CHEMPHYSCHEM ARTICLES DOI: 10.1002/cphc.201301076

Investigation of Dye Regeneration Kinetics in Sensitized Solar Cells by Scanning Electrochemical Microscopy Bingyan Zhang, Xiaobao Xu, Xiaofan Zhang, Dekang Huang, Shaohui Li, Yibo Zhang, Fang Zhan, Mingzhang Deng, Yahui He, Wei Chen, Yan Shen,* and Mingkui Wang*[a] Dedicated to Professor Michael Grtzel on the occasion of his 70th birthday

Sensitizers are responsible for the light harvesting and the charge injection in dye-sensitized solar cells (DSSCs). A fast dye-regeneration process is necessary to obtain highly efficient DSSC devices. Herein, dye-regeneration rates of two DSSC device types, that is, the reduction of immediately formed photo-oxidized sensitizers (ruthenium complex C106TBA and porphyrin LD14, kox’) by iodide ions (I) and [Co(bpy)3]2 + , and the oxidation of formed photo-reduced sensitizers (organic dye P1, kre’) by triiodide ions (I3) and the disulfide dimer (T2) are investigated by scanning electrochemical microscopy

(SECM). We provide a thorough experimental verification of the feedback mode to compare the kinetics for dye-regeneration by using the above mentioned mediators. The charge recombination at the dye/semiconductor/electrolyte interface is further investigated by SECM. A theoretical model is applied to interpret the current response at the tip under short-circuit conditions, providing important information on factors that govern the dynamics of dye-regeneration onto the dye-sensitized heterojunction.

1. Introduction The requirement of renewable energy sources has stimulated interest in the research of efficient, low-cost photovoltaic devices.[1] Dye-sensitized solar cells (DSSCs) have attracted increasing attention as alternatives to conventional silicon-based photovoltaics. DSSCs make use for the first time of a three-dimensional nanocrystalline junction for solar electricity production, separating the sites of light absorption from charge carrier transportation. Sandwich-structure DSSCs are mainly composed of a mesoporous nanocrystalline network of a wideband-gap semiconductor (typically TiO2), a monolayer of dye molecules (e.g. ruthenium dyes) attached to the semiconductor, a redox electrolyte (typically I3/I), and a platinum counter electrode. Upon light absorption, the photo-excited dye injects an electron into the conduction band of the n-type semiconductor nanocrystals, and the resulting oxidized dye is regenerated by a redox mediator in the surrounding electrolyte.[2, 3] In contrast, for p-type DSSCs, hole injection takes place from the excited sensitizers into the valence band of the p-type semiconductor nanocrystals (photocathode), such as NiO and CuCrO2.[4–9] Recently the development of sensitizers or redox shuttles used in DSSC devices have led to remarkable solar [a] B. Zhang, X. Xu, X. Zhang, D. Huang, S. Li, Y. Zhang, F. Zhan, M. Deng, Y. He, Prof. W. Chen, Prof. Y. Shen, Prof. M. Wang Wuhan National Laboratory for Optoelectronics School of Optoelectronic Science and Engineering Huazhong University of Science and Technology Luoyu Road 1037, Wuhan, 430074 (P.R. China) Fax: (+ 86) 27-86692225 E-mail: [email protected] [email protected] Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cphc.201301076.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

photovoltaic power conversion efficiencies (PCE) in the range of 11–12.3 %, rendering this technique a credible alternative to other thin-film photovoltaic cells.[10] Such state-of-the-art cells can be improved by careful optimization of kinetic efficiencies in each step, including sensitizer regeneration.[11] In an efficient n-type DSSC, the excited-state energy level of the dye molecule should be more negative than that of the conduction-band edge [(vs. the standard hydrogen electrode (NHE)] of the n-type semiconductor (photoanode) to ensure efficient electron injection from dye molecules into the semiconductor; and similarly, the oxidized state energy level of the dye must be more positive than that of the redox electrolyte (vs. NHE) to guarantee efficient regeneration of the dye. Several previous studies have analyzed the effects of electrolytes on dye regeneration that are related to the photovoltaic performance of DSSCs.[12–15] A fast regeneration step can suppress decomposition reactions and the back transfer of electrons from the conduction band of the semiconductor, which is a major recombination route and hence a loss mechanism in DSSCs.[16] Various redox couples have been investigated aiming for a fast dye-regeneration. Up to now, I3/I has been demonstrated to be one of the best systems. However, it is limited by the relatively high over-potential for dye regeneration,[17] complex two-electron redox chemistry,[18] and competitive light absorption by the triiodide.[19, 20] Therefore, efforts to find alternate redox electrolytes have been growing, the results of which are ferrocene,[15] 1-methy-1-H-tetrazole-5-thiolate and its dimer,[21] cobalt tris(2,2-bipyridine),[22, 23] and so forth. Several redox mediators have been examined as alternative redox couples in DSSCs in order to avoid the problems associated with ChemPhysChem 2014, 15, 1182 – 1189

1182

CHEMPHYSCHEM ARTICLES

www.chemphyschem.org

Figure 1. Molecular structure of the sensitizers a) C106TBA, b) LD14, c) P1 and redox shuttles d) Co3 + /Co2 + and e) 1-methy-1-H-tetrazole-5-thiolate (T) and its dimer (T2) used in this study.

