Ostwald Isolation to Determine the Reaction Order for TiO2(e−)|S+→
TiO2|S Charge Recombination at Sensitized TiO2 Interfaces
Erinn C. Brigham and Gerald J. Meyer*
Department of Chemistry, Johns Hopkins University, 3400 N. Charles Street Baltimore, Maryland 21218, United States
S Supporting Information

ABSTRACT: Kinetic isolation conditions were identified that
enabled determination of the reaction order for interfacial
charge recombination at a sensitized mesoporous TiO2 thin
film. An external bias was used to maintain a fixed and known
number of oxidized sensitizers, S+, or TiO2 electrons,
TiO2(e−)s. Pulsed laser excitation resulted in excited state
injection and the subsequent TiO2(e−)|S+ → TiO2|S reaction
was quantified spectroscopically. The data provide compelling
evidence that the rate law for charge recombination under
reverse bias is r = k[S+]1[TiO2(e−)]1 with k = 5.0 × 10−16 cm3
s−1 (∼3 × 105 M−1 s−1). Under forward bias, the data were more complex. A recombination mechanism that incorporates a pre-
equilibrium diffusional encounter between injected electrons and oxidized sensitizers is proposed. This and previously reported
data indicate that diffusion limits recombination when the number of TiO2(e−)s is small and electron transfer becomes more
dominant when the number is large.

Recombination of electrons injected in mesoporous nano-
is a dearth of experimental support for this assumption.
Chemical intuition would suggest that the order of the electron
crystalline TiO2 thin films with oxidized sensitizer molecules should also be 1. In practice the experimental data rarely report
has been of theoretical,1−3 experimental,4−9 and practical10 directly on the reaction order, yet values between 2 and 3 have
interest for some time. The reaction is of practical importance been published.17
as it can lower the efficiency of dye sensitized solar cells, In fluid solution, reaction orders are typically determined by
particularly near the power point and open circuit con- quantifying rates when one or more of the reactants are present
ditions.11,12 The reaction is also of intellectual interest as the in large excess such that kinetic isolation conditions apply.18
injected electrons and oxidized sensitizers do not undergo This phenomenally successful experimental technique was
significant geminate recombination. Instead, the charges diffuse originally termed “Ostwald isolation” but is more often referred
through the three-dimensional volume of the mesoporous TiO2 to simply as the “isolation method” or “flooding”.18,19
network or the quasi-two-dimensional molecular layer of the Regardless of the name, the approach is not so much a
sensitizer before interfacial charge recombination occurs. A mathematical approximation as it is an experimental condition
proper mechanistic analysis of kinetic data could in principle that must be identified so that the reaction order (and rate
directly provide rate constants for the fundamental interfacial constants) can be determined. In fluid solution, these
electron transfer event and diffusion of the injected electron conditions are well established, and the rule of thumb
and the oxidized sensitizer. The most basic reaction that can be commonly prescribed is a 10-fold stoichiometric excess of all
written for charge recombination is given in eq 1 reactants but one.18 One purpose of this study was to
determine whether Ostwald isolation could be applied to
TiO2 (e−)|S+ → TiO2 |S (1) sensitized-semiconductor interfaces where “concentration” is
inherently ill-defined, yet intrinsic rate constants for the
The rate law for the reaction is expressed in eq 2, where n and elementary steps of this recombination reaction are of great
m are the order of the reaction in electrons and oxidized importance.

sensitizers, respectively.
rate = k[TiO2 (e−)]n [S+]m (2) EXPERIMENTAL SECTION
Until now, to our knowledge the kinetic order of each Materials. The following substances were used as received
reactant has not been experimentally determined, probably from the provider: titanium(IV) isopropoxide (Aldrich, 97%),
because most, but not all,6,7,13−16 prior studies employed
conditions where the number of oxidized sensitizers and Received: February 20, 2014
injected electrons were equal. The order of the oxidized Revised: March 15, 2014
sensitizer has quite reasonably been assumed to be 1, but there Published: March 18, 2014

© 2014 American Chemical Society 7886 | J. Phys. Chem. C 2014, 118, 7886−7893

Experiments were anatase TiO2 particles deposited on fluorine doped tin oxide conducted with a BAS CV-50W potentiostat and 0. The typical TiO2|S to yield TiO2|S+. The appropriate Sensitization of TiO2. the absorbance at S (black) and TiO2|S+ (red) at each potential with an overlaid 800 nm was used to measure the electron concentration in the sigmoidal fit to the modified Nernst eq 3 from which an E0(TiO2|S+/0) TiO2 film using a TiO2(e−) molar absorptivity of 1000 M−1 = 1630 mV versus NHE and an ideality factor i = 1. C 2014. Phys. Chem.2′-bipyridine and dcb = [2. oriented perpendicular to the probe beam. mesoporous thin films of ∼20 nm diameter Spectroelectrochemical Measurements. Inc.doi. cm−1. The sample was positioned such that the probe and producing FTO|TiO2|Ru(bpy)2(dcb). ■ Cary 50 spectrophotometer at ambient temperature under an argon atmosphere. Nanocrystalline. a 1:1 tert-butanol/CH3CN solution of [Ru(bpy)2(dcb)][ClO4]2 