Application of Degenerately Doped Metal Oxides in the Study of

Jan 26, 2015 - Degenerately doped In2O3:Sn semiconductor nanoparticles (nanoITO) have been used to study the photoinduced interfacial electron-transfe...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCB

Application of Degenerately Doped Metal Oxides in the Study of Photoinduced Interfacial Electron Transfer Byron H. Farnum, Zachary A. Morseth, M. Kyle Brennaman, John M. Papanikolas, and Thomas J. Meyer* Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290, United States S Supporting Information *

ABSTRACT: Degenerately doped In2O3:Sn semiconductor nanoparticles (nanoITO) have been used to study the photoinduced interfacial electron-transfer reactivity of surface-bound [RuII(bpy)2(4,4′-(PO3H2)2-bpy)]2+ (RuP2+) molecules as a function of driving force over a range of 1.8 eV. The metallic properties of the ITO nanoparticles, present within an interconnected mesoporous film, allowed for the driving force to be tuned by controlling their Fermi level with an external bias while their optical transparency allowed for transient absorption spectroscopy to be used to monitor electron-transfer kinetics. Photoinduced electron transfer from excited-state -RuP2+* molecules to nanoITO was found to be dependent on applied bias and competitive with nonradiative energy transfer to nanoITO. Back electron transfer from nanoITO to oxidized -RuP3+ was also dependent on the applied bias but without complication from inter- or intraparticle electron diffusion in the oxide nanoparticles. Analysis of the electron injection kinetics as a function of driving force using Marcus−Gerischer theory resulted in an experimental estimate of the reorganization energy for the excited-state -RuP3+/2+* redox couple of λ* = 0.83 eV and an electronic coupling matrix element, arising from electronic wave function overlap between the donor orbital in the molecule and the acceptor orbital(s) in the nanoITO electrode, of Hab = 20−45 cm−1. Similar analysis of the back electron-transfer kinetics yielded λ = 0.56 eV for the ground-state -RuP3+/2+ redox couple and Hab = 2−4 cm−1. The use of these wide band gap, degenerately doped materials provides a unique experimental approach for investigating single-site electron transfer at the surface of oxide nanoparticles.



INTRODUCTION

ket =

Heterogeneous electron transfer between molecules and metallic or semiconductor electrodes is at the core of electrochemistry,1−4 photoelectrochemistry,5−8 catalysis,9−11 and energy conversion.12−16 As in any electron-transfer process, an interfacial electron-transfer event occurs with energy conservation and reaction rates that are dictated by the Franck−Condon density of states in the molecule and their overlap with occupied/unoccupied electronic levels in the electrode.5,6,17−21 Marcus−Gerischer theory provides a mathematical framework to describe these kinetics with eq 1 showing the semiclassical expression for the heterogeneous electron-transfer rate constant (ket) for reductive electron transfer from an electrode to a surface-bound molecule. In this equation, the distribution of occupied electronic levels in the electrode as a function of energy E is given by g(E)f(E, EF) where g(E) is the total distribution of electronic levels and f(E, EF) is the Fermi−Dirac function which describes the occupancy of electronic levels relative to a given Fermi level (EF). In the classical limit, the free-energy distribution function describing the electron-transfer barrier is given by W(E), and Hab(E) is the electronic coupling matrix element arising from electronic wave function overlap between the acceptor orbital in the molecule and the donor orbital(s) in the electrode. © XXXX American Chemical Society

2π ℏ

W (E ) =



∫−∞ g(E)f (E , EF)|Hab(E)|2 W (E) dE

(1)

⎛ −(ΔG(E) + λ)2 ⎞ 1 exp⎜ ⎟ 4λkBT ⎝ ⎠ 4πλkBT

(2)

From eq 1, ket is maximized when both g(E)f(E, EF) and W(E) are large. For a metal electrode, g(E) is nearly constant with energy4,22 and W(E) can be defined in the classical, harmonic limit, as initially described by Marcus and Hush,17−20 by the Gaussian distribution in eq 2. In this relationship, ΔG(E) (= −nF(Eo′ − E)) is the free-energy change for electron transfer from an energy level E in the electrode to the surfacebound molecule with reduction potential Eo′ where n is the number of transferred electrons and F is Faraday’s constant. The total reorganization energy for electron transfer is given by λ which includes both intramolecular (λi) and medium/solvent (λo) contributions. Scheme 1 shows an energy diagram illustrating g(E)f(E, EF) and W(E) at an electrode surface with an externally applied bias (Eapp) used to control the position of the Fermi level with respect to the reduction Special Issue: John R. Miller and Marshall D. Newton Festschrift Received: December 18, 2014 Revised: January 23, 2015

A

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

ITO and related transparent conducting oxides (TCOs), such as SnO2:F (FTO) and SnO2:Sb (ATO), are ubiquitous in photoelectrochemistry with TCO films deposited on glass substrates used as transparent electrodes in optoelectronic devices such as light-emitting diodes and solar cells.29−32 The metallic behavior of planar TCO films arises from degenerate doping (ND > 1020 cm−3) of the wide band gap parent metal oxide where the electronic levels of dopant atoms merge with those of the conduction band and free electron properties emerge; characterized by high conductivities (10 2 −10 4 Ω−1 cm−1) and electron mobilities (10−100 cm2 V−1 s−1).29 At this condition, the Fermi level lies within the conduction band with EF < Ecb and is defined as the highest occupied electronic level. Such high doping densities also allow for band bending on the order of nanometers over a wide range of applied potentials. The resulting small depletion widths allow for band bending to be sustained even in nanoparticle TCOs which retain the conductive properties of traditional planar films.28,33−36 Mesoporous films of such nanoparticles are used as high surface area electrodes with nanoITO and nanoATO used in spectroelectrochemical,37 electrochromic,34,38−41 and electrocatalytic42,43 applications. Because of the rapid transfer of electrons throughout the films, electron transfer at the semiconductor/liquid interface is limited only by electrontransfer kinetics at the interface. Given their wide band gaps, high conductivities, high surface areas, and ease of surface derivatization, nanoTCO films represent a new class of electrode materials for investigating photoinduced electrontransfer reactions at metal oxide surfaces. These reactions are of importance in photoelectrochemical solar energy conversion in dye-sensitized solar cells12,13,44 and in dye-sensitized photoelectrosynthesis cells.14,45−47 Here, we present further details and discussion on the earlier report of photoinduced electron transfer between surfacebound [RuII(bpy)2(4,4′-(PO3H2)2-bpy)]2+ and nanoITO as a function of applied bias in acetonitrile solvent. Further analysis of the electron injection dynamics reveals that energy-transfer pathways become competitive with excited-state injection at negative applied biases. Additional details of the Marcus− Gerischer analysis for electron injection and back electrontransfer kinetics are also given with a focus on the estimation of λ and Hab for both electron-transfer reactions.

Scheme 1. Gerischer Diagram Depicting Isoenergetic Electron Transfer (Green Dashed Region) from Occupied Electronic Levels in a Metallic Electrode (g(E)) to Unoccupied Levels in a Molecular Species at the Electrode Interface (W(E))

potential of the molecule. The standard free-energy change for electron transfer is ΔGeto′ = −F(Eo′ − EF) < 0. The Fermi− Dirac term has been approximated in the low-temperature limit as a step function at EF for simplification. Shown in the green dashed region of Scheme 1 is the overlap between g(E) and W(E) where electron-transfer events can occur with energy conservation at the electrode interface. Equation 1 reflects the contributions from all of these electrontransfer events by integrating over all energies. For interfacial electron transfer at an electrode, this leads to differences between predictions by the Marcus−Gerischer model for heterogeneous electron transfer and Marcus theory for homogeneous electron transfer. The Marcus−Gerischer model does not predict an inverted region for electron transfer because of the continuous distribution of electronic levels in the metal below the Fermi level. Electron transfer involving these levels is always favorable. Furthermore, at the condition −ΔGeto′ = λ, the rate constant for electron transfer is not maximized, as it is for solution reactions, but is equal to ketmax/2. This is illustrated in Scheme 1 where EF = Eo′ − λ or −ΔGeto′ = λ. At this condition, only half of the W(E) distribution function is overlapped with occupied electronic levels in the metal leading to the condition ket = ketmax/2. The maximum rate constant is reached when −ΔGeto′ ≫ λ. Experimental verification of the Marcus−Gerischer model has been demonstrated by Chidsey3,23,24 and others25−27 for gold electrodes with self-assembled alkanethiols terminated with redox active molecules. The insulating alkanethiol layer was used to decrease the electronic coupling between the electrode and the molecule with electron-transfer rate constants measured on the millisecond time scale by using conventional electrochemical techniques. In a recent communication, we reported an important experimental advance in the evaluation of photoinduced, molecular interfacial electron-transfer reactions.28 The method is based on the conductive properties of optically transparent, wide band gap, degenerately doped semiconductors which were used to investigate photoinduced electron-transfer reactions as a function of driving force controlled by an externally applied bias. Transient absorption measurements were used to monitor both electron injection and back electron-transfer dynamics following laser flash excitation of a Ru(II) polypyridyl complex, surface-bound to degenerately doped In2O3:Sn nanoparticles (nanoITO).