the I3/I system,[18, 24, 25] and devices based on these new electrolytes exhibited good energy conversion efficiencies. To name one example, a cobalt complex with negligible absorption in the visible spectral range has been reported, yielding an efficiency of 12.3 % under full sunlight.[10] The kinetics of dye-regeneration at the dye-sensitized nanocrystal/electrolyte heterojunction is a fast interfacial chargetransfer process in the nanosecond to sub-microsecond time domain, which makes it difficult to characterize. Taking iodidebased electrolyte as an example, the I3/I system is a twoelectron redox couple, requiring the reaction to proceed through several intermediate states. In the past years, the dyeregeneration process in individual electrodes or in complete cells has been characterized by photoelectrochemistry methods.[13, 26–28] The kinetics of dye-regeneration reactions can also be studied by nanosecond transient absorption spectroscopy, which provides valuable information for understanding DSSCs.[29] Spiccia et al. used the nanosecond laser transient absorption spectroscopy to measure the kinetics rates for six organic carbazole-based dyes regenerated by nine ferrocene derivatives.[30] The regeneration kinetics rates for Z907 sensitizer at the maximum power point and at open circuit in DSSCs were also studied by combination transient absorption technique with impedance spectroscopy, showing that inefficient regeneration limits the devices’ efficiency.[31] Scanning electrochemical microscopy (SECM) is an effective technique to determine charge-transfer kinetics at various interfaces such as solid/liquid and liquid/liquid interfaces. Recently, the application of SECM to investigate charge-transfer kinetics between I and photo-oxidized dye molecules (Eosin Y + ) adsorbed on ZnO has been firstly reported by Wittstock et al., demonstrating the viability of this method for under 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

standing DSSCs.[16] Compared to the nanosecond laser transient absorption spectroscopy measurement, the SECM is suitable to monitor fast interfacial charge-transfer processes in devices under working condition.[16, 32–34] In this work, we extended this approach to investigate the dye-regeneration processes of three widely used sensitizers, including C106TBA, LD14, and P1 (see Figure 1) which were strained onto semiconductor nanocrystalline mesoporous films (n-type TiO2 and p-type CuCrO2 semiconductors). The measurements were based on monitoring the feedback current, which is related to the very small

Figure 2. Basic arrangement for probing the heterogeneous reaction at the a) n-type dye-sensitized semiconductor (TiO2) and b) p-type dye-sensitized semiconductor (CuCrO2) interface in the feedback mode of SECM under short-circuit conditions. The mediator couple is Co3 + /Co2 + and T2/T , respectively.

ChemPhysChem 2014, 15, 1182 – 1189

1183

CHEMPHYSCHEM ARTICLES change in the concentration of the active species, caused by the regeneration of one species of the redox shuttle under the active area of an ultramicroelectrode probe (see Figure 2). The dye-regeneration kinetics were investigated by SECM: on the photoanode (i.e. TiO2/C106TBA and TiO2/LD14) in the presence of the oxidized species of the redox couple and on the photocathode (CuCrO2/P1) with the reduced species of the redox couple. The aim was to exploit the influence of different redox couples and sensitizers on the dye-regeneration kinetics. The sensitizers as well as the redox couples were chosen explicitly for the purpose of highlighting the limits of dye-regeneration. In addition, we investigated recombination under working conditions by SECM.

2. Results and Discussion In the present work, three redox shuttles, including triiodide and iodide (I3/I), Co(bpy)3(PF6)2 and Co(bpy)3(PF6)3 (Co3+/Co2+), and 1-methy-1-H-tetrazole-5-thiolate and its dimer (T2/T) were used in combination with three sensitizers, namely, the high molar extinction coefficient ruthenium complex C106TBA1, the push–pull porphyrin (LD14), and the organic dye 4-(bis-{4-[5-(2,2-dicyano-vinyl) thiophene-2-yl]phenyl}amino) benzoic acid (P1).[5, 21, 35–39] The molecular structure of the sensitizers and the redox shuttles are given in Figure 1. C106TBA, LD14, P1, and two types of semiconductor nanocrystals (n-type TiO2 and p-type CuCrO2) were used to fabricate the photoanodes (fluorine-doped tin oxide (FTO)/TiO2/C106TBA and FTO/TiO2/LD14) and the photocathode (FTO/CuCrO2/P1). The UV/Visible absorption spectra of C106TBA, LD14, and P1 in acetonitrile solution are presented in Figure 3 a. Three absorption bands were observed in the 300–800 nm regions for C106TBA. The intense absorption band around 300 nm could be attributed to the interligand p–p* charge transition between 4, 4’-dicarboxylic acid-2,2’-bipyridine (dcbpy) and the ancillary bipyridine ligands. Another absorption band in the UV

Figure 3. a) UV/Visible absorption spectra of LD14, C106TBA, and in acetonitrile solution and b) IPCE responses of device A (TiO2 with C106TBA), device C (TiO2 with LD14), device E, and device F (CuCrO2 with P1). The composites of electrolytes used in these devices are shown in Table 1.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemphyschem.org region (348 nm) could be attributed to the p–p* transition within the ancillary bipyridine ligands and the higher-energy metal-to-ligand charge-transfer (MLCT) transitions. The electronic absorption spectra of C106TBA showed the characteristic MLCT absorption band in the visible region centered at 550 nm.[36] The absorption for the LD14 porphyrin showed an S band at 460 nm and Q band at 670 nm, corresponding to the S0 !S2 transition for the higher vibration mode Q(1,0) and the S0 !S1 transition for the lowest energy vibration mode Q(0,0), respectively.[2] The UV/Visible adsorption of P1 sensitizer presented an adsorption peak at 470 nm, which could be attributed to a p–p* transition mixed with charge transfer from the amine to the dicyanovinyl unit.[38] The photovoltaic data obtained with six DSSC devices containing the three different sensitizers and redox shuttles are listed in Table 1. The efficiency of the photo-electrochemical

