Chen, P. Y.; Meyer, T. J. Inorg. Chem. 1996, 35, 2898−2901. (29) Liu, G.-k. Interaction between Molecule and Metal Probed by Vibrational Spectroscopy; Xiamen University, 1996. (30) Pan, S.; Castner, D. G.; Ratner, B. D. Langmuir 1998, 14, 3545− 3550. (31) Anfuso, C. L.; Snoeberger, R. C.; Ricks, A. M.; Liu, W. M.; Xiao, D. Q.; Batista, V. S.; Lian, T. Q. J. Am. Chem. Soc. 2011, 133, 6922− 6925. (32) Ghosh, A.; Smits, M.; Bredenbeck, J.; Dijkhuizen, N.; Bonn, M. Rev. Sci. Instrum. 2008, 79. (33) Bonn, M.; Hess, C.; Miners, J. H.; Heinz, T. F.; Bakker, H. J.; Cho, M. Phys. Rev. Lett. 2001, 86, 1566−1569. (34) Asbury, J. B.; Wang, Y.; Lian, T. Bull. Chem. Soc. Jpn. 2002, 75, 973. (35) Zhuang, X.; Miranda, P. B.; Kim, D.; Shen, Y. R. Phys. Rev. B 1999, 59, 12632−12640. (36) Lambert, A. G.; Davies, P. B.; Neivandt, D. J. Appl. Spectrosc. Rev. 2005, 40, 103−145. (37) Rao, Y.; Comstock, M.; Eisenthal, K. B. J. Phys. Chem. B 2006, 110, 1727−1732. (38) Bredenbeck, J.; Helbing, J.; Hamm, P. J. Chem. Phys. 2004, 121, 5943−5957. (39) Wang, Z.; Carter, J. A.; Lagutchev, A.; Koh, Y. K.; Seong, N.-H.; Cahill, D. G.; Dlott, D. D. Science 2007, 317, 787−790. (40) Carter, J. A.; Wang, Z.; Dlott, D. D. Acc. Chem. Res. 2009, 42, 1343−1351.

T1 relaxation time of the a′(1) CO stretching mode was observed for the complex on the SiO2 surface compared to bulk solution. It was suggested that the IVR rate is controlled by the intramolecular anharmonic coupling between the CO stretching modes and the phenanthroline ligand and is sensitive to the solvent environment. The difference between these systems is not yet understood and may point to the dependence of the intramolecular anharmonic coupling strength on the nature of the ligand (bipyridine vs phenanthroline) and immobilization chemistry.

■

CONCLUSION In conclusion, we have used broadband time-resolved vibrational sum frequency generation spectroscopy to monitor the vibrational relaxation dynamics for a rhenium bipyridyl CO2reduction catalyst chemisorbed to a Au electrode surface. For comparison, we used standard transient absorption techniques to monitor the relaxation dynamics of a similar rhenium bipyridyl complex in DMF. Both systems exhibited biexponential relaxation from the v = 1 state consisting of an ultrafast (subpicosecond) initial relaxation followed by a complete recovery of the ground vibrational state within tens of picoseconds. The ultrafast decay is assigned to rapid population equilibration between the CO stretching modes, whereas the slower decay is assigned to the vibrational population decay from these coupled modes by intramolecular vibrational energy relaxation. These results indicate that coordination of the catalyst to the Au electrode surface through long alkane thiol spacers does not significantly affect the mechanism for vibrational relaxation of the CO stretching modes at the catalytic rhenium center in these systems. This study also demonstrates the ability of time-resolved VSFG to probe ultrafast dynamics and selectively monitor the time evolution of vibrational energy transfer processes at the electrode surface in relevant electro- and photoelectro- catalytic systems.

■

ASSOCIATED CONTENT

S Supporting Information *

Details of the experimental methods and spectra for IR-pump/ IR-probe measurement. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS T.L. acknowledges the financial support by the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, Office of Science, U.S. Department of Energy (DE-FG02-07ER-15906).

■

REFERENCES

(1) Hawecker, J.; Lehn, J. M.; Ziessel, R. J. Chem. Soc., Chem. Commun. 1984, 328−330. (2) Hawecker, J.; Lehn, J. M.; Ziessel, R. Helv. Chim. Acta 1986, 69, 1990−2012. (3) Juris, A.; Campagna, S.; Bidd, I.; Lehn, J. M.; Ziessel, R. Inorg. Chem. 1988, 27, 4007−4011. (4) Yam, V. W. W.; Lau, V. C. Y.; Cheung, K. K. Organometallics 1995, 14, 2749−2753. 26383

dx.doi.org/10.1021/jp3089098 | J. Phys. Chem. C 2012, 116, 26377−26384

The Journal of Physical Chemistry C

Article

(41) Beckerle, J. D.; Casassa, M. P.; Cavanagh, R. R.; Heilweil, E. J.; Stephenson, J. C. Chem. Phys. 1992, 160, 487−497. (42) Golonzka, O.; Khalil, M.; Demirdoven, N.; Tokmakoff, A. J. Chem. Phys. 2001, 115, 10814−10828. (43) Yan, S. X.; Seidel, M. T.; Zhang, Z. Y.; Leong, W. K.; Tan, H. S. J. Chem. Phys. 2011, 135. (44) Tokmakoff, A.; Sauter, B.; Fayer, M. D. J. Chem. Phys. 1994, 100, 9035−9043. (45) Tokmakoff, A.; Sauter, B.; Kwok, A. S.; Fayer, M. D. Chem. Phys. Lett. 1994, 221, 412−418. (46) Guyotsionnest, P. Phys. Rev. Lett. 1991, 67, 2323−2326. (47) Hirose, C.; Akamatsu, N.; Domen, K. J. Chem. Phys. 1992, 96, 997−1004. (48) Bandara, A.; Kubota, J.; Onda, K.; Wada, A.; Kano, S. S.; Domen, K.; Hirose, C. Surf. Sci. 1999, 427−28, 331−336. (49) Lagutchev, A.; Hambir, S. A.; Dlott, D. D. J. Phys. Chem. C 2007, 111, 13645−13647. (50) McCrea, K. R.; Somorjai, G. A. J. Mol. Catal. A: Chem. 2000, 163, 43−53. (51) Rosenfeld, D. E.; Gengeliczki, Z.; Smith, B. J.; Stack, T. D. P.; Fayer, M. D. Science 2011, 334, 634−639.

26384

dx.doi.org/10.1021/jp3089098 | J. Phys. Chem. C 2012, 116, 26377−26384