EXPERIMENTAL SECTION Materials. All materials were used as received without further purification. LiClO4 (99.999%, trace metal basis) and hydroxypropyl cellulose (average MW = 80 000, 20 mesh particle size) were obtained from Sigma-Aldrich. Ethanol (200 Proof) was obtained from Decon Laboratories. Acetonitrile (MeCN; Optima LC/MS) and methanol (MeOH, ACS grade) were purchased from Fisher Scientific. In2O3:Sn (ITO) nanoparticles (TC8 DE; 20 wt % dispersion in ethanol) were purchased from Evonik Industries. ZrO2 nanocrystalline thin films deposited on glass slides were prepared by a literature method.48 [RuII(bpy)2(4,4′-(PO3H2)2-bpy)](Cl)2 (RuP2+; bpy = 2,2′-bipyridine) was available from a previous study.49 nanoITO Preparation and Characterization. A 10 wt % suspension of hydroxypropyl cellulose (HPC) in ethanol was prepared by adding 1.3 g of HPC to 15 mL of ethanol followed by stirring overnight. A volume of 5 mL was transferred from the 20 wt % ITO dispersion in ethanol to a clean 22 mL scintillation vial and bath sonicated for 20 min followed by B

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Steady-state photoluminescence studies were performed on ZrO2−RuP2+ films using an Edinburgh FLS920 spectrometer. ZrO2−RuP2+ derivatized films were prepared by soaking ZrO2 films in methanol solutions containing 0.5 mM [RuP](Cl)2 overnight. ZrO2−RuP2+ slides were immersed in MeCN with 0.1 M LiClO4 solution at a 45° angle in a 1 cm2 glass cuvette whose top had been adapted with a #15 O-ring sealing joint, side arm, and Kontes valve. Samples were purged with N2 for at least 30 min prior to experimentation. Spectra were collected at room temperature (22 ± 2 °C) using monochromatic excitation at 420 nm. Photoluminescence was passed through a 495 nm long-pass filter and a single Czerny−Turner monochromator before being detected by a Peltier-cooled Hamamatsu R2658P photomultiplier tube. Spectroelectrochemical Transient Absorption (TA) Spectroscopy. All samples utilized in TA measurements consisted of FTO/nanoITO-RuP2+ electrodes immersed in MeCN (0.1 M LiClO4) solution and connected in the same fashion as those discussed for steady-state spectroelectrochemistry. Samples were purged with N2 for at least 30 min prior to experimentation. During experimentation, the externally applied bias was held constant for the duration of each TA experiment. Picosecond TA measurements were conducted using a pump−probe technique based on a Ti:sapphire chirped pulse amplification (CPA) laser system (Clark-MXR CPA2210). The amplifier produced 150 fs pulses at 775 nm and 1 kHz repetition rate. The 420 nm pump pulse was generated in a 2 mm BBO crystal by sum frequency generation of the 775 nm fundamental and the second harmonic of the 1840 nm idler from an optical parametric amplifier (Light Conversion TOPAS-C). The femtosecond probe pulse is generated by focusing 3 mW of the 775 nm amplifier output into a translating 5 mm thick CaF2 window. The pump beam is focused onto the sample using a 300 mm lens, and the probe beam is focused and overlapped with a 250 mm spherical aluminum mirror. The delay between the pump and probe pulses was determined by a computer-controlled stage. Spectra were collected on a shot-by-shot (1 kHz) basis over the range of 350 to 820 nm with a sensitivity of up to 0.1 mOD. The angle between the pump and probe polarization vectors was set to the magic angle (∼54.7°) to avoid polarization effects and ensure that only excited-state population dynamics were being monitored. The sample was raster scanned to provide for a fresh sample between laser pulses. Following data collection, the frequency chirp in the probe pulse was characterized using the optical Kerr response of a glass slide in a 1 cm cuvette in a polarization gating geometry. The spectra were chirp-corrected using a data processing program written in LabVIEW. Nanosecond TA measurements were performed with an Ultrafast Systems EOS spectrometer, in which the probe pulse was produced by continuum generation from a photonic crystal fiber and detected by a fiber optic coupled multichannel spectrometer with a CMOS sensor. The pump−probe delay was electronically controlled. The kinetic window ranged from 500 ps to 400 μs, and the time resolution of the instrument was around 500 ps, dictated by the width of the probe pulse and the timing electronics. The laser fluence for both TA experimental setups was adjusted to be 0.7 mJ cm−2.

ultrasonication for 5 min with a sonicating horn. To this freshly mixed ITO dispersion, 5 mL of the 10 wt % HPC suspension were added, and the final mixture was stirred overnight. NanoITO films were deposited onto conducting SnO2:F (FTO) glass substrates (Hartford Glass, Inc., 15 Ω cm−2) by a doctor blading technique. Film thickness was controlled by the numbered layers of scotch tape used to define the exposed area and measured by profilometry using a Bruker Dektak XT instrument. Thermal annealing of the FTO/nanoITO films was performed in two steps. (1) Doctor bladed thin films were placed in a box oven (Thermo Scientific Lindberg/Blue M) and annealed at 500 °C for 1 h in air. (2) Films were then placed in a tube furnace (Thermo Scientific Lindberg/Blue M) and annealed at 300 °C for 1 h under steady H2/N2 gas flow (5% H2/N2, Airgas). Oxidized nanoITO (nanoITO(ox)) films were produced by only step 1 of the annealing procedure. Reduced nanoITO (nanoITO(red)) films were produced by steps 1 and 2 of the annealing procedure. Scanning electron microscopy (SEM) using a Hitachi S-4700 Cold Cathode FESEM instrument was used to image the mesoporous structure of nanoITO films. Sheet resistance measurements were obtained by a four-point probe method using a Signatone 1160 Series Probe Station. X-ray photoelectron spectroscopy studies were performed using a Kratos Axis Ultra DLD X-ray photoelectron spectrometer. Ultraviolet (UV)−visible−near-IR spectra of nanoITO(ox) and nanoITO(red) were recorded in air using a Cary 5000 spectrophotometer with an integrating sphere attachment. The absorbance for each film was calculated from the transmittance and reflectance spectra of FTO/nanoITO films corrected for the FTO glass. nanoITO-RuP2+ Preparation and Characterization. Derivatized nanoITO-RuP2+ films were obtained by soaking FTO/nanoITO glass slides in MeOH solutions of 0.5 mM [RuP](Cl)2 overnight to achieve maximum surface coverages of Γ = 3.1 × 10−8 and 2.7 × 10−8 mol cm−2 for oxidized and reduced nanoITO films, respectively. Surface coverages were calculated from UV−visible absorption spectra (Cary 50 spectrophotometer) of nanoITO-RuP2+ films recorded in MeCN (0.1 M LiClO4) using the equation Abs(λmax) = 1000ε(λmax)Γ. Here, the extinction coefficient for RuP2+ in H2O (0.1 M HClO4) was used, ε453 nm = 13 400 M−1 cm−1, because of the insolubility of RuP2+ in MeCN solution.49 Steady-state spectroelectrochemical studies on nanoITORuP2+ films were performed in MeCN (0.1 M LiClO4) using a Pine WaveNow potentiostat to control the externally applied bias and a Cary 50 spectrophotometer to record the UV− visible absorption spectrum. FTO/nanoITO-RuP2+ glass slides were immersed in electrolyte solution at a 45° angle in a custom-made three-arm spectroelectrochemical cell based on a 1 cm2 glass cuvette and connected as the working electrode while Pt mesh was used as the counter electrode. A pseudoreference electrode (Ag wire/KCl (satd.)/H2O) was installed in the third arm. The reference potential was calibrated before and after all experiments using ferrocene (Fc) in MeCN with 0.1 M LiClO4. All applied potentials were referenced versus SCE using the conversion factor Fc+/0 = 0.380 V vs SCE.50 Coincidentally, the pseudoreference electrode was found to be stable at the same potential as SCE such that Eapp (V vs ref) = Eapp (V vs SCE). Samples were purged with N2 for at least 30 min prior to experimentation.



RESULTS nanoITO Characterization. Sn-doped In2O3 nanoparticles of 10−20 nm diameter were purchased as a blue-colored C

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B dispersion in ethanol and worked into a sol−gel paste through addition of a surfactant. NanoITO films were deposited onto conductive FTO coated glass slides by a doctor blading technique and annealed in a two-step process that has been reported previously.28,36 The first step occurred by heating nanoITO films at 500 °C under atmospheric conditions to remove the surfactant leaving behind a high surface area mesoporous structure of interconnected ITO nanocrystallites. Note the SEM image in Figure 1. This annealing step resulted

previous literature results.51,52 Reduced films showed similar binding energies for the O 1s and Sn 3d5/2 (within 10 meV); however, the In 3d5/2 binding energy shifted by 100 meV to 444.2 eV. Elemental analysis with XPS revealed that the extent of Sn-doping was 9.7% for both types of films, close to the manufacturer’s estimate of 8%. The ratio of oxygen to total metal ion (O/(In+Sn)) was 1.4 for both films. The expected ratio based on the In2O3 stoichiometry is 1.5, showing that both films were slightly deficient in oxygen. UV−visible−near-IR absorption spectra of the oxidized and reduced nanoITO films were recorded in air to monitor the observed difference in film colors (Figure 2). An integrating

Figure 1. Scanning electron micrograph image of a nanoITO film. The chemical structure of the surface-bound -RuP2+ chromophore is shown in the inset.

Figure 2. UV−visible−near-IR absorption spectra for oxidized and reduced nanoITO collected in air. The LSPR feature in the near-IR was modeled by using Drude theory to estimate the electron density of each film; see the text.

in a noticeable color change of nanoITO films from blue to yellow. The second annealing step occurred at 300 °C under a H2(5%)/N2 gas flow and resulted in a color change from yellow back to the characteristic blue color of the starting ITO nanoparticles. Profilometry measurements revealed film thicknesses of 2.5 ± 0.1 μm after the first and second annealing steps, suggesting that no large-scale degradation of the films occurred in the high-temperature processing steps. Sheet resistance (Rs) measurements were performed by using the four-point probe method. The resistance of films exposed to the first annealing step was RS = 3800 ± 330 Ω cm−2. For films undergoing both annealing steps, RS = 940 ± 90 Ω cm−2 demonstrating that the second annealing step under a reducing H2 environment decreased the sheet resistance by a factor of ∼4. In the descriptions that follow, single-annealed films under atmospheric conditions are referred to as oxidized nanoITO or nanoITO(ox). Doubly annealed films with the final step under H2 are referred to as reduced nanoITO or nanoITO(red). Conductivities for oxidized and reduced nanoITO were calculated from RS values using the expression σ = 1/ (RSAtp). In this expression, σ is the conductivity, RS the sheet resistance, A the film area, t the film thickness, and p the porosity. By taking A = 1 cm2 and assuming p = 50%, we estimate σ = 2.2 ± 0.2 and 8.8 ± 1.3 Ω−1 cm−1 for oxidized and reduced nanoITO, respectively. X-ray photoelectron spectroscopy (XPS) was performed on oxidized and reduced samples to explore in more detail differences between the oxidative and reductive annealing processes. Results are summarized in Table S1 in the Supporting Information. Binding energies arising from In 3d5/2, O 1s, and Sn 3d5/2 levels appeared at 444.1, 529.8, and 486.4 eV, respectively, for oxidized nanoITO, consistent with

sphere was used to obtain spectra that accounted for transmittance and reflectance of the FTO glass substrate and nanoITO film. Both oxidized and reduced films were largely transparent in the visible region, consistent with the wide band gap of the parent In2O3 semiconductor.31 From the data in Figure 2, direct band gaps for oxidized and reduced films were measured by linear extrapolation of (hνAbs)2 versus photon energy plots, Figure S1 in the Supporting Information, to be 3.59 and 3.68 eV, respectively. The increase in band gap for the reduced films is qualitatively consistent with conduction band filling due to a higher electron density, known as the Burstein− Moss effect. It should be noted that scattering from ionized impurities has also been invoked in the literature to explain these shifts.31 In the near-IR region of the absorption spectra, an intense localized surface plasmon resonance (LSPR) feature was observed for reduced nanoITO with maximum at ∼1700 nm. A similar feature was observed for oxidized films at longer wavelengths but was unresolved because of limitations in the spectrophotometer. The appearance of these features is indicative of a large concentration of free electrons within the ITO nanoparticles. Electron densities in the oxides could be obtained by spectral modeling of the LSPR feature by application of Drude theory.53,54 Following the method of Nutz et al.,53 spectral modeling of the LSPR feature resulted in electron densities of Ne = 3.1 and 7.8 × 1020 cm−3 for oxidized and reduced nanoITO, respectively. Details of the fitting procedure are given in the Supporting Information. In this analysis, the electron density is determined primarily by the peak position of the LSPR feature, and its width arises from the mean scattering time of free electrons.53,54 Because the D