and the excitation fluence was measured with a thermopile (where bpy = 2. the steps began at 100 mV and gradually decreased to 10 mV near the reduction potential measured in the CV. The excitation beam was a Q-switched Nd:YAG mesoporous thin films were stored in an oven at ∼75 °C until (Quantel USA (BigSky) Brilliant B) pulsed at 1 Hz and use. measured at open circuit and at the indicated each potential.5 ns 450 °C under a flow of O2 at ∼1 atm. All electronic absorption (generally ∼10 min). The instrument response time was dicarboxylic acid) for several days to anchor the sensitizer ∼10 ns. using clear cellophane Spex monochromator that was optically coupled to a R928 tape as a spacer (∼10 μm thick). TiO2. The probe beam was incident at a 45° angle RESULTS AND DISCUSSION on the glass side of the FTO|TiO2 sample. The experiment then proceeded by monitoring the absorbance of TiO2|Ru(bpy)2(dcb) with increasingly negative or positive applied potential. Company. a minimum of 3 min in the negative direction. Inset: The fractional abundance. and Ag/AgCl reference electrode. and 5 min in the positive direction.. as previously described. The absorbance of TiO2(e−) bisphenol A epichlorohydrin copolymer 15 000−20 000 Da was converted to concentration in cm−3 by calculating the (Sigma). Spectroelectro- electrode was calibrated with the ferrocenium/ferrocene couple chemistry was employed to determine the concentrations of in 0. a reaction known to proceed by lateral magnitude of the peak current density was 150 μA/cm2 or more. This data was fit to eq 3 in Origin 8. polyethyleneglycol geometric surface area using eq 4. nitric acid (Macron Fine Chemicals. potential step. ACS Reagent volume of the film using the modified Beer−Lambert law (eq grade). 99. Each potential was held until the absorbance was stable.3 M LiClO4 CH3CN before and after each experiment. No absorbance potential.99%). excitation beams were both incident at a 45° angle on the glass in. a potential of ∼0 mV versus NHE was applied until the absorbance spectrum returned to match the initial spectrum. The TiO2 electrode was immersed in nonlinear optics to produce 532 nm excitation were employed. The reference studies and is abbreviated as TiO2|S within. The absorbance of the film at 800 nm was related to the lithium perchlorate (Aldrich. and a film thickness of 10 μm. fluorine-doped tin oxide coated glass (Hartford Glass 4). The electrode was then rinsed with acetonitrile and side of the sample. while in the case of positive potential. 2. In the case of negative applied potential. Nanosecond Transient Absorption Spectroscopy. spectrophotometric grade). increased positive applied potential evinces the oxidation of ment. TiO2|S and TiO2|S+ in 300 mM LiClO4 CH3CN at the positive The measured potentials were converted to NHE using the applied potentials indicated in Figure 1. of TiO2| In the case of negative applied potentials.1021/jp501814r | J. This Figure 1. In the case of positive applied potentials. The bleach of the previously reported21 potential of ferrocene in CH3CN (+630 metal-to-ligand charge-transfer (MLCT) absorption of S with mV versus NHE).org/10.4′.5−2 h before use. a CV was taken at a scan rate of 100 mV/s. power meter (Molectron). Measure- Titanium Dioxide Thin Film Electrode Preparation.The Journal of Physical Chemistry C Article acetonitrile (Burdick and Jackson. positive applied potentials. and becoming larger as the potential exceeded the measured reduction potential. χ. The signal was recorded by an min. the absorbance spectra of the Ru(II) species and the Ru(III) species were used to perform a global fit of the absorbance data at each potential in Wolfram Mathematica 9 using a Least Squares analysis. a platinum disk counter sensitized thin film has been the subject of many previous electrode. (Applied Photophysics) pulsed at 1 Hz served as the probe The sol was doctor bladed onto glass slides with a conductive beam was directed through a long-pass 385 nm filter toward a layer of fluorine-doped tin oxide (FTO). 15 Ω/□. Prior to spectroelectrochemical measure.20 This TiO2 serving as the working electrode. only after which transient absorbance measurements were recorded in a quartz cuvette on a Varian measurements were recorded. a porosity of 50%. The absorbance of a [Ru(bpy)2(dcb)]2+ sensitized TiO2 thin operation yielded fractional abundance of Ru(II) and Ru(III) at film. 118. For measurements made with an applied immersed in acetonitrile for 0. a 3-electrode cell (vide supra) was used. After drying in the dark for 30 Hammatsu photomultiplier tube. abbreviated TiO2|S here.22 A 150 W xenon arc lamp isopropoxide as previously described20 and stored in the dark. abbreviated TiO2|S. ments were made under argon at ambient temperature in a anatase nanocrystallites were prepared from titanium(IV) previously described apparatus. After measurement in each direction. At each change was observed during immersion. 7887 dx. the current was monitored until it stabilized Spectroscopic Measurements. 7886−7893 . the cellophane was removed and the films were sintered at optically triggered LeCroy 9450 oscilloscope with 2. The resulting resolution.2′-bipyridine]-4. the potential was stepped by ∼100 mV.76 were abstracted.3 mm thick).3 M LiClO4 coated glass (FTO) were sensitized to visible light by CH3CN electrolyte in a standard 3-electrode cell with the FTO| [Ru(bpy)2(dcb)][ClO4]2.

