Effect of TiO₂ Photoanodes Morphology and Dye Structure on Dye-Regeneration Kinetics Investigated by Scanning Electrochemical Microscopy

Abstract

The dye regeneration in dye-sensitized solar cells (DSSCs) is improved by optimizing the charge separation at the level of the sensitized semiconductor treatment of the mesoporous electrode by TiCl₄ that passivates the surface for back electron transfer reactions. The dye-regeneration kinetics is analyzed for DN216- and D358-sensitized porous TiO₂ electrodes with and without a TiCl₄ treatment by means of scanning electrochemical microscopy (SECM). Different mass transport limitation of the [Co(bpy)₃]³⁺ mediator through the porous electrode is found for the comparison of the structurally similar dyes but cannot be detected for the thin layer introduced by the TiCl₄ treatment. Phototransient measurements are conducted directly in the SECM cell without any intermediated sample manipulation. The results from those measurements corroborate the findings from steady state SECM measurements.

1. Introduction

The dye-sensitized solar cells (DSSCs) were originally developed by O’Regan and Grätzel in 1991 [1]. TiO₂-based devices were reported to have photoconversion efficiencies of up to 12.7% [2], a value halfway of the theoretical maximum estimated to be around 30% [3,4]. The mechanism of photon-conversion into electrons takes place at the sensitized semiconductor [5]. The coupled processes are depicted and numbered in Figure 1. A dye molecule (D) absorbs solar light (1) and injects an electron in the conduction band (CB) of the TiO₂ film (2), on which it is adsorbed [5,6]. Inside the semiconductor, the electron is transported by a diffusion process (3) to the back contact, during which the conduction band electrons e_CB temporarily reside in traps near to the CB edge, from which they can be thermally activated [7,8]. The photo-oxidized dye is thereafter regenerated by a mediator (4) in the electrolyte [9]. Due to space requirements, the redox mediator [Co(bpy)₃]^2+/3+, in its reduced and oxidized form is represented either by R/O or $,$/- in Figure 1. In an illuminated photoanode, the photo-oxidized dye (D⁺) and the e_CB can be considered as an electron-hole pair that can either recombine (5) or separate [10]. In general, the high dielectric permittivity of TiO₂ supports the charge separation [11,12]. During the transport to the back contact (3), e_CB can be lost by “back electron transfer reaction” which are the heterogeneous electron transfer (ET) reactions of e_CB to the oxidized mediator (6) in the electrolyte or, less frequently, to D⁺ (5) [12]. These processes are also called “recombination” because they restore the electronic state before the photo-excitation. The unwanted loss processes 5 and 6 limit the effective life time of e_CB [13,14]. If electrons are retained in trap states located near the TiO₂|solution interface, the probabilities for an ET to the oxidized form of the mediator (6′) or to D⁺ (5′) are dramatically increased and may limit the overall device performance [6]. A common approach to increase the device performance consists in diminishing the trap concentration near the inner surface of the nanoporous photoanodes.

In this work, the dye-regeneration kinetics of TiO₂-based photoanodes were analyzed for two different nanoporous TiO₂ electrodes and two all-organic dyes as sensitizers. The aim was to quantify the influence of reduced back electron transfer reactions from near-surface trap states on the overall dye-regeneration efficiency, which is expected to enhance the overall photovoltaic performances. Mesoporous TiO₂ films are commonly treated by an aqueous TiCl₄ solution in order to improve the efficiency of dye-sensitized solar cells [15]. This treatment covers the surface of mesoporous TiO₂ film by rutile TiO₂ and increases the light scattering property of the film [16]. Several studies have been reported to explain the effects of the post-TiCl₄ treatment. The particle size of TiO₂ increased from 13 nm for the primary TiO₂ nanoparticles to ca. 15 nm by the covering layer [17]. Impurities at the surface of the initial mesoporous TiO₂ film are cleaned during the process and a pure TiO₂ surface is obtained [15]. Although the surface area of the mesoporous TiO₂ film (as measured by gas adsorption) decreased by this treatment, the dye loading increases since the area accessible to the dye molecules and the cleanness of the particles are improved [18,19]. A shift of the CB to a lower potential was also observed and it leads to an enhanced electron injection from the excited dye [17,20]. At last, a stronger bond was observed between the adsorbed dye and TiO₂ from infrared absorption spectroscopy [21]. Such a strengthen bonding and a better surface coverage of the TiO₂ particles by dye molecules inhibit the ET of e_CB oxidized mediator in the electrolyte solution. This effect is measured as an enhanced open-circuit photovoltage [22]. Due to those positive effects, the post-TiCl₄ treatment is commonly applied for the fabrication of dye-sensitized solar cells [22,23] and its effect has been summarized as passivation against losses of e_CB.