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 3. (a) UV−visible absorption spectra of nanoITO(ox)-RuP2+ recorded as a function of applied bias in MeCN with 0.1 M LiClO4. Inset shows the mole fraction of -RuP2+ (χRuII) plotted versus Eapp. Data were fit to eq 3 to obtain Eo′(RuP3+/RuP2+) = 1.30 V vs SCE with a nonideality factor of α = 1.53; see text. (b) Molar extinction coefficient difference spectrum calculated for -RuP3+ on the surface of nanoITO.

absorption maximum for oxidized nanoITO was unresolved, our estimate for electron density in the oxidized films is unreliably large. Cyclic voltammetry (CV) measurements at a scan rate of v = 1 mV s−1 were used to estimate conduction band edges (Ecb) in acetonitrile (MeCN) with 0.1 M LiClO4. Measured currents (i) were converted into capacitance (C) by dividing by scan rate, C = i/v. Mott−Schottky plots of C−2 versus the applied bias (Eapp) are shown in Figure S2 in the Supporting Information. Linear extrapolation to C−2 = 0 mF−2 gave the intercepts Eint = −0.80 and −0.77 V vs SCE for oxidized and reduced nanoITO, respectively. The conduction band edge was calculated from Ecb = Eint + kBT/e, which gave Ecb = −0.77 and −0.74 V vs SCE for oxidized and reduced nanoITO, respectively. nanoITO-RuP2+ Characterization. Surface derivatization of nanoITO films with the molecular chromophore [RuII(bpy)2(4,4′-(PO3H2)2-bpy)](Cl)2 (RuP2+; Figure 1), was achieved by soaking nanoITO films overnight in methanol solutions 0.5 mM in RuP2+. UV−visible absorption spectra of nanoITO-RuP2+ films in MeCN with 0.1 M LiClO4 were reported previously and yielded saturated surface coverages of Γ = 3.1 and 2.7 × 10−8 mol cm−2 for oxidized and reduced nanoITO, respectively.28 These values were calculated from the equation, Absλmax = 1000Γελmax, with the extinction coefficient of RuP2+ in H2O with 0.1 M HClO4 (ε453nm = 13 400 M−1 cm−1) used because of the insolubility of RuP2+ in MeCN.49 The spectroelectrochemical behavior of nanoITO-RuP2+ films immersed in MeCN with 0.1 M LiClO4 was investigated to determine the ground-state -RuP3+/-RuP2+ (RuIII/RuII) redox potential and the visible spectrum of oxidized -RuP3+ on nanoITO. Figure 3a shows UV−visible absorption spectra recorded as a function of applied bias for nanoITO(ox)-RuP2+ from 0.82 to 1.83 V vs SCE uncorrected for changes in the nanoITO background. Spectral changes were near instantaneous with changes in Eapp. Complete oxidation of -RuP2+ to -RuP3+ was obtained by holding Eapp >1.5 V vs SCE as seen by the decrease in the metal-to-ligand charge transfer (MLCT) absorption at ∼450 nm. Return of the MLCT feature was observed as Eapp was reversed and stepped to negative potentials. Single-wavelength absorption data at 454 nm were used to calculate the mole fraction of -RuP2+ (χRuII) as a function of Eapp, Figure 3a inset, by taking the spectrum at 0.82 V to be 100% -RuP2+ and the spectrum at 1.83 V to be 100% -RuP3+.

These data were analyzed by using the rearranged Nernst equation solved for χRuII shown in eq 3. Here, Eo′ is the standard reduction potential for the -RuP3+/2+ couple and α is a nonideality factor; a measure of deviation from Nernst behavior which commonly appears for surface-bound redox couples.55,56 The best fit to the experimental data yielded Eo′ (-RuP3+/2+) = 1.30 V vs SCE with α = 1.53. A dashed line simulation is shown for Eo′ = 1.30 V and α = 1.00 for comparison. χRuII = 1/(1 + 10(Eapp − E

o′

)/ α 0.059

)

(3)

The difference absorption spectrum (ΔAbs) for -RuP3+ was calculated by subtracting the spectrum recorded at 0.82 V from that at 1.83 V, Figure S3 in the Supporting Information. Features in the spectrum arising from -RuP3+ were highlighted by subtracting the difference spectrum from that for an underivatized nanoITO(ox) film recorded over the same voltage range. The molar extinction coefficient difference (Δε) spectrum for -RuP3+ shown in Figure 3b was then calculated using the known surface coverage of -RuP2+. From this spectrum, Δε458nm = −13 000 M−1 cm−1 was calculated for nanoITO-RuP3+. The excited-state reduction potential for the -RuP3+/-RuP2+* (RuIII(bpy)3+/RuIII(bpy•‑)2+*) redox couple was calculated from the expression Eo′(-RuP3+/2+*) = Eo′(-RuP3+/2+) − ΔGes, where ΔGes is the excited-state free-energy content above the ground state. The excited-state free energy was estimated from emission measurements of -RuP2+* on a mesoscopic ZrO2 film given that ZrO2 is inert toward excitedstate injection. The room-temperature emission spectrum recorded after 420 nm excitation in MeCN with 0.1 M LiClO4 is shown in Figure S4 in the Supporting Information. Linear extrapolation of the high energy photoluminescence edge to zero gave ΔGes = 2.16 eV.57 The excited-state reduction potential was then calculated to be Eo′(-RuP3+/2+*) = −0.86 V vs SCE. Transient Absorption. Transient absorption experiments were performed on both nanoITO(ox)-RuP2+ and nanoITO(red)-RuP2+ in MeCN with 0.1 M LiClO4 to investigate photoinduced electron transfer. In these experiments a constant external bias was applied to the electrode for the duration of each experiment. The applied bias was varied in 0.2 V increments from 1.0 to −0.8 V vs SCE. This range was chosen to avoid oxidation or reduction of ground-state -RuP2+ molecules at the nanoITO electrode. UV−visible absorption E

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 4. Transient absorption difference spectra obtained after 420 nm laser pulsed excitation of nanoITO(ox)-RuP2+ in MeCN with 0.1 M LiClO4 at the indicated delay times and Eapp = 1.0 (a), 0.0 (b), and −0.8 (c) V vs SCE. Overlaid on the data are simulations (blue dashed lines) based on the known ΔAbs spectra for -RuP2+*, -RuP3+, and nanoITO(e−). Insets show the mole fraction of -RuP2+* (black circles) and -RuP3+ (red triangles) calculated from spectral modeling of transient spectra.

can be observed in Figure 4a,b by the decrease in the ΔAbs feature at 375 nm indicative of the π → π*(bpy•‑) transition of -RuP2+* and the shift in isosbestic point from ∼500 nm for -RuP2+* to ∼600 nm for -RuP3+. The spectral signature for nanoITO(e−) has been reported previously as a tailing bleach in the UV consistent with a Burstein−Moss shift as the electron concentration is increased.28,31,36 Spectral modeling of the transient spectra using a linear summation of the known Δε and ΔAbs spectra for -RuP2+*, -RuP3+, and nanoITO(e−) shown in Figure S5 in the Supporting Information confirmed the electron injection mechanism in eq 5a. The individual Δε difference spectrum for -RuP2+* was obtained by measurements with a ZrO2− RuP2+ sample with Δε450nm = −10 000 M−1 cm−1 at the bleach minimum.58 The Δε spectrum for -RuP3+ and ΔAbs spectrum for nanoITO(e−) were obtained from the spectroelectrochemical experiments discussed above. Simulations at 1 ps and 1 ns are overlaid on the experimental data as blue dashed lines in Figure 4. Insets in Figure 4 show the mole fractions of -RuP2+* and -RuP3+ as a function of time calculated from spectral modeling. Also shown on the right axis are extracted coefficients for nanoITO(e−) obtained from spectral modeling normalized to the -RuP3+ data because an extinction coefficient for nanoITO(e−) is unavailable. Clearly the kinetics for -RuP2+* decay match those of -RuP3+ and nanoITO(e−) growth. From these data we estimate that the apparent injection yield at 1 ns for Eapp = 1.0 and 0.0 V was 0.83 and 0.63, respectively.

spectra recorded before and after transient absorption experiments showed no signs of irreversible degradation of -RuP2+. Representative transient absorption difference spectra are shown in Figure 4 for nanoITO(ox)-RuP2+ recorded following 420 nm excitation at 1 ps and 1 ns delay times for Eapp = 1.0, 0.0, and −0.8 V vs SCE. Over the entire range of applied biases excited-state -RuP2+* formation by eq 4 was complete by 1 ps. Using the estimate of Δε450nm = −10 000 M−1 cm−1, reported for [Ru(bpy)3]2+* in MeCN solution by Yoshimura et al.,58 we calculate that 1.9 × 10−9 mol cm−2 -RuP2+* were generated from each laser pulse at 1 ps. On the basis of the total surface coverage of 3.1 × 10−8 mol cm−2 given above for nanoITO(ox)RuP2+, this amounts to ∼6% of the total -RuP2+ surface concentration. hv

nanoITO−RuP2 + → nanoITO−RuP2 +*

(4)

k inj

nanoITO−RuP2 +* → nanoITO(e−)−RuP3 +

Eapp ≥ 0V (5a)

kq

nanoITO−RuP2 +* → nanoITO−RuP2 +

Eapp < 0V (5b)

k bet

nanoITO(e−)−RuP3 + ⎯→ ⎯ nanoITO−RuP2 +

(6)