Table 1. Performance [short-circuit photocurrent densities (Jsc), open-circuit voltage (Voc), fill factor (FF) and power conversion efficiency (PCE)] of the DSSC devices. Device A B C D E F

[a]

TiO2/C106TBA TiO2/C106TBA[b] TiO2/LD14[a] TiO2/LD14[b] CuCrO2/P1[c] CuCrO2/P1[d]

Voc [V]

Jsc [mA cm2]

FF

PCE [%]

0.718 0.75 0.73 0.925 0.141 0.309

19.58 14.6 17.38 15.78 0.8 1.43

0.72 0.7 0.71 0.72 0.43 0.38

10.1[3] 7.67 9.01[2] 10.5 0.05[9] 0.17[9]

[a] The conventional iodide-based electrolyte was used: 1.0 m 1,3-dimethylimidazolium iodide (DMII), 50 mm LiI, 30 mm I2, 0.5 m tert-butylpyridine, and 0.1 m guanidinium thiocyanate in a mixture of acetonitrile and valeronitrile (85:15 v/v). [b] The cobalt-based electrolyte was used: 0.22 m Co(bpy)3 (PF6)2, 0.05 m Co(bpy)3(PF6)3, 0.1 m LiClO4, and 0.8 m 4-tert-butylpyridine in acetonitrile. [c] The iodide-based electrolyte was used: 0.3 m I2 and 1.2 m LiI in a mixture of acetonitrile and propylene carbonate (7:3 v/v) with 0.1 m LiTFSI. [d] The thiolate-based electrolyte was used: 0.3 m T2 and 0.9 m T with the tetramethylammonium cation in a mixture of acetonitrile and propylene carbonate (7:3 v/v) with 0.1 MLiTFSI.

cells was strongly affected by the nature of the redox shuttles and sensitizers. Device A (TiO2/C106TBA with an iodine-based electrolyte) showed a PCE of up to 10.1 %.[3] Device B (TiO2/ C106TBA with a cobalt complex-based electrolyte) showed a PCE of about 7.6 %. Device C (TiO2/LD14 with an iodinebased electrolyte) showed a PCE of about 9 %.[2] Device D (TiO2/LD14 with a cobalt complex-based electrolyte) showed a PCE of about 10.5 %. Device E (CuCrO2/P1 with an iodinebased electrolyte) showed a PCE of about 0.05 %.[9] Device F (CuCrO2/P1 with a thiolate-based electrolyte) showed a PCE of about 0.17 %.[9] Figure 3 b presents the monochromatic incident photon-to-electron conversion efficiency (IPCE) performance of the DSSC devices (A, C, E, and F). Devices C, E, and F showed lower IPCE values compared with device A. This result is in agreement with their absorption characteristics. According to the results obtained by UV/Visible spectroscopy and IPCE measurements on these sensitizers, light-emitting diodes (LEDs) that emit an energy centered on specific waveChemPhysChem 2014, 15, 1182 – 1189

1184

CHEMPHYSCHEM ARTICLES lengths were selected to back-illuminate the dye-sensitized nanocrystal mesoporous films (TiO2/C106TBA, TiO2/LD14, and CuCrO2/P1) as indicated in the Experimental Section. Herein, SECM was employed to scrutinize the effect of the redox couples on the dye-regeneration with a feedback model, which was generated under short-circuit condition at the nanocrystalline/electrolyte heterojunction through the reduction of the oxidized dye cations for the photoanodes (n-type) or the oxidation of the reduced dye anions for the photocathode (p-type). Figure 4 presents a typical approach curve (tip

www.chemphyschem.org above an illuminated dye-sensitized film; symbols) shows much higher values, indicating a positive feedback in this case. The normalized tip current IT decreases as the redox mediator concentration increases. The shape of the approach curve is indicative of the electro-activity of the substrate. Based on the normalized apparent heterogeneous electron transfer rate constant k, the effective heterogeneous rate constant (keff, in cm s1) can be evaluated with keff = kD/rT, where D is the diffusion coefficient for the redox couple in the electrolyte solutions. The diffusion coefficients were determined from microelectrode steady-state currents.[32] The data are tabulated in Table 2. Table 2. Diffusion coefficients (D) of the redox active species in the various electrolytes used.

Figure 4. Normalized SECM feedback approach curves for the approach of a Pt ultra-microelectrode towards a FTO/TiO2/C106TBA film in electrolytes containing Co3 + under illumination by a red-light LED at a constant intensity Jhv of 14.7  109 mol cm2 s1. The concentration of Co3 + was changed from 0.03 to 1.0 mm. Scan rate = 1 mm s1, E(T, Co3+) = 0.2 V (vs. Ag/Ag +), rT = 12.5 mm. Solid lines are calculated curves for the approach of a UME with RG = 10 towards an inert insulating surface (curve 7),[32] and towards samples with first-order kinetics of mediator recycling using the normalized rate constant k:[32] curve 1) 0.332, curve 2) 0.154, curve 3) 0.086, curve 4) 0.042, curve 5) 0.037, and curve 6) 0.029.