These fits yielded E0 = 1. Excess S+ collected from a sample with surface coverage Γ = 8.00 V versus Fc+/0 (approximately 1. As the number Figure 2A shows the absorption change monitored at 402 nm of TiO2(e−)s present in the dark at this positive applied after pulsed 532 nm laser excitation of TiO2|S with a reverse potential was expected to be negligibly small.The Journal of Physical Chemistry C Article Figure 2. potentials as determined by standard addition of the known Overlaid on this data are fits to the Kohlrausch−Williams− absorption spectra of TiO2|S and TiO2|S+. This summarizes the surface coverages. was investigated by nanosecond transient absorption experi. TiO2(e−) concentrations. Γ. (A) Absorbance changes of TiO2|S with an applied potential of 1650 mV monitored at 402 nm after 532 nm pulsed laser excitation. and the observed rate constants measured at each excitation ments where an incident laser pulse was used to generate fluence for a potentiostatically controlled sensitized thin film. In an attempt to achieve the flooding conditions necessary for Ostwald isolation. (B) The observed rate constants of recombination versus initial concentration of TiO2(e−) in units of e−/cm3.47 × 10−8 mol/cm2. Surface Coverages. the number bias of +1650 mV and varied excitation fluence. hole hopping from sensitizers anchored to the FTO ground-excited state isosbestic point that enables character- substrate. with β fixed at 0. Table 1 A(λ) = Γ·ε(λ) ·1000 (4) −1 −1 Table 1. where E is the applied potential. The surface coverage of the sensitizer in mol/cm2.1021/jp501814r | J. Overlaid on this data Watts function26−28 (KWW. and i is the ideality factor25 ΔA = A 0e(−k 0·t ) (6) 1 The first moment of this function was taken as an “average” 21 χ (E ) = 0 1 + 10 E − E / i·59 (3) observed rate constant. of complications from the excited state.76. χ.24 ization of the charge recombination reaction without The inset of Figure 1 shows the fractional abundance. The data in Figure 1 were Constant. NHE)22 for the TiO2|S+/0 reduction with an ideality factor fitting procedure as the raw experimental data were super- i = 1. and where ε is the molar absorptivity in M cm . TiO2(e−)s and S+ in equal numbers far smaller than that At this applied potential. kobs.20 This point was TiO2|S (black) and TiO2|S+ (red). this is likely an artifact of the V vs.20 are overlaid in gray. Inset: the same data normalized by dividing by the absolute value of the minimum. approximately 60% of the sensitizers present before laser excitation. TiO2(e−) Concentrations. The present after light excitation was taken to be equal to the monitoring wavelength of 402 nm was chosen because it is a number of photo-oxidized sensitizers measured spectroscopi- 7888 dx. It was thus necessary to show experimentally that kobs = k[S+]m was indeed a constant.23. an applied potential was used to maintain a high surface coverage of S+ such that the [S+] ≫ [TiO2(e−)] and the rate law of eq 2 could be reduced to that given in eq 5 rate = kobs[TiO2 (e−)]n (5) + m where kobs = k[S ] .doi. Fits to the KWW function with β = 0. posable (Figure 2A inset). 7886−7893 . eq 6) are fits to eq 3. all excitation fluences utilized (∼3−7 mJ/pulse). Phys. Knowledge of χ(E) and Γ enabled determination of the TiO2|S and the TiO2|S+ surface coverage at any desired potential. C 2014. E0 is the β reduction potential.06) × 10−8 mol/cm2 at controlled by the laser fluence. Chem. Figure 2B shows plots of kobs versus the initial concentration of TiO2(e−). was It is instructional to consider the “concentrations” that were determined with the modified Beer’s law relation given in eq 4 used to ensure Ostwald isolation conditions. such that the number of S+ was were oxidized corresponding to S+ surface coverages that essentially constant. 118. (±0. and A is the absorbance. λ is the Kinetic Parameters for the Pseudo First-Order Condition of wavelength.63 While some scatter does exist. The initial number of TiO2(e−)s was remained relatively constant at 5.20. The arrow indicates increasing excitation fluence (∼3−7 mJ/pulse). present at the indicated especially relevant under forward bias conditions (vide infra).