For the condition of Eapp ≥ 0 V, ΔAbs spectra from 1 ps to 1 ns were consistent with electron injection from -RuP2+* to nanoITO, kinj in eq 5a, giving -RuP3+ and nanoITO(e−). This F

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

A summary of the KWW kinetic data for ΔAbs375 nm is given in Table 1 where the characteristic lifetime was found to

As an aside, spectral modeling also revealed that direct excitation of the nanoITO film at 420 nm occurred as a function of Eapp, contributing to the transient spectra. This was evidenced by the fact that the transient spectra at 1 ps required the inclusion of a nanoITO(e−) component for Eapp ≥ 0 V. The spectrum for the excited-state nanoITO film would presumably include a bleach of the ground-state UV absorption feature, and including the spectrum for nanoITO(e−) in the transient data was found to adequately model the spectral changes. Electron injection more rapid than 1 ps was not a competitive process because a corresponding amount of -RuP3+ was not detected by spectral modeling. At Eapp < 0 V, the appearance of -RuP3+ was less evident in the transient spectra over time. As can be seen in Figure 4c, the transient spectral features are more consistent with -RuP2+* excited-state decay than electron injection. Spectral modeling of the transient spectra at Eapp = −0.8 V revealed no spectral evidence for -RuP3+ and only minor contributions from nanoITO(e−). The decay of -RuP2+* at Eapp = −0.8 V was found to be accelerated, τ = 380 and 290 ps for oxidized and reduced nanoITO, respectively, compared to that observed for -RuP2+* on ZrO2 under similar conditions (τ = 840 ns).28 This observation is consistent with an additional excited-state quenching pathway, kq in eq 5b, that gives no observable photoproducts. Figure 5 shows single-wavelength kinetic traces at 375 nm at selected applied biases used to monitor electron injection

Table 1. Summary of Transient Absorption Kinetic Data for ΔAbs375 nma nanoITO(ox) Eappb 1.0 0.8 0.6 0.4 0.2 0.0 −0.2 −0.4 −0.6 −0.8

kKWW = [τ /β Γ(1/β)]−1

(8)

kKWWd (s−1)

τc (ps)

βc

0.60 0.70 0.64 0.63 0.64 0.60 0.68 0.77 0.84 0.89

× × × × × × × × × ×

20 21 27 28 33 42 65 140 450 290

0.59 0.59 0.63 0.68 0.66 0.68 0.71 0.80 1.00 0.94

2.1 2.5 2.3 2.1 1.9 1.2 9.3 6.0 3.8 2.4

10

10 1010 1010 1010 1010 1010 109 109 109 109

kKWWd (s−1) 3.3 3.1 2.6 2.8 2.2 1.8 1.2 6.2 2.2 3.3

× × × × × × × × × ×

1010 1010 1010 1010 1010 1010 1010 109 109 109

Data collected in MeCN with 0.1 M LiClO4. bReported as volts versus SCE. cEstimated from eq 7. dCalculated from eq 8.

increase monotonically as Eapp was decreased from 1.0 to −0.8 V. The value of β was nearly constant from 1.0 to −0.2 V with β = 0.64 ± 0.04 and 0.65 ± 0.05 for oxidized and reduced nanoITO, respectively. Beyond Eapp = −0.2 V, β increased to near unity with ΔAbs375nm decay traces being nearly first-order. On time scales longer than 1 ns, recombination of electrons in nanoITO with oxidized -RuP3+ occurred by back electron transfer, kbet, in eq 6. Figure 6 shows single-wavelength kinetic

Figure 6. Transient absorption kinetic traces recorded at 460 nm following 420 nm laser excitation of nanoITO(ox)-RuP2+ as a function of Eapp in MeCN with 0.1 M LiClO4.

dynamics. The data were fit to the Kolrausch−Williams−Watts (KWW) distribution function in eq 7, which is commonly used to define highly dispersive kinetics including interfacial electron transfer at nanoparticle semiconductors where multiple rate processes or local heterogeneities contribute to the overall kinetics.13,59−61 In eq 7, τ is the characteristic lifetime of the distribution and β is the distribution width with β = 1 for single exponential decay. An average rate constant for electron injection at each applied bias was calculated from eq 8, with Γ the gamma function and τ and β derived from KWW fits to the kinetic traces. (7)

31 32 32 33 39 56 83 140 240 380

nanoITO(red)

βc

a

Figure 5. Transient absorbance kinetic traces recorded at 375 nm following 420 nm laser excitation of nanoITO(ox)-RuP2+ as a function of Eapp in MeCN with 0.1 M LiClO4. Black lines are kinetic fits to the KWW function in eq 7.

ΔAbs(t ) = ΔAbs0 exp[−(t /τ )β ]

τc (ps)

traces at 460 nm indicative of back electron transfer via the conversion of -RuP3+ to -RuP2+ for selected applied biases. Attempts to fit the data to the KWW function were unsuccessful with typical values of β < 0.2 indicative of highly nonexponential kinetics giving unreliable estimates of average rate constants. Rather the decay dynamics were described as the rate constant for 1/2 of the ΔAbs460nm change to occur, k1/2 (= 1/τ1/2). A summary of k1/2 values for oxidized and reduced nanoITO as a function of Eapp is given in Table 2. An overall summary of all rate constants measured from 375 and 460 nm ΔAbs kinetic traces is shown in Figure 7 as a function of Eapp for both nanoITO(ox)-RuP2+ and nanoITO(red)-RuP2+ films. G

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

on the thermal preparation methods under air or H2/N2 environments for oxidized and reduced forms of the oxide, respectively. The bixbyite crystal structure of ITO is known to be particularly susceptible to changes in O atom content.52,62,63 Removal of O atoms under reductive conditions leads to oxygen vacancies and reduced sites in the lattice either at nominally InIII or SnIII sites creating energy levels below the conduction band edge. Support for this hypothesis comes from measured electron densities for reduced nanoITO films that were 1.3 times greater than the estimated Sn-doping level indicating that additional dopant states must be present. On the basis of XPS data, there is no difference in the O/(In + Sn) ratio between oxidized and reduced films; however, only small fractional changes in O atom composition would be sufficient to influence the electron density. Thermal treatment in air appears to fill open O atom sites such that Ne should be roughly equal to ND, although the measured electron density of 3.1 × 1020 cm−3 was lower than the estimated Sn-doping density. The difference between the two may arise from incomplete ionization of Sn atoms or from the uncertainties inherent in analysis of the LSPR feature to give Ne. With such large electron densities, band bending in nanoITO occurs on the order of a few nanometers at the surface of ITO nanoparticles with application of an external bias. Albery and Bartlett showed that for spherical semiconductor particles of radius r, the depletion width (w) induced by band bending is proportional to the potential drop within the semiconductor (ΔE) as described by eq 9.33 The potential drop is equal to the difference between EF in the bulk, which is controlled by Eapp, and the flat band potential or conduction band edge at the surface, ΔE = Eapp − Ecb. In the present study, the conduction band edges for oxidized and reduced nanoITO were found to be within the limit of error the same at Ecb = −0.77 and −0.74 V vs SCE, respectively. For ΔE = 1 V, a doping density of ND = 6.0 × 1020 cm−3, and a particle radius of 10 nm, the depletion width calculated from eq 9 is w = 1.3 nm. A depletion width of this magnitude is effectively too small to impede electron transfer between the oxide bulk and surface molecules. With the small depletion widths mesoporous nanoITO films behave like metal electrodes with Eapp = EF at the nanoparticle surface throughout the film structure.

Table 2. Summary of Transient Absorption Kinetic Data for ΔAbs460nma nanoITO(ox)

nanoITO(red)

Eappb

k1/2 (s−1)

k1/2 (s−1)

1.0 0.8 0.6 0.4 0.2 0.0 −0.2 −0.4 −0.6 −0.8

× × × × × × × × × ×

1.1 7.1 1.6 2.0 2.3 2.2 2.5 5.0 3.2 4.1

7

10 107 108 108 108 108 108 108 109 109

3.7 1.9 3.4 3.6 3.8 4.3 5.8 5.9 7.9 2.0

× × × × × × × × × ×

107 108 108 108 108 108 108 108 108 109

a

Data collected in MeCN with 0.1 M LiClO4; bReported as volts versus SCE.

Figure 7. Rate constants reported as kKWW (375 nm) or k1/2 (460 nm) measured from single-wavelength transient absorption kinetic traces following excitation of nanoITO(ox)-RuP2+ (closed circles) and nanoITO(red)-RuP2+ (open triangles) in MeCN with 0.1 M LiClO4.



DISCUSSION nanoITO Characterization. In2O3:Sn is a member of a family of n-type transparent conductive oxide materials that includes, among others, SnO2:F, SnO2:Sb, and ZnO:Al.29−32 Their transparency in the visible is due to the wide band gap of the parent oxide with high conductivities due to the large doping density (ND) of donor atoms in the material. With sufficiently large doping densities, the electronic levels of the donor atoms merge with those of the conduction band and the metal oxide is “degenerately doped.” As defined by Mott’s criterion, degenerate doping occurs at ND = (0.25/ao*)3 where ao* = h2εoεIn2O3/πe2mc* is the effective Bohr radius.31 Using the static dielectric constant for In2O3 of εIn2O3 = 8.9 and the effective mass of conduction band electrons as mc* = 0.35me, we calculate an effective Bohr radius of ao* = 1.4 nm and a critical doping density of ND = 6.4 × 1018 cm−3 for ITO. The doping density for the nanoITO films studied here can be estimated to be 6.0 × 1020 cm−3 based on 9.7% Sn-doping as calculated from XPS data assuming one electron per Sn atom and an In atom density of 6.2 × 1021 cm−3 for In2O3. On the basis of our analysis of the LSPR absorption feature in the near-IR, we estimated electron densities for oxidized and reduced nanoITO films to be Ne = 3.1 and 7.8 × 1020 cm−3, close to the estimated Sn-doping densities. The differences in electron density between the two films is proposed to arise from removal or addition of O atoms in the crystal lattice based

⎛ eN ⎞⎛ 2w ⎟⎞ 2 ΔE = ⎜ D ⎟⎜3 + w r ⎠ ⎝ 6εεo ⎠⎝

(9)