Redox species

Concentration [mm]

Limited current[a] [nA]

Diffusion coefficient[b] [cm2 s1]

Co2 + Co3 + T T2 I I3

0.1 0.1 0.1 0.1 0.1 –

0.594 0.571 1.061 1.698 0.659 –

1.22  105 1.17  105 2.2  105 1.76  105 1.86  105 1.37  105[c]

[a] Radius of the UME, rT = 12.5 mm. [b] Diffusion coefficients (D) of the redox active species in the various electrolytes can be determined from diffusion-limited currents on the UME (iT,1) using the following equation: D = iT,1/(4 nF[C]  rT), where n is the number of exchange transfer electrons for the redox reaction on the surface of the tip, F being the Faraday constant. [c] See ref. [33].

In this study, we evaluated for the first time the influence of redox couples on the dye-regeneration for n-type DSSC (TiO2 sensitized with C106TBA and LD14) and p-type DSSC (CuCrO2 sensitized with P1) devices. Figure 5 presents the keff values for the C106TBA- and LD14-sensitized TiO2 films in electrolyte solutions of Co3 + and I3 at a constant illumination intensity provided by a red LED. The detailed approach curves for the FTO/ TiO2/C106TBA and FTO/TiO2/LD14 in Co3 + or I3 electrolytes for various concentrations or light intensities are shown in Figures S1–S9. As illustrated in Figure 5 and Sections S8.1 and

current-distance relationships) for the case of TiO2/C106TBA in an electrolyte containing various concentrations of Co3 + illuminated by a red LED at a constant intensity serving as an example. Other approach curves obtained for various sensitizers, redox couples, semiconductors, and incident light intensities are shown in Figures S1–S11 (in the Supporting Information). In Figure 4 the normalized tip current (IT = iT/ iT,1) is related to the tip current (iT) and the steadystate current (iT,1) at semi-infinite height. L = (d/rT) is the normalized tip height, where d is the distance between the tip and the substrate electrode and rT is the radius of the active area of the SECM tip. As shown in Figure 4, the interfacial charge transfer dominates the current curves when the tip approaches the sample surface. The important kinetic parameters (such as the unitless, normalized apparent heterogeneous electron transfer rate constant, k) can be obtained by fitting the approach curves with  3+ the Equations S1–S5 in the Supporting Information.[40] Figure 5. Plot of keff versus different concentrations of [Co ] and [I3 ] for a) C106TBAand b) LD14-sensitized TiO2 photo-electrochemical electrodes in acetonitrile. The full data Compared to the current recorded with an inert insusets are presented in the Supporting Information Sections S8.1 and S8.2. The solid lines lating surface, ITins , (c at the bottom in Figure 4), represent different fits of parameters to the data using Equation (1). The fitting paramethe normalized tip current IT (i.e. the current recorded ters are given in Table 3.  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemPhysChem 2014, 15, 1182 – 1189

1185

CHEMPHYSCHEM ARTICLES

www.chemphyschem.org

S8.2 in the Supporting Information, keff decreases with the increase in Co3 + or I3 concentration. It was observed that keff for Co3 + is larger than that for I3 in the case of C106TBA and LD14 dyes, in which the TiO2/C106TBA/electrolyte interface showed a faster apparent regeneration process. This result supports the argument that cobalt redox couple complexes can be used as effective electrolytes in DSSCS.[10, 23] The influence of photon flux Jhv (caused by the red LED used in our experiments) on the effective heterogeneous rate constant keff was also examined to compare the associated kinetics for dye-regeneration at the C106TBA- and LD14-sensitized TiO2/electrolyte interfaces. The results are illustrated in Figure 6