10 45 5.2 (±0. 10 μm thickness. 7886−7893 .7) × 1018 cm−3 (see Table with β = 0.1 solution. semiconductor interface.2 fold change in the initial TiO2(e−) concentration and the number represented 2−4% of the 1610 0. such that the dark TiO2(e−) concentration was in large excess The excess of S+ was potentiostatically varied to ascertain the and essentially constant and eq 2 reduces to rate = k′obs[S+]m dependence of kobs on [S+].59 5. the kinetic data was only influenced by the M−1s−1. 118.1) × 1012 Γ−1s−1.54 4. albeit with more scatter (see Supporting controlled electrode were repeated except with a forward bias Information).2 × 1018 cm−3 Table 2 displays the observed rate constants. the dark TiO2(e−) concentration was 31. (B) The observed rate constants of charge recombination plotted against initial surface coverage of S+. Plots of still indicate a ∼10 fold or greater excess of S+. fractional at this potential. satisfying the <10% requirement 1620 0. This is conditions. ergo the order of mechanism at lower surface coverage.22) versus ΓS+ with an overlaid linear fit that yielded 3). Fractional Abundance of S+. 10 mol/cm2 where ΓS+ refers to the total surface coverage of S+. Chem. which implies that the 3B are nonzero. the lifetime as a function of initial [TiO2(e−)] lead to the same The same transient absorption studies of a potentiostatically conclusion. electrons under conditions where the oxidized sensitizer concentration remained essentially constant and in large excess is telling. and Observed Rate Constants for Simple division yielded the concentration of injected electrons the Pseudo First-Order Condition of Varied.11 37 It shows that kobs = k[S+]m is indeed a constant and a second-order rate constant of k = 1.1 mJ/pulse.1021/jp501814r | J. b 4 −1 10 s . Figure 4A absorbance of TiO2|S at 402 nm at the indicated applied shows an absorbance change that corresponds to interfacial potentials after pulsed 532 nm laser excitation.37 0. With the Ostwald isolation condition with some critical assumptions. Surface Coverages of S+. thus Ostwald isolation has been achieved at a sensitized This value could also be expressed as 5.22 0. The initial number of TiO2(e−)s was 20−100 abundance and surface coverages of S+ corresponding to the fold larger than the initial number of S+ created by the laser data presented in Figure 3. C 2014.79 0.0 1640 0. On the basis of spectroelectrochemical measure- baseline more rapidly as the potential (and ΓS+) increased. pulse while the concentration of TiO2(e−)s varied only slightly Figure 3B shows kobs (abstracted from the KWW parameters with excitation fluence at 32. Figure 3A shows the change in where k′obs is a constant equal to k[TiO2(e−)]n.8 (±0.38 3.18 The number of S+ was thus determined to be 20−60 1650 0.15 22 2. measured geometrical area. in the more familiar units of satisfied. It was not possible to TiO2(e−) must be 1 (see Supporting Information for investigate this latter possibility while also maintaining isolation expressions of lifetime for various reaction orders).3 number of oxidized sensitizers. Only systematic error in determination of the oxidized sensitizer first-order kinetics permit constancy of the lifetime with surface coverage or a change in the charge recombination changing initial reactant concentration. ∼ 3 × 105 M−1s−1 (Supporting Information). the resulting ΓS+ values would the error and assumptions introduced by kinetic fitting. If it were assumed that the deviation from zero was particularly compelling evidence because it is independent of solely due to systematic error. Applied Potential Eapp. Overlaid on the charge recombination measured after pulsed 532 nm laser kinetic data are fits to the KWW function. The pulsed laser fluences employed Eapp χ (S+) Γ’S+a Γ’S+a S+:TiO2(e−) kobsb resulted in a 2. Even with changing the initial be noted that the intercepts of the linear fit of the data in Figure [TiO2(e−)] the data were superposable. cally. 7889 dx.21 0. The inset shows the excitation of a sensitized thin film held at a forward bias of normalized data. The overlaid linear fit yielded a second order rate constant of 1.doi.8 fold larger than the number of injected electrons in the range of a −8 excitation fluences utilized.1) × 1012 cm2 mole−1 s−1. Excess S+ in units of electrons/cm3. Fits to the KWW function are overlaid. Possible explanations for this include a lifetime was constant with respect to initial [TiO2(e−)].0 × 10−16 cm3 s−1 or. Excitation fluence was 3. It should concentration of TiO2(e−).13 29 3.The Journal of Physical Chemistry C Article Figure 3.48 4.6 The TiO2 volume was determined based on the film’s Table 2. (A) The absorption change after pulsed 532 nm laser excitation of TiO2|S monitored at 402 nm with the indicated applied potentials.68 0.8 (±0.19 27 5. The independence of kobs on the number of injected while Γ′S+ refers to the contribution of the photo-oxidized sensitizers. ments. that were not superposable and returned to −400 mV. and 50% porosity.5 often asserted as necessary for the isolation method in fluid 1630 0.43 3. Inset: The same data normalized by dividing by the absolute value of the minimum. Phys.