Variation of the thickness of photoanodes can increase the light absorption. However, this effect may be counteracted and even overcompensated by increased possibilities for loss processes at a larger inner surface [24] and mediator mass transport limitation inside the porous network [25,26] as was found from the analysis of films with thicknesses below 6 µm [24,27,28]. There are indications that, on the one hand, thicker photoanodes produce high open-circuit photovoltages resulting in enhanced photoconversion efficiency [28,29] and on the other hand exhibit a higher area for back electron transfer reactions [30]. It has been reported that the use of thin films leads to a more effective dye-regeneration kinetics favored by the limited back electron transfer reaction and faster mediator diffusion through the porous film [29].

Despite extensive work to improve photovoltaic characteristics of the TiO₂ photoanode, little information has been obtained about the quantitative effect of post-TiCl₄ treatment and sensitization processes on the dye-regeneration kinetics (process (4) in Figure 1). In this regard, scanning electrochemical microscopy (SECM) provides an evaluation of the dye regeneration rate constant (k_ox) for process 4 in Figure 1, that measures the ability of the mediator to reduce the photo-oxidized dye adsorbed on the photoanode [29,31,32]. In this work, SECM analysis is applied to passivated TiO₂ photoanodes sensitized with the all-organic dye molecules D358 or DN216 (Figure 2) [33,34]. These molecules have an aliphatic chain terminated by a carboxylic group that attaches it to the semiconductor surface. A long anchoring chain prevents dye-aggregation [35] and reduces ET of e_CB to the oxidized form O of the mediator [36,37] by limiting the proximal approach of O, e.g., [Co(bpy)₃]³⁺, to the semiconductor surface [2,38]. However, sterically demanding dyes can hinder the diffusion of the mediator in the pore space (internal diffusion), a trend that is expected to become more relevant for sterically demanding mediators like [Co(bpy)₃]^2+/3+ and photoanodes with smaller pores [39]. This would result in less effective dye-regeneration kinetics. The steady-state approach curves in the SECM feedback modes combined with a kinetic model introduced before for the evaluation of the dye regeneration in sensitized TiO₂-based photoanodes by cobalt-based mediators [32] provides a rate constant k_ox that is again an effective constant bearing a relation to structural features that influence the internal diffusion of the mediator in the porous photoanode.

In addition, the SECM approach has been expanded by performing transient photocurrent measurements inside the SECM cell on TiO₂ photoanodes with and without post-TiCl₄ treatment to qualitatively investigate the influence of the structural modification as well as the effect of the dye steric hindrance on the transient profile shape. The photocurrent recorded in on-off light cycles is indeed influenced by the mediator mass transport from the inner layers of the photoanodes towards the counter electrode [40], the porosity of the electrode [41], the eventual passivation with nanostructures [22], and the thickness of the porous film [39].