where [C] is the concentration of the redox shuttle in the electrolyte [mol cm3], l is the film thickness [cm], [S] is the concentration of the sensitizer on the film [mol cm3], Jhn is the incident photon flux [mol cm2 s1], and fhn is the excitation crosssection of the sensitizer molecule [cm2 mol1]. The values of kox’ and fhn can simultaneously be obtained by fitting the relationship of keff and the redox shuffle concentration with Equation (1). Similarly, the constants kre’ for the regeneration of the photo-excited dye by the oxidized species of the redox shuttle in the p-type DSSC device at a given incident light intensity can be obtained by switching kox’ to kre’ in Equation (1). The heterogeneous rate constants kox’ for the regeneration of the photo-excited dyes C106TBA and LD14 reduced by either I or Co2 + , and the excitation cross-section of the sensitizer molecule fhn was obtained by fitting the keff versus [C] (Figure 5) and the keff versus Jhn curves (Figure 6) based on Equation (1). The results are listed in Table 3. The dye-regeneration rate value of kox’ obtained with Co3 + was about two times larger than with I3 for the case of the C106TBA and LD14 sensitized TiO2/electrolyte interfaces. This Figure 6. Plot of keff versus Jhn (red LED) for a) C106TBA- and b) LD14-sensitized TiO2 photo-electrochemical elecresult indicates that the redox  3+ trodes in acetonitrile with constant electrolyte concentrations of Co and I3 . The full data sets are presented in shuttle has a significant effect the Supporting Information Sections S8.3 and S8.4. The solid lines represent different fits of parameters to the on the dye regeneration rate data using Equation (1). The fitting parameters are given in Table 3. kox’. However, it was observed that the redox species had negliand Sections S8.3 and S8.4 in the Supporting Information. The gible influence on the excitation cross-section (fhn ) of the substrates were illuminated with a red LED while the redox same sensitizers. This observation is reasonable because the species were kept at a constant concentration of 0.03 mm. electron transfer rate depends on the molecular structure of Without illumination, both mediators yielded normalized apthe sensitizers and the redox mediators. It has been reported proach curves corresponding to the hindered diffusion toward that the size of the donor and acceptor affect the reorganizathe microelectrode probe. It was observed that under weak tion energy and electronic coupling.[20] According to the 2 1 red LED intensities (less than 2.1 mol cm s ) the keff sensitizer Marcus theory of electron transfer, the faster dye-regeneration values were very small (about 1.0 cm s1). This result is consiscan be rationalized in terms of an increase in the outer-sphere tent with those obtained in the case of hindered diffusion apreorganization energy (ls) for the cobalt complex redox couple compared to the iodide-based one, the driving force of which proach curves during the SECM measurements. The keff values for C106TBA and LD14 in Co3 + and I3 solutions increased with is about 100 mV less for the former redox shuttle.[17] This indithe light intensity. A significant enhancement in keff was obcates that the electron transfer occurs in the inverted Marcus region. For a given mediator concentration and light intensity, served when using Co3 + electrolytes compared to I3 electrosignificantly larger kox’ values were obtained for C106TBA comlytes in the cases of C106TBA and LD14 at a given mediator concentration and light intensity. Similar to the observations pared to LD14 with either of the redox mediators Co3 + /Co2 + shown in Figure 5, the keff values for C106TBA were found to be larger than for LD14 within the same mediator concentrations and light intensities. Table 3. Fitting results of the experimental values of keff versus [C]* and Constants kox’ (n-type) for the regeneration of the photo-exkeff versus Jhν values (Figures 5 and 6), l = 6.3  104 cm, [C106TBA] = cited dye by the reduced species of the redox shuttle in the 2.32  105 mol cm3, and [LD14] = 2.30  105 mol cm3. n-type DSSC device at a given incident light intensity can thus kox [mol1 cm3 s1] Redox shuttles fhn [cm2 mol1] be obtained with the Equation (1):[33] C106TBA

k0 ox ¼

2 keff fhn Jhn 6 keff ½C  3 l½Sfhn Jhn

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

3+

ð1Þ

Co /Co I3/I

2+

6

2.34  10 2.32  106

LD14

C106TBA

5

8.60  10 7.13  105

5

3.43  10 1.04  105

LD14

1.55  105 6.90  104

ChemPhysChem 2014, 15, 1182 – 1189

1186

CHEMPHYSCHEM ARTICLES or I3/I . This observation can be explained by the big difference between the fhn of the sensitizer molecule of C106TBA and LD14 under the same illumination. The value of fhn for C106TBA (approximately 2.34  106 cm2 mol1) was about three times larger than that for LD14 (8.60  105 cm2 mol1). It can thus be said that the cobalt complex redox couple boosts the DSSC device performance due to a faster dye-regeneration. However, device B with the cobalt complex electrolyte showed less photocurrent than device A, which operates with the conventional iodide-based electrolyte, but there was an increase in photovoltage for device B. This result can be explained by the slower mass transfer for the cobalt complex in the nanocrystal mesoporous film[41] and a faster interfacial charge recombination between the electrode and the electrolyte as discussed later. The dye-regeneration process based on the CuCrO2/P1 (ptype) photo-electrochemical cell was also investigated by SECM with the two redox shuttles I3/I and T2/T . The approach curves to evaluate the influence of light intensity (produced by a blue-light LED) and of the redox couple concentration on dye-regeneration are presented in Figures S10 and S11. Figure 7 and Section S8.5 in the Supporting Information presents the effective heterogeneous rate constant keff versus light intensity. The keff for P1 regeneration increased with the light intensity, regardless of the redox shuttle. At a given mediator concentration and light intensity, the keff values for the thiolate-based electrolyte were significantly larger than those for the iodine-based one. The fitting results of Figure 7 with Equation (1) are listed in Table 4. It was observed that the dye regeneration rate kre’ for P1 in the T-based electrolyte was much larger than that in the I-based one. In contrast, the fhn values of P1 was similar in both electrolytes. Our previous report showed that the low PCE of p-type DSSC devices could be due to a relative slow dye-regeneration.[39] Herein, by comparing the results in Tables 3 and 4, one can see that the regeneration rate kre’ for a dye with a p-type CuCrO2 is higher than kox’ for a dye with a n-type TiO2 with the same redox mediator I3/I , although different wavelengths of illumination

www.chemphyschem.org Table 4. Fitting results for the experimental keff versus Jhv values (in Figure 7), l = 1.3  104 cm, and [P1] = 1.23  105 mol cm3. Redox shuttles

Parameters hn [cm2 mol1]