Surface Coverages. Figure 5A shows the absorbance change measured at the indicated negative applied potentials after pulsed 532 nm laser excitation of TiO2|S.1021/jp501814r | J. Observed Rate Constants for the Pseudo First-Order these data imply that the order of the recombination reaction Condition of Constant. The arrow indicates increasing excitation fluence (∼2−7 mJ/pulse). 7886−7893 . Another explanation for the superlinear normalized (Figure 4A inset).The Journal of Physical Chemistry C Article Figure 4.20 for all the kinetic data. 7890 Fits to the KWW function with β = 0.35 are overlaid in gray. Phys. (A) The absorption change after pulsed 532 nm laser excitation of TiO2|S monitored at 402 nm with the indicated applied potentials. Plausible explanations for the nonlinear dependence of k′obs on the [TiO2(e−)] include the very real possibility that Ostwald isolation was not achieved under all the reverse bias conditions reported in Figure 5 (see Table 4).20 are overlaid. TiO2(e−) Concentrations. Inset: The same data normalized by dividing by the absolute value of the minimum. (B) The k′obs of charge recombination plotted on a log scale against initial concentration of TiO2|(e−). and presented for the complementary data in Figure 2 (vide supra). See Supporting Information for plots of lifetime as a function of initial surface coverage of S+. in accord with the results presented in Figure 3B. Table 3. Following the same arguments increase in k′obs is that electron transfer contributes more to Figure 5. Figure 4B shows that observed rate constants abstracted from Demonstration of isolation over a larger potential range was the KWW fit parameters (eq 7 with β = 0. as can be seen in the log−linear plot in Figure 5B.35) were frustrated by sensitizer desorption and/or irreversible chemistry independent of ΓS+. However this data did not produce a linear relationship with k′obs. 118. The kinetic data were noticeably noisier that occurred particularly at applied potentials more negative at lower fluences but were otherwise superposable when than −400 mV. Excess [TiO2(e−)] with respect to S+ was 1. The inset shows that the normalized data were nonsuperposable with more negative potentials inducing a more rapid recombination with the oxidized sensitizer.1 mJ/pulse. Excitation fluence was 3. Overlaid are fits to the KWW model with common β equal to 0. monitored at 402 nm after 532 nm pulsed laser excitation.doi. Chem. (B) The average observed recombination rate constants versus the initial S+ surface coverage. (A) Absorbance changes of TiO2(e−)|S at −400 mV. Inset: the same data normalized by dividing by the absolute value of the minimum. C 2014. Fits to the KWW function with β = 0.

k‑D. Indeed.9 −400 32.35.The Journal of Physical Chemistry C Article Table 4.30 and transfer rate constant. the second-order rate constant for recombina- tion would be 4.33 Demonstration that charge recombination has a first-order This interpretation would correspond to the kinetic limit where dependence on the concentration of both the oxidized diffusion was much slower than electron transfer such that kobs sensitizers and the TiO2(e−)s was an important outcome of = Kket. artifacts can be eliminated. 7 −100 5. but the nonlinear dependence of k′obs precluded Kket. electron injection and is the result of more rapid ambipolar diffusion that also occurs thermalization of the oxidized sensitizer and the injected at higher excitation fluences as was revealed previously by electron are expected to be complete. and Scheme 1.doi.15 10 15. the observed rate constants oxidized sensitizer. The reaction rate also increased dramatically.7 0.7. C 2014.5 −300 17.18 5 2. recombination provides a means by which such experimental A steady-state approximation yields eq 7 where K = kD/k‑D.1 0.8 Such behavior corresponds to the kinetic recombination rate clearly increased with [TiO2(e−)] (see limit where the reaction is controlled by diffusion. this study. physical location5. Recombination of TiO2(e−) and S+ Excess [TiO2(e−)] Eapp [TiO2(e−)]a Γ’S+b TiO2(e−):S+ k′obsc 100 3.1021/jp501814r | J.3 0 3. This finding was enabled by steady-state and The data reported here suggest that the intermediate case.6 0. It is worthwhile to briefly review previously reported the relative weight of these two contributors can be controlled kinetic data for charge recombination within the framework of a experimentally.13 It is probable that this acceleration On the nanosecond time scale. will similarly increase overlap is highest. The encounter occurs through are expected to more closely reflect the true interfacial electron lateral hole hopping by the oxidized sensitizer29. If the rate constant for diffusion away from the “dark” TiO2(e−) concentration and average observed rate this close contact.1 −200 9. may be the general case of the oxidized sensitizers or the TiO2(e−) were systematically when the number of injected electrons is small. a potential where Ostwald isolation was clearly demonstrated. ket. In some cases.32 Although absorption experiments. the recombination mechanism as the potential decreases due to the increased mobility of the electrons in TiO2 (vide infra).org/10. Scheme 1.31. such as an intense probe beam or high electron transfer will occur over a range of distances. 