The metallic behavior of nanoITO films was reflected in sheet resistance measurements with calculated conductivities of σ = 2.2 ± 0.2 and 8.8 ± 1.3 Ω−1 cm−1 for oxidized and reduced films, respectively. In general these values are 2−3 orders of magnitude lower than the conductivity of planar ITO films used commercially as transparent electrodes.29 This difference is attributed to the mesoscopic structure of nanoITO films that results in indirect pathways for electron transfer and a higher number of grain boundaries at nanoparticle intersections. The difference in conductivities between oxidized and reduced films is largely a result of their difference in electron density. Based on the relation σ = eμNe where e is the elementary electron charge, μ is the electron mobility through the film, and Ne = 3.1 and 7.8 × 1020 cm−3 for oxidized and reduced nanoITO, μ = 0.04 and 0.07 cm2 V−1 s−1, respectively. The estimated mobilities are similar to each other with the H

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

nanoITO and λ* is the total reorganization energy for excitedstate electron transfer. The integral form of eq 10 shows that multiple electron injection events into a range of unoccupied energy levels in nanoITO are able to contribute to the observed rate constant, kinj.

larger mobility for reduced films possibly arising from the greater degree of sintering in the second annealing step. nanoITO-RuP2+ Characterization. Derivatization of nanoITO films with RuP2+ molecules occurred by well-known binding of acid functional groups to metal oxide surfaces.64,65 Spectroelectrochemical experiments on nanoITO(ox)-RuP2+ in MeCN with 0.1 M LiClO4 showed that stepping the external bias through the wave for the -RuP3+/2+ couple at Eo′(RuP3+/2+) = 1.30 V vs SCE led to rapid spectral equilibration at the applied potential. Conversion to -RuP3+ was complete by 1.5 V and allowed for estimation of Δε458nm = −13,000 M−1 cm−1 for -RuP3+ on nanoITO. Analysis of the spectroelectrochemical data by the modified Nernst relation in eq 3 resulted in a nonideality factor of α = 1.53. Nonideality factors of this magnitude have been observed for other surface-bound redox couples and appear to arise from nearest neighbor interactions across the electrode surface.55,56 Electron Injection. As found for other wide band gap, ntype metal oxides derivatized with surface-bound chromophores, excitation of -RuP2+ on the surface of nanoITO resulted in electron injection on the picoseconds time scale to give -RuP3+ and nanoITO(e−), eq 4-5a.12,66−68 Transient absorption measurements were used to monitor the resulting dynamics as a function of applied bias with monitoring at characteristic wavelengths. In general, electron injection was most efficient at positive applied potentials and became less efficient as the applied bias was stepped in a negative direction. These observations are qualitatively consistent with an expected dependence on the driving force for electron injection with ΔGinjo′ = -F(EF − Eo′(-RuP3+/2+*)). Based on Eo′(RuP3+/2+*) = −0.86 V vs SCE and EF = Eapp for nanoITO, ΔGinjo′ = −1.86 eV at an applied bias of 1.0 V resulting in an injection yield of 83%. The driving force for electron injection at Eapp = 0 V was calculated to be ΔGinjo′ = −0.86 eV resulting in an injection yield of 63%. Past Eapp < 0 V, electron injection became competitive with an alternate quenching pathway as discussed in detail in a later section. Discussion of electron injection kinetics is limited to data in the range Eapp = 1.0 to 0 V. The rate constants for interfacial electron injection were determined by fitting absorbance−time decay traces at 375 nm, a characteristic wavelength for -RuP2+*, to the KolrauschWilliams−Watts distribution function in eq 7-8 such that kinj = kKWW. From these data, electron injection is maximized at Eapp = 1.0 to 0.6 V, giving average rate constants of kinjmax = 2.3 and 3.0 × 1010 s−1 for oxidized and reduced nanoITO, respectively. As the applied bias was stepped to more negative potentials, kinj decreased monotonically for both oxidized and reduced films to roughly half of its maximum value by Eapp = 0 V with the decrease due to the reduction in driving force for electron injection. Electron injection rate constants were analyzed as a function of Eapp by rewriting the Marcus−Gerischer expression in eq 1 for oxidative electron transfer with the result shown in eq 10. In eq 10, g(E)(1 − f(E, EF)) is the distribution of unoccupied electronic states in nanoITO, Hab(E) is the electronic coupling matrix element that describes the overlap of donor and acceptor wave functions, and W2+*(E) is the distribution function that describes the free-energy barrier to electron transfer contributed by the -RuP3+/2+* couple. W2+*(E) is further defined in eq 11 as a Gaussian distribution in the typical Marcus form. ΔGinj(E) = -F(E - Eo′(-RuP3+/2+*)) is the driving force for injection from -RuP2+* to an unoccupied energy level E in

k inj =

2π ℏ



∫−∞ g(E)(1 − f (E , EF))|Hab(E)|2 W2+*(E) dE (10)

W2 +*(E) =

⎛ −(ΔG (E) + λ*)2 ⎞ 1 inj ⎟ exp⎜⎜ ⎟ 4λ*kBT 4π λ*kBT ⎝ ⎠

(11)

Equation 10 can be simplified by assuming that both Hab(E) and g(E) are constant with respect to energy. The latter assumption is based on the expected large distribution of conduction band levels in the oxide consistent with its metallic electronic properties, as discussed above. Taking the lowtemperature limit of the Fermi−Dirac distribution function, where all levels at E ≥ EF are occupied and all levels at E < EF are vacant, leads to eq 12 with kinjmax defined in eq 13. In this limit, the integral is determined only by W2+*(E) and can be solved directly to give the cumulative distribution function in eq 14. ΔGinjo′ is the standard free-energy change for electron injection defined above with the Fermi level of nanoITO controlled by the applied bias. It follows from eq 14 that kinj = kinjmax at -ΔGinjo′ ≫ λ* and kinj = kinjmax/2 at -ΔGinjo′ = λ*. As noted above, these characteristics of the Marcus−Gerischer model are in contrast to intermolecular electron transfer in solution between discrete molecules for which ket = ketmax at -ΔGet = λ and is a result of the contribution toward electron transfer from a range of energy levels within the oxide.3,5,6 k inj =

max k inj

EF

∫−∞

⎛ −(ΔG (E) + λ*)2 ⎞ 1 inj ⎟ dE exp⎜⎜ ⎟ 4λ*kBT 4π λ*kBT ⎝ ⎠ (12)

max k inj =

k inj max k inj

=

2π Hab 2g ℏ

(13)

⎡ ⎛ ΔG o ′ + λ* ⎞⎤ 1⎢ inj ⎟⎥ 1 − erf⎜⎜ ⎟ 2 ⎢⎣ ⎝ 2 λ*kBT ⎠⎥⎦

(14)

Figure 8 shows electron injection rate constants normalized to kinjmax values given above for oxidized and reduced nanoITO, respectively. The solid blue line in Figure 8 shows a collective fit of both data sets to eq 14. The appearance of the expected free-energy dependence for injection is notable and consistent with injection into unfilled levels at energies E < EF. From the fit we estimate that λ* = 0.83 eV for excited-state electron injection. This value is close to Sutin’s estimate of λ* = 0.71 eV for the [Ru(bpy)3]3+/2+* couple.21 The value derived here from the surface injection studies is unique in providing insight into the activation barrier by measurements on a single member of a redox couple. In self-exchange measurements both members of the redox couple contribute to the overall barrier.21,69 It follows from eqs 12 and eq 14 that the W2+*(E) distribution can be directly calculated from the derivative of the fit to the experimental data in Figure 8. This is shown as the dashed blue line in Figure 8. Note that the peak of the Gaussian distribution occurs at -ΔGinjo′ = λ* or Eapp = Eo′ + λ*. I

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

clearly in Figure 7 where the rate constants measured for 460 nm rose suddenly at Eapp < 0 V until they were almost identical with those measured for 375 nm at Eapp < −0.6 V where spectral modeling showed that only -RuP2+* was present on the nanoITO surface. The excited-state lifetime of -RuP2+* on ZrO2, which is inert toward electron injection, was previously reported to be 840 ns.28 This value is similar to the lifetime of -RuP2+* in fluid solution measured in the absence of excitedstate quenching.49 By comparison, based on kKWW rate constants measured at 375 nm, the lifetime of -RuP2+* on oxidized and reduced nanoITO at Eapp = −0.8 V was 420 and 300 ps, respectively. The large decrease in lifetime points to one or more additional quenching pathways for nanoITO-RuP2+* at Eapp < 0 V. The most likely mechanisms are energy transfer with ITO nanoparticles or reductive quenching of -RuP2+* by excess electrons present in ITO. Energy transfer between molecular chromophores and metal surfaces is well-known in the literature and is due to the large distribution of electronic levels and associated electronic states in the metal which provide pathways for energy transfer.6,70−72 The nanoITO films used in these experiments are largely transparent in the spectral range for -RuP2+* emission (600−900 nm); however, there is overlap between the emission spectrum of the chromophore and the tail end of the LSPR absorption that peaks in the nearIR. The intensity of the LSPR feature for ITO and ATO nanoparticles is also known to be enhanced at large negative potentials.34,35,40,73 Enhanced overlap of the -RuP2+* photoluminescence spectrum with the surface plasmon resonance of nanoITO at negative potentials therefore may contribute to rapid energy-transfer quenching of the excited-state. The largely triplet nature of the -RuP2+* MLCT excited-state suggests that the exact nature of energy transfer is more likely to be rooted in a Dexter mechanism as opposed to Förster energy transfer.74−77 The second alternative quenching mechanism requires reductive electron transfer between excess electrons in nanoITO and -RuP2+* followed by rapid recombination of the charge separated state, eqs 15-16. The excited-state reduction potential for the -RuP2+*/-RuP+ (RuIII(bpy•‑)2+*/ RuII(bpy•‑)+) couple can be estimated from the expression Eo′(RuP2+*/+) = Eo′(-RuP2+/+) + ΔGes. Taking the ground-state reduction potential Eo′(-RuP2+/+) = −1.5 V vs SCE reported by Hoertz et al.37 and ΔGes = 2.16 eV reported here, we calculate Eo′(-RuP2+*/+) = 0.66 V vs SCE. The driving force for reductive quenching of -RuP2+* was therefore ΔGo = −0.86 eV at Eapp = −0.2 V vs SCE and became more favorable at more negative applied potentials. The lack of observation for the Ru+ product of the reductive quenching mechanism in the transient absorption spectra at Eapp < 0 V could point to a rapid recombination step, eq 16, to yield ground-state products. If the rate constant for the recombination reaction is greater than that of excited-state quenching, eq 15, then rate-limited excited-state decay would be observed in the transient spectra. Further investigations into these alternative quenching mechanisms are currently underway.