6

T2/T I3/I

5.28  10 5.31  106

k0 re [mol1 cm3 s1] 1.66  106 3.96  105

were used in this study. Therefore, we believe that the cause for its low PCE is rather an ineffective charge separation after photo-excitation in the p-type DSSC device.[39] This would explain the low PCE observed in DSSC devices using p-type CuCrO2 (devices E and F in Table 1) and indicate that with a suitable electrolyte large improvements for p-type DSSCs can be obtained. In order to confirm that the interfacial charge recombinations between FTO/TiO2 or FTO/CuCrO2 and the redox mediators play a key effect towards the device photovoltaic performance, the kinetics of charge recombination were investigated by SECM technology.[42] The SECM approach curves for evaluation of the potential dependence of the rate constants on interfacial recombination are shown in Sections S5 and S7 in the Supporting Information. In this experiment, a bare nanocrystal semiconductor mosoporous film (FTO/TiO2 or FTO/ CuCrO2) was used as the substrate and I3/I , Co3 + /Co2 + , and T2/T were used as redox mediators. A thin compact layer of TiO2 or NiO deposited by spray pyrolysis technique was used to coat the FTO conducting glass substrates in order to prevent electron–hole recombination arising from direct contact between the redox shuttle and the highly doped SnO2 layer. The SECM experimental approach curves were fitted, which can relate the tip current to the surface’s heterogeneous electron or hole transfer kinetics and eventually obtain the normalized apparent charge-transfer constant k (Sections S7.1–7.4, and S8.6 and S8.7 in the Supporting Information). The keff (kD/ rT) values were plotted in Figure 8 as a function of their corresponding overpotential (h, where h = EsubstrateEredox1/2) in order to determine the apparent standard heterogeneous rate constant k0. Herein, the k0 [cm s1] stands for the rate constant of the interfacial charge recombination process. A detailed evaluation of k0 is given in ref. [42]. The calculated k0 values for various mediators are listed in Table 5. It was observed that the apparent standard heterogeneous electron transfer kinetics constant k0 for FTO/TiO2 was slower in the I3/I electrolyte than that in the Co3 + /Co2 + electrolyte, and the hole transfer kinetics constant k0 for FTO/CuCrO2 in the case of T2/T was

Table 5. Apparent heterogeneous charge-transfer rate constants for the mediators from the SECM measurements. Electrode Figure 7. Plot of keff versus Jhn (blue LED) for P1 sensitized CuCrO2 photoelectrochemical electrodes in acetonitrile. The full data set is presented in the Supporting Information Section S8.5. The solid lines represent different fits of parameters to the data by using Equation (1). The fitting parameters are given in Table 4.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

FTO/TiO2 FTO/CuCrO2

Mediator Co I T2 I3

2+

E1/2 [V vs. Ag/Ag + ]

K0 [104 cm s1]

0.05 0.15 0.05 0.15

4.2955 1.4196 7.5977 15.79

ChemPhysChem 2014, 15, 1182 – 1189

1187

CHEMPHYSCHEM ARTICLES

www.chemphyschem.org

DSSC devices were prepared as outlined previously.[8, 43, 44] Detailed procedures can be found in the Supporting Information. SECM experiments were performed on a CHI 920C electrochemical workstation (CH Instruments, Shanghai). A homemade Teflon cell (with a volume of 2 mL) was used to hold a Pt wire counter electrode, and an Ag/Ag + reference electrode. The photoanodes FTO/TiO2/ C106TBA, FTO/TiO2/LD14 (n-type) and the photocathode FTO/ CuCrO2/P1 (p-type) sample films were attached to the cell bottom and sealed with an O-ring as work Figure 8. Plot of ln(keff) versus h for a) FTO/TiO2 electrodes in acetonitrile corresponding to the reduction with I and Co2 + and for b) FTO/CuCrO2 electrodes in acetonitrile corresponding to the oxidation with T2 and I3 . The full electrode. In the SECM measuredata sets are presented in the Supporting Information Sections S8.6 and S8.7. The fitting parameters are given in ment, the dye-sensitized transparTable 5. ent semiconductor nanocrystal films (TiO2 and CuCrO2 with a thickness of about 6.3 mm and 1.3 mm, respectively) were used as working electrodes. An extra Pt wire   slower than that of I3 /I (see Table 5). Further comparison connected the FTO substrate with the electrolyte in order to opershowed that the recombination rate constant in p-type DSSCs ate the photoelectrochemical cell in a short-circuit setup (Figure 2). was much higher than n-type devices. The transfer kinetics A 25 mm diameter Pt wire (Goodfellow, Cambridge, UK) was sealed constants, however, for the interfacial charge recombination of into a 5 cm glass capillary prepared by a Vertical pull pin instrument (PC-10, Japan). The ultra-microelectrode (UME) was polished the FTO/TiO2/electrolyte or the FTO/CuCrO2/electrolyte are all by a grinding instrument (EG-400, Japan) and micro-polishing smaller than the dye-regeneration kinetics constants of the cloths of 1.0, 0.3, and 0.05 mm alumina powder grains. The UME corresponding systems. This ensures effective charge collection was sharpened conically to a RG of 10, where RG is the ratio bein the DSSC devices. tween the diameters of the glass sheath and the Pt disk. All experiments were carried out at room temperature. The irradiation was focused onto the backside of the photoanode or photocathode Conclusions from different light emitting diode (blue and red, Lumileds Lightin, USA). Dye regeneration kinetics at dye-sensitized the semiconductor

(n-type and p-type)/electrolyte interface was studied by SECM approach curves based on the feedback mode. Rate constants kox’ or kre’ with different mediator concentrations and light intensities were determined for the dye regeneration process. The SECM approach curves showed that the kinetics of dye regeneration depends on the nature of the dyes and the redox mediators constituting the DSSCs. For n-type sensitizers (C106TBA and LD14), the regeneration rate (kox’) in the case of a Co3 + /2 + -based electrolyte is about two times larger than that of the I3/I redox electrolyte. The kox’ of C106TBA is about two times larger than that of LD14 in the same redox mediator electrolyte. For p-type sensitizers, the kre’ for P1 in the T2/T electrolyte is much higher than that in I3/I . We reported for the first time the investigation of the interfacial charge recombinations by SECM technology, showing that cobalt complexbased electrolytes accelerate the process. This work will offer some new complementing aspects to establish methods that can be used to characterize DSSCs by testing different redox mediator electrolytes and sensitizers with a single dye-sensitized electrode.