2. Materials and Methods

2.1. Materials and Chemicals

Tris-(2,2′-bipyridine)cobalt(III) tri(hexafluorophosphate) [Co(bpy)₃](PF₆)₃ (Dyenamo, Stockholm Sweden), dye 5-[3-(carboxymethyl)-5-[[4-[4-(2,2-biphenylethenyl) phenyl] -1, 2, 3, 3a, 4, 8b-hexahydrocyclopent [b]indol-7-yl] methylene]-4-oxo-2-thiazolidinylidene]-4-oxo-2-thioxo-3-thiazolidine-hexanoic acid (DN216) and dodecanoic acid (D358) (Dyenamo), the co-adsorbant lithocholic acid (>95%, Sigma-Aldrich, Steinheim, Germany), tert-butanol (≥99.5%, Sigma-Aldrich); tetrabutylammonium hexafluorophosphate (TBAPF₆, Sigma-Aldrich), acetonitrile (≥99.5%, Sigma-Aldrich), gold wire (99.999%, Goodfellow, Friedberg, Germany) were used as received.

2.2. TiO₂ Film Production

The TiO₂ films were prepared according to the procedure detailed before [34]. Shortly, a ca. 20 nm-TiO₂ paste was screen-printed on an FTO slide already covered with a compact TiO₂ layer of circa 0.6 µm. The resulting porous layers had thicknesses of 3, 4, and 6 μm. Two specimens for each thickness were sensitized by D358 or DN216 without post-TiCl₄ treatment and two specimens were subjected to a post-TiCl₄ treatment and later sensitized by D358 or DN216. The post-TiCl₄ treatment involved the exposure of the porous TiO₂ film to an aqueous solution of 53 mM TiCl₄ for 30 min. As a result, the internal surface of the porous TiO₂ film was covered by a layer of TiO₂ nanoparticles, that form a passivation layer. The thickness of the porous films was determined by profilometry (Alpha-Step 10-00020, Tencor Instruments, Milpitas, CA, USA).

2.3. Sensitization of the Photoanodes

The electrodes were sensitized in 2 mL of a solution of a 0.5 mM DN216 or D358 and 1 mM lithocholic acid in 1:1 (v:v) acetonitrile:tert-butanol for 1 h at room temperature. The amount of dye molecule loaded on a porous semiconductor film Γ_D was calculated by measuring the attenuation of the absorption of the sensitized films at 455 nm according to the Lambert-Beer equation with the decadic molar extinction coefficient of 77,600 L mol⁻¹ cm⁻¹ for the dye DN216 and 67,000 L mol⁻¹ cm⁻¹ for D358 dye [33]. The complete details of samples A to E are summarized in Table S1. That includes samples of 3, 4, and 6 µm thickness with and without passivation and sensitized with either DN216 or D358.

2.4. SECM Instrumentation

The setup for scanning electrochemical microscopy has been detailed before [32,34,42] and was run by the program SECMx [43]. The setup is shown in Figure 3 and included a piezoelectric actuator and the control electronics (P780.20 and E665, Physik Instrumente, Karlsruhe, Germany) for vertical approach, a digital potentiostat (Ref600, Gamry Instruments, Warminster, PA, USA), three SPI motors (SPI Robot, Oppenheim, Germany) to coarsely move the microelectrode (ME) relative to the sample and the light beam, and a two-dimensional positioning system (Nexcact N661.21A with controller E681, Physik Instrumente) for moving the sample horizontally at a fixed relative position between the microelectrode and the light beam. The samples were illuminated on an area of 1.5 mm² by a blue LED (455 nm) at photon flux J_hv ranging from 0 to circa 2 × 10⁻⁷ mol cm⁻² s⁻¹. The illuminated area was selected far off the edge of the photoanode. The precise location of the analyzed area did not influence the measurements because possible local variations of properties are average on the length scale of the used microelectrode. The LED was powered by an XPOT potentiostat with a photodiode (both Zahner Elektrik, Kronach, Germany) in the beam to measure and control the actual light intensity. The light intensity was calibrated with a power meter (Gigahertz-Optik, Türkenfeld, Germany). The light beam is focused with a lens on the sample (10×, Carl Zeiss, Jena, Germany). A schematic of the setup is provided in Figure 3.