Figure 8. Normalized electron injection rate constants for oxidized (closed circles) and reduced (open triangles) nanoITO-RuP2+ obtained from transient absorption measurements at room temperature in MeCN with 0.1 M LiClO4. The solid blue line shows a fit of the data to eq 14 with λ* = 0.83 eV. The derivative of the fit was used to obtain the free-energy barrier distribution function for -RuP2+* shown as the blue dashed line.

An estimate of the electronic coupling matrix element, Hab, between -RuP2+* and nanoITO was obtained by using eq 13, which requires knowledge of the distribution of electronic states in nanoITO. For metals, this value is assumed constant and can be calculated from the expression, g = nDm/dm, with Dm = (3/2)*(Ne/ΣF) the density of electronic states, dm the density of metal atoms, and n the number of metal atoms capable of accepting the transferred electron.22 Calculation of Dm using the measured values for Ne and calculated Fermi energies according to ΣF = (ℏ2/2mc*)(3π2Ne)2/3 for oxidized and reduced nanoITO, gave 2.8 and 3.8 × 1021 cm−3 eV−1, respectively. Using the atomic density of In atoms in ITO dm = 6.2 × 1021 cm−3 and assuming n = 1, gives g = 0.45 and 0.61 eV−1 for oxidized and reduced nanoITO, respectively. Notably, the ratio of g(red)/g(ox)= 1.36 is close to kinjmax(red)/ kinjmax(ox) = 1.27 indicating that the difference in electron injection rate constants between nanoITO films is determined largely by the difference in electron densities and that the electronic coupling is similar for both oxidized and reduced films. Employing the constants derived above and using eq 13, gives Hab = 45 cm−1 for both oxidized and reduced nanoITO assuming n = 1. It is likely that more than one metal atom is capable of accepting the injected electron given the close interatomic distance in the nanocrystalline solid. Royea et al. have given a more detailed calculation of the electronic distribution that uses an electron-transfer distance and interatomic spacing to determine n.22 Here, we estimate that for an upper limit of n = 5 metal atoms Hab = 20 cm−1. The magnitude of Hab is consistent with nonadiabatic electron transfer which is typically observed for electron injection from Ru(II) polypyridyl excited-states into metal oxide semiconductors.12,13 Quenching Mechanisms at Negative Potentials. With Eapp < 0 V it was evident that electron injection was not the only excited-state process occurring at the nanoITO surface. This conclusion was based on the rapid decay of -RuP2+* and decrease in the characteristic features arising for the injection products, -RuP3+ and nanoITO(e−). This can also be seen

nanoITO−RuP2 +* → nanoITO(h+)−RuP+(e−)

(15)

nanoITO(h+)−RuP+(e−) → nanoITO−RuP2 +

(16)

Back Electron Transfer. Back electron transfer between nanoITO(e−) and -RuP3+, eq 6, occurs on the nanosecond time scale. These kinetics were more complex than for injection and J

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B were characterized by the half time (τ1/2) for the absorbance change at 460 nm to occur with k1/2 = 1/τ1/2. From Figure 7, k1/2 was found to be bias dependent, increasing rapidly as Eapp was decreased from 1.0 V vs SCE before reaching a plateau near 0 V. From these data, k1/2max = 2.3 and 4.0 × 108 s−1, respectively for oxidized and reduced nanoITO. These changes are consistent with the decrease in driving force for back electron transfer with ΔGbeto′ = -F(Eo′(-RuP3+/2+) - EF). Based on Eo′(-RuP3+/2+) = 1.30 V vs SCE, the driving force at the most positive bias of 1.0 V was −0.3 eV reaching −1.3 eV at a bias of 0 V. Beyond Eapp < 0 V, k1/2 increased further because of contributions from excited-state -RuP2+* decay as discussed above. These data were therefore neglected in the Marcus− Gerischer analysis discussed below. Analysis of back electron-transfer kinetics by Marcus− Gerischer theory follows directly from the discussion of electron injection kinetics developed above but involving the ground-state -RuP3+/2+ couple. In applying eqs 1-2, for reductive electron transfer to -RuP3+ it was assumed that g(E) and Hab(E) are constant with energy and applying the low-temperature limit of the Fermi−Dirac distribution function gives eq 17. In this equation, ΔGbet(E) = -F(Eo′(-RuP3+/2+) - E) is the driving force for back electron transfer from an occupied energy level E in nanoITO to -RuP3+ and λ is the total reorganization energy for ground-state electron transfer. In this limit, the maximum rate constant for back electron transfer, kbetmax, is given by eq 18. Solution of the integral in eq 17 results in the cumulative distribution function shown in eq 19. max k bet = k bet

∫E



F

Figure 9. Normalized back electron-transfer rate constants for oxidized (closed circles) and reduced (open triangles) nanoITORuP2+ obtained from transient absorption measurements in MeCN with 0.1 M LiClO4 at room temperature. The solid blue line is a fit to eq 19 with λ = 0.56 eV. Differentiation of the fit resulted in the dashed line free-energy barrier distribution function for -RuP3+.

oxidized molecule at the surface.79 However, most attempts to control ΔGbeto′ by either adjusting Ecb using pH changes at the solid/liquid interface or applying a negative bias to the TiO2 electrode have shown no dependence of back electron-transfer kinetics on ΔGbeto′.80,81 These results were attributed to slow, trap-state limited diffusion of electrons from the nanoparticle bulk to the surface that masked interfacial electron-transfer kinetics.80−83 Perhaps the best evidence for Marcus controlled back electron-transfer kinetics comes from detailed temperature dependence studies on a series of Ru(II) and Os(II) polypyridyl complexes on the surface of TiO2.84 In these studies the reorganization energy was found be unique to the particular molecule but ranged from λ = 0.55−1.18 eV. These values were found to be larger than calculated estimates for λ based on solvent reorganization leading the authors to suggest that contributions to λ from the TiO2 semiconductor may be present. In the present study, the large dopant densities in nanoITO allow for band bending within the nanoparticles without rate limitations from inter- or intraparticle electron diffusion. In addition, the wide distribution of dopant levels below the conduction band edge allows for efficient electron transfer from electronic levels far from EF with no appearance of an inverted region. Our estimate of λ = 0.56 eV for the -RuP3+/2+ couple at the surface of nanoITO is therefore a direct measurement of contributions to the electron-transfer barrier of only the molecular component with contributions from nanoITO expected to be minimal. Electronic coupling constants for back electron transfer were calculated by following the methods used for electron injection. Using the same estimates for the distribution of electronic states in oxidized and reduced nanoITO, Hab = 4.7 and 5.3 cm−1, respectively for electronic coupling with one metal atom in ITO (n = 1) and Hab = 2.1 and 2.4 cm−1 for n = 5. These values are similar to those found from temperature-dependent back electron-transfer measurements at TiO2 discussed above with Hab = 5−9 cm−1 for RuIII(4,4′-(CO2H)-bpy)2(CN)2 and RuIII(4,4′-(CO2H)-bpy)2(SCN)2.84 The lower magnitude of electronic coupling for back electron transfer compared to electron injection is an expected consequence of the nature of the electron-transfer processes

⎛ −(ΔG (E) + λ)2 ⎞ 1 bet ⎟dE exp⎜ 4λkBT 4πλkBT ⎝ ⎠ (17)

max k bet

2π = Hab 2g ℏ

(18)

⎡ ⎛ ΔG o ′ + λ ⎞⎤ k bet 1⎢ bet ⎜⎜ ⎟⎟⎥ 1 erf = − max 2 ⎢⎣ k bet 2 λkBT ⎠⎥⎦ ⎝ max

(19) max

Figure 9 shows plots of kbet/kbet (= k1/2/k1/2 ) for both oxidized and reduced nanoITO-RuP2+ as a function of Eapp and ΔGbeto′. The experimental data were fit to eq 19 to give λ = 0.56 eV for the total reorganization energy of the -RuP3+/2+ couple. This value is close to values of 0.4 and 0.57 eV reported for the [Ru(bpy)3]3+/2+ couple in aqueous solution.21,69 In these earlier examples, the total reorganization energy was found to be dominated by solvent reorganization (λo). Our results point to a local solvent environment at the surface of the oxide that is not significantly different from that in fluid solution. This observation is contrary to models for the reorganization energy at solid-solution interfaces that predict a decrease in λo due to partial desolvation at the surface and may be a consequence of the phosphonate binding.4,78 Back electron transfer at the surface of nanoITO is uniquely different from recombination at intrinsic metal oxide nanoparticles such as TiO2, SnO2, and ZnO. For these semiconductors, the Marcus−Gerischer model for back electron transfer is not applicable because of a lack of band bending on the nanoparticle scale and the low intrinsic doping densities in the oxide. Therefore, the standard Marcus model is expected to control back electron-transfer kinetics with some reports citing inverted region behavior due to the large ΔGbeto′ present for electron transfer from the conduction band edge to the K