118. However. first reported by Durrant. recombination classical time-of-flight and intensity modulated photocurrent is proposed to occur by a mechanism that incorporates a pre.33 and/or the determined in the constant reverse bias condition. these results do not necessarily conflict with the conclusion of a the observed rate constants were directly related to electron first-order dependency. like in laser repetition rates (that do not allow complete recombina- fluid solution. is comparable to the electron transfer constants abstracted from such data will be larger. kobs. Furthermore. Phys. sensitized TiO2 interfaces remains unclear as to which For this to be realized. Further studies with a larger number 1 1 1 of sensitized materials may one day directly provide the = + critically needed rate constants for the fundamental interfacial kobs kD Kket (7) electron transfer event and diffusion of the injected electron Interestingly. Therefore.1 0.34 of the molecular sensitizer. contains here that Ostwald isolation can be utilized to quantify charge contributions from both diffusional and electron transfer steps. kD ≤ Table 4). the vast literature on charge recombination at and the oxidized sensitizer.19 2 2. TiO2(e−) Concentrations.8 0. early diffusion of TiO2 electrons to the oxidized dye molecules was charge recombination studies of dye sensitized TiO2 revealed 7891 dx.08 33 139 a 1018 cm−3. that is. transient absorption experiments of a potentiostatically where the observed rate constant is comparable to diffusion and controlled sensitized TiO2 thin film where the concentrations the electron transfer rate constants. it is most favored at close contact where orbital tion before the subsequent laser pulses). the observed rate constant. Surface Coverages.23 1 0. other investigators found that charge recombi- corroboration of the kinetic order of n = 1 for TiO2(e−) as nation was sensitive to the redox potential5.9 0.21 1 0. Chem.3 × 10−14 cm3 s−1 (or ∼3 × 107 M−1 s−1).36 Hence as equilibrium encounter between the injected electron and the the number of TiO2(e−)s increases. The most important. It should be emphasized that this kinetic data was not easily interpreted yet is consistent with the reports of a power-law dependence7 (see Supporting Information). The finding rate constant.32 Experimental variables in transient ambipolar diffusion of the injected electron. c104 s−1. b10−8 mol/cm2. However. studied. However. Proposed Mechanism of Interfacial Charge Observed Rate Constants for the Condition of Varied. an additional issue is that the kinetics mechanistic step(s) the observed “average” recombination for charge recombination display complex kinetics that are not rate constants correspond to. studies of operational dye sensitized solar cells. transfer rate constants analyzed in terms of Marcus theory. if pseudo first-order behavior was assumed for the data measured at −400 mV. Kinetic isolation measurements at more simplified mechanism that links experimentally measured negative potentials clearly revealed a first-order dependence observed rate constants to the elementary steps in the on [S+]. Early studies concluded that well described by first-order kinetic models. as was recombination reaction. 7886−7893 .

(16) Kamat. D. C. 24196−24205. Am. K. J.. M. Determination of Sensitizer straightforward provided that the sensitizer is stable in its one. 8352−8355. Patterson. CONCLUSIONS B 1999. J. Science 2011. ■ (13) Haque. and 1 for S+ under conditions where the number 629−634. P. Graetzel. McGraw-Hill: New York. 25.. Lutz. Chem. Chem.. R. H. J. details of data analysis and unit (14) Hu. 7886−7893 . O’Regan. Am. Phys. Moser. 118. G. Berlinguette.. Molecular Control of in equal numbers. Regeneration Efficiency in Dye-Sensitized Solar Cells. Mater. S... ■ assigned to recombination. Since the derivation of the KWW function by Scher (3) Nelson. Modu- *E-mail: meyer@jhu. Tachibana. M..24 (2) Bisquert. Experimental methods. 117. Soc... 2012. J. J. C. TiO2(e−) in the forward bias condition was not verified. J.. tion of the full recombination reaction often required a sum of (6) Haque..20 This model is consistent with the report Recombination Dynamics in Dye-Sensitized Nanocrystalline TiO2 here if the dark concentrations of injected electrons and Films: Free Energy vs. We acknowledge Ke Hu for his help with data with multiexponential kinetics. K. J. J. Hopping Transport of Electrons in Dye-Sensitized with the first moment often taken as an “average” observed rate Solar Cells. Moser.. H. J.. Grätzel. Durrant. R. G.. X. 59.. The authors declare no competing financial interest. C 2014... Heterogeneity may ultimately be Dioxide Films. Phys. K. Semiconductor-Based Interfacial Electron- is rate limited by transport/diffusion of the charges. E. Chem. D. Berlinguette. The Solar Cell. Chem. J. of TiO2(e−) was high (forward bias). Phys. Phys. Chem. This Interfaces. Grätzel. 