Acknowledgements

Experimental Section

Keywords: dye-regeneration · dye-sensitized solar cells · feedback mode · redox shuttle · scanning electrochemical microscopy

The potential of the UME (ET) was selected well in the region of the steady diffusion current after recording the cyclic voltammogram of the redox mediators with a fixed concentration. The feedback mode of the SECM was used in this work. The basic arrangement for probing the heterogeneous reaction at dye-sensitized n-type TiO2 and the dye-sensitized p-type CuCrO2 interfaces in the feedback mode of SECM under short-circuit conditions are shown in Figure 2. The reaction mechanism has been reported by other researchers.[34]

We gratefully acknowledge the 973 Program of China (2014CB643506, 2013CB922104, and 2011CBA00703), the NSFC (21103578, 21161160445, and 201173091), and the Fundamental Research Funds for the Central Universities (HUST: CXY12Q022, 2012YQ027, and 2013YQ051). The authors thank the Analytical and Testing Center of Huazhong University of Science and Technology for support.

Tetrabutyl ammonium perchlorate (TBAP) was used as supporting electrolyte. LiI and I2 were purchased from Sigma (USA).  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemPhysChem 2014, 15, 1182 – 1189

1188

CHEMPHYSCHEM ARTICLES [1] G. Yu, J. Gao, J. Hummelen, F. Wudl, A. Heeger, Science 1995, 270, 1789 – 1791. [2] J. Lu, X. Xu, K. Cao, J. Cui, Y. Zhang, Y. Shen, X. Shi, L. Liao, Y. Cheng, M. Wang, J. Mater. Chem. A 2013, 1, 10008 – 10015. [3] X. Xu, D. Huang, K. Cao, M. Wang, S. Zakeeruddin, M. Grtzel, Sci. Rep. 2013, 3, 1489. [4] J. He, H. Lindstro, A. Hagfeldt, S. Lindquist, J. Phys. Chem. B 1999, 103, 8940 – 8943. [5] A. Nattestad, A. Mozer, M. Fischer, Y. Cheng, A. Mishra, P. Buerle, U. Bach, Nat. Mater. 2010, 9, 31 – 35. [6] P. Qin, H. Zhu, T. Edvinsson, G. Boschloo, A. Hagfeldt, L. Sun, J. Am. Chem. Soc. 2008, 130, 8570 – 8571. [7] J. Bai, X. Xu, L. Xu, J. Cui, D. Huang, W. Chen, Y. Cheng, Y. Shen, M. Wang, ChemSusChem 2013, 6, 622 – 629. [8] a) D. Xiong, Z. Xu, X. Zeng, W. Zhang, W. Chen, X. Xu, M. Wang, Y. Cheng, J. Mater. Chem. 2012, 22, 24760 – 24768; b) Z. Xu, D. Xiong, H. Wang, W. Zhang, X. Zeng, L. Ming, W. Chen, X. Xu, J. Cui, M. Wang, S. Powar, U. Bach, Y. Cheng, J. Mater. Chem. A 2014, 2, 2968 – 2976. [9] a) X. Xu, B. Zhang, J. Cui, D. Xiong, Y. Shen, W. Chen, L. Sun, Y. Cheng, M. Wang, Nanoscale 2013, 5, 7963 – 7969; b) X. Xu, J. Cui, J. Han, J. Zhang, Y. Zhang, L. Luan, G. Alemu, Z. Wang, Y. Shen, D. Xiong, W. Chen, Z. Wei, S. Yang, B. Hu, Y. Cheng, M. Wang, Sci. Rep. 2014, 4, 3961. [10] A. Yella, H. Lee, H. Tsao, C. Yi, A. Chandiran, M. Nazeeruddin, E. Diau, C. Yeh, S. Zakeeruddin, M. Grtzel, Science 2011, 334, 629 – 634. [11] K. Cao, M. Wang, Front. Optoelectron. 2013, 6, 373 – 385. [12] E. Barea, J. Ortiz, F. Pay, F. Fernndez-Lzaro, F. Fabregat-Santiago, A. Sastre-Santos, J. Bisquert, Energy Environ. Sci. 2010, 3, 1985 – 1994. [13] A. Anderson, P. Barnes, J. Durrant, B. O’Regan, J. Phys. Chem. C 2011, 115, 2439 – 2447. [14] K. Robson, K. Hu, G. Meyer, C. Berlinguette, J. Am. Chem. Soc. 2013, 135, 1961 – 1971. [15] T. Daeneke, A. Mozer, T. Kwon, N. Duffy, A. Holmes, U. Bach, L. Spiccia, Energy Environ. Sci. 2012, 5, 7090 – 7099. [16] Y. Shen, K. Nonomura, D. Schlettwein, C. Zhao, G. Wittstock, Chem. Eur. J. 2006, 12, 5832 – 5839. [17] A. Hagfeldt, M. Grtzel, Acc. Chem. Res. 2000, 33, 269 – 277. [18] T. Daeneke, T. Kwon, A. Holmes, N. Duffy, U. Bach, L. Spiccia, Nat. Chem. 2011, 3, 211 – 215. [19] G. Boschloo, A. Hagfeldt, Acc. Chem. Res. 2009, 42, 1819 – 1826. [20] a) M. Wang, C. Grtzel, S. Zakeeruddin, M. Grtzel, Energy Environ. Sci. 2012, 5, 9394 – 9405; b) M. Marszalek, F. Arendse, J. Decoppet, S. Babkair, A. Ansari, S. Habib, M. Wang, S. Zakeeruddin1, M. Grtzel, Adv. Energy Mater. 