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

(4) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications, 2nd Ed.; Wiley: New York, 2001. (5) Gerischer, H. Electrochemical Techniques for the Study of Photosensitization. Photochem. Photobiol. 1972, 16, 243−260. (6) Gerischer, H.; Willig, F. Reaction of Excited Dye Molecules at Electrodes. Top. Curr. Chem. 1976, 61, 31−84. (7) Nozik, A. J.; Memming, R. Physical Chemistry of Semiconductor - Liquid Interfaces. J. Phys. Chem. 1996, 100, 13061−13078. (8) Hagfeldt, A.; Grätzel, M. Light-Induced Redox Reactions in Nanocrystalline Systems. Chem. Rev. (Washington, DC, U.S.) 1995, 95, 49−68. (9) Saveant, J.-M. Electron Transfer, Bond Breaking, and Bond Formation. Acc. Chem. Res. 1993, 26, 455−461. (10) Savéant, J.-M. Molecular Catalysis of Electrochemical Reactions. Mechanistic Aspects. Chem. Rev. (Washington, DC, U.S.) 2008, 108, 2348−2378. (11) Liu, M.; Zhang, R.; Chen, W. Graphene-Supported Nanoelectrocatalysts for Fuel Cells: Synthesis, Properties, and Applications. Chem. Rev. (Washington, DC, U.S.) 2014, 114, 5117−5160. (12) Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Dye-Sensitized Solar Cells. Chem. Rev. (Washington, DC, U.S.) 2010, 110, 6595−6663. (13) Ardo, S.; Meyer, G. J. Photodriven Heterogeneous Charge Transfer with Transition-Metal Compounds Anchored to TiO2 Semiconductor Surfaces. Chem. Soc. Rev. 2009, 38, 115−164. (14) Youngblood, W. J.; Lee, S.-H. A.; Maeda, K.; Mallouk, T. E. Visible Light Water Splitting Using Dye-Sensitized Oxide Semiconductors. Acc. Chem. Res. 2009, 42, 1966−1973. (15) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi, Q.; Santori, E. a; Lewis, N. S. Solar Water Splitting Cells. Chem. Rev. (Washington, DC, U.S.) 2010, 110, 6446−6473. (16) Bard, A. J.; Fox, M. A. Artificial Photosynthesis: Solar Splitting of Water to Hydrogen and Oxygen. Acc. Chem. Res. 1995, 28, 141− 145. (17) Hush, N. S. Adiabatic Theory of Outer Sphere ElectronTransfer Reactions in Solution. Trans. Faraday Soc. 1961, 57, 557− 580. (18) Marcus, R. A. On the Theory of Oxidation-Reduction Reactions Involving Electron Transfer. I. J. Chem. Phys. 1956, 24, 966−978. (19) Marcus, R. A. Exchange Reactions and Electron Transfer Reactions Including Isotopic Exchange. Discuss. Faraday Soc. 1960, 29, 21−31. (20) Marcus, R. A. Chemical and Electrochemical Electron-Transfer Tehory. Annu. Rev. Phys. Chem. 1964, 15, 155−196. (21) Sutin, N. Nuclear, Electronic, and Frequency Factors in Electron-Transfer Reactions. Acc. Chem. Res. 1982, 15, 275−282. (22) Royea, W. J.; Fajardo, A. M.; Lewis, N. S. Fermi Golden Rule Approach to Evaluating Outer-Sphere Electron-Transfer Rate Constants at Semiconductor/Liquid Interfaces. J. Phys. Chem. B 1997, 101, 11152−11159. (23) Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. Coadsorption of Ferrocene-Terminated and Unsubstituted Alkanethiols on Gold: Electroactive Self-Assembled Monolayers. J. Am. Chem. Soc. 1990, 112, 4301−4306. (24) Smalley, J. F.; Feldberg, S. W.; Chidsey, C. E. D.; Linford, M. R.; Newton, M. D.; Liu, Y.-P. The Kinetics of Electron Transfer Through Ferrocene-Terminated Alkanethiol Monolayers on Gold. J. Phys. Chem. 1995, 99, 13141−13149. (25) Terrettaz, S.; Becka, A. M.; Traub, M. J.; Fettinger, J. C.; Miller, C. J. ω-Hydroxythiol Monolayers at Au Electrodes. 5. Insulated Electrode Voltammetric Studies of Cyano/Bipyridyl Iron Complexest. J. Phys. Chem. 1995, 99, 11216−11224. (26) Li, T. T.-T.; Weaver, M. J. Intramolecular Electron Transfer at Metal Surfaces. 4. Dependence of Tunneling Probability upon DonorAcceptor Separation Distance. J. Am. Chem. Soc. 1984, 106, 6107− 6108. (27) Finklea, H.; Hanshew, D. D. Electron-Transfer Kinetics in Organized Thiol Monolayers with Attached Pentaamine(pyridine) ruthenium Redox Centers. J. Am. Chem. Soc. 1992, 114, 3173−3181.

involved and the orbital character of the donor or acceptor orbitals in -RuP2+* or -RuP3+. Injection from the excited-state involves electron donation from a largely ligand-based π*(bpy) orbital extensively mixed with the phosphonate linking group promoting electronic coupling with the ITO surface. Back electron transfer to -RuP3+ involves dπ acceptor levels largely localized on the metal with a corresponding decrease in overlap with ITO donor orbitals.



CONCLUSION We have described in detail the photoinduced interfacial electron-transfer reactions between the surface-bound Ru(II) polypyridyl chromophore [RuII(bpy)2(4,4′-(PO3H2)2-bpy)]2+ and the degenerately doped metal oxide nanoITO. A detailed analysis of the driving force dependence of excited-state electron injection and ground-state back electron-transfer reactions has provided experimental estimates for the total reorganization energies and electronic coupling matrix elements for the two processes. For electron injection, λ* = 0.83 eV was obtained for the -RuP3+/2+* couple with Hab ∼ 20−45 cm−1. For back electron transfer, λ = 0.56 eV was estimated for the -RuP3+/2+ couple with Hab ∼ 2−4 cm−1. Our results demonstrate that high surface area, transparent conductive oxides can be used effectively to study interfacial photoinduced electron transfer. The approach used here should be germane to any molecular redox couple surface-bound to nanoITO and useful for the evaluation and quantitation of heterogeneous electron transfer for a wide range of organic and inorganic chromophores and molecular assemblies.



ASSOCIATED CONTENT

S Supporting Information *

Summary of XPS data, Drude analysis, (hνAbs)2 versus photon energy plots, Mott−Schottky plots of C−2 versus the applied bias (Eapp), difference absorption spectrum (ΔAbs) for -RuP3+, room-temperature emission spectrum recorded after 420 nm excitation in MeCN with 0.1 M LiClO4, spectral modeling of the transient spectra. 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 This material is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Award DE-FG02-06ER15788. The authors thank Dr. Carrie Donley for assistance with XPS studies as well as Tim Garvey and Dr. Rene Lopez for assistance in sheet resistance measurements.



REFERENCES

(1) Murray, R. W. Chemically Modified Electrodes. Acc. Chem. Res. 1980, 117, 135−141. (2) Fajardo, A. M.; Lewis, N. S. Rate Constants for Charge Transfer Across Semiconductor-Liquid Interfaces. Science 1996, 274, 969−972. (3) Chidsey, C. E. Free Energy and Temperature Dependence of Electron Transfer at the Metal-Electrolyte Interface. Science 1991, 251, 919−922. L

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Dye-Sensitized Photoelectrochemical Cell. J. Am. Chem. Soc. 2009, 131, 926−927. (48) Heimer, T. A.; D’Arcangelis, S. T.; Farzad, F.; Stipkala, J. M.; Meyer, G. J. An Acetylacetonate-Based Semiconductor−Sensitizer Linkage. Inorg. Chem. 1996, 35, 5319−5324. (49) Hanson, K.; Brennaman, M. K.; Ito, A.; Luo, H.; Song, W.; Parker, K. A.; Ghosh, R.; Norris, M. R.; Glasson, C. R. K.; Concepcion, J. J.; et al. Structure−Property Relationships in PhosphonateDerivatized, RuII Polypyridyl Dyes on Metal Oxide Surfaces in an Aqueous Environment. J. Phys. Chem. C 2012, 116, 14837−14847. (50) Pavlishchuk, V. V.; Addison, A. W. Conversion Constants for Redox Potentials Measured versus Different Reference Electrodes in Acetonitrile Solutions at 25° C. Inorg. Chim. Acta 2000, 298, 97−102. (51) Donley, C.; Dunphy, D.; Paine, D.; Carter, C.; Nebesny, K.; Lee, P.; Alloway, D.; Armstrong, N. R. Characterization of Indium−Tin Oxide Interfaces Using X-Ray Photoelectron Spectroscopy and Redox Processes of a Chemisorbed Probe Molecule: Effect of Surface Pretreatment Conditions. Langmuir 2002, 18, 450−457. (52) Fan, J. C. C.; Goodenough, J. B. X-Ray Photoemission Spectroscopy Studies of Sn-Doped Indium-Oxide Films. J. Appl. Phys. 1977, 48, 3524−3531. (53) Nütz, T.; zum Felde, U.; Haase, M. Wet-Chemical Synthesis of Doped Nanoparticles: Blue-Colored Colloids of N-Doped SnO2:Sb. J. Chem. Phys. 1999, 110, 12142−12150. (54) Mendelsberg, R. J.; Garcia, G.; Li, H.; Manna, L.; Milliron, D. J. Understanding the Plasmon Resonance in Ensembles of Degenerately Doped Semiconductor Nanocrystals. J. Phys. Chem. C 2012, 116, 1226−12231. (55) Hu, K.; Robson, K. C. D.; Johansson, P. G.; Berlinguette, C. P.; Meyer, G. J. Intramolecular Hole Transfer at Sensitized TiO2 Interfaces. J. Am. Chem. Soc. 2012, 134, 8352−8355. (56) Brown, A. P.; Anson, F. C. Cyclic and Differential Pulse Voltammetric Behavior of Reactants Confined to the Electrode Surface. Anal. Chem. 1977, 49, 1589−1595. (57) Adamson, A. W. Properties of Excited States. J. Chem. Educ. 1983, 60, 797−802. (58) Yoshimura, A.; Hoffman, M. Z.; Sun, H. An Evaluation of the Excited State Absorption Spectrum of Ru(bpy)32+ in Aqueous and Acetonitrile Solutions. J. Photochem. Photobiol., A 1993, 70, 29−33. (59) Lindsey, C. P.; Patterson, G. D. Detailed Comparison of the Williams−Watts and Cole−Davidson Functions. J. Chem. Phys. 1980, 73, 3348−3357. (60) Williams, G.; Watts, D. C. Non-Symmetrical Dielectric Relaxation Behaviour Arising from a Simple Empirical Decay Function. Trans. Faraday Soc. 1969, 66, 80−85. (61) Nelson, J.; Chandler, R. E. Random Walk Models of Charge Transfer and Transport in Dye Sensitized Systems. Coord. Chem. Rev. 2004, 248, 1181−1194. (62) Frank, G.; Kostlin, H. Electrical Properties and Defect Model of Tin-Doped Indium Oxide Layers. Appl. Phys. A: Mater. Sci. Process. 1982, 27, 197−206. (63) Gassenbauer, Y.; Klein, A. Electronic and Chemical Properties of Tin-Doped Indium Oxide (ITO) Surfaces and ITO/ZnPc Interfaces Studied in-Situ by Photoelectron Spectroscopy. J. Phys. Chem. B 2006, 110, 4793−4801. (64) Gillaizeau-Gauthier, I.; Odobel, F.; Alebbi, M.; Argazzi, R.; Costa, E.; Bignozzi, C. a; Qu, P.; Meyer, G. J. Phosphonate-Based Bipyridine Dyes for Stable Photovoltaic Devices. Inorg. Chem. 2001, 40, 6073−6079. (65) Anderson, S.; Constable, E. C.; Dare-Edwards, M. P.; Goodenough, J. B.; Hamnett, A.; Seddon, K. R.; Wright, R. D. Chemical Modification of a Titanium (IV) Oxide Electrode to Give Stable Dye Sensitisation without a Supersensitiser. Nature 1979, 280, 571−573. (66) Ardo, S.; Sun, Y.; Staniszewski, A.; Castellano, F. N.; Meyer, G. J. Stark Effects after Excited-State Interfacial Electron Transfer at Sensitized TiO2 Nanocrystallites. J. Am. Chem. Soc. 2010, 132, 6696− 6709.