An in-house designed electrochemical cell was used. It was custom-made to work in Ar atmosphere and allowed the simple localization of the illuminated area. A soft latex finger-cot was used to close the cell against the ambient atmosphere and insert the microelectrode while allowing its translation in the three-dimensions. The cell contained 1 mL redox electrolyte solution and facilitated the exchange of the redox electrolyte and the microelectrode.

The SECM working electrodes were Au microdisk electrodes with a radius r_T ≈ 25 µm and RG = r_glass/r_T = 14. The exact radius was determined from approach curves to a glass slide and fitting the obtained curve to an analytical model [44] The specific r_T value is given in the figure captions. The auxiliary electrode and the quasi-reference electrodes were Pt wires. The mediator [Co(bpy)₃](PF₆)₃ was used in a concentration of 0.1 mM for the approach curves and 0.2 mM for the transient measurements. The redox mediator was dissolved in acetonitrile with 0.1 M TBAPF₆ as supporting electrolytes. The approach curves were recorded at a translation rate of v_T = 0.5 μm s⁻¹. The transient measurements were recorded on the sample in the SECM setup short-circuited to the redox electrolyte solution. The selected area was illuminated intermittently with 10 s on–off light cycles with a photon flux in units of 10⁻⁸ mol cm⁻² s⁻¹ of 0.76, 1.51, 2.26, 3.01, 3.77, 6.02, and 7.52. This corresponds to a flux in units of 10¹⁵ photons cm⁻² s⁻¹ of 4.58, 9.09, 13.6, 18.1, 22.7, 26.3 and 45.3. During the transient measurements the microelectrode was positioned in a distance of 10 µm from the sample and reduced [Co(bpy)₃]³⁺ under diffusion-controlled condition to [Co(bpy)₃]²⁺.

3. Result

3.1. SECM Approach Curves

SECM approach curves were recorded in the feedback mode on TiO₂ photoanodes with and without passivation to determine the dye (D) regeneration rate constant k_ox of D358- and DN216-sensitized TiO₂ film by [Co(bpy)₃]³⁺ mediator. The passivation layer obtained by post-TiCl₄ treatment is expected to improve the charge separation by inhibiting ET to the oxidized mediator (which would diminish the ME current) and decreasing the probability of ET from e_CB to D⁺ so that interfacial concentration of D⁺ available for reaction with the reduced form of the mediator is increased compared to the non-passivated electrodes with the same dye. The variation of the k_ox was also tested on TiO₂ photoanodes of variable thickness to expand the considerations on the effect of mediator mass transport through the porous film.

3.1.1. Treatment of the Approach Curves

The photoanodes were positioned in an adapted sample-holder in the SECM setup, which enabled the illumination of a selected area of 1.5 mm² of the electrode through the transparent back contact. The samples were short-circuited to the solution by a Pt wire. Initially, the working solution contains only the oxidized form of the mediator [Co(bpy)₃]³⁺ which is continuously reduced under a diffusion-controlled condition at the ME. A suitable ME potential was found from a cyclic voltammogram in the working solution. The current was controlled by the hindered diffusion of the mediator to the ME. During illumination, the dye molecules were photo-excited and injected electrons in the CB of the semiconductor. The photo-oxidized dye molecules were regenerated according to Equation (1) by the redox reaction with the reduced form of the mediator.

$${\mathrm{D}}^{+}+[\mathrm{Co}(\mathrm{bpy}{)}_3{]}^{2+}\stackrel{k_{\mathrm{ox}}}{\to }{\mathrm{D}}^0+[\mathrm{Co}(\mathrm{bpy}{)}_3{]}^{3+}$$

(1)