2014, DOI: 10.1002/aenm.201301235. [21] M. Wang, N. Chamberland, L. Breau, J. Moser, R. Baker, B. Marsan, S. Zakeeruddin, M. Grtzel, Nat. Chem. 2010, 2, 385 – 389. [22] S. Feldt, E. Gibson, E. Gabrielsson, L. Sun, G. Boschloo, A. Hagfeldt, J. Am. Chem. Soc. 2010, 132, 16714 – 16724. [23] H. Nusbaumer, S. M. Zakeeruddin, J. E. Moser, M. Grtzel, Chem. Eur. J. 2003, 9, 3756 – 3763.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemphyschem.org [24] D. Li, H. Li, Y. Luo, K. Li, Q. Meng, M. Armand, L. Chen, Adv. Funct. Mater. 2010, 20, 3358 – 3365. [25] H. Tian, X. Jiang, Z. Yu, L. Kloo, A. Hagfeldt, L. Sun, Angew. Chem. 2010, 122, 7486 – 7489; Angew. Chem. Int. Ed. 2010, 49, 7328 – 7331. [26] I. Montanari, J. Nelson, J. Durrant, J. Phys. Chem. B 2002, 106, 12203 – 12210. [27] T. Heimer, E. Heilweil, C. Bignozzi, G. Meyer, J. Phys. Chem. A 2000, 104, 4256 – 4262. [28] C. Nasr, S. Hotchandani, P. Kamat, J. Phys. Chem. B 1998, 102, 4944 – 4951. [29] S. Pelet, J. Moser, M. Grtzel, J. Phys. Chem. B 2000, 104, 1791 – 1795. [30] T. Daeneke, A. Mozer, Y. Uemura, S. Makuta, M. Fekete, Y. Tachibana, N. Koumura, U. Bach, L. Spiccia, J. Am. Chem. Soc. 2012, 134, 16925 – 16928. [31] F. Li, J, Jennings, Q. Wang, ACS Nano 2013, 7, 8233 – 8242. [32] Y. Shen, U. Tefashe, K. Nonomura, T. Loewenstein, D. Schlettwein, G. Wittstock, Electrochim. Acta 2009, 55, 458 – 464. [33] a) U. M. Tefashe, K. Nonomura, N. Vlachopoulos, A. Hagfeldt, G. Wittstock, J. Phys. Chem. C 2012, 116, 4316 – 4323; b) U. Tefashe, M. Rudolph, H. Miura, D. Schlettwein, G. Wittstock, Phys. Chem. Chem. Phys. 2012, 14, 7533 – 7542. [34] M. Wang, G. Alemu, Y. Shen, Current Microscopy Contributions to Advances in Science and Technology, Vol. 2 (Ed.: A. Mndez-Vilas), Formatex Research Center, Badzjoz, 2011, pp. 1377 – 1386. [35] Y. Liu, J. Jennings, Y. Huang, Q. Wang, S. Zakeeruddin, M. Grtzel, J. Phys. Chem. C 2011, 115, 18847 – 18855. [36] Y. Cao, Y. Bai, Q. Yu, Y. Cheng, S. Liu, D. Shi, F. Gao, P. Wang, J. Phys. Chem. C 2009, 113, 6290 – 6297. [37] Y. Chang, C. Wang, T. Pan, S. Hong, C. Lan, H. Kuo, C. Lo, H. Hsu, C. Lin, E. Diau, Chem. Commun. 2011, 47, 8910 – 8912. [38] P. Qin, M. Linder, T. Brinck, G. Boschloo, A. Hagfeldt, L. Sun, Adv. Mater. 2009, 21, 2993 – 2996. [39] J. Cui, J. Lu, X. Xu, K. Cao, Z. Wang, G. Alemu, H. Yuan, Y. Shen, J. Xu, Y. Cheng, M. Wang, J. Phys. Chem. C 2014, DOI: 10.1021/jp410829c. [40] R. Cornut, C. Lefrou, J. Electroanal. Chem. 2008, 621, 178 – 184. [41] J. Nelson, T. Amick, C. Elliott, J. Phys. Chem. C 2008, 112, 18255 – 18263. [42] N. Ritzert, J. Rodrguez-Lpez, C. Tan, H. AbruÇa, Langmuir 2013, 29, 1683 – 1694. [43] a) M. Wang, C. Grtzel, S. Moon, R. Humphry-Baker, N. Rossier-Iten, S. Zakeeruddin, M. Grtzel, Adv. Funct. Mater. 2009, 19, 2163 – 2172; b) X. Zhang, Z. Zhang, F. Huang, P. Buerle, U. Bach, Y. Cheng, J. Mater. Chem. 2012, 22, 7005 – 7009. [44] M. Wang, X. Li, H. Lin, P. Peter, S. Zakeeruddin, M. Grtzel, Dalton Trans. 2009, 10015 – 10020.

Received: November 15, 2013 Revised: February 17, 2014

ChemPhysChem 2014, 15, 1182 – 1189

1189