(28) Farnum, B. H.; Morseth, Z. a; Brennaman, M. K.; Papanikolas, J. M.; Meyer, T. J. Driving Force Dependent, Photoinduced Electron Transfer at Degenerately Doped, Optically Transparent Semiconductor Nanoparticle Interfaces. J. Am. Chem. Soc. 2014, 136, 15869−15872. (29) Chopra, K. L.; Major, S.; Pandya, D. K. Transparent Conductors - A Status Review. Thin Solid Films 1983, 102, 1−46. (30) Ellmer, K. Past Achievements and Future Challenges in the Development of Optically Transparent Electrodes. Nat. Photonics 2012, 6, 809−817. (31) Hamberg, I.; Granqvist, C. G. Evaporated Sn-Doped In2O3 Films: Basic Optical Properties and Applications to Energy-Efficient Windows. J. Appl. Phys. 1986, 60, R123−R159. (32) Robertson, J.; Falabretti, B.Electronic Structure of Transparent Conducting Oxides (Ch. 2). In Handbook of Transparent Conductors; Ginley, D. S., Ed.; Springer: New York, 2011; pp 27−50. (33) Albery, W. J.; Bartlett, P. N. The Transport and Kinetics of Photogenerated Carriers in Colloidal Semiconductor Electrode Particles. J. Electrochem. Soc. 1984, 131, 315−325. (34) Boschloo, G.; Fitzmaurice, D. Spectroelectrochemistry of Highly Doped Nanostructured Tin Dioxide Electrodes. J. Phys. Chem. B 1999, 103, 3093−3098. (35) Zum Felde, U.; Haase, M.; Weller, H. Electrochromism of Highly Doped Nanocrystalline SnO2:Sb. J. Phys. Chem. B 2000, 104, 9388−9395. (36) Farnum, B. H.; Morseth, Z. A.; Lapides, A. M.; Rieth, A. J.; Hoertz, P. G.; Brennaman, M. K.; Papanikolas, J. M.; Meyer, T. J. Photoinduced Interfacial Electron Transfer within a Mesoporous Transparent Conducting Oxide Film. J. Am. Chem. Soc. 2014, 136, 2208−2211. (37) Hoertz, P. G.; Chen, Z.; Kent, C. A.; Meyer, T. J. Application of High Surface Area Tin-Doped Indium Oxide Nanoparticle Films as Transparent Conducting Electrodes. Inorg. Chem. 2010, 49, 8179− 8181. (38) Hanson, K.; Losego, M. D.; Kalanyan, B.; Ashford, D. L.; Parsons, G. N.; Meyer, T. J. Stabilization of [Ru(bpy)2(4,4′(PO3H2)bpy)]2+ on Mesoporous TiO2 with Atomic Layer Deposition of Al2O3. Chem. Mater. 2013, 25, 3−5. (39) Hou, K.; Puzzo, D.; Helander, M. G.; Lo, S. S.; Bonifacio, L. D.; Wang, W.; Lu, Z.-H.; Scholes, G. D.; Ozin, G. a. Dye-Anchored Mesoporous Antimony-Doped Tin Oxide Electrochemiluminescence Cell. Adv. Mater. (Weinheim, Ger.) 2009, 21, 2492−2496. (40) Llordés, A.; Garcia, G.; Gazquez, J.; Milliron, D. J. Tunable Near-Infrared and Visible-Light Transmittance in Nanocrystal-in-Glass Composites. Nature 2013, 500, 323−327. (41) Schwab, P. F. H.; Diegoli, S.; Biancardo, M.; Bignozzi, C. A. Novel Ru-Dioxolene Complexes as Potential Electrochromic Materials and NIR Dyes. Inorg. Chem. 2003, 42, 6613−6615. (42) Chen, Z.; Vannucci, A. K.; Concepcion, J. J.; Jurss, J. W.; Meyer, T. J. Proton-Coupled Electron Transfer at Modified Electrodes by Multiple Pathways. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, E1461− E1469. (43) Chen, Z.; Concepcion, J. J.; Luo, H.; Hull, J. F.; Paul, A.; Meyer, T. J. Nonaqueous Catalytic Water Oxidation. J. Am. Chem. Soc. 2010, 132, 17670−17673. (44) Grätzel, M. Photoelectrochemical Cells. Nature 2001, 414, 338− 344. (45) Alibabaei, L.; Brennaman, M. K.; Norris, M. R.; Kalanyan, B.; Song, W.; Losego, M. D.; Concepcion, J. J.; Binstead, R. A.; Parsons, G. N.; Meyer, T. J. Solar Water Splitting in a Molecular Photoelectrochemical Cell. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 20008−20013. (46) Gao, Y.; Ding, X.; Liu, J.; Wang, L.; Lu, Z.; Li, L.; Sun, L. Visible Light Driven Water Splitting in a Molecular Device with Unprecedentedly High Photocurrent Density. J. Am. Chem. Soc. 2013, 135, 4219−4222. (47) Youngblood, W. J.; Lee, S. A.; Kobayashi, Y.; Hernandez-pagan, E. A.; Hoertz, P. G.; Moore, T. A.; Moore, A. L.; Gust, D.; Mallouk, T. E. Photoassisted Overall Water Splitting in a Visible Light-Absorbing M

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B (67) Anderson, N. A.; Lian, T. Ultrafast Electron Transfer at the Molecule-Semiconductor Nanoparticle Interface. Annu. Rev. Phys. Chem. 2005, 56, 491−519. (68) Tachibana, Y.; Haque, S. A.; Mercer, I. P.; Moser, J. E.; Klug, D. R.; Durrant, J. R. Modulation of the Rate of Electron Injection in DyeSensitized Nanocrystalline TiO2 Films by Externally Applied Bias. J. Phys. Chem. B 2001, 105, 7424−7431. (69) Young, R. C.; Keene, F. R.; Meyer, T. J. Measurement of Rates of Electron Transfer between Ru(bpy)33+ and Fe(phen)32+ and between Ru(phen)33+ and Ru(bpy)32+ by Differential Excitation Flash Photolysis. J. Am. Chem. Soc. 1977, 99, 2468−2473. (70) Chance, R. R. Lifetime of an Emitting Molecule near a Partially Reflecting Surface. J. Chem. Phys. 1974, 60, 2744−2748. (71) Hale, J. M. Electrode Reactions Involving Electronically Excited States of Molecules. J. Phys. Chem. 1968, 73, 3196−3201. (72) Kuhn, H. Classical Aspects of Energy Transfer in Molecular Systems. J. Chem. Phys. 1970, 53, 101−108. (73) Garcia, G.; Buonsanti, R.; Runnerstrom, E. L.; Mendelsberg, R. J.; Llordes, A.; Anders, A.; Richardson, T. J.; Milliron, D. J. Dynamically Modulating the Surface Plasmon Resonance of Doped Semiconductor Nanocrystals. Nano Lett. 2011, 11, 4415−4420. (74) Ito, A.; Meyer, T. J. The Golden Rule. Application for Fun and Profit in Electron Transfer, Energy Transfer, and Excited-State Decay. Phys. Chem. Chem. Phys. 2012, 14, 13731−13745. (75) Allen, G. H.; White, R. P.; Rillema, D. P.; Meyer, T. J. Synthetic Control of Excited-State Properties. Tris-Chelate Complexes Containing the Ligands 2,2′-Bipyrazine, 2,2′-Bipyridine, and 2,2′Bipyrimidine. J. Am. Chem. Soc. 1984, 106, 2613−2620. (76) Chen, P.; Meyer, T. J. Medium Effects on Charge Transfer in Metal Complexes. Chem. Rev. (Washington, DC, U.S.) 1998, 98, 1439− 1478. (77) Caspar, J. V.; Meyer, T. J. Photochemistry of (Ru(bpy)32+). Solvent Effects. J. Am. Chem. Soc. 1983, 105, 5583−5590. (78) Hamann, T. W.; Gstrein, F.; Brunschwig, B. S.; Lewis, N. S. Measurement of the Dependence of Interfacial Charge-Transfer Rate Constants on the Reorganization Energy of Redox Species at n-ZnO/ H2O Interfaces. J. Am. Chem. Soc. 2005, 127, 13949−13954. (79) Moser, J. E.; Grätzel, M. Observation of Temperature Independent Heterogeneous Electron Transfer Reactions in the Inverted Marcus Region. Chem. Phys. 1993, 176, 493−500. (80) Yan, S. G.; Hupp, J. T. Semiconductor-Based Interfacial Electron-Transfer Reactivity: Decoupling Kinetics from pH-Dependent Band Energetics in a Dye-Sensitized Titanium Dioxide/Aqueous Solution System. J. Phys. Chem. 1996, 100, 6867−6870. (81) Haque, S. A.; Tachibana, Y.; Klug, D. R.; Durrant, J. R. Charge Recombination Kinetics in Dye-Sensitized Nanocrystalline Titanium Dioxide Films under Externally Applied Bias. J. Phys. Chem. B 1998, 102, 1745−1749. (82) Nelson, J.; Haque, S.; Klug, D.; Durrant, J. Trap-Limited Recombination in Dye-Sensitized Nanocrystalline Metal Oxide Electrodes. Phys. Rev. B 2001, 63, 205321. (83) Barzykin, A. V.; Tachiya, M. Mechanism of Charge Recombination in Dye-Sensitized Nanocrystalline Semiconductors: Random Flight Model. J. Phys. Chem. B 2002, 106, 4356−4363. (84) Kuciauskas, D.; Freund, M. S.; Gray, H. B.; Winkler, J. R.; Lewis, N. S. Electron Transfer Dynamics in Nanocrystalline Titanium Dioxide Solar Cells Sensitized with Ruthenium or Osmium Polypyridyl Complexes. J. Phys. Chem. B 2001, 105, 392−403.

N

DOI: 10.1021/jp512624u J. Phys. Chem. B XXXX, XXX, XXX−XXX