The electrolysis current i_T at the ME is increased compared to the situation in the dark due to the additional flux of [Co(bpy)₃]³⁺ caused by reaction (1) from the TiO₂ photoanode to the ME. During an approach curve, the distance d between the ME and the photoanode is steadily decreasing, thus changing the relative contributions of the hindered diffusion of [Co(bpy)₃]³⁺ form the bulk phase and the flux generated by the dye regeneration in reaction (1). This changing contribution of hindered mediator diffusion and dye regeneration reaction at the sample is well described by continuum simulation of the reaction-diffusion problem. Such simulation results have been condensed to an analytical approximation for an irreversible first-order reaction [45], which we used in this work. Details are provided in our notation as Supporting Material SM-2. The approach curves were plotted in normalized coordinates I_T vs. L, in Figure 4, where the normalized microelectrode current I_T = i_T/i_T,∞ is the microelectrode current i_T(d) at a specific distance d normalized by the microelectrode current i_T,∞ in the bulk (i.e., quasi-infinite distance). The normalized distance L = d/r_T was obtained from the working distance d and the microelectrode radius r_T. A fit of the experimental curve to the analytical approximation yields a dimensionless pseudo-first-order rate constant κ, from which an effective heterogeneous rate constant k_eff [cm s⁻¹] for the overall dye regeneration process was obtained from Equation (2)

k_eff = κD/r_T,

(2)

where D is the diffusion coefficient of the mediator [Co(bpy)₃]³⁺ in the bulk solution and r_T is the radius of the ME. The diffusion coefficient D for the [Co(bpy)₃]³⁺ in the bulk was calculated as 8.3 × 10⁻⁶ cm² s⁻¹ from the diffusion-limited steady-state current of the reaction at the ME (Figure S4). The k_eff values are further related to the parameters of the photoanode in Section 4.1.

3.1.2. Effect of Post-TiCl₄ Treatment

Each TiO₂ photoanode was investigated under eight different light intensities J_hv by SECM approach curves in the feedback mode. Figure 4 shows the exemplary approach curves over a 4 µm thick TiO₂ photoanodes C-DN216, D-DN216, C-D358, and D-D358, which will be further discussed in detail. The approach curves for samples with other thicknesses are reported in the Supporting Material (SM) in Figures S1 and S2.

The approach curves recorded in the dark overlapped with the hindered-diffusion behavior (Supporting Material (SM), SM-2 with ĸ = 0). This led to the calculation of k_eff values very close to zero because no photo-driven reaction took place. The very small k_eff results from the mediator diffusion through the porous film that hinders the diffusion slightly less than an impermeable flat sample surface. However, this contribution is very small and does not affect the applicability of the model. The negligible current enhancement also proves the effectiveness of the blocking layer between the thin conducting glass and the porous TiO₂ electrode. If there were significant density of defect in the blocking layer processes 7 and 8 of Figure 1 would cause enhanced ME currents as demonstrated before [46].

The approach curves were fitted with the model for first order, finite and irreversible kinetic mediator regeneration by Cornut and Lefrou [45]. Details are provided in SM-2. The effective rate constant k_eff for the mediator redox reaction with the dye at the photoanode for each J_hv was calculated from κ (obtained by fitting the approach curve) and D (from Figure S3) according to Equation (2). All rate constants are collected in Tables S2 and S3. The approach curves recorded during illumination reach higher normalized currents on samples with post-TiCl₄ treatment which translates to higher k_eff values for an identical photon flux.

Naturally, the k_eff values increase with the photon flux until they reached a saturation behavior at a photon flux of J_hv > 0.5×10⁻⁷ mol cm⁻² s⁻¹ (Figure 5). Similar data are reported for all TiO₂ photoanodes of different thicknesses and sensitized with D358 or DN216 dye molecules in Figure S5. An enhancement of k_eff was detected every time that a photoanode with post-TiCl₄ treatment was used (compared to the corresponding non-treated counterpart). The enhancement was particularly strong when the dye DN216 with the small anchoring chain had been used for the sensitization. The increase of k_eff will be related to the structure of the photoanode in Section 4.1.

3.2. Transient Photocurrents

Photoinduced processes at sensitized porous semiconductor|electrolyte interface were investigated in literature with transient photocurrents and photovoltages [8,22,23,24,47,48]. For the experiments described below, region of the identical samples from Table S1 (3 and 4 µm thickness) were investigated directly in the SECM cell that had not been used before for approach curves. A similar solution was used as for the SECM approach curves. The photocurrents at the sample are recorded according to the scheme in Figure 3, while the microelectrode is constantly generating [Co(bpy)₃]²⁺ for dye regeneration.