7. W. 100.. fit to this function in itself does not indicate that recombination (4) Yan. ACS Nano electron oxidized form. Phys. J. Li. ACKNOWLEDGMENTS T. J. 334. Durrant. (5) Clifford. J. More Transfer Reactivity: Decoupling Kinetics from pH-Dependent Band recently. B. Lee. L. M. the underlying cause of this behavior and identification of the (7) Nelson.. 1745−1749. 538−547.. Chem. A. B 1998. C. Tsao. R. 2013. Q. J. Electron Dynamics in Nanocrystalline ZnO and TiO2 between surface anchored oxidized dye molecules and electrons Films Probed by Potential Step Chronoamperometry and Transient injected into mesoporous nanocrystalline TiO2 thin films. Trap-Limited origin(s) of this heterogeneity represent important goals for Recombination in Dye-Sensitized Nanocrystalline Metal Oxide future research. Hupp. Rev.. application of this approach to other sensitized materials is (12) Li. 17163−17168. The order of the (11) Robson. Tachibana. J. 205321.. H. R.28 it has become a paradigm for modeling charge Transport in Nanocrystalline TiO2 Electrodes.. lation of Electron Transfer between Excited Ruthenium Complex and Notes SnO2 Nanocrystallites with an Externally Applied Bias. Intramolecular Hole Transfer at Sensitized TiO2 orders in terms of initial reactant concentration and k. G. Hotchandani. material is available free of charge via the Internet at http:// (15) O’Regan. 1990. J. J. N. N. 7671−7675. and expressions for lifetimes of several reaction Meyer. 100. T. R. A.. Y. J. Yi. 7605−7613. 106. Chem. K.. R. D. Phys. G. I. 94. Durrant. J.38 A kinetic model widely (1) Maggio. Diau.. 1996. B 2000. J. Hu. K. 8720−8726.. S. Continuous-Time Random-Walk Model of Electron and Montroll. C. G. 135. J. 2002. Meyer. Phys. of Chemical Sciences. Haque. a statistically good 15374−15380. T.. J. S. D. Meyer. Robson. Distance Dependence. E. B 1999. Klug. 8233−8242. 2013. B 2002.. Chemical Kinetics and Reaction Mechanisms. a second-order equal concentration kinetic model has Energetics in a Dye-Sensitized Titanium Dioxide/Aqueous Solution been used that yielded rate constants that were independent of System. 1996.. E. C 2013. J. Jennings. K. Wang.. Johansson. Electron Injection into Transparent Semiconductor Membranes and ■ Electric Field Effects on the Dynamics of Light-Induced Charge AUTHOR INFORMATION Separation. Anderson. transport in heterogeneous media. Soc. Parameters Influencing Charge distributions of rate constants to model charge recombination Recombination Kinetics in Dye-Sensitized Nanocrystalline Titanium data is not well understood. ■ (8) Hasselmann. Phys. C 2007.7. 6867−6870.. A. P. Gratzel. A.37 The data reported here and elsewhere20 demonstrate that such modeling is of very limited REFERENCES value unless the concentration dependence is explicitly accounted for. Klug. J. Nazeeruddin.. Willis. R. Phys. X.. M. Office of (18) Espenson. Porphyrin-Sensitized Solar Cells with Cobalt (II/III)− under conditions where the number of TiO2(e−) was low Based Redox Electrolyte Exceed 12% Efficiency..-G. J. S. However. 1881− E. A.-W..acs. Klug. Charge ASSOCIATED CONTENT Recombination Kinetics in Dye-Sensitized Nanocrystalline Titanium Dioxide Films under Externally Applied Bias.. R. A.The Journal of Physical Chemistry C Article that recombination occurred on the 10−9 to 10−2 s time scale 96ER14662). Troisi. P. Adv. Yeh. F. Office of Basic Energy Sciences. mechanistic tool for kinetic analysis of charge recombination Durrant. 111. based on a skewed Lèvy distribution of rate constants3. Energy Research.. Am. G. two second-order rate constants.. the initial concentration of TiO2(e−) and S+ always generated Nelson.S. M. 1961−1971. M. S. K. 63. J. R. Palomares... ■ (17) Barnes.. J. (reverse bias). However. Interpretation of Optoelectronic Transient and Charge Extraction Measurements in Dye-Sensitized The authors acknowledge support by a grant from the Division Solar Cells. Bessho. C. R. L. U. Chem. Nelson... Phys. Y. constant. and the slow component was analysis and experimental methods. P. quantifica.-Y. Y. C. Olson. Phys. Ostwald isolation has been shown to be applicable as a (9) Willis. established to be 1 in both oxidized sensitizer and TiO2(e−) Nazeeruddin.. L. Department of Energy (DE-FG02. J. Long. O’Regan. K. Diffusion-Limited Interfacial Electron Transfer with Large Apparent Driving Forces.. Chem.. 5225−5233. R. V. 126. order of the recombination reaction TiO2(e−)|S+→ TiO2|S was (10) Yella. D... oxidized sensitizers are negligibly small. Soc. Li. Miettunen. M. Durrant. likely Atomic Level Resolution of Dye Regeneration in the Dye-Sensitized as a result of concentration dependent electron diffusion. B. 104.20 The need for sums or M. Theory of the Charge Recombination Reaction at the Semiconductor−Adsorbate Interface in the Presence of used for charge recombination is the KWW function that is Defects. K. 103. J. 2004. Bedja. Zakeeruddin. Grätzel. J. 7892 dx. M. Phys.. as was recently done. Electrodes. J. * S Supporting Information 102. B 2001. Rev. Vectorial pubs.. J. N.. M.. The Absorption Spectroscopy. 2013.doi. Anderson. conversion. F. Chem.1021/jp501814r | J. 4900−4908. Chandiran. Chem. C. Chem.. 134. Corresponding Author Photosensitization of Nanocrystalline Semiconductor Films. B.6. C. S. R. Chem.