Figure 6 shows the photocurrent transient recorded at a D358- and DN216-TiO₂ photoanodes with and without post-TiCl₄ treatment. When the light is switched on, the photocurrent rises. The slope is determined by filling the trap states in TiO₂. As the trap occupancy is successively increased, the transport processes of e_CB in the interconnected TiO₂ particles accelerate because the density of unoccupied trap states decreases relative to that of mobile e_CB. Higher light intensity also shortens the time until a significant fraction of the traps is filled. This can be qualitatively seen from both traces in Figure 6. The post-TiCl₄ treatment is assumed to reduce the density of near-surface trap states. In agreement, the current rise is faster for the sample TiCl₄-treated sample at each light intensity. Effects of limited mediator availability have to be considered as well, especially at high light intensity, narrow porous channels, sterically demanding mediator, and low mediator concentrations as used in these experiments [49]. These effects can contribute to a decline after an initial photocurrent spike, which is detected for the TiCl₄-treated sample at high light intensities [50,51]. Eventually, the photocurrents reach steady-state values i_ss under constant illumination. These i_ss values will be discussed in Section 4.3.

4. Discussion

4.1. Determination of Regeneration Rate Constant

A very simplified model that relates structural properties of DSSC photoanodes to their SECM response has been derived before and is summarized by Equation (3) [32].

$$k_{\mathrm{eff}}=\frac{{\Gamma }_{\mathrm{D}}{\varphi }_{hv}{J}_{hv}{k}_{\mathrm{ox}}}{c*k_{\mathrm{ox}}+{\varphi }_{hv}{J}_{hv}}$$

(3)

Here k_ox [cm³ mol⁻¹ s⁻¹] is the (heterogeneous) bimolecular regeneration rate constant for the reaction in Equation (1), Γ_D [mol cm⁻²] is the dye loading per geometric area evaluated from independent measurements, c* [mol cm⁻³] is the [Co(bpy)₃]³⁺ mediator bulk concentration, J_hν [mol cm⁻² s⁻¹] is the photon flux used to illuminate the samples, and φ_hv [cm² mol⁻¹] is the dye excitation cross-section. Fitting the k_eff-J_hv values to Equation (3) yielded estimates of k_ox and φ_hv. The rate constant k_ox is not an elementary rate constant because individual dye molecules in the porous electrode of finite thickness are potentially exposed to different photon fluxes and different diffusional fluxes of the mediator. For this reason, k_ox decreased with increasing thickness, because Equation (3) corrects for the higher dye loading with thicker samples but does not correct for the less efficient mediator diffusion to the dye molecules in different depth of the photoanode.

Figure 7 shows the k_ox values for each photoanodes as a function of the thickness of the porous photoanode. The detailed data set is reported in Tables S4 and S5. The data suggests that the k_ox steadily decreases as the thickness of the photoanodes increases, as highlighted by the linear fit of the semilogarithmic plot. In this case, the mass transport limitation in thick electrodes [4,5] becomes dominant over the dye-regeneration kinetics and can be particularly strong when the sterically demanding mediator such as [Co(bpy)₃]³⁺ mediator was used [49]. In addition to the porous layer thickness, also the nature of the dye molecule appears to consistently influence the dye-regeneration kinetics. The samples sensitized with the DN216 dye molecule showed a regeneration rate constant k_ox that was up to 2.4 times higher than the ones recorded over D358-sensitized electrodes despite an almost identical chromophore in DN216 and D358. This was due to the smaller anchoring chain of DN216 which reduces its steric hindrance towards the mediator diffusion inside the porous photoanode. This information covers the role played by the steric hindrance of the dye structural features on the dye regeneration efficiency [9,10].

The post-TiCl₄ treatment of TiO₂ photoanodes strongly enhanced the dye regeneration rate constant of both, D358- and DN216-sensitized samples, and [Co(bpy)₃]³⁺ as mediator. This effect is due to the inhibition of the ET of e_CB to the oxidized form of the mediator and the photo-oxidized dye D⁺ probably be decreasing the number of near-surface defects that act as traps from which such ET processes can start. The enhanced dye regeneration efficiency as measured by SECM for passivated photoanodes is in line with the results of other working groups that passivate the TiO₂ photoanodes for enhancing the DSSCs photovoltaic features [11,12].

4.2. Light-Harvesting Efficiency

The dye absorption cross-section ${\Phi }_{hv}$ is obtained from the fitting of Equation (3) and is used for the calculation of the light-harvesting efficiency according to Lambert-Beer law. All the TiO₂ samples show an LHE of 100% which is in agreement with what is commonly reported for TiO₂-based photoanodes in literature [52].

4.3. Transient Photocurrent

Different processes shape the photocurrent transients at different time scales. The rate of the trapping-detrapping transport of e_CB in the semiconductor [7,48,53], back electron transfer processes and mediator mass transport from counter electrode to the inner layers of the photoanodes [25] have been reported to influence the current transients in different time regimes. Consequently, the porosity of the electrode [26], the thickness of the porous film [10], and the passivation of the electrode with nanostructures [11] were all reported to exert an influence the photocurrent behavior [38,44]. The transport effects in the electrolyte will become more important when the mediator concentration is lower as it is the case for SECM experiments and the photocurrent transients measured under those conditions.

The data shown in Figure 8 suggested that photoanodes with post-TiCl₄ treatment strongly enhance the steady-state photocurrents i_ss by decreasing the rat of back electron transfer processes compared to untreated TiO₂ porous film in agreement with Ref. [20]. These findings are also in agreement with the enhanced dye-regeneration rate constant k_ox obtained DN216-sensitized TiO₂ photoanodes post-TiCl₄ treatment from SECM approach curves in the feedback mode.

Figure 9 shows i_ss as a function of the photoanode thickness and passivation. It appears that the steady-state photocurrent is negligibly affected by the sample thickness at low light intensities. However, i_ss strongly decreases at high light intensities and in thicker photoanodes as a result of processes 5 and 6 and mediator mass transport limitation compared to thin samples. The mediator diffusion plays a dominant role also in thicker passivated photoanodes, where i_ss is smaller than the values recorded on thinner samples. Overall, the highest i_ss values are recorded on 3 µm passivated TiO₂ photoanodes. These results are in agreement with a decrease of k_ox values as a function of increased thickness as reported in Figure 7.

5. Conclusions

This work reports a SECM investigation of the dye regeneration kinetic optimized by passivating the TiO₂ photoanodes by the post-TiCl₄ treatment as commonly used in dye-sensitized solar cells. The passivation layer is known to improve the charge separation by reducing the density of near-surface trap states, from which back electron transfer processes may commence to the oxidized mediator or the photo-oxidized dye. Therefore, this SECM investigation of the dye-regeneration reaction rate adds valuable information on the overall characterization of the DSSCs. The use of dye molecule with variable steric demand, the structure of which is a source of a hindrance for the mediator diffusion through the porous sensitized film, identify mediator mass transport limitations as a factor influencing the dye-regeneration kinetic. This is also supported by the comparison of several photoanode thicknesses. The analysis of photocurrent transients recorded directly in the SECM cell at sensitized TiO₂ photoanodes with and without post-TiCl₄ treatment supports the SECM data. In addition to the known effects, the post-TiCl₄ treatment decreases the pore diameters slightly. It could be expected that this increases the diffusional hindrance for mediator transport in the mesoporous electrode. However, this effect, if present at all, is overcompensated by the improved charge separation. The SECM method can detect effects on mediator regeneration invoked by small changes in the spacer length such as between the dyes D358 and DN216, where the dye with the shorter spacer and thus less constriction of diffusion path yields higher k_ox values that contain a contribution from the internal mass transport.