J. Soc. 1970. F. Electron Transport in Porous Nanocrystalline TiO2 Photoelectrochemical Cells. P..-E. H. 118.. 2011. A.. J. W. (34) Johansson. (21) Pavlishchuk. and Interfacial Charge Transfer in the Nanocrystalline TiO2/Electrolyte System by Impedance Methods. S. Phys. Lenzmann. 97−102. 2455−2477.. N. W.. Barbé. C.. 2011. Phys. C. S. 56−82. Kitamura. J. 7886−7893 . Grätzel. Soc.. J. Chem. 15384−15396. J. 91. (38) Knauf. Direct Observation of Photodriven Intermolecular Hole Transfer across TiO2 Nanocrystallites: Lateral Self-Exchange Reactions and Catalyst Oxidation. N. W. Salvador. Phys.1021/jp501814r | J. Vlachopoulos. T. S. M. 104. Kopecky. H. A. J. Nakade. B 1998. Addison. M. Bignozzi. Comte. J. C. Alibabaei.. Wada. Farzad. L. Phys. M. Engelmann: Electronic Structure and Electron Transfer Dynamics at Metal Oxide− Leipzig. Chem. Morris. Y. (27) Williams.. M. (35) Cao... (24) Heimer. Inorg. W. C. Inorg. Phys. Chromophore Interfaces.. J. Meyer. P. Soc. Theorie Des Elektrischen Rückstandes in Der Leidener Flasche.. P. 17021−17027. (Leipzig) 1854. 9283−9285. Electron Transfer Dynamics in Nanocrystalline Titanium Dioxide Solar Cells Sensitized with Ruthenium or Osmium Polypyridyl Complexes. P. V. Ann. G. Brennaman. 15. T.. Chem. Watts. A. Chem. J. Long-Lived Photoinduced Charge Separation and Redox-Type Photochromism on Mesoporous Oxide Films Sensitized by Molecular Dyads.. F. N. Revealing the Relationship between Semiconductor 7893 dx. 334−339. Winkler. N. 1498− 1507. Meyer. (30) Ardo. Lewis. J. Heimer. J. 1324−1336. 33. Zaban. F. Chem. J.. A. S. 132.. 135. (36) Fabregat-Santiago. G. Charge Storage. S. Phys. (33) Kuciauskas.. R. Faraday Soc. 7047−7054.. Grätzel. J. M.. (25) Ardo. B 1975. 66.. D’Arcangelis. Phys. S. J. 5741− 5749.. Chim. (31) Kopidakis. T.. 2010. J. A. Am. S. Chem. J... Phys.. Decoupling of Transport. G.2. Meyer. 8331−8341. G. J. (26) Kohlrausch.... S. Vol. Frank. E. 106. 106. L. Influence of the Electrolytes on Electron Transport in Mesoporous TiO2−Electrolyte Systems. M. Walder. Stipkala. 1994. Lehrbuch der allgemeinen chemie.. T. R. (29) Ardo. (28) Scher. Chem. Anomalous Transit-Time Dispersion in Amorphous Solids. Farzad. Abrahamsson.. F. Chem.. Characterization of Photoinduced Self- Exchange Reactions at Molecule−Semiconductor Interfaces by Transient Polarization Spectroscopy: Lateral Intermolecular Energy and Hole Transfer across Sensitized TiO2 Thin Films. G. 2. Chem. Meyer. B 2000. Rev. 133. 100. A. J. J. R. Thompson. M. C. Meyer. W. E. J. J. G. S.. 2967−2972. 25259−25268. V. 1902. Zakeeruddin. Am. Chem. Meyer. Soc. 2013. 1996. Garcia-Belmonte.. Soc.. Chem. (22) Argazzi. 1999. Bisquert. 121.. G.... Phys. R.. Am. Non-Symmetrical Dielectric Relaxation Behaviour Arising from a Simple Empirical Decay Function. P. Inorg. Nanoporous TiO2. Meyer. 80−85. Searson. Chem.. J. Montroll.. Stipkala. D. F. Conversion Constants for Redox Potentials Measured versus Different Reference Electrodes in Acetonitrile Solutions at 25°C. (23) Bonhôte. Distance Dependent Electron Transfer at TiO2 Interfaces Sensitized with Phenylene Ethynylene Bridged Ru(II)−Isothiocyanate Com- pounds... G. Park. M. G. Efficient Lateral Electron Transport inside a Monolayer of Aromatic Amines Anchored on Nanocrystalline Metal Oxide Films.. K. R. 117. Chem. D. J. Galoppini. C 2013. J. 105.. Acta 2000. 392−403.. B 2002.. Norris. A.. Chem. A. J. B. Ambipolar Diffusion of Photocarriers in Electrolyte-Filled. C 2014. S. Meyer. Phys. Langmuir 1999. 35. J. G. 1996. J.doi. A. Yanagida. (37) Bonhôte. N. J. P.. L. 12.. 5319−5324. R. 3930−3936. 102. An Acetylacetonate-Based Semiconductor−Sensitizer Linkage. Castellano. 133. B 2001. Vlachopoulos. F. G.. (32) Kambe. Am. 298. Humphry-Baker. B 2002. Chem. (20) Kelly. S. J. van de Lagemaat. Freund. Gray. D.The Journal of Physical Chemistry C Article (19) Ostwald. G. Enhanced Spectral Sensitivity from Ruthenium(II) Polypyridyl Based Photovoltaic Devices. M. R. Am. Oskam. Trans.. N..-G.. E. Chem. Schiff.. Non-Nernstian Two-Electron Transfer Photocatalysis at Metal- loporphyrin−TiO2 Interfaces. D.. A. 16572− 16580. Dempsey. Tingry. E... Gogniat.. Achey. Cation-Controlled Interfacial Charge Injection in Sensitized Nanocrystalline TiO2.